A FISH CONSUMPTION SURVEY OF
THE NEZ PERCE TRIBE
December 2016
-------
4* RIDOLFI _
Environmental J """" -.1^
A Fish Consumption Survey
of the
Nez Perce Tribe
Final Report
This report was prepared under
EPA Contract EP W14 020 Task Order 10
and Contract EP W09 Oil Task Order 125
with SRA International.
Nayak L Polissar, PhDa
Anthony Salisbury5
Callie Ridolfi, MS, MBAC
Kristin Callahan, MSC
Mom Neradilek, MSa
Daniel S Hippe, MSa
William 11 Beck ley, MSC
aThe Mountain-Whisper-Light Statistics
bPacific Market Research
cRidolfi Inc.
December 31, 2016
-------
Contents
Volume I—Heritage Fish Consumption Rates of the Nez Perce Tribe
Volume II—Current Fish Consumption Survey
Volume III—Appendices to Current Fish Consumption Survey
-------
PREFACE TO VOLUMES I-III
This report culminates two years of work—preceded by years of discussion—to characterize the
current and heritage fish consumption rates and fishing-related activities of the Nez Perce Tribe.
The report contains three volumes in one document. Volume I is concerned with heritage rates and
the methods used to estimate the rates; Volume II describes the methods and results of a current
fish consumption survey; Volume III is a technical appendix to Volume II. Each volume has its
own page numbering and Table of Contents. The foreword to Volumes I-III has been authored by
the Nez Perce Tribe and EPA. All other sections of this report have been authored by the members
of the contractor team listed on the title page.
-------
Foreword to Volumes I-III: Background and Context for the Nez Perce Tribal Fish
Consumption Survey (Authored by the Nez Perce Tribe and EPA)
The Native American tribal governments in the State of Idaho collaborated with the U.S.
Environmental Protection Agency (EPA) Region 10, and tribal consortia to gather data on tribal
fish consumption rates (FCRs) in Idaho. One objective of this effort was to support the effort to
assess risks posed by contaminants in fish for populations who consume large quantities of fish.
More generally, this effort was intended to enhance tribal environmental capacity in the area of
water quality. The tribes and EPA met with the State of Idaho to develop tribal surveys that
supported Idaho's efforts to develop ambient water quality criteria (AWQC) protective of high
fish consumers.
This report presents survey methodology and results, specifically FCRs, for the Nez Perce Tribe.
The survey is focused on both current and heritage rates. Heritage rates are tribal FCRs that existed
prior to modern environmental and social interferences with historic tribal fishing and fish
consumption practices. Within this report, current rates are discussed in Volume II, with supporting
material provided in Volume III. Heritage rates are discussed in Volume I.
For tribes and tribal members, fish are an important food and economic resource. The harvest and
consumption of fish also figure significantly in tribal culture and spirituality. The Nez Perce have
many concerns about water quality. However, the effect of water quality on fish and fisheries
resources is of particular importance to the Nez Perce Tribe. Water quality affects the health of fish
populations, the level of contaminants in fish and the consequent health risks posed by these
contaminants to tribal members when they consume fish. Water quality also impacts fishing and
fish consumption aspects of tribal culture and spirituality.
This report shows that a substantial portion of the diet of the Nez Perce Tribe consists of fish and
shellfish1, which research has shown acquire contaminants from water. This report's results are
consistent with findings that Puget Sound and Columbia River Basin tribes have much higher
FCRs than the general U.S. population. (CRITFC 1994, Toy et al. 1996, Suquamish Tribe 2000,
Polissar et al. 2014). As a result of higher tribal fish consumption relative to the general
population, tribal people suffer disproportionate exposure and risks associated with contaminants
in fish. As the FCRs for populations consuming fish increase, the water must become cleaner in
order to keep human exposures to toxic chemicals in fish at acceptable levels, with consequences
for target water quality. EPA Region 10 is supporting Idaho's tribal governments in identifying
appropriate FCRs to use in protecting the health of the Idaho tribes. Current FCR statistics (i.e.,
averages and percentiles) included in Volume II of this report are reported in terms of usual
consumption: the average daily grams of the edible mass of uncooked fish and shellfish
consumed by a tribal member.
A fish consumption study fits into a larger context. There are three eras of importance for such a
study: the past, the present, and the future. Considering the past, over an extended period of time
the Nez Perce Tribe has experienced environmental and social changes that have reduced fish
abundance, access to fish, safety of fish consumption, and fish consumption itself. During the
design phase of the current study, the Tribe expressed its goals to increase fish availability, reduce
1 Hereafter, "fish" will refer to fish and shellfish.
-------
contamination of fish, and increase fish consumption in the future. Thus, current consumption does
not reflect the Tribe's past, nor its goals. Assessing consumption through a current cross-sectional
survey will provide relatively precise information about current consumption only.
A complete understanding of tribal fish consumption issues thus requires not only consideration of
current fish consumption rates, but also tribal goals and heritage fish consumption. Assessing past
consumption involves review of historical materials and, potentially, interviews with some older
individuals whose memories span a long lifetime (and whose memories may carry stories passed
down from earlier generations).2 Assessment of past consumption is likely not as precise as current
surveys because derivation of heritage rates does not employ the same methodology as modern
surveys of current fish consumption, and involves longer-term recall and unknown quality and
completeness of past documentation. Further, heritage surveys can only provide average estimates
of fish consumption as opposed to distributions of fish consumption that can be obtained by current
fish consumption survey methodologies. Nonetheless, heritage rates are valid data that have been
developed with defensible, rational, and accepted research methods (e.g. ethnographic observation,
caloric intake, etc.). There have been many studies of historic rates and suppression of fish
consumption in the past, but their isolation from a report on current rates may have denied them the
attention they deserve.
Multiple studies using different methods have demonstrated that heritage FCRs exceeded current
FCRs. Nez Perce heritage and current FCRs documented in Volumes I and II of this report are
consistent with these findings. In other words, current FCRs are reduced or suppressed relative to
heritage FCRs. The Tribe is concerned that development of water quality criteria based on
suppressed fish consumption rates may not allow restoration of water quality to support safe
consumption of fish at the higher rates the Tribe desires and that are of cultural importance, rates
informed by treaties between the Tribe and the U.S. government that guaranteed tribal rights to
practice subsistence fishing.
The concept of suppression was discussed in depth in a publication by the National Environmental
"3
Justice Advisory Council (NEJAC, 2002) . Specifically, a "suppression effect" occurs when a fish
consumption rate for a given population, group, or tribe reflects a current level of consumption that
is artificially diminished from an appropriate baseline level of consumption for that population,
group, or tribe. Suppression effects can arise from at least the following three factors:
• First, a suppression effect may arise when an aquatic environment and the fish it supports
have become contaminated to the point that humans refrain from consuming fish caught
from particular waters. Were the fish not contaminated, these people would consume fish at
more robust baseline levels.
• Second, a suppression effect may arise when fish upon which humans rely are no longer
available in historical quantities (and kinds), such that humans are unable to catch and
consume as much fish as they had or would. Such depleted fisheries may result from a
variety of causes, including an aquatic environment that is contaminated, altered (due,
2 It should be noted that suppressed fish consumption has likely occurred prior to the birth of almost all tribal elders alive today, and
hence no firsthand accounts of unsuppressed consumption are possible.
3National Environmental Justice Advisory Council (NEJAC). Fish Consumption and Environmental Justice: A Report Developed from
the National Environmental Justice Advisory Council Meeting of December 3-6, 2001. 2002.
https://www.epa.gov/sites/production/files/2015-02/documents/fish-consump-report_1102.pdf
-------
among other things, to the presence of dams), overdrawn, and/or overfished. Were the fish
not depleted, these people would consume fish at more robust baseline levels.
• Third, a suppression effect may occur from loss of access to fisheries resources and
changes in social structure such that individuals no longer harvest fish to the same extent as
before, or do not harvest at all.
Another concern in assessing suppression is how to define the more robust "baseline" level for the
particular group affected. In some cases, a tribe will be able to cite a historical "point of reference"
that would describe an appropriate baseline in terms of environmental quality, geographic
delineation, and treaty rights. In each case, there may be important questions of history, culture,
and aspiration to be considered in determining an appropriate baseline; that is to say, an
appropriate baseline might mean examination of what people had consumed as well as aspiration
for what people would consume were there "fair access for all to a full range of resources,"
(NEJAC, 2002) or were the conditions fulfilled for full exercise of treaty- and trust-protected rights
and purposes.
The strength of the current rates is that they are derived by a technically defensible methodology,
and these rates can be compared to those of other populations. The strength of the heritage rates is
their relevance to the goals of the Tribe. The website of the Nez Perce Department of Fisheries
Resource management states, "Our vision is to recover and restore all species and populations of
anadromous and resident fish within the traditional lands of the Nez Perce Tribe."
Development of the survey design involved informational visits to the Nez Perce Tribe, including
an open exchange of interests, concerns, and ideas; collection of relevant information on culture,
history, fisheries, environment, and Tribal objectives; investigation of statistical methods and
issues; development of an appropriate statistical methodology for the current fish consumption
survey and an approach for documentation of heritage rates; preparation of a multi-part survey
questionnaire, including screening, two 24-hour dietary recalls, and food frequency questionnaire;
calculations to support a statistically valid design; and coordination with involved agencies, tribes,
consortia, and consultants.
The survey was implemented largely consistent with the Nez Perce final survey design report
(Appendix H in Volume III). Some design modifications were made while the survey for current
rates was underway to improve response rates without introducing bias. The final survey design
report also includes a description of the Nez Perce Tribe's story about suppression, based primarily
on existing literature and supplemented with input directly from the Tribe. Historical fish harvest
and fish consumption by Tribal members is presented, as well as causes of decline in the fish
populations, and goals for the future.
4 http://www.nptfisheries.org, accessed September 17, 2015.
-------
ACKNOWLEDGEMENTS
The authors wish to thank the following for invaluable collaboration and support in designing and
implementing the current fish consumption survey, the heritage rate studies and in preparation of
this report.
Region 10, EPA
Region 10, EPA
Headquarters, EPA
Nez Perce Tribe
Nez Perce Tribe
Nez Perce Tribe
Columbia River Inter-Tribal Fish Commission
Lon Kissinger
Mary Lou Soscia
Deborah Dalton
Michael Lopez
James Holt
Joseph Oatman
Dianne Barton
The authors also wish to thank the following colleagues in the contractor team. Greg Frey of SRA
International; Elizabeth McManus, Gerald Boese and Ken Ghalambor of Ross Strategic; Tambria
Cox, Penny Lamb and the interviewers of Pacific Market Research; Mayuri Mandel of The
Mountain-Whisper-Light Statisties.
Most of all we wish to thank the members of the Nez Perce Tribe who served as respondents to this
survey, patiently sitting through long interviews and sharing important and personal aspects of their
lives. Without their stories, this report could never have been written.
We are grateful to all of those mentioned above, and to others who helped carry this project to its
completion. An important addition to any acknowledgment such as this is the authors' affirmation
that any errors of fact, method, numeric values or interpretation in this report belong only to the
authors and not to any of the people, organizations or sources that were consulted or cited.
-------
Volume I:
Heritage Fish Consumption
Rates of the Nez Perce Tribe
-------
TABLE OF CONTENTS
1.0 INTRODUCTION 1
1.1 Purpose and Obj ectives 1
1.2 Study Approach 2
2.0 BACKGROUND 3
2.1 Summary of Historical Fish Harvest and Consumption 3
2.2 Summary of Causes of Decline in Fish Populations 5
2.3 Vision for the Future 7
3.0 HERITAGE FISH CONSUMPTION RATES (FCRs) 12
3.1 Defining Fish Consumption 12
3.2 Defining Factors Influencing Consumption Rates 12
3.2.1 Migration Calorie Loss Factor 13
3.2.2 Waste Loss Factor 14
3.2.3 Other Assumptions used to Develop Consumption Rates 14
3.3 Columbia Basin-Wide Heritage Rates 14
3.3.1 Craig and Hacker, 1940 15
3.3.2 Swindell, 1942 15
3.3.3 Hewes, 1947 16
3.3.4 Griswold, 1954 17
3.3.5 Walker, 1967 17
3.3.6 Boldt, 1974 17
3.3.7 Hunn, 1981 18
3.4 Nez Perce Tribe Heritage Rates 18
3.4.1 Walker, 1967 18
3.4.2 Hewes, 1973 19
3.4.3 Marshall, 1977 19
3.4.4 Walker, 1985 19
3.4.5 Schalk, 1986 20
3.4.6 Hunn and Bruneau, 1989 20
Nez Perce Tribe
Volume I: Heritage Fish Consumption Rates
December 2016 Page i
-------
4.0 RATE EVALUATION AND DISCUSSION 21
4.1 Factors Influencing Consumption Rates 21
4.1.1 Migration Calorie Loss Factor 21
4.1.2 Waste Loss Factor 22
4.1.3 Other Assumptions used to Develop Consumption Rates 23
4.2 Heritage Fish Consumption Rates (FCRs) 24
4.2.1 Columbia Basin-Wide Heritage Rates 24
4.2.2 Nez Perce Tribe Heritage Rates 25
5.0 REFERENCES FOR VOLUME 1 26
6.0 TABLES 28
Nez Perce Tribe
Volume I: Heritage Fish Consumption Rates
December 2016 Page ii
-------
LIST OF TABLES
Table 1. Average Heritage Fish Consumption Rates for Columbia Basin Tribes
Table 2. Average Heritage Fish Consumption Rates for the Nez Perce Tribe
Table 3. Spawning Migration and Calorie Loss (Fraser River)
LIST OF ABBREVIATIONS AND ACRONYMS
EPA U.S. Environmental Protection Agency
ESA Endangered Species Act
FCR Fish Consumption Rate
ICC Indian Claims Commission
IDFG Idaho Department of Fish and Game
NPT Nez Perce Tribe
LIST OF UNITS
% percent
cal/d calories per day
g/d grams per day
kCal kilocalories
km kilometers
lb/d pounds per day
lb/yr pounds per year
Nez Perce Tribe
Volume I: Heritage Fish Consumption Rates
December 2016 Page iii
-------
1.0 INTRODUCTION
A study of heritage Fish Consumption Rates (FCRs) was conducted for the Nez Perce Tribe. The
study was done as part of a larger fish consumption survey of federally recognized Tribes in Idaho,
which was initiated by the U.S. Environmental Protection Agency in 2013. This report presents the
results of the Nez Perce Tribe's heritage rate research, which was based upon an evaluation of
available ethnographic literature on aboriginal fish consumption by Columbia Basin Tribes and
other influential studies that have supported previous estimates of heritage rates.
1.1 Purpose and Objectives
Tribal Governments in the State of Idaho are working closely with the U.S. Environmental
Protection Agency (EPA) Region 10, the State of Idaho, and other stakeholders to gather data on
FCRs. The overarching goal of this process is to obtain information on fish consumption to enable
Tribal governments to set water quality standards for tribal waters, and to allow Tribes to
meaningfully participate as informed partners in Idaho DEQ's ambient water quality criteria
review process that impacts tribal interests. A Tribal heritage rate study was conducted as part of
this effort.
Recognizing that current Tribal fish consumption is suppressed due to a number of factors (e.g.
decreased fish populations due to physical habitat modifications and adverse effects of chemical
contamination, loss of Tribal access to fisheries resources, fears of exposure to contaminants in
fish, and changes in fish harvesting by Tribal members associated adaptation to economic and
cultural shifts), this study compiled and evaluated available data to determine heritage FCRs for
the Nez Perce Tribe (NPT). Knowledge of past rates may help determine how current FCRs might
increase in the future if current fisheries resources are improved and fish consumption is restored to
past, higher levels. Information about FCRs may be used to support development of water quality
standards that protect human health.
Water quality is of great importance to the Nez Perce Tribe, since a substantial portion of their diet
is derived from aquatic sources, and water and aquatic resources are of great cultural and spiritual
significance. As part of the survey effort, discussions with the Tribe highlighted the issue of
suppression of current fish consumption and its causes. Therefore, the survey team agreed to
review and evaluate heritage rates available in the literature, which may be more relevant than
current suppressed rates to the long-term restoration goals of the Tribe.
The Nez Perce Tribe has treaty reserved fishing rights within the Columbia Basin and Snake River
basins. In the Snake River Basin, the Nez Perce Tribe has quite possibly the largest number of
tributary salmon and steelhead fisheries which can often occur year- round across the states of
Washington, Oregon and Idaho. The NPT has usual and accustomed fishing places throughout 13
million+ acres that have been found to been exclusively used and occupied by the Tribe (including
the major portions of the Snake, Tucannon, Imnaha, Grande Ronde, Salmon and Clearwater Rivers
and their drainages); the mainstem Columbia River; and other locations in the Columbia/Snake
River Basin.
The Nez Perce Tribe's primary objective for the fish consumption survey is to support
development of more stringent water quality standards that are protective of tribal members'
consumption of fish. The Tribe's culture is and always has been intimately tied to fish, which is a
Nez Perce Tribe
Volume I: Heritage Fish Consumption Rates
December 2016 Page 1
-------
staple of their diet and an integral part of their society; poor water quality impedes fish survival and
can affect both the quantity and availability of fish that can be harvested and safely consumed by
tribal members. The NPT has a vision of restoring fish species native to the Nez Perce Treaty
Territory. To accomplish this vision, the Tribe has engaged in managing the resident and
anadromous fish species in the streams, lakes, and watersheds within their management authority
in an effort to rebuild habitat and restore opportunities for fish harvest. Their goal is that fish will
be found in all available habitats and will provide fishing opportunities for present and future
generations. Increased fisheries resources will support higher fish consumption.
1.2 Study Approach
The approach for estimating heritage rates was based on a comprehensive review and evaluation of
literature that is relevant to heritage rates, including historical accounts and modern studies of
heritage consumption. For Tribes that harvest fish from the Columbia Basin, there is a significant
volume of literature to form the basis for a range of quantitative estimates of fish consumption.
Information includes ethnographic studies, personal interviews, historical harvest records,
archaeological and ecological information, and nutritional and dietary information. The
quantitative assessment includes compilation and analysis of historic and heritage information
across the region of the Columbia Basin.
The survey team compiled and evaluated available information regarding heritage consumption
rates relevant to the Nez Perce Tribe. The development of estimates of heritage rates presented
here includes a discussion of the available information, including methodologies used to develop
the fish consumption estimates and factors affecting the uncertainty associated with the estimates.
Based on available information, a quantitative range of heritage FCRs is presented for the Tribe.
Certain key geographic features referred to in the following discussion are mapped in Figure 1.
Nez Perce Tribe
Volume I: Heritage Fish Consumption Rates
December 2016 Page 2
-------
2.0 BACKGROUND
The Nez Perce Tribe has relied extensively on fish resources and fishing activities throughout time.
A summary of the fish harvest and extensive use and consumption of fish historically, as well as
the causes of decline in fish availability over time, is provided for context.
2.1 Summary of Historical Fish Harvest and Consumption
The Nez Perce are a large Northwest tribe with a culture tied closely to fish. Since time
immemorial, the Tribe occupied a territory covering more than 13 million acres that included what
is today north central Idaho, southeastern Washington, and northeastern Oregon. The Nez Perce
subsistence cycle involved traveling year to year on the same well-traveled routes through the
canyons of the Snake, Tucannon, Clearwater, Grande Ronde, Imnaha and Salmon Rivers,
primarily to follow the salmon runs. In addition to those rivers and their tributaries, the Nez Perce
historically took part in the fishing and trading that occurred between several of the region's tribes
at Celilo Falls on the Columbia River, among other locations of the Columbia Basin.
The Tribe has always fished. Their economy and culture evolved around Northwest fish runs.
Their persistence can be attributed in large part to the abundance of fish, which has served as a
primary food source, trade item and cultural resource for thousands of years. Settlement by others
in the last 150 years has disrupted people of the Tribe and the natural resources (NPT, 2005). The
degree to which the Tribe is culturally coupled to fish was recognized in treaties signed between
the Tribe and the United States Government. The same treaties that confined the Tribe to a fraction
of their former territory also guaranteed their access to fishery resources. Article III of the Treaty
of 1855 guarantees to the Tribe:
"The exclusive right of taking fish in all the streams running through or bordering
said reservation ...as also the right of taking fish at all usual and accustomed
places in common with citizens of the Territory. " Treaty with the Nez Perces, 12
Stat. 957 (1859).
The 1855 Treaty Council at Walla Walla and the Treaty negotiations reflect the Tribe's inherent
tribal sovereignty and its "aboriginal title" to land. At the Treaty Council, the United States sought
to clear title to lands; the Nez Perce sought to reserve and maintain a homeland ("Reservation")
and reserve its aboriginal rights and way of life. The Nez Perce would not have signed this treaty
without first receiving assurances that these rights, including the right to fish, would be protected
into the future. Additional treaties between the two sovereigns have been made, but the reserved
fishing right has remained unchanged since 1855.
In its 1855 Treaty, the Nez Perce reserved a significant portion of their aboriginal land (about 8
million acres). And, this Nez Perce homeland contained, as the United States recognized, many of
the best fisheries:
Gov. Stevens said: "Here (showing a draft on a large scale) is a map of the
Reservation. There is the Snake River. There is the Clear Water river. Here is the
Salmon river. Here is the Grande Ronde river. There is the Palouse river. There is
the El-pow-wow-wee. This is a large Reservation. The best fisheries on the Snake
River are on it...
Nez Perce Tribe
Volume I: Heritage Fish Consumption Rates
December 2016 Page 3
-------
Moreover, in addition to this homeland, Nez Perce leaders insisted on reserving off-reservation
hunting, fishing, gathering, and pasturing rights. The minutes of the treaty negotiations reflect
Governor Stevens' repeated assurances, on behalf of the United States, that the treaty would
reserve these off-reservation rights to the Nez Perce Tribe:
You will be allowed to pasture your animals on land not claimed or occupied by
settlers, white men. You will be allowed to go on the roads, to take your things to
market, your horses and cattle. You will be allowed to go to the usual and
accustomedfishing places andfish in common with the whites, and to get roots
and berries and to kill game on land not occupied by the whites; all this outside
the Reservation:"
Gov. Stevens said: "I will ask of Looking Glass whether he has been told of our
council. Looking Glass knows that in this reservation settlers cannot go, that he can
graze his cattle outside of the reservation on lands not claimed by settlers, that he
can catch fish at any of the fishing stations, that he can kill game and can go to
Buffalo when he pleases, that he can get roots and berries on any of the lands not
occupied by settlers...
Fish, as a staple of the Nez Perce diet, have always been an integral part of the Nez Perce society.
Principal to the Nez Perce diet were the anadromous fish species that inhabit the rivers of the
inland northwest. This is corroborated by other existing information such as those from federal
court proceedings.
For example, in its 1967 decision concerning the Nez Perce Tribe, the Indian Claims
Commission (ICC) made comprehensive findings based on detailed anthropological evidence
from both the United States and the Nez Perce Tribe, of the Tribe's area of "exclusive use and
occupancy" and "aboriginal ownership." The ICC determined that the Nez Perce had "exclusive
use" and occupancy of 13,204,000 acres of land and "that salmon fishing was one of the major
sources of subsistence since the main rivers through the area, which include the Snake, the
Clearwater, the Salmon, and their branches, were well supplied with this fish in aboriginal
times." It also concluded that their seasonal "cycle consists of specific times of the year for
fishing for salmon, digging camas and other roots, hunting the game"; this "economic cycle can
generally be summarized as ten months salmon fishing and two months berry picking, with
hunting most of the year."5
During the time that the treaty was negotiated, the salmon resource reserved by the Nez Perce
came from ".. .river systems that were biologically functional and fully productive..." (Meyer
Resources, 1999). The decline of salmon productivity since the mid-1800s to present, does not
alter, change, or abrogate the Nez Perce treaty right to take fish. This right to take fish represents
an inherent right that the Nez Perce have held since time immemorial. The fishing right is as
important to the Nez Perce today as it was before contact with non-Indians.
5 The ICC was created by Congress in 1946 to hear claims by Indian tribes for, among other things, compensation for the taking of
aboriginal lands by the United States without fair payment. Compensable aboriginal title was required to be based on "actual and
exclusive use and occupancy 'for a long time' prior to the cession, transfer, or loss of the property." It provided historical information
regarding Nez Perce village sites, uses of natural resources, and range and extent of natural resource use.
Nez Perce Tribe
Volume I: Heritage Fish Consumption Rates
December 2016 Page 4
-------
The Nez Perce governed where fishing occurred, how many fish were to be harvested, who could
participate, how to use the resource, and ways to honor and perpetuate the resource. They
developed ways to harvest large amounts of fish. These were documented as proven methods to
catch the substantial numbers of salmon and steelhead (as well as other species of fish). The
complex, elaborate, and efficient Nez Perce fishing techniques described below document the
extent of their reliance on this valuable resource and the importance of fish to its society and
cultural identity.
Whenever possible, the Nez Perce historically and contemporarily have regularly fished for the
following species: Chinook, Coho, and Sockeye varieties of salmon; Dolly Varden, Cutthroat,
Brook, Lake, and Rainbow varieties of trout; several species of suckers, white fish, sturgeon,
squawfish (Northern pikeminnow), lampreys, and some shellfish (freshwater clams). In order to
harvest these fish species, the Nez Perce developed a number of fishing techniques and methods:
weirs and traps; dipping platforms (either natural or man-made); fish walls and dams; canoes;
spears; hook and line; gaffs; and variety of nets (dipnets, set nets, and throw nets).
The expansive territory of the Nez Perce people was rich in rivers and streams abundant in fish life.
Bands fished from the Snake, Salmon, Clearwater, Imnaha, Grand Ronde, Selway, Tucannon,
Rapid River and many other rivers within and outside its homeland and territory. As with other
tribes, the Nez Perce did not limit their fishing to salmon. Research has been conducted by a
number of people in an effort to determine how many fish were historically harvested by the Nez
Perce. There are a number of methods to estimate amount of fish harvested and consumed by the
Nez Perce (commonly expressed in numbers of fish harvested and annual per capita consumption).
In addition to salmon and steelhead, the Tribe has traditionally harvested Snake River white
sturgeon for subsistence purposes. Tribal elders confirm the historical presence of white sturgeon
throughout the Snake River, mainstem Salmon River, the Clearwater River from its mouth to
above Orofino, Idaho, as well as seasonal migrations into the Grande Ronde River (Elmer Crow,
Nez Perce Tribe Department of Fisheries Resources Management, Personal Communication,
2014). In addition to being an important food source, white sturgeon served many purposes in the
culture of the Tribe. White sturgeon blood was used to make glue; the hides were used for bow
cases and quivers, and for water proofing footwear. However, subsistence fishing has been
severely limited as a result of low white sturgeon numbers between Hells Canyon and Lower
Granite dams (NPT, 2005).
The traditional way of life for the Nez Perce (e.g. gathering, harvesting, ceremonies, and traditions)
depends on continuance of the circle of life for all native species (plants and animals). To the Nez
Perce, the rights reserved under the Treaty of 1855 must be protected such that the enjoyment of
these rights resembles that envisioned by the treaty signers and Nez Perce leaders.
2.2 Summary of Causes of Decline in Fish Populations
Nez Perce tribal elders believe that one of the greatest tragedies of this century is the loss of
traditional fishing sites and Chinook salmon runs on the Columbia River and its tributaries. They
believe the circle of life has been broken and ask us to consider what the consequences of breaking
that circle may mean for future generations. In many ways the loss of the salmon mirrors the plight
of the Nez Perce people. The elders remind us that the fates of humans and salmon are linked
(Landeen and Pinkham, 1999). This dependence on fish to meet dietary, spiritual, and basic
Nez Perce Tribe
Volume I: Heritage Fish Consumption Rates
December 2016 Page 5
-------
subsistence needs is still a prevailing necessity of Nez Perce life. To this day, the right to a "fair
share" of the salmon harvest by the Nez Perce Tribe does not occur because of the impacts to these
fish by non-Indian activities and development in the Columbia and Snake basins.
The Nez Perce lived in the heart of salmon country - along the Salmon, Snake, Grande Ronde,
Imnaha, Clearwater and Tucannon rivers; which historically were major salmon and steelhead
producers. The Nez Perce have lived through and experienced the extirpation of entire populations
of fish by blocking and altering of thousands of miles of rivers and streams as result of dams. The
Hells Canyon, Oxbow, and Brownlee dams on the Snake River, Wallowa Lake Dam on the
Wallowa River, Dworshak Dam on the North Fork Clearwater, the eight major dams on the
Columbia and Snake rivers, and the many other smaller projects, have individually and collectively
impacted fish, and thus the Nez Perce ability to fish for them.
The environment and water that support fish has been altered due to human development and
enterprise over the past century and a half. This human progress has come at a cost to the fish
species and "salmon people." Current productivity of salmon- producing streams is much lower
than it was historically. Many of the fish species either face extinction or are in seriously depressed
conditions. As a result, tribal harvest in the present day is only a very small fraction of what the
Nez Perce harvested in the mid- 1800s. Although hard to quantify, it is probable that until recently
harvest has been less than 1% of historic harvest levels prior to 1855.
Causes contributing to salmon and steelhead decline encompass a variety of human activities and
anthropogenic and natural phenomena. These include the following: commercial, recreational, and
subsistence fishing; freshwater and estuarine habitat alteration due to urbanizing, farming, logging,
and ranching; dams built and operated for electricity generation and flood control; water
withdrawals for agricultural, municipal, or commercial needs; stream and river channel alterations;
hatchery production; predation by marine mammals, birds, and other fish species; competition with
other fish species; diseases and parasites; and reduction in annual nutrient distribution from
spawned-out salmon to the local ecosystem. These activities continue to affect fish.
Salmon and steelhead runs in the Snake Basin are not as abundant or productive as they were
historically. Snake River Chinook salmon (spring, summer, and fall runs), sockeye, and steelhead
are listed under the Endangered Species Act (ESA). Coho and Chinook salmon were extirpated in
the Clearwater River subbasin in the 1990s, and steelhead were at very depressed levels.
Snake River spring/summer Chinook salmon were historically found spawning in the Snake River
tributaries of the Clearwater, Salmon, Weiser, Payette, and Boise Rivers. A review of run size for
Snake River of spring/summer Chinook salmon is provided by Matthews and Waples (1991).
Their summary of research on run size reports historic runs in the Snake River probably exceeded
one million fish annually in the late 1800s. By the mid-1900s, the abundance of adult spring and
summer Chinook salmon had greatly declined to near 100,000 adults per year in the 1950s. Since
the 1960s, counts of spring and summer Chinook salmon adults have declined considerably at the
lower Snake River dams (IDFG, 2013).
The construction of hydroelectric dams on the main stem Snake and Columbia Rivers blocked
access to nearly half of the historic spawning habitat and reduced survival of juveniles and adults
migrating to and from the ocean. Additional effects from hydroelectric dams and water storage
projects have resulted in altered hydrographs and water temperature regimes affecting run timing
of juveniles and adults. Diversions in spawning and rearing streams have caused direct mortality,
Nez Perce Tribe
Volume I: Heritage Fish Consumption Rates
December 2016 Page 6
-------
loss of habitat and migration barriers. Land management activities have resulted in degraded
habitat with the loss of riparian cover, sedimentation and artificial barriers to passage. The addition
of hatchery programs to mitigate for lost habitat and survival of fish have introduced genetic
concerns about effects to wild stocks. Declining water quality from increasing development in and
along river and tributary streams can affect fish populations. Introductions of non-native fish in
some waters can increase predation and competition with juvenile fish (IDFG, 2013).
Salmon runs in the Clearwater River Subbasin were virtually eliminated by the construction of
hydroelectric dams (Matthews and Waples, 1991). In 1910, the Harpster Dam, constructed on the
lower South Fork Clearwater River, prevented all fishes from returning upstream of Harpster, ID,
and eliminated access to over 95% of the watershed and its high quality spawning grounds
(Schoning, 1940). In 1927, the Washington Water Power Diversion Dam constructed just above
the mouth of the Clearwater River eliminated all upriver salmon runs (Parkhurst, 1950; USFWS,
1962). A crude fish ladder was built on the lower Clearwater River dam, which allowed steelhead
passage during higher flow periods, but proved almost impassible during lower flows when salmon
arrived (Parkhurst, 1950). The ladder was not modified for a period of 12 to 14 years; eliminating
all late returning fish, like coho and fall Chinook salmon (all as cited in Everett, et al, 2006).
The cumulative loss of anadromous fish to the Nez Perce Tribe as a result of these two dams was
substantial (Cramer, et al., 1993). The Harpster Dam was removed in 1963 and the lower
Clearwater River dam was removed in 1972, making available most of the salmon production
areas in the drainage. However in 1971, Dworshak Dam was built just upstream of the mouth of
the North Fork Clearwater River. Dworshak Dam lacks fish passage, resulting in the permanent
loss of productive salmonid spawning aggregates and high quality habitat. The lower Clearwater
River temperature regime continues to be altered by Dworshak Dam, resulting in warmer water in
the winter and cooler water in the summer (Arnsberg, et al., 1992, Arnsberg and Statler, 1995; all
as cited in Everett et al., 2006).
Currently, a majority of the fisheries that occur in the Snake River basin are supported by hatchery
programs. All of the anadromous fish hatcheries in the Snake River basin are mitigation hatcheries
for the development of hydroelectric dams. All of the returns from these hatcheries pass through or
return to the Nez Perce Tribe's usual and accustomed fishing places.
2.3 Vision for the Future
The Nez Perce Tribe has a vision of restoring all fish species native to the Nez Perce Treaty
Territory. To that end, the Tribe has engaged in management of all fish species- both resident
and anadromous - for all streams, lakes and watersheds within their management authority.
The Tribe is involved in these efforts to protect implementation of treaty rights, to restore
species and conditions consistent with the treaty, and to protect the long-term productivity of
their natural resources.
Today, maintaining a healthy 13-plus million acre watershed and improving survival of salmon
and steelhead under the auspices of the 1855 Treaty, rests with the Tribe's Department of Fisheries
Resources Management program and policy direction from the Nez Perce Tribal Executive
Committee (NPTEC), the governing body of the Nez Perce Tribe. Native fish within the Nez Perce
Country depend on healthy habitats, healthy watersheds, and healthy ecosystems. Sound fisheries
and habitat management actions will be implemented to improve survival, production, recovery
Nez Perce Tribe
Volume I: Heritage Fish Consumption Rates
December 2016 Page 7
-------
and restoration of all populations of native anadromous and resident fish species and their habitats
throughout the Nez Perce Tribe's usual and accustomed fishing places. It is the Tribe's desire that
all species and populations of anadromous and resident fish and their habitats will be healthy and
harvestable throughout the Nez Perce Tribe's usual and accustomed fishing places.
As described in the Department's Strategic Management Plan (NPT, 2013), Tribal member use of
and access to all treaty rights and resources guaranteed under the Treaty of 1855 guide's the
department's restoration program and actions:
• All native anadromous fish and resident fish have had long-standing cultural significance to
the Nimiipuu, including: subsistence value, ceremonial and spiritual value, medicinal
value, economic or commercial value, and intrinsic value.
• Native fish populations thrive best under natural or normative conditions to which they are
best adapted.
• Natural ecosystems have been and will continue to be increasingly stressed and altered by
human activities and population levels.
• When historic natural conditions are not achievable, altered ecosystems should function
adequately enough to maintain harvest opportunities.
• The entire life cycle of a species must be successfully carried out (from egg through
adulthood) for that species or population to persist.
• Failure to serve a species' needs, at any life history stage, can lead to extirpation of
populations.
• Federal governmental agencies have treaty trust responsibilities; their actions must
recognize the treaties as federal commitments and their actions must be taken in support of
a tribe's ability to exercise rights guaranteed in the treaties.
The following goals seek to secure the integrity of populations and habitat features essential to
anadromous and resident fish:
• Achieve and maintain fish abundance in tributary-specific areas at levels sufficient to
support: 1) population persistence, 2) harvest, and 3) ecological processes.
• Achieve and maintain diverse and productive ecosystems with species composition and
productivity consistent with historic conditions.
• Achieve and maintain adult spawner distribution consistent with historically utilized
tributaries (includes within and across tributary spatial scales).
• Achieve and maintain fish population genetic diversity at levels adequate for population
persistence and consistent with historic conditions.
• Ridge top to ridge top watershed protection and restoration for rearing and spawning
habitats and protection of water quality.
• Supplementation approach "putting fish in the rivers" with hatchery tool.
Nez Perce Tribe
Volume I: Heritage Fish Consumption Rates
December 2016 Page 8
-------
• Protection and providing flows, water quality and passage for upstream and downstream
migrants.
• Participate in Pacific Salmon Treaty and US v Oregon for ocean and in-river harvest
management.
• Allow an abundance of spawners to protect the resource for future generations.
• Monitor our activities and the runs to determine how things are faring.
• Harvest opportunities currently available will be protected and enhanced.
The Nez Perce Tribe continues to protect and enhance abundance of fish through natural
production and artificial production in the form of hatcheries. Hatcheries for salmon and steelhead
in the Columbia Basin were developed as a necessary mitigation tool to compensate for the fishery
losses that resulted from the impacts of increased human settlement that began soon after
ratification of the Treaty of 1855.
Accordingly, hatcheries represent a promise to those who have always depended on the salmon for
culture, sustenance, and livelihood to replace the fish that are and were diminished as a result of
human development of salmon habitats. In the Snake River Basin, all but one of the hatcheries
(Kooskia), were built specifically to mitigate for the impacts of the development and operation of
hydroelectric dams (Dworshak, Brownlee, Hells Canyon, Oxbow, Lower Granite, Little Goose,
Lower Monumental, Ice Harbor, McNary, John Day, The Dalles, and Bonneville dams). These
hatchery programs play a very important role in meeting congressionally mandated mitigation
obligations and treaty trust responsibility to protect and maintain tribal treaty reserved fisheries.
The Department has been a leader in implementing supplementation programs and hatchery
reform. Tribal goals for supplementation programs are: increased abundance (both total and natural
origin) and spatial structure; maintenance of culturally and economically important tribal salmon
fisheries; contribution to non-Indian fisheries; and restored ecosystem processes and health.
The Fisheries program has over 150 employees and operates on a budget derived from more than
50 contracts. There are 7 divisions within the program: Administration, Conservation Enforcement,
Harvest, Production, Research, Resident Fish and Watershed. The Fisheries program works
throughout the ceded lands and has offices in Powell, Red River, Grangeville, Orofino, McCall,
Sweetwater, Lapwai and Joseph, OR. Tribal staff coordinate and interact with State, Federal and
Tribal agencies and committees and private entities in assessing and implementing fish recovery
and restoration plans and actions.
The Department has engaged in a significant body of work throughout its U&A areas -
implementing more restoration actions within the Snake River basin than perhaps any other single
entity or agency. The aquatic habitat is subject to a diverse array of natural and anthropogenic
influences and impacts and given the synergistic effect of watershed health on aquatic habitat
quality, the Department employs a "ridge-top to ridge-top" approach to restoration.
The Department adopted abundance-based reference points (thresholds) for certain anadromous
fish to assist in development of long-term management strategies and to guide the implementation
of short-term management actions to achieve both broad and population-specific salmon rebuilding
goals. Adult salmon abundance (or escapement) objectives are our primary measure for
Nez Perce Tribe
Volume I: Heritage Fish Consumption Rates
December 2016 Page 9
-------
quantifying goals and are generally defined as the number of adults and jacks in each population
that return to their river of origin.
These identified abundance thresholds serve as useful decision criteria that trigger specific
actions (e.g. harvest rates or initiation and other management actions). Populations at very
depressed to critically low levels require "more aggressive actions and demand a more rapid
population response than populations fluctuating at higher, less risky levels of abundance."
Reference abundances or population designations specified in this section include the designated
escapement objective, and the ecological escapement objective for four focal species,
spring/summer Chinook, steelhead, and fall Chinook (see Table below). The following are
descriptions for each threshold type.
• Viable abundance thresholds are considered the size at which a population maintains
essential genetic diversity, and at which there is negligible risk of long-term extinction
given contemporary levels of environmental variability. They are the minimum abundance
for a healthy population.
• Sustainable Escapement Objectives describe the numbers of returning adults that would
annually sustain substantial spawning as well as harvest for tribal and non-tribal fisheries. It
is assumed that escapement sizes reflecting these values would also encompass healthy
tribal and non-tribal fisheries downriver.
• Ecological Escapement Objectives refer to the escapement level at which sustainable
spawning abundance is maximized within a population, the full utilization of available
spawning and rearing habitat is promoted, and the ecosystem-level processes (e.g., nutrient
redistribution) for multiple species are fostered. Historical salmon and steelhead
escapement to the Columbia and Snake river basins was 8-16 million and 500,000 - 2
million, respectively (NPPC, 1986; CBFWA, 1990; Chapman, 1986; Fulton, 1968).
According to tribal knowledge, escapement at those historic levels to tributary-specific
areas resulted in "fish so thick you could walk across their backs."
The following table depicts the aggregated abundance thresholds for certain fish species.
Table 2-1. Abundance Thresholds for Certain Snake River Anadromous Fish
Species
# Major
Population
Group
#
Population(s)
Viable
Abundance
Sustainable
Harvest Goal
Ecological
Escapement
Goal
Spring/Summer
Chinook
7
41
31,500
215,900
669,000
Fall Chinook
1
1
3,000
39,110
120,000
Steelhead
6
25
25,500
330,200
602,000
Nez Perce Tribe
Volume I: Heritage Fish Consumption Rates
December 2016 Page 10
-------
The Nez Perce Tribe intends to increase and expand the level of harvest or fishing areas for salmon
and steelhead at all Nez Perce usual and accustomed places, including those in the Snake Basin, in
a way that balances conservation needs of the fish with the right to take fish. This can be achieved
through a biologically-sound harvest management philosophy and harvest rate schedules keyed to
the status and trends in abundance and productivity of fish resources. Generally, abundance-based
tribal harvest strategies can be designed to account for annual variation in total fish run size and
run composition. This is illustrated in the Figure below.
Figure 2-1. Abundance-Based Tribal Harvest Goals
High
Moderate
Low
Minirral
i ¦
Eitintion Viable
Sustainable
Ecological
Adult Escapement Threshold
As returns increase, the Nez Perce Tribe expects to increase the relative magnitude of tribal harvest
and fishing effort and fish consumption.
When restoration efforts result in sustainable returns, the Tribe anticipates that Tribal harvest will
increase and fish consumption rates will rise when fish populations attain "sustainable abundance"
and "ecological abundance" levels of adult escapement. Ultimately, the goal is to achieve a harvest
consistent with pre-Treaty harvest levels. Simply put, the Tribe's goal is to rebuild the Snake River
fishery to healthy, self-sustaining levels that will in turn support sustainable treaty fisheries.
Nez Perce Tribe
Volume I: Heritage Fish Consumption Rates
December2016 Pagell
-------
3.0 HERITAGE FISH CONSUMPTION RATES (FCRs
A summary of the primary source literature reviewed for this heritage rate study is provided here,
including a definition offish consumption," as used differently by various authors, and certain
factors and other assumptions that have been used to adjust and/or calculate consumption rates.
Also presented below are the average aboriginal per capita FCRs estimated for the Columbia
Basin Tribes (summarized in Table 1) and rates for the Nez Perce Tribe specifically (summarized
in Table 2).
3.1 Defining Fish Consumption
The focus of this effort is to compile, summarize, and evaluate estimates of Tribal fish
consumption during the period when Tribes had full access to their traditional fisheries, which we
refer to here as "heritage rates." This effort is intended to provide Tribes with information that may
be useful in establishing water quality criteria for the protection of human health. The information
supporting heritage rates is on a per capita basis that can be used to estimate average FCRs,
however this information is not suitable for development of FCR distributions or percentiles of fish
consumption.
As evident in review of the documentary record, the definition of fish consumption as fish
ingestion is not necessarily shared by the various researchers who have attempted to estimate
aboriginal FCRs for various Tribal groups. Several researchers include all uses of fish in what they
describe as a "total consumption rate." For example, one researcher (Schalk, 1986), suggested that
a previously calculated consumption estimate was too low because it "only considers human
dietary demands." Another (Griswold, 1954) stated that "[t]he tribes here required salmon for fuel
as well as for food. Consequently, it may be inferred that their per capita consumption was
considerably greater than that of the tribes [downstream] below." Still another, (Walker, 1967)
discussed "exceptional areas of unusually high consumption, up to 1000 lbs. per capita, per year"
which are "caused not only by the high calorie demands typical of colder climates, but also by the
use of fish for dog food or for fuel."
Estimates by various researchers, therefore, may include as part of a total FCR that portion of the
overall fish harvest that was used for trade, for fuel, for animal feed, or may include the inedible
portion of fish not actually ingested. To the extent that it is discussed in the literature, this report
attempts to describe the assumptions involved in estimating a consumption rate, and, where
possible and appropriate, identify that portion that was actually ingested.
3.2 Defining Factors Influencing Consumption Rates
Many sources of information providing estimates of heritage FCRs for Tribal groups in the
Columbia Basin tend to refer to or build upon previous work, in some cases revising or adjusting
rates from previous reports based on new knowledge, new data, or new approaches for interpreting
consumption information. Some authors have attempted to revise earlier estimates of fish
consumption, particularly those estimates based on caloric intake, to account for the caloric losses
that occur as a result of salmon spawning migration ("migration calorie loss factor") and to account
for the fact that not all of an individual fish is consumed ("waste loss factor"). Each of these factors
and their effect on consumption estimates, as well as other variables that influence the calculation
of consumption rates, are discussed below.
Nez Perce Tribe
Volume I: Heritage Fish Consumption Rates
December 2016 Page 12
-------
3.2.1 Migration Calorie Loss Factor
Eugene Hunn (1981) appears to be the first author to suggest modifying the calorie-based fish
consumption estimates originally developed by Gordon Hewes (1947, 1973). While Hunn
considered Hewes' estimates of salmon consumption to be "the most comprehensive attempted to
date for the region" he contends that "his interpretation of the nutritional factors is misleading."
Specifically, Hewes's caloric calculations did not account for the calories that salmon lose during
spawning migration (since migrating salmon no longer feed once they re-enter freshwater).
Citing a study by Idler and Clemens (1959), who determined that sockeye salmon lose 75% of
their caloric potential during spawning migration in the Fraser River watershed, Hunn proposed
the following approach, as transferred to the Columbia River watershed: the "migration calorie
loss factor" is computed as a ratio of (a) the distance in river-kilometers (km) from the mouth of
the Columbia River to the approximate middle of each group's territory, to (b) the entire length
of the Columbia River (1,936 km). This ratio was then multiplied by the average value for
calorie loss during salmon migration, 75% (0.75), and the product was subtracted from one. For
example, a salmon harvested halfway to the headwaters of the Columbia River is assumed to
have lost half of 75%, or 37.5% (0.375) of its beginning caloric potential, and, therefore, would
retain 62.5% of its beginning caloric potential (1 - 0.375 = 0.625), which is considered the
migration calorie loss factor. Based in part on this adjustment, Hunn suggested that Hewes likely
overestimated the calories provided by salmon, and therefore salmon's contribution to the overall
diet, and that "vegetable resources" likely played a larger dietary role than assumed by other
authors. In fact, he concluded that the food collecting societies of the southern half of the
Columbia-Fraser Plateau "obtained in the neighborhood of 70% of their food energy needs from
plant foods harvested by women."
Other authors (e.g., Scholz et al., 1985; Schalk, 1986) have taken a different approach and assumed
that Hewes was correct about the proportion of the diet supplied by salmon (on average 50%, or
about 1,000 calories), but by not accounting for migration calorie loss, Hewes likely
underestimated salmon consumption rates, particularly for upriver Tribes (as Schalk, 1986, stated,
"some adjustment should have been made for distance traveled upstream"). To account for this,
Schalk divided the consumption estimates developed by Hewes by a specific migration calorie loss
factor determined for each Tribal group, following the approach described above.
Again using the example of a salmon harvested halfway to the headwaters of the Columbia River,
Hewes's estimate for average per capita consumption for the Columbia Basin tribes of 365 pounds
per year would be revised in the following manner: assuming a salmon has lost 21.5% of its initial
caloric potential during spawning migration, 62.5% of its caloric potential would remain (the
migration calorie loss factor). Dividing 365 pounds per year by 62.5% (0.625) gives a revised
estimate of 584 pounds per year - a 60% increase. In other words, a person harvesting salmon
halfway up the Columbia River would need to consume 584 pounds of salmon to get the same
amount of calories as someone consuming 365 pounds of salmon harvested at the mouth of the
Columbia. As Schalk (1986) noted, "the total annual per capita estimate for fish consumed rises
significantly when a migration calorie loss factor is included."
Nez Perce Tribe
Volume I: Heritage Fish Consumption Rates
December 2016 Page 13
-------
3.2.2 Waste Loss Factor
In addition to considering calorie loss from migration, Hunn (1981) also appears to be the first
author to suggest modifying the calorie-based fish consumption estimates originally developed by
Hewes (1947, 1973) based upon the fact that some portion of a fish is not edible. Hunn (1981)
stated that Hewes "does not allow for the fact that the edible fraction of whole salmon is generally
considered to be approximately 80% of the total weight." Since many authors providing estimates
of historical Tribal fish consumption did so for the purpose of estimating historical harvest rates,
this factor (if accurate) was likely an important consideration. For example, if only 80% of each
salmon harvested is edible (i.e., 20% is "waste"), then a person consuming 100 pounds of salmon
per year would need to harvest 125 pounds of salmon to support that consumption rate.
Schalk (1986) incorporated this "waste loss factor" into his estimates of annual salmonid catch in
the Columbia Basin by revising Hewes's consumption estimates for various Tribes and Tribal
groups. Schalk stated that "the revised estimate involves dividing the per capita consumption
estimate by a waste loss factor of 0.8 to get the gross weight of fish utilized. This figure is also
derived from Hunn's (1981) suggestion that 80% of the total weight of a salmon is edible." While it
appears that the main objective in using this factor is in estimating total catch ("the gross weight of
fish utilized"), the terms "total catch" and "total consumption" are sometimes used
interchangeably. Some subsequent authors have incorporated this waste loss factor into their
estimates of actual fish ingestion when estimating aboriginal FCRs.
3.2.3 Other Assumptions used to Develop Consumption Rates
In addition to the rate adjustment factors discussed above, there are a number of other assumptions
that various authors have made to develop consumption rate estimates, including the following
(discussed in more detail in Section 4.1.3).
• Fish ingestion versus harvest and other uses (i.e., definition of "consumption")
• Percent of diet (calories) provided by fish (versus other food items)
• Salmon (anadromous) and/or resident fish consumption
• Historical Tribal population estimates
• Number of fishing sites, fishing methods, and fishing efficiency
3.3 Columbia Basin-Wide Heritage Rates
Below is a summary of the primary source information reviewed on aboriginal FCRs of Columbia
Basin Tribes. Relevant information is presented from each of the following publications, including
fish consumption estimates and associated assumptions (and summarized in Table 1).
• Craig and Hacker, 1940
• Swindell, 1942
• Hewes, 1947
• Griswold, 1954
• Walker, 1967
• Boldt, 1974
• Hunn, 1981
Nez Perce Tribe
Volume I: Heritage Fish Consumption Rates
December 2016 Page 14
-------
3.3.1 Craig and Hacker, 1940
In 1940, Joseph Craig and Robert Hacker of the U.S. Bureau of Fisheries estimated an aboriginal
per capita salmon consumption rate of 1 pound per day (lb/d), which equates to 365 pounds per
year (lb/yr) (or 454 grams per day [g/d]6) for Columbia Basin Tribes (Table 1). This estimate is
based on historical ethnographic observations of extensive salmon harvest and use. The authors
stated that, based on accounts of early explorers:
"Without doubt salmon, either fresh or dried, was the chief single factor in the diet
of the Indians of the Columbia Basin in their native state. " (p. 140)
Other species were identified as consumed as well, including sturgeon, trout, and other fish;
however, salmon was the primary species consumed. While the authors noted that it was "not
possible to make an accurate estimate of the amount of salmon used by the Indians," at the time, an
approximation could serve "to illustrate the possible magnitude" of fish caught and consumed,
with a wide margin of error (p. 141).
The authors stated that since significant quantities of salmon were available in the Columbia River
and its tributaries during at least 6 months of the year, the Indians likely harvested and consumed
large quantities of fresh salmon during this period and then consumed dried salmon for the
remainder of the year. Therefore, "it appears to be well within the realms of probability that these
Indians had an average per capita consumption of salmon of 1 pound per day during the entire
year" (p. 142).
3.3.2 Swindell, 1942
In 1942, Edward Swindell of the U.S. Department of the Interior's Office of Indian Affairs
estimated an aboriginal per capita salmon consumption rate of 322 lb/yr (or 401 g/d) for Columbia
Basin Tribes, specifically in the Celilo region prior to the installation of the Dalles Dam and
flooding of Celilo Falls (Table 1). This estimate is based on field survey interviews (and published
affidavits) with local Indian families.
Swindell agreed that the estimate reported by Craig and Hacker (1940) of per capita salmon
consumption of 1 pound per day was "not unreasonable" (p. 13) and that while "the poundage of
the fish used for subsistence purposes cannot be definitely ascertained... the importance of this
article of food as shown by a survey of 55 representative families is shown..." in his report (p.
147). As part of this study, the author presented and compared results obtained from interviews
conducted with the heads of the 55 selected families, which represented a total of 795 Indian
families present "under the jurisdiction of the Yakima, Umatilla, and Warm Springs" (p. 13-14).
These interviews determined an average consumption rate of 1,611 lb/yr per family. Assuming a
family unit was comprised of 5 members, Swindell calculated this to be a per capita rate of 322
lb/yr. This value accounted for both fresh and cured salmon, where the dried weights were
converted to wet (fresh) weights. The affidavits given by participants of the survey supported
Swindell's aboriginal fish consumption estimates.
6 Most sources present rates in pounds per day; this report applies a conversion to grams per day (1 pound = 454 grams) for the reader
and for applicability to water quality standards.
Nez Perce Tribe
Volume I: Heritage Fish Consumption Rates
December 2016 Page 15
-------
An affidavit provided by Tommy Thompson (age 79), of the Wyam Tribe of Indians residing at
Celilo, Oregon, stated that "each family of Indians, when he was a boy,7 would dry and put away
for their own future use, about 30 sacks of fish... each sack would contain about 10 or 12 fish
which weighed almost 100 pounds [total]... each fish after it had been cleaned, the head and tail
removed, and then dried, would only weigh between 6 and 8 pounds" (p. 153). Another affidavit
provided by Chief William Yallup (age 75), a Klickitat Indian of Rock Creek, stated that "when he
was a boy... during the [fish] runs, they would eat fresh fish three times daily and the surplus they
caught would be dried for use when no fresh ones were available" and "that in those days each
family would dry for its own personal use approximately 30 sacks of fish, each of which contained
about six large salmon weighing, after they had been cleaned for drying, about six pounds; that for
purposes of trading, each family would put away about 10 sacks of fish" (p. 165). Further, the
affidavit noted that fishing rights "have a value to the Indians which cannot be measured in the
terms of dollars and cents of the white man; that the subsistence value to the Indians as a whole is
enormous..." (p. 167).
3.3.3 Hewes, 1947
In 1947, as part of his dissertation required for a Ph.D. in Anthropology, Gordon Hewes developed
an estimate reflective of Craig and Hacker's (1940) per capita salmon consumption estimate of 1
lb/d (365 lb/yr or 454 g/d) for aboriginal Columbia Basin Tribes (Table 1). The justification for
this estimate was based on the average human caloric requirements of 2,000 calories per day
(cal/d), the assumption that nearly 50% of the Indian diet was salmon, and that the caloric value of
o
salmon was approximately 1,000 calories per pound (p. 213-215). This assumed that salmon
provided nearly all dietary protein (primary source of energy) and that other food sources (such as
plants) contributed minimal caloric value to the diet.
Hewes presented various consumption rate estimates for Tribal groups in different regions of
Alaska and the Pacific Northwest compiled from various sources, stating that "while we have very
few quantitative hints for the regions south of Alaska, it is reasonable to suppose that per capita
consumption among intensive fishing peoples in parts of the Plateau.. .reached amounts equivalent
to at least the lower estimates..provided for Alaska and the Pacific Northwest by other authors
(p. 223), including the estimate of 365 lb/d for the Columbia Basin presented by Craig and Hacker
(1940). Acknowledging the guesswork involved, the author made every effort to develop
reasonable rates, based on available ethnographic data for the various Tribes in the Pacific
Northwest and Alaska, weighing salmon consumption by group or area accordingly. Tribe-specific
rates are further discussed in Hewes, 1973 (Section 3.4.1).
7 Based on the year of the publication (1942) and the age of Tommy Thompson at the time of the affidavit (79 years), the period
discussed here equates to the mid to late 1800s.
8 Calculation: 2000 cal/d * 0.5 * 1 lb/1000 cal = 1 lb/d
Nez Perce Tribe
Volume I: Heritage Fish Consumption Rates
December 2016 Page 16
-------
3.3.4 Griswold, 1954
In 1954, as part of his dissertation required for a Master of Arts, Gillett Griswold cited
Swindell's survey of Indian families in the Celilo region of the Columbia Basin, specifically
noting the input factors that, when applied together, would result in an aboriginal per capita
salmon consumption rate of 800 lb/yr (or 995 g/d). This rate was not presented in his publication
per se (and, therefore, not listed in Table 1), only the factors used to calculate the rate.
Referring to affidavits presented in Swindell's study, Griswold assumed that each family cured
and stored 30 sacks of salmon for their own use and an additional 10 sacks of salmon for trade
each year, with each sack weighing 100 pounds. This equates to 4,000 lb/yr per family harvested.
Assuming 5 individuals per family (as stated by Swindell), this equates to a per capita rate of 800
lb/yr. It should be noted that this rate considers all salmon that was harvested for both ingestion
as well as trade (i.e., not eaten). While this consumption rate was not presented by Griswold in
his dissertation, his input factors (4,000 lb/yr per family of 5 individuals) were used in the rate
calculation by another author (Walker, 1967, discussed below) to estimate a range of
consumption rates.
3.3.5 Walker, 1967
In 1967, Deward Walker conducted research on behalf of the Nez Perce Tribe and estimated an
average per capita salmon consumption rate of 583 lb/yr (or 725 g/d) for aboriginal Tribes of the
Columbia Plateau in general (Table 1). This estimate was based on the median value of two
previously reported estimates: 365 lb/yr (estimated by Craig and Hacker, 1940) and 800 lb/yr
(calculated from assumptions in Griswold, 1954). Walker also estimated a rate specifically for the
Nez Perce Tribe, which is discussed in Section 3.4.1 below.
Walker stated that "in light of the known annual dietary dependence on fish among aboriginal
societies of the Plateau, it seems safe to conclude that the range was between 365 and 800 lbs. per
capita with the average probably close to the median, i.e., 583 lbs." (p. 19). It should be noted that
the higher value of this range was calculated from Griswold, which, as discussed above, includes
salmon harvested for ingestion as well as other uses such as trade. Walker noted that a typical use
of fish in the Celilo region was for fuel. He also noted that determining a rate for particular groups
in the Plateau would "require substantial, additional research" (p. 19).
3.3.6 Boldt, 1974
In the 1974 decision, Senior District Judge George H. Boldt ruled in the case regarding Treaty
fishing rights in Washington State. The Judge stated that salmon "both fresh and cured, was a
staple in the food supply" of the Columbia River Tribal fishers, and that salmon was consumed
annually "in the neighborhood of 500 pounds per capita" (or 622 g/d) (p. 72) (Table 1). This case
decision reaffirmed the reserved right of Native Americans in Washington State to harvest fish
from their traditional use areas.
Nez Perce Tribe
Volume I: Heritage Fish Consumption Rates
December 2016 Page 17
-------
3.3.7 Hunn, 1981
In 1981, Eugene Hunn from the University of Washington, Department of Anthropology, re-
evaluated the assumptions associated with Hewes' (1947 and 1973) salmon consumption estimates
for Columbia Basin Tribes, suggesting that salmon likely did not provide as many calories as
originally estimated in the aboriginal diet. Although Hunn did not present FCRs in his publication
(and, therefore, no estimate is included in Table 1), he first introduced the concept of migration
calorie loss and waste loss factors, as discussed in Section 3.2 above, and as later applied to fish
consumption estimates by other authors (e.g., Scholz, et al., 1985, and Schalk, 1986).
While Hunn considered Hewes' estimates to be the most comprehensive to date, Hunn contended
that the caloric calculations were based on commercial fish, which are generally the fattest species,
and which are typically harvested prior to upstream migration. Hunn cited Idler and Clemens
(1959), which concluded that migrating salmon in the Fraser River "lose on average 75% of their
caloric potential during this migration" (p. 127). It may be assumed that fewer calories per pound
of salmon upstream results in people consuming more salmon to meet their daily caloric
requirements. However, Hunn stated that other foods, such as roots and bulbs, likely provided a
large caloric percentage of traditional diets. In addition to migration loss, Hunn determined that
only about 80% of the total weight of salmon was edible, therefore introducing the concept of the
"waste loss" factor, later applied by other authors to adjust consumption rates.
3.4 Nez Perce Tribe Heritage Rates
Below is a summary of the primary source information reviewed on heritage FCRs specific to the
Nez Perce Tribe. Relevant information is presented from each of the following publications (and
summarized in Table 2), including fish consumption estimates and associated assumptions.
• Walker, 1967
• Hewes, 1973
• Marshall, 1977
• Walker, 1985
• Schalk, 1986
• Hunn and Bruneau, 1989
3.4.1 Walker, 1967
In 1967, Deward Walker, in the same publication discussed above, estimated an average per capita
salmon consumption rate of 300 lb/yr (or 373 g/d) for the Nez Perce Tribe (Table 2). This estimate
was based on the following assumptions: a minimum of 300 fish harvested on a peak day, a
minimum of 10 peak days per year, a minimal average fish weight of 10 pounds per fish, and a
total of 50 historical fishing sites or villages (this last assumption was made from Spalding in 1936,
as noted in Walker, 1967). Multiplied together, this value was divided by the total estimated
population at the time of 5,000 people, yielding a total of 300 lb/yr.
9 Calculation: (300 fish/site x 10 peak days/year x 10 lb/fish x 50 fishing sites) 5,000 people
Nez Perce Tribe
Volume I: Heritage Fish Consumption Rates
December 2016 Page 18
-------
Walker's (1967) assumptions were identified as minimum estimates. His informants, for example,
estimated 10 to 20 peak days of fish harvest, and Hewes (1947) reported a total population of 4,000
(which would increase the per capita consumption estimate).
3.4.2 Hewes, 1973
In 1973, continuing on his previous dissertation work, Gordon Hewes presented updated aboriginal
per capita salmon consumption rates for specific Tribes in Alaska, British Columbia, and the
Pacific Northwest, including a rate of 300 lb/yr (or 373 g/d) for the Nez Perce Tribe (Table 2). This
rate is based on caloric content and daily requirements, population estimates, and ethnographic
accounts of the importance of salmon; it is also based on human dietary demands only, not
including other non-ingestion uses.
Hewes initially published a general rate for salmon consumption by Columbia Basin Tribes based
on assumptions about dietary caloric requirements and the contribution of salmon to aboriginal
diets (see discussion of Hewes, 1947, in Section 3.3.3 above). In this report, Hewes again presents
an average per capita estimate of 365 lb/yr (or 454 g/d) for the Columbia Basin Tribes as well as
rates for individual Tribes. The Tribe-specific rates account for variability in salmon dependence
between regions and population groups, and they reflect population numbers available at the time
for each Tribe.
3.4.3 Marshall, 1977
In 1977, working on his dissertation for the Washington State University Department of
Anthropology, Alan Marshall estimated an aboriginal per capita salmon consumption rate of 560
lb/yr (or 697 g/d) for the Nez Perce, based on total fish harvest (Table 2).
Marshall (1977) estimated the Nez Perce rate based on the following assumptions, the majority
which originated from Walker's "informants" (1967): a minimum of 300 fish harvested on a peak
day, a minimum of 10 peak days per year, a minimal average fish weight of 10 pounds per fish,
and a total of 94 historical fishing sites or villages. This last assumption (fishing sites) was
increased from Walker's estimate of 50 (according to information from Schwede, 1966, as cited in
Marshall, 1977).10 Multiplied together, this value was divided by the total estimated population at
the time of 5,000 people, yielding a total of 564 lb/yr, which the author presents as "roughly 560
pounds" that "reasonably approximates the figure" from Walker (1967) for Columbia Basin
Tribes.
3.4.4 Walker, 1985
In 1985, Deward Walker conducted ethnographic research that included information about the Nez
Perce Tribe; however, the report was never published and remains unavailable due to the
sensitivity of the information it contained. The data presented here is based upon citations in
Scholz, et al. (1985), in which the author included estimates and quotes and, therefore, apparently
had access to Walker's (1985) report. Walker calculated an average per capita total (anadromous
10 Calculation: (300 fish/site x 10 peak days/year x 10 lb/fish x 94 fishing sites) 5,000 people
Nez Perce Tribe
Volume I: Heritage Fish Consumption Rates
December 2016 Page 19
-------
and resident) FCR of 1,000 lb/yr (or 1,244 g/d) for the Nez Perce Tribe (Table 2). Note that this
rate intended to include both salmon and resident fish consumption combined in the estimate.
According to Scholz (1985), Hewes "checked Walker's new figures for populations and per capita
consumption and agrees with Walker's revisions" (Scholz, 1985, p. 73). Scholz also stated that
Walker's (1985) estimates were significantly different from those of Schalk (1986), discussed
below, primarily because Walker assumed higher Tribal population totals (and also includes
resident fish with salmon consumption). Without the original document, however, it is unclear if
Walker's estimates represent fish ingestion only or include fish used for other purposes, such as
trade and fuel.
3.4.5 Schalk, 1986
In 1986, Randall Schalk calculated salmon consumption estimates for specific Tribes based on
Hewes' (1947 and 1973) original estimates, including a rate of 647 lb/yr (or 804 g/d) for the Nez
Perce Tribe (Table 2). This rate includes migration and waste loss factors applied to Hewes' Tribe-
specific values. Schalk contended that many of Hewes' original estimates were biased low because
they were based on:
• A caloric content of fish representing salmon as they enter freshwater in prime condition
(i.e., having more calories than upstream salmon). Schalk stated that "since salmonids lose
an average of 75% of their caloric content during migration (Idler and Clemens 1959),
some adjustment should have been made for distance traveled upstream" (i.e., applying a
migration loss factor).
• The assumption that salmon were eaten in their entirety. Schalk states that assuming the
entire fish was consumed was "unrealistic" and cited Hunn (1981) to state that only "about
80% of the weight of a salmon is edible" (p. 17).
Schalk, therefore, adjusted (increased) Hewes' consumption rates by applying a migration loss
factor (variable by Tribe depending on how far upstream they harvested salmon) of 58% (0.58) for
the Nez Perce Tribe. Schalk also applied a waste loss factor of 80% (0.80), citing Hunn (1981),
therefore, including inedible fish parts in the fish consumption estimate.
3.4.6 Hunn and Bruneau, 1989
In 1989, Eugene Hunn and C. Bruneau of Pacific Northwest Laboratory (on behalf of the U.S.
Department of Energy at the Hanford Site) estimated an anadromous fish (including salmon,
steelhead, and lamprey) consumption rate of 320 lb/yr (or 398 g/d) for the Nez Perce Tribe
(Table 2).
Based on the "educated guesses" of previous authors, including Craig and Hacker (1940), Hewes
(1947, 1973), and Walker (1967), Hunn and Bruneau (1989) estimate 400 pounds per person per
year as a "reasonable traditional gross harvest rate" for the Nez Perce. Assuming that the actual
consumption was only 80% of the total harvest, the authors adjusted (reduced) this value (i.e.,
multiplied by 0.80) to account for the edible fraction only.
Nez Perce Tribe
Volume I: Heritage Fish Consumption Rates
December 2016 Page 20
-------
4.0 RATE EVALUATION AND DISCUSSION
This section further evaluates and discusses the information presented above, including the
uncertainty associated with the rate adjustment factors and other assumptions influencing rate
calculations.
4.1 Factors Influencing Consumption Rates
The migration calorie loss factor and waste loss factor are considered here, particularly regarding
the uncertainty associated with applying these adjustment factors to heritage rates. Other factors
that influence the calculation of heritage rates and that may also increase uncertainty of the
estimates include population size estimated at the time, number of fishing sites, and reliability of
ethnographic data in general.
4.1.1 Migration Calorie Loss Factor
For a number of reasons, the application of the migration calorie loss factor as described above
introduces a high degree of uncertainty into the revised estimates of tribal fish consumption. The
study that forms the basis of this adjustment (Idler and Clemens, 1959) is based on one year's run
of one species of salmon (sockeye) in one watershed (the Fraser River). The conclusions of this
study are then broadly applied to all salmon species within a different watershed (the Columbia
River), even though it is estimated that sockeye accounted for only 7% of the Upper Columbia
salmon harvest (Beiningen, 1976 as cited in Scholz, et al., 1986). The degree to which different
salmon species lose calories at different rates or in different proportions during spawning
migration, and the degree to which the Columbia River and Fraser River watersheds differ (in
length, elevation change, etc.) all affect the degree of uncertainty associated with the calculation
and application of a migration calorie loss factor.
The migration calorie loss factor is based on a gross percentage of calories lost by a sockeye
salmon during spawning migration in the Fraser River (i.e., ending calories compared to beginning
calories). However, the factor is applied in revising consumption rates as though it represents the
amount of calories lost per pound consumed, which is not the same; salmon not only lose calories
during migration, they also lose weight. Based on measurements collected by Idler and Clemens
(1959), the average overall weight loss during spawning migration was 25%, and the loss in caloric
density (calories per gram) was therefore about 65%, as opposed to 75%. Table 3 provides the total
calories, total weight (in grams), and caloric density (in calories per gram) of sockeye salmon
measured at various stages in the Fraser River (from Idler and Clemens, 1959).
Further, the overall decrease in caloric potential was based on measurements of sockeye salmon
that have spawned and died in headwater streams. Michael Kew (1986) describes the results of the
Idler and Clemens study as follows:
"As a general rule, the further from the sea a salmon is, the less fat and protein it
carries. The loss is considerable. Total caloric value of a sockeye, measured at the
river mouth, will be reduced to nearly one-half when it reaches the Upper Stuart
spawning grounds, one thousand kilometers from the sea. After the enriched
gonads have been expended in spawning and the fish die on these upper streams,
they will have lost over 90 percent of their fat and one-half to two-thirds of their
protein (Idler and Clemens, 1959; reviewed in Foerster, 1968: 74-6). "
Nez Perce Tribe
Volume I: Heritage Fish Consumption Rates
December 2016 Page 21
-------
As Kew notes, there is a significant difference in caloric potential between the time a salmon
reaches its spawning grounds and the time it has spawned and died. Based on measurements
collected by Idler and Clemens (1959), the average sockeye loses almost 15% of its caloric density
(calories per pound) between the time it reaches its spawning grounds and the time it has spawned
and died. At the time a sockeye salmon reaches its spawning grounds in the upper Fraser River
watershed, it has lost about 50% of its caloric density (Table 3).
Still further, the derivation of the migration calorie loss factor relies on the assumption that the
salmon harvest location is at "the approximate middle of each group's territory" (Hunn, 1981). To
the extent that a majority of salmon harvest occurs either downstream or upstream of this point, the
migration calorie loss factor would either overestimate or underestimate, respectively, the effect on
the consumption rate.
Mullan, et al. (1992) note that caloric losses in salmon are generally related to mileage of
migration, but not directly. "Idler and Clemens (1959) show much higher energy expenditures by
sockeye in some river reaches than others, and higher rates for females than males. In other words,
caloric content is not linear in relation to distance." Further, Mullan notes that in migration and
maturation the fish tend to mobilize fat reserves and resorb organs (e.g., gastro-intestinal tract), and
"[t]hus they lose weight, but not necessarily caloric content, between cessation of ocean feeding
and nominal freshwater capture."
While the idea of adjusting calorie-based consumption estimates to account for migration calorie
loss does not seem unreasonable, based on the uncertainty described above, it most likely tends to
overestimate salmon consumption relative to Hewes' original estimates (because it likely
overestimates calorie loss per pound). Since sockeye salmon lose approximately 50% of their
caloric density upon reaching their spawning grounds, a maximum migration calorie loss factor of
50%), as opposed to 75%, may be more consistent with the supporting research (although the
existing research is limited to a single species of salmon). Hewes's diet and calorie-based
consumption estimate for the Columbia Plateau Tribes is identical to that proposed by Craig and
Hacker (1940), which is not based on caloric intake but on observation and review of the ethno-
historical literature (although it is "admittedly liable to a wide margin of error").
4.1.2 Waste Loss Factor
Incorporating a waste loss factor to revise Hewes's fish consumption estimates has the effect of
increasing the consumption rate (relative to Hewes's estimate) by 25%. If the interest is in
understanding how much individuals consumed (ingested), as opposed to "used," then the use of a
waste loss factor is not appropriate. Essentially, this factor adjusts a consumption rate, increasing it
by 25%), to account for the portion of fish NOT consumed. Consumption estimates that have been
revised to account for a waste loss factor (as in Scholz et al., 1985, and Schalk, 1986) would tend
to overestimate consumption (ingestion) by 25%, relative to the "unrevised" rates.
Some estimates of consumption by Tribal groups are based on an estimate of total harvest and total
population. For example, some authors estimate a total harvest (in pounds) based on the number of
fishing sites, number of fishing days, efficiency of fishing techniques, average weight of fish, etc.,
and simply divide the total estimated harvest by the total estimated tribal population to arrive at an
annual per capita consumption rate. However, this type of estimate does not account for the fact
Nez Perce Tribe
Volume I: Heritage Fish Consumption Rates
December 2016 Page 22
-------
that only a portion of each fish may be edible (i.e., 80%), and may tend to overestimate the amount
that people are actually consuming.
Mullan, et al. (1992) suggested that, because many Tribal groups prepared and consumed most
parts of the salmon, including organs, eyes, eggs, etc., the inedible waste was much less than 20%,
arguing that "waste factor of a salmon amounted to bones only, under 10% of body weight."
4.1.3 Other Assumptions used to Develop Consumption Rates
In addition to the rate adjustment factors discussed above, other assumptions that various authors
have made in developing consumption rates introduce varying degrees of uncertainty to the
estimates, including those discussed below.
Ingestion, Harvest, and Consumption
As discussed in Section 3.1, the effort here is to summarize estimates of fish ingestion which may
be relevant to the development of Tribal water quality standards. The degree to which estimates of
Tribal fish consumption in the various studies include uses in addition to ingestion may affect their
applicability to Tribal regulatory or policy development.
Percent of Diet Supplied by Fish
The calorie-based consumption estimates developed by Hewes, which form the basis for a
number of subsequent estimates, are based on the assumption that salmon account for about 50%
of the average Columbia Basin aboriginal diet. Many authors have made similar estimates, while
others have assumed either higher or lower dietary estimates. While 50% of the diet (i.e., 50% of
total calories) is among the most common estimates, the degree to which a specific Tribe has a
higher or lower percentage of diet supplied by fish can affect the accuracy of the calculated
consumption rate.
Salmon and Resident Fish Consumption
Because of the importance of salmon to the Columbia Basin Tribes, and because many studies
have attempted to evaluate the impact of the hydroelectric system on anadromous fisheries, a
majority of the studies evaluated focused exclusively or primarily on the harvest and consumption
of salmon. The degree to which individual Tribal groups relied on resident fish, either to
supplement or to substitute for salmon consumption, will affect the accuracy of consumption
estimates included in these studies relative to total fish consumption.
Tribal Population Estimates
Some authors have estimated total fish consumption for various Tribal groups by estimating an
overall harvest rate and dividing that rate by the total Tribal population to develop an average per
capita estimate. Therefore, the accuracy of population estimates may directly affect the accuracy of
consumption estimates developed using this approach.
Number of Fishing Sites, Fishing Methods, and Fishing Efficiency
Some authors have developed consumption estimates based on assumptions about the type and
effectiveness of Tribal fishing methods and the number of harvest locations utilized by individual
Tribes or Tribal groups. The degree to which these assumptions are accurate will directly affect the
accuracy of consumption estimates using this approach.
Nez Perce Tribe
Volume I: Heritage Fish Consumption Rates
December 2016 Page 23
-------
4.2 Heritage Fish Consumption Rates (FCRs)
The heritage rates estimated for the Columbia Basin Tribes and, specifically, the Nez Perce Tribe,
introduced in Sections 3.3 and 3.4 above, are evaluated in more detail below, including discussion
of the assumptions and uncertainty associated with the estimates.
4.2.1 Columbia Basin-Wide Heritage Rates
Craig and Hacker (1940) presented the first estimate of per capita salmon consumption for
aboriginal Tribes of the Columbia Basin of 365 lb/yr (or 454 g/d), which was based on historical
ethnographic observations, although acknowledged by the authors as likely having a wide margin
of error. Hewes (1947) validated this rate with additional assumptions related to average dietary
caloric requirements, the contribution of salmon to the aboriginal diet, and a caloric value for
salmon. These assumptions (a 2,000 calorie diet, 50% of the diet was salmon, and salmon
contained 1,000 calories per pound), while generalized, provided additional justification for this
rate. Hunn (1981) later re-evaluated Hewes' assumptions by suggesting that migration calorie loss
and inedible waste loss factors should be considered. While variability exists in how many calories
each salmon contained and how much of each salmon was eaten, the method for developing and
applying such "adjustment factors" (discussed in Section 4.1 above), as done to aboriginal rates by
other authors (Scholz, et al., 1985, and Schalk, 1986), may have added a level of uncertainty to
those estimates.
Shortly after Craig and Hacker (1940) published the first aboriginal salmon consumption estimate,
Swindell (1942) published a very similar estimate of per capita salmon consumption of 322 lb/yr
(or 401 g/d) for the Tribes of the Celilo Falls region. This value was based on interviews with
Indian families, including affidavits of extensive salmon consumption and use, and total harvest
(according to sacks of fish and average weights per fish). Griswold (1954) later cited Swindell's
work, referring to these affidavits, to calculate a total annual harvest of 4,000 pounds per family.
Although Griswold did not calculate a per capita consumption rate in his publication, Walker
(1967), by assuming 5 individuals per family, calculated a per capita rate of 800 lb/yr (or 995 g/d)
for an upper range of fish consumption. Based on per capita FCRs ranging from 365 lb/yr
(presented in Craig and Hacker, 1940, and Hewes, 1947) to 800 lb/yr (calculated from Griswold,
1954), Walker (1967) calculated an average (median) per capita salmon consumption rate of 583
lb/yr (or 725 g/d). A few years later, Boldt (1974) stated that Columbia River Tribes consumed (as
food supply) a comparable rate of about 500 lb/yr (or 622 g/d) of salmon.
It is important to remember that the rate calculated from Griswold's (1954) information reflects
salmon that was harvested for both consumption as well as trade (i.e., salmon not ingested). If all
other assumptions hold true, based on Swindell's (1942) information (3,000 lb/yr harvested per
family for consumption, 5 individuals per family11), a more accurate per capita upper range for fish
consumption as defined for this report would be 600 lb/yr (or 746 g/d). If this alternate value is
used from Griswold (1954), calculating an average rate similar to Walker's approach would result
in an average rate of 483 lb/yr (or 600 g/d). See Table 1.
11 If the10 sacks of salmon that were harvested for trade are removed from the equation, the 30 sacks offish consumed at 100 pounds
= 3,000 pounds (per family).
Nez Perce Tribe
Volume I: Heritage Fish Consumption Rates
December 2016 Page 24
-------
4.2.2 Nez Perce Tribe Heritage Rates
In addition to estimating an average consumption rate for aboriginal Tribes of the Columbia
Basin in general, Walker (1967) also estimated a rate specific to the Nez Perce Tribe. He
estimated an average per capita salmon consumption rate of 300 lb/yr (373 g/d) based on
estimates of fish harvest on peak days, number of fishing sites, average fish weight, and total
population. Hewes (1973), continuing his earlier dissertation research from 1947, published his
estimates for various Tribes, including the Nez Perce, based on fish caloric content and daily
requirements, population estimates, and ethnographic accounts of the importance of salmon
among different Tribes. He estimated an average per capita salmon consumption rate identical to
Walker (1967) of 300 lb/yr (or 373 g/d) for the Nez Perce Tribe. Marshall (1977) believed
Hewes' rate to be a minimum estimate; he calculated an average per capita salmon consumption
rate of 560 lb/yr (or 697 g/d) based on the same assumptions as Walker (1967), but assuming
nearly twice the number of fishing sites.
Schalk (1986) later applied migration and waste loss factors to Hewes' estimate (dividing Hewes'
rate of 300 lb/yr by 0.58 and 0.80), yielding a higher salmon consumption rate of 647 lb/yr (or 804
g/d) for the Nez Perce Tribe. Taking a slightly different approach, Hunn and Bruneau (1989)
removed the inedible fraction from a total harvest estimate (multiplying a harvest rate of 400 lb/yr
by the 0.80 waste loss factor), yielding a lower anadromous FCR (including consumption of
salmon, steelhead, and lamprey) of 320 lb/yr (or 398 g/d).
In 1985, Walker expanded upon his previous work from 1967 and calculated Tribe-specific per
capita total FCRs for individual tribes, including 1,000 lb/yr (or 1,244 g/d) for the Nez Perce Tribe.
Although this study remains unpublished, the estimates were presented (with supporting
information) by Scholz (1985). Walker's estimates appear to be the only rates (of those presented
here) that reflect use of both anadromous and resident fish; however, since the report is
unavailable, it cannot be verified if these estimates account for only fish ingested or include fish
used for other purposes (such as trade). See Table 2.
Nez Perce Tribe
Volume I: Heritage Fish Consumption Rates
December 2016 Page 25
-------
5.0 REFERENCES FOR VOLUME I
Boldt, G.H. 1974. The Boldt Decision. United States District Court for the Western District of
Washington, Tacoma Division. Civ. No. 9213. Boldt, Senior District Judge. February 12.
Center for Columbia River History (CCRH). 2013. Nez Perce, Nee-Me-Poo. Available:
http://www.ccrh.org/comm/river/profile/nezperce.php. Accessed September 19.
Craig, J. A. and R.L. Hacker. 1940. The History and Development of the Fisheries of the Columbia
River. United States Bureau of Fisheries. Bulletin No. 32. Approved for publication August
27, 1938.
Everett, S., C. Beasley, R. Johnson, C. Davenport, andR. Larson. 2006. "Nez Perce Tribal
Hatchery Project; Coho Salmon Master Plan, Clearwater River Basin", 2003-2004 Annual
Report, Project No. 198335000, 134 electronic pages, (BPA Report DOE/BP-00004035-1).
FishPro Division of HDR Engineering.
Foerster, R.E., 1968. The Sockeye Salmon, Oncorhynchus nerka. Fisheries Research Board of
Canada, Biological Station, Nanaimo, B.C.
Griswold, G. 1954. Aboriginal Patterns of Trade Between the Columbia Basin and the Northern
Plains. Presented for Master of Arts, Montana State University.
Hewes, G.W. 1947. Aboriginal Use of Fishery Resources in Northwestern North America.
Doctoral dissertation, University of California, Berkeley. February.
Hewes, G.W. 1973. Indian Fisheries Productivity in Precontact Times in the Pacific Salmon Area.
Northwest Anthropological Research Notes 7:133-155.
Hunn, E.S. 1981. On the Relative Contribution of Men and Women to Subsistence Among Hunter-
Gatherers of the Columbia Plateau: A Comparison with Ethnographic Atlas Summaries.
University of Washington, Department of Anthropology. Seattle, WA.
Hunn, E.S. and C.L. Bruneau. 1989. Estimations of Traditional Native American Diets in the
Columbia Plateau. Pacific Northwest Laboratory. Work supported by the U.S. Department
of Energy. Richland, Washington. August.
Idaho Department of Fish and Game (IDFG). 2013. Chinook Salmon (Snake River Spring/Summer
Run) Conservation Status. Available:
http://fishandgame.idaho.gov/ifwis/cwcs/pdf/Chinook%20Salmon%20%28Snake%20Rive
r%20spring_summer%20run%29.pdf. Accessed September 13.
Idler, D R. andW.A. Clemens. 1959. The Energy Expenditures of Fraser River Sockeye Salmon
During the Spawning Migration to Chilko and Stuart Lakes. Intern. Pacific Salmon Fish.
Comm., Prog. Rept. No. 6, New Westminster, B.C.
Nez Perce Tribe
Volume I: Heritage Fish Consumption Rates
December 2016 Page 26
-------
Kew, Michael. 1986. Salmon Availability, Technology, and Cultural Adaptation in the Fraser
River Watershed. In: A Complex Culture of the British Columbia Plateau: Traditional
Stl'Atl'Imx Resource Use. edited by Brian Hayden. University of British Columbia Press
(August 1992).
Landeen, D. and A. Pinkham. 1999. Salmon and His People: Fish and Fishing in Nez Perce
Culture.
Marshall, A.G. 1977. Nez Perce Social Groups: An Ecological Interpretation. Dissertation for
Doctor of Philosophy. Washington State University, Department of Anthropology.
Copyright by Alan Gould Marshall and the Nez Perce Tribal Executive Committee.
Matthews, G.M. andR.S. Waples. 1991. Status Review for Snake River Spring and Summer
Chinook Salmon. NOOA Technical Memorandum NMFS F/NWC-200. National Oceanic
and Atmospheric Administration, National Marine Fisheries Service. June.
Meyer Resources, Inc. 1999. Tribal Circumstances and Impacts from the Lower Snake River
Project on the Nez Perce, Yakama, Umatilla, Warm Springs, and Shoshone Bannock
Tribes. Prepared for CRITFC, funded by US ACE.
Mullan, J.W., K.R. Williams, G. Rhodus, T.W. Hillman, and J.D. Mclntyre. 1992. Production and
Habitat of Salmonids in Mid-Columbia River Tributary Streams. U.S. Fish and Wildlife
Service, Monograph I.
Nez Perce Tribe (NPT). 2005. White Sturgeon Management Plan in the Snake River Between
Lower Granite and Hell's Canyon Dams. Department of Fisheries Resource Management.
Lapwai, ID. September.
Schalk, R.F. 1986. Estimating salmon andsteelhead usage in the Columbia basin before 1850: the
anthropological perspective. Northwest Environmental Journal. Vol. 2 (2), p. 1-26.
Scholz, A., K. O'Laughlin, D. Geist, D. Peone, J. Uehara, L. Fields, T. Kleist, 1. Zozaya, T. Peone,
and K. Teesatuskie. 1985. Compilation of Information on Salmon and Steelhead Total Run
Size, Catch and Hydropower Related Losses in the Upper Columbia River Basin, above
Grand Coulee Dam. Fisheries Technical Report No. 2. Upper Columbia United Tribes
Fisheries Center, Eastern Washington University, Department of Biology. Cheney,
Washington 99004. December.
Swindell, E.G., Jr. 1942. Report on Source, Nature, and Extent of the Fishing, Hunting, and
Miscellaneous Related Rights of Certain Indian Tribes in Washington and Oregon
Together with Affidavits Showing Location of a Number of Usual and Accustomed Fishing
Grounds and Stations. United States Department of the Interior, Office of Indian Affairs,
Division of Forestry and Grazing. Los Angeles, California. July.
Walker, D.E., Jr. 1967. "Mutual cross-utilization of economic resources in the plateau: an example
from aboriginal Nez Perce fishing practices." In: Washington State University, Laboratory
of Anthropology, Report of Investigations. No. 41, p. 1-70.
Nez Perce Tribe
Volume I: Heritage Fish Consumption Rates
December 2016 Page 27
-------
6.0 TABLES
Notes/Footnotes for Tables:
1 Includes a migration calorie loss factor (based on Hunn, 1981, citing Idler and Clemens, 1959) to
adjust estimates based on caloric intake.
Waste loss may be accounted for either in direct observation (i.e. the author is citing consumption
of fish that had been prepared for consumption, as was done by Craig and Hacker and Swindell) or
by adjusting the amount of fish harvested by a waste loss factor loss factor (0.8, based on Hunn,
1981) to translate from amount consumed to amount harvested. For consumption rates derived
using caloric analysis, waste loss is inherently accounted for, as calories consumed are converted
into edible fish mass consumed.
Estimates based on ethno-graphic observation sometimes appear to be based on amounts actually
consumed (e.g. Craig and Hacker; Swindell) and sometimes based on amounts harvested (e.g.
Walker; Marshall). Those based on the amount harvested would include the inedible (waste loss)
portion, and would likely overestimate consumption. They may also include harvest for other uses,
although that is not specifically stated in most studies.
Different studies address "waste loss" differently. Most that use the "waste loss factor", like
Schalk and Scholz, use the factor to translate from a consumption rate to a harvest rate, so they
tend to inflate the consumption rate (by dividing by 0.8). Other studies (e.g. Hunn and Bruneau,
1989) use the same factor to translate from a harvest rate to a consumption rate (by multiplying by
0.8). So both studies "account" for waste loss, but they do so to opposite effect.
Here is an excerpt from Hunn and Bruneau:
"Based on these educated guesses, I use 500 pounds per person per year as a
reasonable traditional gross harvest rate for "River Yakima" and 400 pounds for
the Nez Perce (cf Walker 1973:56) and the Colville. Actual consumption is
estimated at 80% for the edible fraction (thus 400 and 320 pounds respectively). "
Nez Perce Tribe
Volume I: Heritage Fish Consumption Rates
December 2016 Page 28
-------
Table 1: Average Heritage Fish Consumption Rates for the Columbia Basin Tribes
Reference
Methodology
Species
Evaluated
Rate in
g/day
Rate Derivation
Includes
(Note: +/-/U indicates whether the way in
which a particular factor was addressed
causes an increase, decrease, or unknown
impact on the FCR)
Uses Besides
Consumption
Migratory
Caloric Loss
Factor1
Accounting
for inedible
portion2
Craig &
Hacker 1940
Ethnographic
Observation
Salmon,
sturgeon,
trout
454
Not presented
No (+)
No (-)
Yes (U)
Swindell 1942
Ethnographic
Observation
Salmon
401
1611 lb salmon/year + 5 people/family x 454 g salmon/lb salmon + 365 days/year
No (+)
No (-)
Yes (U)
Hewes1947
Caloric Analysis
Salmon
454
2000 calories/day x 50% of diet as salmon x 1000 calories/lb salmon x lb salmon/454 g salmon
Yes (-)
No (-)
Yes (U)
Griswold 1954
Ethnographic
Observation
Salmon
746
30 sacks salmon/year/family x 10 lb salmon/sack x family/5 people x 454 g salmon/lb salmon x year/365 days
Griswald cited 40 sacks of salmon per family were obtained with 30 retained for family use and 10 used for other
purposes.
No (+)
No (-)
No (U)
Walker 1967
Evaluation of Craig &
Hacker 1940 and
Griswold 1954
Salmon
725
Average of 454 g/day (from Craig and Hacker, 1940) and 995 g/day (from Griswold 1954). The Griswold value
was based on families obtaining 40 bags of salmon, 30 for consumption and 10 for trade.
995 g/day = 40 sacks salmon/year/family x 100 lb salmon/sack x family/5 people x 454 g salmon/lb salmon x
year/365 days
Yes (+)
No (-)
No (U)
Boldt 1974
Undocumented, (United
States v. Washington,
384 F. Supp. 312
Salmon
622
500 lb salmon/person/year x 454 g salmon/lb salmon x year/365 days
Unknown (U)
No (-)
Unknown (U)
Nez Perce Tribe
Volume I: Heritage Fish Consumption Rates
December 2016 Page 29
-------
Table 2. Average Heritage Fish Consumption Rates for the Nez Perce Tribe
Reference
Methodology
Species
Evaluated
Rate in
g/day
Rate Derivation
Includes
(Note: +/-/U indicates whether the way in
which a particular factor was addressed
causes an increase, decrease, or unknown
impact on the FCR)
Uses Besides
Consumption
Migratory
Caloric Loss
Factor1
Accounting
for inedible
portion2
Walker 1967
Ethnographic
observation citing
Spalding 1936
Salmon
373a
466b
300 fish/peak day/fishing site x 10 peak days/year x 10 lb tissue/fish x 50 fishing sites + 5000 total population
(from Spalding 1936)
a: assumes population of 5000
b: assumes population of 4000 (Hewes 1947)
Unknown (U)
No (-)
Unknown (U)
Hewes1973
Caloric
Analysis/Ethnographic
Observation
Salmon
373
No (+)
No (-)
No (U)
Marshall 1977
Ethnographic
Observation citing
Walker
Salmon
701
300 fish/peak day/fishing site x 10 peak days/year x 10 lb salmon/fish x 94 fishing sites x 454 g salmon/lb
salmon + 5000 total population
Note: fishing sites increased from 50 to 94 based on Schwede 1966
Unknown (U)
No (-)
No (U)
Walker 1985
Ethnographic
Observation,
unpublished by cited by
Scholz 1985
Salmon &
Resident
1,244
Methodology not presented
Unknown (U)
Unknown (U)
Unknown (U)
Schalk 1986
Ethnographic
Observation citing
Hewes 1947 and 1973
Salmon
804
300 lb salmon/year/person x 454 g salmon/lb salmon x year/365 days + 0.58 caloric loss factor + 0.8 edible
fraction.
Modified consumption rates of Hewes 1947 and 1973. Hewes (1973) assumed a consumption rate of 300
lb/year. Assumed that caloric content of fish was reduced during migration. For the Nez Perce, there was a
58% reduction in caloric value. Further, not all parts of the salmon are edible. Schalk assumed 80% of the fish
was consumed.
Unknown (U)
Yes (+)
Yes (+)
Hunn and
Bruneau 1989
Ethnographic
Observation, derived
from: Craig and Hacker
1950; Hewes 1947 &
1973; Walker 1967
Salmon,
Steelhead,
Lamprey
398
400 lb salmon/year/person x 454 g salmon/pound of salmon x year/365 days x 0.8 edible fraction
Based on review of references cited in the methodology column, Hunn and Bruneau estimated the annual
salmon harvest per person at 400 lb/year
Unknown (U)
No (-)
Yes (-)
Nez Perce Tribe
Volume I: Heritage Fish Consumption Rates
December 2016 Page 30
-------
Table 3. Spawning Migration and Calorie Loss (Fraser River)
Fraser River Location
Total Calories1
(kCal)
Total Weight1
(grams)
Caloric Density
(calories/ gram)
At River Mouth
5,173
2,585
2.00
At Spawning Grounds
2,248
2,363
0.95
After Spawning and Death
1,334
1,917
0.70
Percent Loss at Spawning Grounds
57%
9%
52%
Percent Loss After Spawning and
Death
74%
26%
65%
Notes for table 3:
All values are based on Idler and Clemens, 1959.
'Based on average of male and female values.
Nez Perce Tribe
Volume I: Heritage Fish Consumption Rates
December 2016 Page 31
-------
Figure 1: Key geographic features referred to in this report.
Legend
*
Dam
«
Waterfall
Townsite
ID AH 0
CtescJWto.
NiMdiu)
OREGON
Nez Perce Reservation
SfiiMxj H- i
zjtMk tut* - Harpster
Wallowa Hells Canyon1 '
* *
Main
B t.
Dworshafc -
Nez Perce Tribe
Volume I: Heritage Fish Consumption Rates
December 2016; Page 32
-------
Volume II:
Current Fish
Consumption Survey-
Nez Perce Tribe
-------
Table of Contents
1.0 Preface to Volume II 1
2.0 Acronyms and Abbreviations 2
3.0 Executive Summary 3
3.1 Introduction and Purpose 3
3.2 Survey Methods 3
3.3 Results 5
3.4 Discussion 8
3.5 Conclusion 12
4.0 Introduction 13
4.1 Purpose of the Overall Fish Consumption Survey Effort 13
4.2 Putting the Survey of Current Fish Consumption in Context 13
4.3 A Brief Description of the Nez Perce Tribe 13
4.4 Populations 13
4.5 Guide to Report Sections 14
5.0 Methods 15
5.1 Overview 15
5.2 Sample Selection 16
5.3 Inclusion/Exclusion Criteria 19
5.4 Geographic Sample Selection Criteria 20
5.5 Stratification and Drawing the Sample 22
5.6 Questionnaire Development 23
5.7 Portion Models, Photos, Portion-to-Mass Conversions 25
5.8 CAPI (Computer-Assisted Personal Interviewing) 26
5.9 Interviewer Recruitment and Training, Pilot Tests 27
5.10 Calculation of FFQ Consumption Rates 29
5.11 Species Groups 31
5.12 Demographic Groups 33
5.13 Response Rates 33
5.14 Design Changes 34
5.15 Reinterviews 36
5.16 Reviews and Approvals 36
5.17 Internal Reviews 38
-------
5.17.1 Review by the Tribe and Other Organizations 38
5.17.2 Review of Statistical Computing 38
5.18 Overview of Statistical Analysis 38
5.19 Sampling Probabilities 39
5.20 Non-Response Adjustments to Weights 39
5.21 Consumer/Non-Consumer Determination (Overall and per Species) 41
5.22 Mean, Variance and Percentile Methods for non-NCI analyses 41
5.23 NCI Method 42
5.23.1 Overview 42
5.23.2 Covariate Selection and Assessment of Seasonality 47
5.23.3 Quality Checking of the Model 51
5.23.4 Sensitivity Analyses 52
5.24 Effect of Changes in Study Design on FFQ Rates 53
5.25 Confidence Intervals 54
5.26 Replicate Weight Calculations 55
5.27 Confidence Interval Calculations for a Specific Statistic 55
5.28 Handling Missing Values 56
5.29 Limited Percentiles for Small Sample Sizes 58
5.30 Large Consumption Values 58
5.31 Software and Software Modules 59
6.0 Results 60
6.1 Response Rates 60
6.2 Factors Affecting Response Rates 61
6.3 Consumers, Non-Consumers and Frequency of Consumption 64
6.4 Demographic Characteristics 66
6.5 FFQ Rates for Species and Groups of Species 67
6.6 FFQ Consumption Rates by Demographic Groups 71
6.7 Effect of Changes in Study Design on FFQ Rates 73
6.8 Consumption Rates from the NCI Method 74
6.9 Quality Checking—NCI Method 80
6.10 Sensitivity Analyses—NCI Model 81
6.11 Comparison of FFQ Rates to 24-Hour and NCI-Method Rates 83
6.12 Consumption at Special Events and Gatherings 96
-------
6.13 Fish Parts Eaten, Preparation Methods and Sources 96
6.14 Fishing Activities 99
6.15 Changes in Consumption and Reasons 100
6.16 Reinterviews 102
6.17 Reliability and Cooperation of Respondents—Interviewer's Assessment 103
7.0 Discussion 104
7.1 Overview 104
7.2 Comparison of FFQ Rates to NCI-Method Rates 106
7.3 Comparison of This Survey's Rates to Other Surveys' Rates Ill
7.4 Strengths and Limitations 114
7.5 Characterizing Uncertainty 119
7.6 Next Steps, Lessons Learned 121
7.7 Conclusions 121
8.0 References for Volume II 123
List of Appendices 126
-------
List of Tables
Table SI. Mean, median and selected percentiles of FFQ and NCI method FCRs (g/day, raw weight,
edible portion); consumers only. Estimates are weighted 6
Table S2. Mean, median and selected percentiles of FFQ and NCI method FCRs (g/day, raw weight,
edible portion) for fishers and non-fishers; consumers only. Estimates are weighted 7
Table S3. Total FCRs (g/day, raw weight, edible portion, all species combined) of adults in Pacific
Northwest Tribes (with consumption rates available) and the U.S. general population.
Consumers only 7
Table 1. ZIP codes included for sampling members of the Nez Perce Tribe 21
Table 2. Species groups 32
Table 3. Survey response rate 60
Table 4. Response rates by sampling strata. Estimates are unweighted 62
Table 5. Response rates by demographic factors. Estimates are unweighted 63
Table 6. Frequency of fish consumption based on 472 responders to the screening questionnaire.
Estimates are weighted 65
Table 7. Demographic characteristics of consumers. Estimates are weighted 66
Table 8. Mean, median and selected percentiles of FCRs (g/day, raw weight, edible portion) in the Nez
Perce Tribe, based on the FFQ; consumers only. Estimates are weighted 69
Table 9: Estimated distribution of FFQ consumption rates (g/day, raw weight, edible portion) of
consumers within demographic groups. All rates are for total consumption (Group 1).
Estimates are weighted 72
Table 10. Mean and median of Group 1 (all fish) FFQ consumption rates (g/day, raw weight, edible
portion) by groups according to design variables. Weighted results 73
Table 11. Unadjusted and adjusted differences in mean Group 1 (all fish) FFQ FCRs (g/day, raw weight,
edible portion) by the design variables. Linear regression. Weighted results 74
Table 12. Distribution of the usual fish consumption of species Group 1 (g/day, raw weight, edible
portion) based on the 24-hour recalls. Estimated by the NCI method 77
Table 13. Distribution of the usual fish consumption of species Group 1 (g/day, raw weight, edible
portion) and their 95% confidence intervals based on the 24-hour recalls. Estimated by the NCI
method 78
Table 14. Distribution of the usual fish consumption of species Group 2 (g/day, raw weight, edible
portion) based on the 24-hour recalls. Estimated by the NCI method 79
Table 15. Estimated means and 95th percentiles of consumption (g/day, raw weight, edible portion) by
species group and estimation method 84
Table 16. Number of respondents with consumption on the FFQ and 24-hour recall by species group
and decile of FFQ consumption rate. These show the number of non-zero values included
in the calculations for Tables 17 and 18, and the sample sizes for each cell in Table 19.... 91
-------
Table Title Page
Table 17. Weighted mean consumption (g/day) from the 24-hour recall and FFQ interviews for each
species group, overall and by decile of FFQ consumption rate. Deciles are calculated from the
group 1 FFQ consumption rate. All rows are based on all group 1 consumers. Ratios were not
calculated when a species group was not consumed by the FFQ 92
Table 18. Weighted mean expected frequency (percentage of days) with fish consumption from the 24-
hour recall and FFQ interviews for each species group, overall and by decile of FFQ
consumption rate. Deciles are calculated from the group 1 FFQ consumption rate. All rows are
based on all group 1 consumers. Ratios were not calculated when a species group was not
consumed, based on the FFQ responses 93
Table 19. Weighted mean portion size (grams) from the 24-hour recall and FFQ for each species group,
overall and by decile of FFQ consumption rate. Deciles are the deciles of the group 1 FFQ
consumption rate. Each individual's portions sizes were averaged across species with a weight
according to the species consumption frequency. All calculations are limited to positive portion
sizes. Ratios were not calculated when a species group was not consumed based on the FFQ or
24-hour recall 94
Table 20. Frequency of consumption at special events and gatherings for selected species and groups.
Does not include consumption outside of special events and gatherings. Estimates are
weighted 96
Table 21. Percent of the time other fish parts were consumed for selected species and species groups.
Consumers only. Estimates are weighted 97
Table 22. Percent of the time different preparation methods were used for selected species and species
groups. Consumers only. Estimates are weighted 97
Table 23. Percent of the time selected species and species groups were consumed from different sources.
Consumers only. Estimates are weighted 98
Table 24. Fishing activities during the preceding year as reported by the 283 respondents who reported
fishing at least once. Estimates are weighted 100
Table 25. Changes in consumption and access to fishing in the eligible consumer population. Estimates
are weighted 101
Table 26. Summary of FFQ interview and reinterview responses. All rows are based on all 31
respondents who completed both interviews. Summaries are unweighted 102
Table 27. Descriptive summary of interviewers' ratings of respondents' cooperation and reliability during
the first interview. Summaries are unweighted 103
Table 28. Total FCRs (g/day) of adults in Pacific Northwest Tribes (with consumption rates available)
and the U.S. general population. Consumers only 113
-------
List of Figures
Figure Title Page
Figure 1. Nez Perce reservation and surrounding eligible ZIP codes for
inclusion in the Nez Perce Tribe fish consumption survey 20
Figure 2. Histogram of Group 1 (all fish) FCRs
(g/day, raw weight, edible portion) 59
Figure 3. Estimated cumulative distribution of FFQ FCRs
(g/day, raw weight, edible portion) 70
Figure 4. Distribution of the usual fish consumption
(g/day, raw weight, edible portion) 76
Figure 5. Weighted Group 1 means (g/day) of the 24-hour recall
and of the FFQ consumption rates by Group 1 FFQ deciles 95
Figure 6. Weighted Group 2 means (g/day) of the 24-hour recall
and of the FFQ consumption rates by Group 1 FFQ deciles 95
Figure 7. Mean number of times respondents went fishing each
month among the 283 respondents who reported fishing at least once 99
-------
1.0 Preface to Volume II
This report of current fish consumption rates (FCRs), which includes both finfish and shellfish,
among the Nez Perce Tribe is a step toward quantitatively documenting the role of fish in the life
of the Tribe. The FCRs from this survey can be used by the Tribe, by the State of Idaho and by
other bodies to inform and guide the effort to assess risks posed by contaminants in fish for
populations with a high level of fish consumption. The foreword to Volumes I-III provides much
more comprehensive material on development of this report and the context for use of the
information included within the report.
While the results of this report are numeric, the numbers are only a companion to the Nez Perce
culture, heritage and vision for their future. It may help the reader to know more about the Nez
Perce Tribe, the role of fish in the lives of its members and the activities of the Tribe in relation
to fish and fishing. Volume I of this report on heritage FCRs includes material that provides a
better understanding of the Tribe's longstanding relationship and dependence on fish and fishing.
The Nez Perce Tribe Final Survey Design document (included in Volume III of this report as
an appendix) provides additional information on the Nez Perce Tribe. The design report covers
a number of topics, including the background and purpose of the survey, the survey objectives
for the Tribe, the importance of heritage FCRs to the Tribe, the suppression of fish
consumption over time, the role of the current survey and a historic assessment. The Nez Perce
Tribe Final Survey Design document was written prior to implementation of the survey. The
document covers issues, concepts and planning that were developed or considered in
preparation for implementation. Some aspects of the design were changed after the Nez Perce
design document was completed (document date: February 2014). Changes in design that
might have statistical implications are covered in Volume II of this report (with related
material in Volume III, as needed).
About this volume. Volume II of this report includes the main numeric findings from the survey
data. At various places in the report there are references to Volume III, which is a series of
appendices intended to provide more detail or additional relevant material.
Nez Perce Tribe
Volume II: Current Fish Consumption Survey
December 2016 Page 1
-------
2.0 Acronyms and Abbreviations
AMPM
Automated Multiple Pass Method
AWQC
Ambient Water Quality Criteria
CAPI
Computer-Assisted Personal Interviews
CRITFC
Columbia River Inter-Tribal Fish Commission
EPA
Environmental Protection Agency
FCR
Fish Consumption Rate(s)
FFQ
Food Frequency Questionnaire
g
Grams, as in g/day
HSSRO
Human Subjects Research Review Official
IDDEQ
Idaho Department of Environmental Quality
IRB
Institutional Review Board
NCI
National Cancer Institute
NHANES
National Health and Nutrition Examination Survey
NPT
Nez Perce Tribe
SBT
Shoshone-Bannock Tribes
USRTF
Upper Snake River Tribes Foundation
Nez Perce Tribe
Volume II: Current Fish Consumption Survey
December 2016 Page 2
-------
3.0 Executive Summa
3.1 Introduction and Purpose
This is a report on fish consumption by the Nez Perce Tribe (NPT). The numeric FCRs (edible
mass of uncooked finfish and/or shellfish in grams per day) presented here are based on two
statistical methods and two types of data used to estimate FCRs. One method uses a food
frequency questionnaire (FFQ), wherein survey respondents directly provide estimates per
species of frequency of consumption, portion sizes and duration of their consumption seasons
during the past year. The analysis results provide means and percentiles of FCRs for the Nez
Perce Tribe. The second statistical method uses responses to questions asked on two separate
days about fish consumption "yesterday" (a 24-hour recall period). The 24-hour data along with
additional data from the survey and some accepted and plausible statistical modeling yields,
again, means and percentiles of FCRs. The purpose of the report is to quantitatively describe
current fish consumption and related activities of the Nez Perce Tribe. The FCRs from this
survey can be used by the Tribe, by the State of Idaho and by other bodies to inform and guide
the effort to assess risks posed by contaminants in fish for populations with a high level of fish
consumption, including development of ambient water quality criteria to protect human health.
The data analyzed in this report are based on interviews conducted from May 2014 to May 2015.
The earliest in-person interview (including the FFQ and first 24-hour recall) that supplied
useable data for this report occurred on May 10, 2014. The last in-person interview occurred on
April 24, 2015. Telephone interviews continued through May 4, 2015 to complete the second 24-
hour dietary recall interview.
3.2 Survey Methods
Every aspect of this survey was designed in an extensive, time-consuming and transparent
collaborative process beginning in the Fall of 2012 and lasting until the Fall of 2016 between the
Nez Perce Tribe, the Environmental Protection Agency, tribal consortia and a highly skilled and
experienced team of expert consultants. Efforts were made to incorporate state-of-the-art survey
and analytical methods and tribal cultural and governmental concerns in a study that was
designed to contribute to understanding fish consumption by members of the Nez Perce Tribe
who were surveyed.
This study is unique among tribal surveys in that it included all of the following features: the
interviews covered an entire year; the survey included a FFQ (food frequency questionnaire)
which yielded data to support fish consumption estimates; and the survey simultaneously
included up to two 24-hour recall interviews which were used to calculate fish consumption
estimates using the statistical modeling of the NCI method. The FFQ method has been used
frequently in the past. The NCI method was included in the survey as a more state-of-the-art
method that was designed to improve accuracy in fish consumption estimates.
The survey covered adult tribal members (age 18 and over) residing in ZIP codes falling within
approximately 50 miles of two major tribal centers, Lapwai and Kamiah, which are 60 miles
apart by road. Children and teenagers were not included in the survey due to the additional time
and resources that would have been needed for development of appropriate methodology,
interviewing and analyses for this age group. The geographic scope was selected in consideration
Nez Perce Tribe
Volume II: Current Fish Consumption Survey
December 2016 Page 3
-------
of the logistics of interviewers needing to reach respondents as well as to select a sample that
would represent Nez Perce fish consumers specific to Idaho. A stratified random sample was
drawn from tribal enrollment files, where the strata were defined by gender and age. The sample
size of each stratum was chosen to be in proportion to the size of the stratum in the tribal
enrollment file. Within each stratum, members were drawn randomly. Tribal fishers ("Tribal
members who fish") were identified from a roster maintained by the Tribe; a number of fishers
were included in the sample and were interviewed. A fish consumption rate is reported for the
fishers as a distinct population.
Tribal interviewers were employed and trained by an EPA sub-contractor in charge of
administering the questionnaire. Tribal interviewers (rather than non-tribal interviewers) were
selected, because both tribal representatives and EPA contractors thought that tribal members
would be more likely to accept an interview from and convey more accurate information a fellow
tribal member (and also be more likely to accept a home interview) than from someone outside
the Tribe. In addition, tribal member interviewers have a very wide network of relatives and
friends within the tribal community—which proved to be very helpful in locating members to be
sampled (sometimes the most difficult step) and gaining their cooperation for an interview. The
tribal leadership and staff also expressed, in advance, the importance of using tribal interviewers
for cultural reasons, for tribal capacity-building, to improve the likelihood that tribal members
would participate in the survey, and also to provide income for tribal members. Tribal
interviewers were also used in other Pacific Northwest fish consumption surveys of Native
Americans (CRITFC, 1994, Toy et al., 1996, Suquamish Tribe, 2000). As noted later (Section
5.14, Design Changes), non-tribal interviewers conducted some interviewers under special
circumstances. In order to facilitate coordination and maintain data quality, interviewers worked
under close supervision of the staff of the survey research firm charged with implementing the
survey. Respondents were offered an incentive for participation in the survey, financed entirely
by the Tribe. Incentives included entering respondents who completed interviews into a raffle
drawing (approximately $3,000 worth of prizes were available), t-shirts and paid time off for
tribal employees for participating in the interviews. Tribal officials, EPA staff and contractor
staff met in conference calls several times each month to review progress and resolve evolving
challenges of fieldwork. Meetings were held each week during the summer of 2014 to address
issues of respondent recruitment. Survey progress was reviewed with State of Idaho officials on
a regular basis paralleling the State's own survey effort.
Respondents to the survey answered questions about species consumed (frequency and quantity),
covering consumption over the past year, as well as answering questions about fish consumption
"yesterday" (the 24-hour recall). The questions on 24-hour recall were repeated in a separate
interview (usually by telephone) administered on a later day, chosen with enough lag after the
first interview (at least 3 days) to provide an independent assessment of the respondent's
consumption. More closely spaced interviews might have caused second interview results to be
affected by consumption events covered by the first interview, as, for example, leftovers from a
first interview fish meal might be consumed over the next few days. An attempt was made to
match the first and second interview timing during the seven days of the week so that the two
interviews would both either be on a weekday or a weekend day.
The questions about consumption over the past year followed the format of a food frequency
questionnaire (FFQ), which is common in dietary studies. The analysis of the FFQ data provides
Nez Perce Tribe
Volume II: Current Fish Consumption Survey
December 2016 Page 4
-------
an estimated average daily fish consumption rate in grams/day for each respondent and for any
species or species group referenced in the survey. Data from the two 24-hour recall interviews
were analyzed using the "NCI method"—a methodology developed by the National Cancer
Institute and other researchers. (The NCI method can—and did in this study—also use other
survey data to improve the estimates of fish consumption rates.) The NCI method yields a
distribution of the usual fish consumption rate in grams/day. The results of the NCI method are
also presented here. Both FFQ and 24-hour recall questionnaires can be found in Appendix A.
The statistical analysis included development of appropriate statistical weights in an effort to
provide unbiased estimates of fish consumption for the Tribe. These weights are expected to
correct for some or all of the potential response bias due to differential response rates across
demographic groups of the Tribe. Specifically, the respondents in demographic groups with a
smaller response rate (relative to other groups) needed to be given a greater statistical weight so
that all demographic groups would be appropriately represented in the analysis. The mean,
median and percentiles of fish consumption are reported for all species combined (species Group
1), for near coastal, estuarine, freshwater and anadromous species (species Group 2), and for
other species groups. Additional fish consumption statistics are provided for demographic sub-
groups of the Tribe.
This survey project includes an analysis of heritage rates—the FCRs of the Tribe that were in
place prior to modern environmental and social interference with its fishing practices. The
current consumption rates presented here, combined with the heritage rates (see Volume I),
provide a range of potential future fish-consuming populations (and associated FCRs) to be
considered in the effort to protect people with a high level of fish consumption, including
development of ambient water quality criteria to protect human health.
3.3 Results
A sample of 1,250 adult tribal members (age 18 or older) was drawn from tribal enrollment files,
representing 46% of the 2,727 adult members recorded in the files. Over the course of the
interview period, 460 members were interviewed and provided sufficient information to classify
them as fish consumers or non-consumers and to calculate an FFQ consumption rate for the
consumers. The response rate for the survey is 38%. Only 9 of the respondents were non-
consumers, and, using appropriate survey weighting, this count leads to an estimate of 3% non-
consumers in the Tribe. The FCRs for the Tribe are summarized briefly in Tables SI and S2.
Additional FCRs are provided in the body of this report.
The Tribe's estimated current total fish and shellfish consumption rates are high relative to the
U.S. general population (Table S3), and the rates for the population of fishers in the Tribe is
substantially higher (Table S2). The consumption rates include some relatively high rates for
each of the population and species groups presented in Tables SI and S2; the 95th percentile is
several-fold larger than the median, typically an indication of the presence of relatively high
rates. The mean and percentiles of consumption by the NCI method are smaller than those
calculated by the FFQ method. For example, among the consumption estimates for Group 1
species (all species combined), the mean consumption rate from the NCI method is 39% lower
than the mean rate from the FFQ method. The NCI method median is 30% lower than the FFQ
median, and the NCI method 95th percentile is 47% lower than the corresponding FFQ value. For
Nez Perce Tribe
Volume II: Current Fish Consumption Survey
December 2016 Page 5
-------
Group 2 species, the mean, median and 95th percentile of consumption rates calculated by the
NCI method range from 29% to 41% lower than the corresponding FFQ rates.
The smaller rates from the NCI method than from the FFQ method arise, in part, from the
smaller values of fish consumption frequencies and portion sizes reported in the 24-hour data
than in the FFQ data. For Group 1 species (all species combined), the mean frequency calculated
from the 24-hour data was 85% as large as the mean frequency from the FFQ data. The
corresponding value for Group 2 species was 86%. The Group 1 and Group 2 mean portion sizes
from the 24-hour data were 87% and 89% as large as the mean portion sizes from the FFQ data,
respectively. The relative difference in frequencies and portion sizes was larger for the high
consumers. Among the 10% of consumers with the highest FFQ consumption rate (all species
combined) the 24-hour mean frequency for Group 1 and Group 2 was 64% and 57% of the FFQ
mean frequency, respectively. Again, for these high consumers, the 24-hour data's mean portion
size for Group 1 and Group 2 species was 61% and 64% of the FFQ mean portion size,
respectively.
Table SI. Nez Perce Tribe. Mean, median and selected percentiles of FFQ and NCI method FCRs
(g/day, raw weight, edible portion); consumers only. Estimates are weighted.
Percentiles
Species Group*
No. of Consumers
Mean
50%
90%
95%
Group 1 - FFQ
451
123.4
70.5
270.1
437.4
Group 1 - NCI Method
451
75.0
49.5
173.2
232.1
Group 2 - FFQ
446
104.0
61.3
231.4
327.9
Group 2 - NCI Method
446
66.5
36.0
159.4
233.9
*Group 1 includes all finfish and shellfish. Group 2 includes near coastal, estuarine, freshwater, and anadromous
finfish and shellfish.
Nez Perce Tribe
Volume II: Current Fish Consumption Survey
December 2016 Page 6
-------
Table S2. Nez Perce Tribe. Mean, median and selected percentiles of FFQ and NCI method FCRs
(g/day, raw weight, edible portion) for fishers and non-fishers; consumers only. Estimates are
weighted.
Percentiles
Species Group*
Group
No. of Consumers
Mean
50%
90%
95%
Group 1
Fishers - FFQ
138
171.8
98.0
436.8
543.5
Group 1
Fishers - NCI Method
138
98.2
64.7
229.2
305.0
Group 1
Non-fishers - FFQ
313
107.9
65.5
232.9
337.7
Group 1
Non-fishers - NCI Method
313
67.6
45.6
155.1
206.0
Group 2
Fishers - FFQ
138
156.7
83.5
360.7
507.8
Group 2
Fishers - NCI Method
138
98.4
55.2
238.6
345.0
Group 2
Non-fishers - FFQ
308
86.9
51.0
186.2
261.1
Group 2
Non-fishers - NCI Method
308
55.6
32.0
132.0
189.5
*Group 1 includes all finfish and shellfish. Group 2 includes near coastal, estuarine, freshwater, and anadromous
finfish and shellfish.
Table S3. Nez Perce Tribe. Total FCRs (g/day, raw weight, edible portion, all species combined) of
adults in Pacific Northwest Tribes (with consumption rates available) and the U.S. general
population. Consumers only.
Population
No. of
Consumers
Mean
Percentiles
50%
95%
Nez Perce Tribe - FFQ
451
123.4
70.5
437.4
Nez Perce Tribe - NCI Method
451
75.0
49.5
232.1
Shoshone-Bannock Tribes - FFQ
226
158.5
74.6
603.4
Shoshone-Bannock Tribes - NCI
Method
226
34.9
14.9
140.9
Tulalip Tribes
(Toy, et al, 1996)
73
82.2
44.5
267.6
Squaxin Island Tribe
(Toy, et al, 1996)
117
83.7
44.5
280.2
Nez Perce Tribe
Volume II: Current Fish Consumption Survey
December 2016 Page 7
-------
Population
No. of
Consumers
Mean
Percentiles
50%
95%
Suquamish Tribe
(The Suquamish Tribe, 2000)
92
213.9
132.1
796.9
Columbia River Tribes
(CRITFC, 1994)
464
63.2
40.5
194.0
USA-NCI Method
(U.S. EPA., 2014)
*16,363
23.8
17.6
68.1
* Adults >21 years old; includes both consumers and non-consumers.
The rates for Columbia River Tribes are from CRITFC, 1994, Table 10. The rates for the Suquamish Tribe are from
Suquamish Tribe, 2000, Table T-3 and Liao, 2002. These rates were converted from g/kg/day to g/day by
multiplying by the mean body weight of 79.0 kg, found in Table T-2 of Suquamish, 2000. The rates for the Tulalip
and Squaxin Island Tribes are from Polissar, 2014, Table 2 and Table 3, respectively. The national rates are from
U.S. EPA, 2014, Appendix E, Table E-l. The rates for the Nez Perce and Shoshone-Bannock Tribes are from this
report and the other report released at the same time as this report with virtually the same format, in Table 8 (FFQ
rates) and Table 12 (NCI method rates).
3.4 Discussion
The NPT's fish consumption rates are high compared to the rates of other populations. The NPT
rates can be compared to the rates for other populations, carrying out the comparison among
rates based on a similar survey and analysis methodology (either the FFQ method or the NCI
method). The NPT's NCI-method rates (which are probably more accurate than the FFQ rates)
are several-fold higher than the NCI-method FCRs for the U.S. general population and the
Shoshone-Bannock Tribes. (See Table S3.)
The FCRs of the Nez Perce Tribe—based on the FFQ methodology—are generally higher than
those observed in other Pacific Northwest tribal fish consumption surveys, such as the Columbia
River Inter-Tribal Fish Consumption survey (which included the Nez Perce Tribe—see CRITFC,
1994), with an exception being the survey of the Suquamish Tribe. The FFQ mean and 95th
percentile rates of the Shoshone-Bannock Tribes are numerically larger than those of the Nez
Perce Tribe, but the uncertainty renders the FFQ rates for these two tribes comparable. For
example, the margin of error (95% confidence intervals) for both the SBT's mean and its 95th
percentile consumption rates include the NPT's values for the mean and 95th percentile,
respectively (see Table 8). The level of uncertainty is such that it is difficult to designate either
tribe's actual adult FFQ rates as higher or lower than that of the other tribe. The NPT's FFQ
mean consumption rate is from 50% to 100%) larger and the 95th percentile of consumption is
from 56%o to 125% larger than the FFQ mean consumption rate of the pooled Columbia River
Tribes (CRITFC survey), the Squaxin Island Tribe and the Tulalip Tribes.
The estimated mean consumption rates (Groups 1 and 2) differed between the FFQ-based rates
and the rates based on the 24-hour recalls (which are used in the NCI method), with the 24-hour
Nez Perce Tribe
Volume II: Current Fish Consumption Survey
December 2016 Page 8
-------
mean rates being lower. The survey-weighted 24-hour mean consumption rates of Group 1 and
Group 2 species were 76% and 78% as large as the FFQ means for Groups 1 and 2, respectively.
The other species groups assessed (Groups 3-7) also had lower survey-weighted 24-hour means
than the FFQ means.
It is likely that—compared with the FFQ approach—the rates based on the NCI method are closer
to the actual1 FCRs of the adult Nez Perce population, because the challenge to a respondent's
memory is less than that involved in collecting the type of data used by the FFQ method. The NCI
method, however, contains strong assumptions about the shape of the distribution of usual
consumption, and the fitted shape used to provide the NCI estimates may or may not fit well in the
tails of the distribution, including the important and often-cited 90th and 95th percentiles. At this
point in the history of fish consumption surveys, there is no definitive scientific evidence that the
NCI method yields rates that are closer than FFQ rates to the actual distribution of fish
consumption of the adult Nez Perce population. Invoking the memory issue in favor of the NCI
method provides a type of common-sense piece of evidence, but that evidence alone is not
sufficient to eliminate FFQ rates from serious consideration. It is likely that FFQ surveys will need
to continue into the future in certain situations, such as for small surveys, for surveys with limited
resources, or for surveys assessing fish species (or other foods) with a relatively low frequency of
consumption. Such surveys will address the need for estimates of fish consumption.
The NCI method, using 24-hour recall data, and the FFQ method, using respondents' perceptions
about the past year of consumption, yield a range of estimates, and this range seems highly likely
to include the actual FCR values. It seems likely that the actual consumption rates are closer to
the NCI estimates, since they are based on memory of consumption "yesterday" rather than
memory of the past year of consumption. Both the FFQ and NCI method approaches are,
currently, accepted survey methodologies.
Some factors—including those just discussed—that may help to explain the difference between
the FFQ consumption rates and the rates from the NCI method include the following. Chance:
The days on which the respondents were interviewed about their consumption "yesterday" (24-
hour recall) happened to selectively miss their days of actual fish consumption. Memory and
interpretation: Both the FFQ and 24-hour recall responses require the respondents to exercise
their memory and interpret their fish consumption behavior. The 24-hour recall is less
challenging to memory than the FFQ. Differences in frequency or portion-size reporting: Both
frequency and portion size appear to be either over-reported in the FFQ data or under-reported in
the 24-hour recall data, or both. Modeling: tails of the distribution: As noted earlier in this
section, the rates based on the 24-hour recall and the NCI method may be more accurate in the
middle of the distribution of usual consumption rates than in the upper or lower tails, including
the important 95th percentile of consumption rates.
In summary, the NCI method's rates based on the 24-hour recall interviews are likely to be closer
to the actual rates than the rates from the FFQ analysis, due to the lighter demand on memory
required by the 24-hour recall approach. The NCI method's and the FFQ method's rates provide a
1 Throughout this report, the familiar term "actual" (e.g., "actual adult FFQ rates") is usually used in place of the more statistical term
"true" (e.g., "true adult FFQ rates") to indicate the rates that apply to the population under study. If, for example a rate such as the
95th percentile offish consumption were known for the entire target population, such as the population of adults in the defined ZIP
code area, it would be referred to as the "true 95th percentile" or "the population 95th percentile" for a statistical audience.
2 The NCI method assumes a certain family of shapes derived from the normal distribution by a Box-Cox power transformation.
Nez Perce Tribe
Volume II: Current Fish Consumption Survey
December 2016 Page 9
-------
plausible range of consumption rates. Additionally, the FFQ approach may be the only feasible
method for development of FCRs for narrowly defined fish groups or for small surveys, for which
the data needed to implement the NCI approach would usually not be available. Future fish
consumption surveys utilizing the NCI vs. FFQ methodologies will, hopefully, clarify the precision
and accuracy of these approaches. Unfortunately, the resources required to run surveys, in
particular for the NCI method, will likely result in relatively slow acquisition of new information
that can shed light on the reasons for differences in rates from these two methodologies.
This study is unique in that it used both the FFQ (food frequency questionnaire) and the 24-hour
recall (NCI) methods simultaneously in a survey of tribal consumption of fish over an entire
year. The survey included the two methods in a manner such that both methods could be used to
provide quantitative estimates of fish consumption. No other studies have included all of these
elements. The strength of the current rates is that they are derived by technically defensible
methodologies, and these rates can be compared to those of other populations. The use of two
distinct methods to estimate fish consumption—FFQ and 24-hour recall (combined with the NCI
method)—had multiple benefits, and, taken together, provided a very comprehensive study on
fish consumption. This study is also unique in the length of time over which it was conducted.
No other study of tribal fish consumption has run both the FFQ method and NCI method and
also conducted interviews for a full year, covering multiple periods of fish runs and seasons and
a full annual cycle of cultural activities. The span of the survey allowed evaluation of seasonal
and temporal impacts on FCRs (although the evaluation of the role of time was limited by a
relatively small number of respondents for some months of the survey, particularly during some
months with strong fish runs).
The design and implementation of this study involved a collaborative effort of tribal
governments, the U.S. Environmental Protection Agency, and a team of highly qualified and
experienced cross-disciplinary consultants. Significant financial and in-kind resources, as well as
technical and cultural expertise, were combined to create a unique and comprehensive survey.
The expert contractor team consisted of firms with considerable relevant experience in: survey
fieldwork (Pacific Market Research), conducting surveys of other Native American tribes and
minority ethnic groups (The Mountain-Whisper-Light and Pacific Market Research), conducting
statistical analysis and reporting results of Native American fish consumption surveys (The
Mountain-Whisper-Light) and working with Native Americans on environmental issues
(Ridolfi). This contributed to the rigor of the study design and provided ongoing review and
adaptation as challenges were encountered in the field.
One advantage of the collaboration with the tribal government is that the contractor team was
allowed access to a unique frame for drawing the sample: tribal enrollment records. The use of
the enrollment records avoided a costly effort to develop an alternative frame for sampling. The
random sampling (as opposed to, for example, a convenience sample) conducted from this
complete population listing added to the precision of the survey by using survey resources to
increase the sample size rather than using some of the resources for an alternative and costlier
means of identifying respondents with, inevitably, a reduced sample size. The availability of a
population roster from which to draw the sample along with the availability of a list of fishers
also permitted characterization of population demographics, which supported statistical
weighting of respondent results to ensure that the results represented the target population as
Nez Perce Tribe
Volume II: Current Fish Consumption Survey
December 2016 Page 10
-------
much as possible. Developing the statistical weights would have been far less successful without
access to a population roster.
The use of in-person interviews is a strength of the study. That form of data collection was
expected to generally lead to more accurate and complete responses in this population, due to the
expected better acceptance of a personal approach to potential respondents by tribal interviewers
and because in-person interviews readily allowed the use of physical display models for species
identification and portion sizes. Many of the interviews were conducted in respondents' homes,
which may have provided a more comfortable environment to participate in a long, detailed
personal interview. Advance scheduling of interviews also ensured that interviews were
conducted during times that were convenient for respondents, allowing collection of information
without competing demands. The interviewers could ensure completeness of responses (e.g.,
ensuring topics and questions were not skipped), could question inconsistent responses, and
could clarify questions for respondents. In-person interviews also allowed interviewers to use
portion model displays and photographs of relevant fish species. These visual aids enhanced the
ability of respondents to accurately identify the species consumed, specify portion size, and
correctly identify preparation methods.
Interviews were conducted by using unobtrusive electronic tablets to collect raw interview data;
the data were uploaded frequently and subsequently reviewed by the survey team. The electronic
CAPI system also immediately checked key entry to permit only valid codes. Automated data
uploading eliminated errors associated with manual data entry. The CAPI3 interview model
likely made the data more accurate and complete by assisting the interviewers in following skip
patterns (avoiding inapplicable questions or topics, for example, questions on breast-feeding for
male respondents) and ensured that relevant questions were not missed or left unanswered. The
CAPI also facilitated interview administration and accuracy by including prompts for the
interviewer to use visual aids (i.e., portion size models and species photographs) at relevant
points in the interview. In summary, use of a CAPI allowed for far more accurate administration
of a complex interview than would have been possible using a typical manual approach (e.g.,
paper and pencil).
A minor limitation of the survey is that some respondents could not remember and supply
answers to some questions, such as the typical portion size consumed for a particular species.
The missing data had to be imputed in order to retain the respondent's other related responses for
inclusion in the survey. A sensitivity analysis suggests that the imputations had little impact on
the final results. Another potential limitation of this interview-guided survey (and of any dietary
survey) is the possibility of social desirability bias, where some individuals may have the
tendency to over- or under-report consumption due to perceived social norms (Herbert, et al.,
1995, Tooze et al., 2004).
The survey had a modest response rate of 38%. The four other fish consumption surveys of
Pacific Northwest Indian Tribes have had response rates over 60% (i.e., CRITFC, Suquamish,
Squaxin Island and Tulalip surveys). While the statistical weighting may have addressed the
potential selection bias that may occur when there is a response rate of this magnitude, it is
possible that those in the sample who were not reached and interviewed do have a different
3 See Section 5.8 for a description of the CAPI method of interviewing. CAPI: computer-assisted personal interviewing. In this survey
the CAPI software was installed on electronic tablets.
Nez Perce Tribe
Volume II: Current Fish Consumption Survey
December 2016 Page 11
-------
consumption rate, on the average, than those included. That is an unknown at this time, and the
response rate of 38%, by itself, does not discredit this survey. The 95% confidence intervals
(showing the "margin of error" presented later in this report) allow for interpretation of
uncertainty (due to sample size) in the FCRs presented. However, the confidence intervals do not
show uncertainty due to undetected bias. The range of values in the confidence interval represent
plausible alternatives for the actual FCR, based on the degree of uncertainty. However, a
reported mean FCR or FCR percentile is itself the single best estimate of consumption, because
these estimates are derived through methodologic principles designed to avoid bias.
Because of the small populations of the Tribe, achieving the ideal response rate posed a
challenge that was not easily overcome. The Tribe is scattered over a large, primarily rural
geographic area. Obtaining contact information for many people was difficult and time-
consuming due to the rural nature of the sample. In-person interviews required significant
resources for travel time and costs and may have resulted in fewer interviews than would have
been possible in a more densely populated area. Also, the early period of interviewing coincided
with a time of strong fish runs. During this period interviews were accrued at quite a low rate and
some high fish consumers may have been unavailable due to their absence while out fishing.
Additional advance consideration of these issues might have increased response rates during the
early phase of the interviewing.
This study could not have been designed or completed without the full collaboration of tribal
officials. In order to meet interview quotas, the Tribe had to be creative in encouraging the
participation of its members through various public statements, promotional activities and,
importantly, offering incentives (financed entirely by the Tribe) in the form of a raffle as well as
in-kind services. Tribal enrollment data in itself, while an excellent and helpful source from
which to draw the sample, did not always provide contact information (e.g., phone numbers,
physical addresses) that the interviewers could use to make contact with the respondents. Much
time was spent developing additional methods to reach respondents and arrange interviews.
Tribal cooperation and willingness to think creatively about how to connect a respondent's name
to a contact point was critical to increasing the response rate. Additional time spent up-front in
testing the survey implementation might have discovered faster or less time-consuming ways of
contacting respondents prior to initiation of the field data collection. The frequent status
discussions with all parties involved in the survey enabled creative responses to the challenges of
locating tribal members and providing encouragement and incentives (financed entirely from
tribal resources) for tribal members to participate. These and the aforementioned experiences
should be considered in the design of and preparation for future fish consumption surveys of
Native American tribes.
3.5 Conclusion
The Nez Perce Tribe has FFQ FCRs that are among the highest in the Pacific Northwest, and
both the FFQ and NCI-method means and percentiles are several-fold higher than consumption
rates of the U.S. general population (See Table S3.) FCRs determined using the NCI method
were lower than those determined using the FFQ approach. Mean FCRs for Group 1 species (all
finfish and shellfish) and Group 2 species (near coastal, estuarine, freshwater, and anadromous
finfish and shellfish), based on the NCI method, were, respectively, 39% and 36% lower than
means obtained via the FFQ approach.
Nez Perce Tribe
Volume II: Current Fish Consumption Survey
December 2016 Page 12
-------
4.0 Introduction
4.1 Purpose of the Overall Fish Consumption Survey Effort
The Nez Perce Tribe collaborated with the U.S. Environmental Protection Agency (EPA) Region
10, tribal consortia and the State of Idaho to gather data on tribal FCRs (FCR) in Idaho. One
objective of this effort was to support the effort to assess risks posed by contaminants in fish for
populations who consume fish at high levels. More generally, this effort was intended to enhance
tribal environmental capacity in preserving and improving water quality. This report presents
survey methodology and results, specifically FCRs, for the Nez Perce Tribe.
4.2 Putting the Survey of Current Fish Consumption in Context
A fish consumption study fits into a larger context. There are three eras of importance for such a
study: the past, the present, and the future. Assessing consumption through a current cross-
sectional survey will provide relatively precise information about current consumption only. The
strength of the current rates is that they are derived by a technically defensible methodology, and
these rates can be compared to those of other populations.
The reader is directed to the foreword of this three volume report for a discussion of the
background and purpose of volume II, which discusses current fish consumption rates. The
foreword places the information in volume II in context of the overall survey effort. Specifically,
the foreword discusses how historic fish consumption rates, suppression in historic fish
consumption rates, current fish consumption rates, and tribal treaty rights should be considered
in developing water quality criteria to protect human health.
The survey was implemented largely consistent with the Nez Perce final survey design report
(Appendix H in Volume III).
4.3 A Brief Description of the Nez Perce Tribe
The Nez Perce Tribe of today is a self-governing, Federally Recognized Tribe located on
a reservation in North Central Idaho, which lies primarily in the Camas Prairie region south of
the Clearwater River, covering parts of Nez Perce, Lewis, Idaho, and Clearwater Counties. The
tribal government seat is at Lapwai, which also contains the largest population of Nez Perce, and
the community with the largest population within the reservation boundary is the City
of Orofino. Additional material about the Nez Perce Tribe can be found in Volume I of this
report (Heritage Rates) and in Volume III, Appendix H, "Design of a Survey on Fish
Consumption by the Nez Perce Tribe".
4.4 Populations
The tribal populations described quantitatively in this report are the Nez Perce Tribe as a whole
(adult members) and the population of adult, "documented" fishers within the Tribe.
Identification of tribal members was obtained from confidential tribal enrollment records in close
consultation with tribal officials.
Identification of the fisher group was achieved using a list derived from the Nez Perce
Department of Fisheries Resources Management (DFRM) records of sampling activities that are
conducted annually for certain fisheries. Information is collected and compiled for specific
Nez Perce Tribe
Volume II: Current Fish Consumption Survey
December 2016 Page 13
-------
individual tribal members who fish in certain rivers/areas. Tribal members were observed or
interviewed for their fishing activities at a certain area during a certain fishery season. This fisher
data was either collected during the actual fishery season or collected post-season. This list
represents only those tribal members who provided in-season and/or post-season catch/harvest
data to DFRM staff. Some tribal members who are, in fact, fishers, do not appear on the fishers
list. Thus, the fishers list is not a comprehensive representation of all "fishers" of the Tribe, but,
rather, a "fisher indicator" (i.e., includes a subset) of the actual fisher population. When the term
"fisher" is used in this report, it refers to persons appearing on this fishers list. When there is
reference to a non-fisher, it means a person not on the fishers list. A certain fraction of those not
on the fishers list do, in fact, harvest fish, as discovered through answers to survey questions
regarding fishing activity, cross-referenced to the fishers list. Despite any inaccuracies in
designation of fishers and non-fishers, the fishers list is a useful roster of persons, most of whom
are engaged in fishing and harvesting activities. Those on the fishers list constitute one of the
populations identified in this report, with a presentation of their consumption rates. As noted,
some active fishers are not on the fishers list and will, thus, fall into the category labeled as "non-
fishers." The comparison of consumption rates between persons labeled as fishers or as non-
fishers has some uncertainty because some of the active fishers (and the complement, non-
fishers) among the respondents have not been assigned to the correct fisher/non-fisher category.
4.5 Guide to Report Sections
This document follows the commonly used IMRD format for scientific articles and reports:
Introduction, Methods, Results and Discussion. After this introduction, the methods used to
prepare for and then execute the survey in the field are described, as are the methods used to
analyze the data obtained from the survey. The Results section contains demographic statistics
about the population, the selected sample and the survey respondents, survey response rates,
quantitative fish consumption rates (overall and by demographic subgroups) and other statistics
related to tribal fishing and fish consumption. The Discussion section recaps the main findings
and discusses the strengths and limitations of the survey and its analysis. Appendices include
supporting technical material.
Nez Perce Tribe
Volume II: Current Fish Consumption Survey
December 2016 Page 14
-------
5.0 Methods
5.1 Overview
This section describes the basis for choosing the survey sample, including sample size,
inclusion/exclusion eligibility criteria, and the definition of the geographic area from which
survey-eligible tribal members were selected. It discusses the review and approval process, by
both tribal and external sources, for determining the survey's approach and procedures.
This section also reviews the development of the questionnaire, the methods used to draw the
sample from tribal enrollment records, identification of fishers4 to be used in calculating fisher
consumption rates, allocation of selected tribal members to sample waves of interviewing in
order to provide interviewing throughout the one-year survey period, reinterviewing of initial
respondents, and the relevance to this survey of computer-assisted personal interviewing (CAPI).
Selection and training of interviewers is discussed, along with methods for calculating survey
response rates, methods for weighting the sample to adjust for differential response rates in
different sample strata and for differentials in the probability of response related to demographic
factors. Finally, this section covers methods to convert respondent data on frequency and portion
sizes of consumed species to quantitative consumption rates, and methods to obtain means and
percentiles of fish consumption and their confidence intervals using two different analysis
methodologies. One methodology uses data collected from a food frequency questionnaire
(FFQ). A separate methodology, the "NCI method," uses data collected from the respondents'
recall of fish consumption during one or two 24-hour periods and also uses FFQ data and other
variables as covariates.
The two methods were used in tandem in order to be able to compare consumption estimates
from two very different methodologies. Under the assumption of perfect accuracy of responses
by the interviewed tribal members (and additional assumptions described later), the distribution
of usual consumption (means and percentiles) would have the same expected values. The two
sets of results would differ only by the element of chance that enters through, for example, the
random selection of days on which people were interviewed. An additional reason for using both
methods was the challenge of obtaining the required dataset for the NCI method. The modeling
used in the NCI method may not succeed if there are fewer than 50 respondents who report
having consumed fish on both of the 24-hour recall days.5 At the outset of planning for this
survey, it was not certain that the consumption frequency in the population (and the yet-unknown
total number of successful interviews) would be sufficient to offer adequate assurance of
reaching the 50 double hits. The FFQ method always yields data that can be used to develop
FCRs, though, ideally, FCR estimates from the FFQ method would be accompanied by an
evaluation of uncertainty in the rates. In this survey the NCI method was favored as a
methodology, because its use of recall data from "yesterday" was expected to be more accurate
than the recall of average consumption over the past year. The often-used and previously
accepted FFQ method was run in parallel with the newer NCI method, since the FFQ method can
4 See Section 4.4 for a definition of "fisher" as used in this document.
5 Based on discussions with key developers of the NCI method, Dr. Janet Tooze and Dr. Kevin Dodd, the NCI method may work
(produce a distribution of usual consumption rates) with fewer than 50 double hits. In the contractors' work with the NCI method,
covering this and other projects, the NCI method has sometimes worked and sometimes failed with fewer than 50 double hits. For
planning purposes it is safest to aim for at least 50 double hits.
Nez Perce Tribe
Volume II: Current Fish Consumption Survey
December 2016 Page 15
-------
succeed in yielding a consumption rate distribution even with a quite limited dataset. It also
allows more direct comparison with previous tribal fish consumption surveys. Further, the FFQ
method can provide consumption estimates for species groups with smaller numbers of
consumers, whereas small sample sizes and the associated small number of double hits usually
cannot meet the NCI method's data requirements.
5.2 Sample Selection
The planned sample size was developed to fulfill two goals: (a) a sufficient sample size so that
means and percentiles of FCRs calculated from the FFQ portion of the questionnaire would be
reasonably precise; and, (b) a sufficient sample size to provide reasonable assurance of an
adequate number of respondents with two separate 24-hour recall interviews, both of which
reported some fish consumption during the preceding 24-hour day ("yesterday").
The second goal was considerably more challenging to plan than the first. The criterion of at
least 50 "double hits" from the survey—two separate, independent interviews wherein a
respondent recalled eating fish on the preceding day—is a requirement6 of one of the methods
used to calculate a distribution of usual fish consumption. The "NCI method" refers to a
statistical procedure for calculating the distribution of usual consumption of episodically
consumed foods (Dodd, KW, et al. 2006; Tooze, JA, et al. 2006; Kipnis V, et al. 2009). Fish
consumption would fall into the "episodically consumed" category, since most people do not eat
fish every day. This technical method was designed to exploit data collected about consumption
(or non-consumption) of a food item on two or more independent days. The NCI method has
been used to analyze the data of this survey and the results of the analysis are provided in this
final report.
Part of the challenge in planning the sample size was the lack of relevant data or tabulations on
frequency of fish consumption (expressed in days with fish consumption per week, days per
month, or days per year) for this population. Data of this type were needed in order to estimate
what percentage of respondents who reported about their fish consumption on two independent
days would have fish consumption on both days. A count of 50 of the respondents having these
"double-hits" (two different days with fish consumption) is needed to provide strong assurance
that the NCI method can provide a distribution of consumption rates for a population. Among the
fish consumption survey reports about Native American tribes in the Pacific Northwest, there is
no survey that includes tabulations specifically on the frequency of consumption of fish (all
species combined), with frequency reported as consumption days per week, per month, per year
or per other time unit. The tabulations closest to this framework are in a Columbia River Inter-
Tribal Fish Commission survey report (CRITFC Technical Report 94-3, 1994), which reports on
the frequency of fish meals (not days with fish meals). In order to properly plan use of the NCI
method of estimating fish consumption rates, an estimate of the fraction of days with positive
fish consumption (or the average number of such days per week) is needed. The count of number
of meals per week with fish consumption would not suffice, in case there is a sizeable fraction of
tribal members who consume fish during two or more meals per day, for some days of the week.
6 While analysis by the NCI method might be possible with fewer than 50 double hits, the 50 count provides reasonable assurance
that models used in the analysis will converge on the necessary parameter estimates. The contractors have carried out NCI method
analyses for this and other projects. The analyses with fewer than 50 double hits would sometimes be successful (resulting in a
distribution offish consumption rates with means and percentiles of consumption) and would sometimes fail.
Nez Perce Tribe
Volume II: Current Fish Consumption Survey
December 2016 Page 16
-------
For example, a tribal member who eats three fish meals per week, on average, might typically eat
two fish meals on one day and one fish meal on another day. The respondent would have three
fish meals per week but only two days with fish consumption per week. Thus, the number of
meals per week is 50% larger than the number of days per week with fish consumption.
The CRITFC survey was carried out among four Columbia Basin tribes—the Warm Springs
Confederated Tribes, Yakama Indian Nation, Confederated Tribes of the Umatilla Indian
Reservation and the Nez Perce Tribe of Idaho. The Nez Perce Tribe was the only one of the four
tribe included in the current survey7.
Calculations were carried out on the expected number of double hits with various assumed
sample sizes, and assumptions were made which allowed for the conversion of fish meals per
week, as tabulated in the CRITFC report, to days with fish meals per week. Using these planning
assumptions and the CRITFC input tabular data, it was estimated that a sample of approximately
1,800 tribal members would provide good confidence that those completing the interviews of the
survey would include at least 50 individuals who would report eating fish on both of the two
independent days targeted by a 24-hour recall questionnaire (i.e., 50 double hits). Some notes
and calculations on the methods used to estimate the expected number of double hits under
various scenarios can be found at the end of Appendix D.
During the survey planning phase, five tribes of Idaho (the Kootenai, Shoshone-Paiute, Coeur
d'Alene, Shoshone-Bannock, and Nez Perce) were considering participation in the survey. To
employ the NCI method for each tribe individually, 50 double hits would have been needed for
each tribe. This was not possible given the resources available, the sample size that would be
o
needed per tribe, and, for the Kootenai, the small population size . Consequently, the 1,800
interviews were to be distributed over the five participating tribes with the intention of finding 50
double hits from the pooled results of all participating tribes. Thus, the authors decided to report
separate FCR distributions per participating tribe, using the NCI method, although the data from
multiple tribes would need to be pooled as input to the NCI method. The rates for individual
tribes would be obtained through the use of covariates in the NCI modeling process. The NCI
method includes provisions for the use of covariates (see Section 5.23.2). The covariates can be
used to indicate sub-populations. Thus, the combined tribal samples would represent a
"population" that is created for computational purposes only, and this pseudo-population is
needed in order to reach (or surpass) the 50 double hits collectively. A covariate indicating tribal
membership (with one category per tribe) would then allow for the computation of fish
consumption rates per tribe; each tribe would be a sub-population for computational purposes.
7 See CRITFC, 1994, Appendix 1, pp. 106-107. In the CRITFC survey each of the four tribes received a statistical weight used in the
weighted estimates presented in the report, such as consumption rates. The Nez Perce had a weight of 19%. The value of 19%
statistical weight for the Nez Perce Tribe is calculated as the Nez Perce population divided by the total population of all four tribes as
listed in the CRITFC report, p. 106. The listed population of the four tribes (which determines the statistical weight of each tribe in
calculating the combined CRITFC rates) are as follows: Umatilla, 818; Nez Perce, 1440; Warm Springs, 1531; Yakama, 3872. Total
of the four tribes: 7661
8 The Kootenai Tribe reported an adult population of 85 individuals (data received on October 2, 2013 from the Tribe). It may have
been technically feasible to achieve enough multiple hits to run the NCI method for this tribe analyzed separately, perhaps using
more than two 24-hour recall interviews per fish consumer. However, there was the uncertainty of reaching sufficient multiple hits,
and, further, the analysis would need to statistically accommodate the correlation of consumption among members of a household—
sure to be a feature of a 100% sample of this small population. The available software code for the NCI method does not currently
include an option for analyzing this type of correlation. These issues were a considerable barrier to implementing an NCI-method
data collection and analysis for this Tribe considered by itself.
Nez Perce Tribe
Volume II: Current Fish Consumption Survey
December 2016 Page 17
-------
This computational convenience has no cultural implication and it does not assume that the
distribution of usual consumption is the same for each of the tribes involved.
After further deliberation by the Idaho tribes, the Nez Perce and Shoshone-Bannock Tribes chose
to participate in surveying current fish consumption. Based on discussions with staff of these
Tribes, the planned approximate sample size of 1,800 was allocated as a sample of
approximately 1,200 from the Nez Perce Tribe and 600 from the Shoshone-Bannock Tribes.
Based on available information regarding fisheries and harvest levels, it was thought that the Nez
Perce Tribe had higher FCRs than the Shoshone-Bannock Tribes and, consequently, would
consume fish more frequently. Allocating more interviews to the Nez Perce Tribe improved the
chances of obtaining 50 double hits. The two tribes recognized that they both needed to achieve
the necessary number of "double hits" and that this part of the survey would require a joint effort
to do so—with a greater allocation of available sample size to the tribe expected to have more
frequent consumption. Within each tribe, of course, the sample would be selected by a random
process, and every effort would be made to obtain unbiased responses about consumption. None
of the respondents were aware of the goal of 50 double hits to support the NCI method.
The anticipated percentage of sampled members providing two 24-hour interviews was
calculated as: (a) an anticipated 60% response rate for the first 24-hour interview (and FFQ-
based interview), followed by (b) an anticipated 80% response rate for the second interview
among those participating in the first interview. The 60% for the first interview response rate was
selected as a conservative value given that response rates above 60% have been obtained for
other Northwest tribal fish consumption surveys (see Toy, et al, 1996 and Suquamish Tribe,
2000). The 80% continuation rate for those completing the first interview was simply an
assumed reasonable value for continuation among those who had participated in the first
interview. The net response rate for completion of both interviews would thus be 48%—
approximately half of the sampled members. The method for computing response rates is
covered in Section 5.13 ("Response rates" in the "Methods" section) and the achieved response
rates upon completion of the survey are covered in Section 6.1 ("Response rates" in the
"Results" section).
Nez Perce Tribe
Volume II: Current Fish Consumption Survey
December 2016 Page 18
-------
5.3 Inclusion/Exclusion Criteria
The survey was designed to assess the consumption rate of adults, defined as individuals age 18
and over. Specifically excluded from the survey were any members who were living in an
institutional setting (e.g., a nursing home). The reason for this exclusion is that a person in the
institutional setting would typically not be in control of their diet and might not be living a tribal
lifestyle in terms of diet. The enrollment files did not indicate this status, and such members were
identified during the initial contacts or attempts at contact with potential respondents.
During the interview process, an additional exclusion was incorporated: tribal members who
could not participate in the interview process due to physical, mental or other reasons were
excluded as they were encountered.9 This exclusion was based on practical considerations; in
particular, extra time would be needed to locate a person familiar with the tribal member's fish
consumption, both for a first interview (in person) and for a second interview (by phone). The
interviewers identified eight tribal members whom they encountered as falling in this category.
The tribal interviewers were also excluded from the sample. Their training and their extensive
contact with the contractors had made them very familiar with the potential use of the survey
data in the State of Idaho's deliberations on water quality and health. Even though the
interviewers were well aware of the need for unbiased responses, the contractors chose to
remove them from the pool of potential respondents and avoid any possibility or challenge that
their exceptional knowledge of the purpose of the survey might put them in a meaningfully
different category than the rest of the tribal population. While this may have been excessive
caution, the number of interviewers was small and the exclusion has presumably had a very
minor impact on the final fish consumption estimates. (There was a total of six interviewers from
the Nez Perce Tribe.)
There were no exclusions based on language issues. In advance of the survey, the contractor
team was informed by the tribal authorities that there would be no need to prepare for interviews
in any other language than English. No instances of non-response due to language issues were
reported to the contractors.
9 The specific disposition code that could be used by the interviewers for this status was labeled as "Impairment: hearing, mental
health, other."
Nez Perce Tribe
Volume II: Current Fish Consumption Survey
December 2016 Page 19
-------
5.4 Geographic Sample Selection Criteria
Initial exploration showed that this survey could not use the entire population of enrolled adult
tribal members as a target population for interviews. Data (not containing any personally
identifying information) from the tribal enrollment office showed that tribal members live
throughout the United States, with the greatest concentration on and near the reservation. There
would clearly be a limitation on the travel resources available for interviewing people in
person; persons living very far from the reservation would need to be excluded. Secondly,
there was a concern that members living very far from the reservation and far from the
fisheries used by tribal members might be different in some way from those living close; fish
consumption habits, lifestyle, and other known or unknown factors might substantially differ
from those living closer to or on the reservation. The travel limitations were the deciding factor
in limiting the geographic scope of the survey. A fifty-mile travel limit was considered
acceptable for practical survey operation. The selection of geographic areas was based on ZIP
codes, and the selected ZIP codes for the survey were approved by the Tribe. The selected ZIP
codes are shown in Table 1 and displayed in Figure 1. Areas on the map falling within the 50-
mile limit but with no (zero) population are not color-coded as included in the survey area. Not
all ZIP codes shown in the table and map provided respondents who were interviewed for the
fish consumption survey. Any adult tribal members residing in the noted ZIP codes were
eligible to be selected into the survey sample.
Figure 1. Nez Perce Tribe. Nez Perce reservation and surrounding eligible ZIP codes for inclusion
in the Nez Perce Tribe fish consumption survey.
M.M - 99,28 , , I
\ . rfarmingron
\ 93130
:tcHi tat- * • 83855
Varfa'UI CJO" 33834
Pott&ieh ; !|,jrvor •• !
sin" 99161J3672 83S57
ftofe,. go
-CWkta
83S12
B33D6
Lg&M
Albiii'n *99102
Ell<
.River
99113
I".*
UnliijitfiwSs*
83871
KCjhdriick
83527
B3832 83535
AvItfW 83540 83541
83524
-fcllrti
i
•OfArinW
tljfl'.sthn** « 83524
03545
63548
f!«" ,R - 8354c
«... , • 5523
WliitllWu-i
42? % ~ tr ¦
83544
83543 '.W'TPj--
83538
•Head uiui ftc f s
8354S
83553
luili
Ferdimmf.
83526
83533
jCoULin.viVitl*
83530 E
Grjirigov.'Ke;,
| Wfitte .
Sin.1
83554
•St.uv
Legend
£$« 2014, US Censuj 2010: US Nafidpal Acta 2010
ZIP Codes within 25 Miles of Lapwai or Kamiah
ZIP Codes within 50 Miles of Lapwai or Kamiah
Nez Perce Reservation Boundary
Population Centers
Note: Moscow, ID and Pullman, WA
arc the population centers for two ZIP code areas
RID0LFI
Environmental
Fish Consumption Survey of the Nez Perce Tribe
Prepared for the Nez Perce Tribe
April 2015
Figure J
Ne? Perce ZIP Code Map
for Sample Selection
Nez Perce Tribe
Volume II: Current; Fish Consumption Survey
December 2016 Page 20
-------
Table 1. Nez Perce Tribe. ZIP codes included for sampling members of the Nez Perce Tribe.
ZIP Code Population
Center
83501
Lewiston
83520
Ahsaka
83522
Cottonwood
83523
Craigmont
83524
Culdesac
83525
Elk City
83526
Ferdinand
83530
Grangeville
83533
Green Creek
83535
Juliaetta
83536
Kamiah
83537
Kendrick
83539
Kooskia
83540
Lapwai
83541
Lenore
83542
Lucile
83543
Nezperce
83544
Orofino
83545
Peck
83546
Headquarters
83548
Reubens
83552
Stites
ZIP Code Population
Center
83553
Weippe
83554
White Bird
83555
Winchester
83806
Bovill
83812
Clarkia
83823
Deary
83827
Elk River
83832
Genesee
83834
Harvard
83843
Moscow
83844
Moscow
83855
Potlatch
83857
Princeton
83871
Troy
83872
Viola
99102
Albion
99111
Colfax
99174
Steptoe
99113
Colton
99128
Farmington
99130
Garfield
99161
Palouse
ZIP Code Population
Center
99163
Pullman
99164
Pullman
99179
Uniontown
99347
Pomeroy
99401
Anatone
99402
Asotin
99403
Clarkston
Nez Perce Tribe
Volume II: Current Fish Consumption Survey
December 2016 Page 21
-------
5.5 Stratification and Drawing the Sample
The survey statistical team visited the Nez Perce Tribe on April 3, 2014 to draw the sample for
interviewing, which was carried out on-site in the tribal offices. The Tribe provided the
contractors with a tribal enrollment list of 2,727 adult members. The list contained gender, age,
physical address and mailing address for each tribal member (though a physical address was not
always available).
Members eligible for sampling were determined by first restricting the list to those 18 years or
older and with a physical address ZIP code on the eligible ZIP code list (see Section 5.4.) For
records without a physical address, the ZIP code of the mailing address was used instead. For
records in which both addresses were available, the ZIP codes of the physical and mailing
addresses matched in 2,011 of 2,061 cases, or 98% of them. This close matching supported the
use of mailing address ZIP codes as a surrogate for physical address ZIP codes when needed. Of
the original list of 2,727 adult members eligible for sampling, 68 were missing both physical and
mailing addresses and 1,085 were located outside of the eligible ZIP codes, leaving 1,574
eligible for the sample.
Each eligible member was assigned a unique PMRID (Pacific Market Research Identification
Number). A stratified random sample size of 1,250 was drawn from the 1,574 eligible members,
with strata defined by each combination of gender and age group (18-29, 30-39, 40-49, 50-59,
60+). No other demographic variables were available in the tribal enrollment list. In particular,
fisher status was not stratified on because the fishers list (see Section 4.4) was not available at
the time of the sample draw. Fisher status was determined after members were sampled using the
fishers list. The percentage of each stratum in the population of the 1,574 eligible members was
then determined. The sample size allocated per stratum was determined by multiplying 1,250 by
the population percentage computed for each stratum, thus creating a stratified sample with strata
sizes proportional to the corresponding strata in the original population of interest. The sampled
members were then randomly partitioned into four waves (to be successively allocated to
interviewers approximately every three months) within each stratum. Once a wave of
respondents was released to the interviewers, they could interview any sample member from the
current or any preceding wave. While this expanded access to the waves of respondents may
have introduced a greater possibility of selection bias from interviewer choice of respondents to
approach, it was a necessary step due to the difficulty of locating respondents (Section 5.14,
"Design Changes").
Personally identifying information (PII) was utilized to draw the sample, but all such information
was left with the Nez Perce Tribe after generating the list of sampled members. The Tribe
retained full control of PII and its use for locating respondents
Nez Perce Tribe
Volume II: Current Fish Consumption Survey
December 2016 Page 22
-------
5.6 Questionnaire Development
Every aspect of this survey was designed in an extensive, time-consuming and transparent
collaborative process beginning in the Fall of 2012 and lasting until the Fall of 2016 among the
five tribes in Idaho, the Environmental Protection Agency, two tribal consortia, the State of
Idaho, and a highly skilled and experienced team of expert EPA contractors and sub-contractors.
Efforts were made to incorporate state-of-the-art survey and analytical methods and tribal
cultural and governmental concerns in a study that was designed to contribute to understanding
fish consumption by members of the two tribes surveyed.
This study is unique in that it conducted both the FFQ (food frequency questionnaire) and the 24-
hour recall (NCI) method simultaneously in a survey of tribal consumption of fish over an entire
year. (See Section 5.1, "Overview," for a discussion of the merit of using the NCI method and
the FFQ method together.)
The survey team, in close collaboration with tribal officials and EPA staff, developed an
interview questionnaire to gather information from tribal members to help determine current
tribal FCRs. Questionnaires from several other surveys were reviewed, specifically other Pacific
Northwest regional fish consumption surveys employing a Food Frequency Questionnaire (FFQ)
approach (Suquamish 2000, Toy et al. 1996, Sechena et al. 1999, CRITFC 1994). A draft
questionnaire drew on components of these questionnaires. After several iterations and
refinements, the final FFQ became the critical survey instrument used to ask respondents about
their dietary patterns and activities related to fish consumption over the preceding 12 months.
The questionnaire also covered several other topics, such as demographic characteristics and
changes in fish consumption and access to fishing over time.
Drawing primarily from U.S. national dietary surveys (Johnson, 2013), additional questions were
included in the questionnaire to assess fish consumption during the preceding 24 hours
("yesterday"). These 24-hour recall questions were needed in order to enable use of the NCI
method of determining the distribution of usual fish consumption. At least two independent days
of fish consumption (or non-consumption) need to be assessed for the NCI method. This
requirement was met by conducting two 24-hour dietary recall interviews in addition to the FFQ.
An attempt was made to match the timing of the first and second interview so that the two
interviews would either both be on a weekday or on a weekend day. The reason for matching the
interviews on the period of the week (weekdays or weekend days) was that the matching for
some participants would then yield an estimate of within-person variation in consumption—the
natural day-to-day variation in consumption amount that is independent of the weekday-
weekend. This variation (technical term: within-person variance) is a component that is essential
to and is estimated by the NCI method. Such variation would not generally be affected by other
fixed factors (fixed within an individual), such as age, gender, or whether the two 24-hour
periods are matched, and would also not depend on the specific aspect of fish consumption that
is unique to and differs between weekends and weekdays.
The NCI methodology does provide for (and does include in the modeling) a possible weekend
vs. weekday difference in daily consumption, and the methodology does appropriately handle
data from respondents who have any combination of a weekend and weekday in their two 24-
hour interviews. In the execution of this survey, there was some mixing of weekends and
Nez Perce Tribe
Volume II: Current Fish Consumption Survey
December 2016 Page 23
-------
weekdays for the two interviews. As noted, this mixture is addressed as part of the NCI method
of analysis.
Survey design provided that after first contacting potential respondents through a telephone
screening process, interviewers administered the first 24-hour dietary recall interview and the
FFQ in person to willing participants. The second 24-hour dietary recall interview was intended
for telephone administration from three days up to 4 weeks after the first interview, though a
longer interval was permitted during the later part of the field work. The longer interval was
permitted in order to achieve an increased number of completed second 24-hour interviews and,
thereby, increase the chances of reaching at least 50 double hits to use in the NCI method of
analysis.
Data collected via the questionnaire included fish species consumed, frequency of consumption
and portion size, with additional information gathered about fish parts eaten, preparation
methods and special events and gatherings. Special events and gatherings include ceremonies or
other community events but it was left up to the respondent to decide which events qualified.
Examples of special events include longhouse meals, "Wahlusut" funeral meals, memorials,
potlatches, name-givings and First Fish feast.
With regard to typical portion size and frequency of consumption of a species over the past year,
respondents were allowed to provide this information for a respondent-identified period of higher
fish consumption (along with the respondent's estimate of the period's duration) and for the
balance of the year, a period of lower fish consumption. Alternatively, the respondent could
simply describe a consumption pattern that was relatively constant throughout the year. If two
periods (of higher vs. lower fish consumption) were chosen, the periods may or may not have
coincided with periods of higher vs. lower fish runs and harvest.
Qualitative data were collected regarding both changes in fish consumption patterns as compared
to the past and expectations for future consumption in order to provide additional context around
the quantitative consumption rates. Demographic information was also collected, such as height
and weight (to calculate and check FCRs) and education and income ranges (to determine FCRs
for various population groups). A subset of respondents was reinterviewed by telephone, which
involved asking a subset of the same questions (from the FFQ) a second time. The purpose of the
reinterview was to assess reproducibility.
The FFQ survey questionnaire is presented in Appendix A. The survey team developed this
questionnaire with substantial collaboration, review and input from the Tribe, tribal consortia,
the EPA, discussion with the State of Idaho and review by two Institutional Review Boards
(discussed below in Section 5.16). In addition, the questionnaire was subject to pilot testing,
during which the interviewers tried out the questionnaire on tribal members and provided
feedback to the survey team on any problems with the questionnaire. These pilot interviews were
not used in the analysis for this report. The questionnaire was ultimately transferred to a CAPI
software program on tablets, as described in Section 5.8, to facilitate more efficient and accurate
reporting during the interviews in comparison to the use of a paper questionnaire. The
questionnaire was then used to conduct interviews via CAPI, along with other visual instruments
such as portion models and species identification photographs, as discussed in Appendix B.
Nez Perce Tribe
Volume II: Current Fish Consumption Survey
December 2016 Page 24
-------
5.7 Portion Models, Photos, Portion-to-Mass Conversions
To facilitate questionnaire administration during the survey, interviewers used portion model
displays and species identification photographs (presented in Appendix B). The survey team
selected species and developed these visual representations in collaboration with tribal technical
and cultural staff to reflect the appropriateness of the fish species and preparation methods most
commonly consumed by tribal members.
To aid in accurate determination of portion sizes, three-dimensional (3-D) and two-dimensional
(2-D) model displays were used during the in-person interviews. These models can be broadly
grouped into three types: realistic depictions of the part of an organism consumed (e.g., a fillet),
measures of volume (e.g., bowls of various volumes), or photos of numbers of selected shellfish
species (i.e., crayfish, mussels, and shrimp) consumed. Each interviewer had one full set of
models to bring to the interviews. A set of photographs depicting those same models, printed at
full scale, were left behind with each respondent after the first interview for use during the
follow-up (second 24-hour dietary recall) telephone interview. This allowed respondents to
report portion sizes using the same models consistently throughout the survey.
The survey team developed the following portion model displays for this survey, each of which
included pre-determined serving sizes (as described in Appendix B):
1. A urethane rubber replica of a cooked whole salmon fillet, cut into multiple servings.
2. A flexible plastic replica of a single-serving, cooked trout-like (white fish) fillet.
3. A gray PVC pipe to represent lamprey, marked with portion sizes.
4. A package of salmon jerky to represent dried (or similarly shaped) fish tissue.
5. A set of measuring bowls for different portions of fish soup or volume of fish tissue.
6. Photograph displays of selected shellfish (crayfish, mussels, and shrimp).
Interviewers displayed portion models to respondents in familiar cooked forms (e.g., baked or
dried); however, associated uncooked weights (edible mass) were calculated for application
during data analysis. Each portion model had a specific (unique) code attached to it, and a
separate table was created to show the volume and/or weight per species corresponding to each
portion identified on a display. To maintain interview efficiency, respondents answered the
questions in terms of simple portion marks or codes on each display, saving the interviewer from
having to refer to a look-up table for the species-specific weight of the noted portion. Mass
conversions of each model serving, corrected according to appropriate published moisture loss
factors, were tabulated and used following the interviews to analyze the data and determine
FCRs (see Section 5.10 for FFQ calculations and Section 5.23 for the NCI method, based on the
24-hour recalls). Details of the portion-to-mass calculations are provided in Appendix B.
In addition to the portion models (and the photographs of them which were left with each
respondent), each interviewer had a laminated sheet with illustrations or photographs of each
species to facilitate identification by the respondents, if necessary, during the interviews. The
species identification photographs used to help respondents identify unfamiliar species during
the interviews are also provided in Appendix B.
Nez Perce Tribe
Volume II: Current Fish Consumption Survey
December 2016 Page 25
-------
5.8 CAPI (Computer-Assisted Personal Interviewing)
The survey implementation team explored many modes for data collection. After careful
consideration, the team identified CAPI as the most efficient and best data-collection process for
this survey.
With a CAPI system, the respondent or interviewer uses a computer to answer survey questions.
This is the preferred mode when a questionnaire is long and complex (Groves, Fowler, et al.,
2009) such as in this case, when the in-person portion of the first interview (FFQ plus first 24-
hour recall) lasted over an hour. This is due to the way that computer-assisted interviewing
improves data quality; the computer script increases interviewer efficiency and decreases the
likelihood of human error related to skip-pattern problems (i.e. moving to different sections of
the survey based on the answers to previous questions) or misprinted questionnaires.
Additionally, the CAPI system provides help screens and error checking and messages at the
time of input. This ensures that surveys are completely filled out and enhances the accuracy of
the entered data, decreasing backend data cleaning and processing tasks. Finally, there is no need
to transcribe results.
The survey team selected Confirmit, a globally-recognized leader among online and CAPI
software developers, as the CAPI application because it provides both on-demand resources, via
Software as a Service (SaaS), and on-premises software, two critical requirements for this
project: the survey team used both SaaS and an on-premises product for the interviews. When
interviews were conducted in remote locations without internet or telephone access, the on-
premises application, loaded on the tablets, was integral to the data collection process, allowing
interviewers to conduct interviews and data entry, then synchronizing their data files the next
time their tablets were connected to Wi-Fi.
After the questionnaire was finalized, a programming team built and scripted the computer
version (to be used by the interviewers) within the Confirmit environment. This task, including
thousands of lines of code, was substantial and was reviewed on a daily basis during the initial
programming. All programming reviews were conducted by a programmer who was not directly
involved in this project. After the programmed version was approved by the Lead Programmer
and vetted by the programming review team, it was delivered to the Quality Assurance
Department and the Project Manager for independent review and validation, prior to distribution
to a larger team.
Each interviewer received a Windows 8 tablet for this study. These tablets were selected based
on their reliability, durability, and especially their small and unobtrusive form factor. Not only
was it important that the tablets were easily portable, but also that the technological "footprint"
and the sometimes off-putting nature of a physical barrier between the interviewer and the
respondent were minimized.
Interviewers brought the tablets with them to each in-person interview where the interviewer, not
the respondent, would enter the data. The tablets included detachable screens and keyboards, as
well as touchpad mice and power adapters for AC outlets and car lighters—a necessity in some
rural areas where power was not always guaranteed.
The tablets were password-protected. Survey responses were encrypted and transmitted via
HTTPS to central servers each time a WiFi connection was available and all data files were
automatically removed from the tablets after synchronization with the master database. No
Nez Perce Tribe
Volume II: Current Fish Consumption Survey
December 2016 Page 26
-------
personally identifiable information from respondents was stored either on the tablets or in the
master database.
Confirmit stores data in an optimized database format. Using the Extensible Markup Language
protocol or XML, its database is accessible with many popular software applications. Using
Confirmit's built-in "Export" feature, the data were transferred from the Confirmit database into
a standard SPSS file format (IBM SPSS Statistics, Armonk, NY) in an automated manner. To do
this, Confirmit uses the metadata assigned to all fields when the questionnaire was programmed.
The only configuration needed was to specify certain administrative variables (used internally by
Confirmit—not from the questionnaire itself) to be filtered out of the data file supplied for
statistical analysis. The generated SPSS data file is readable by the statistical software used (see
Section 5.31). This data file contains a row for each respondent or attempted contact and has a
unique ID. Responses to each question in the interview are stored in columns. The testing of
CAPI and verification that data input matches the output is described in the next section.
5.9 Interviewer Recruitment and Training, Pilot Tests
In February 2014, prior to the start of data collection, a widespread recruitment campaign was
initiated by the survey implementation firm to search for local candidates to hire as interviewers.
The contractors worked closely with the Tribes to publicize the survey effort, advertising online,
in the newspaper, on tribal bulletin boards, and using word-of-mouth among the tribal council
and the fisheries and water quality personnel.
Interviewers were required to be currently enrolled members of the Tribes. Tribal interviewers
(rather than non-tribal interviewers) were selected, because it was thought that tribal member
respondents were more likely to accept and open up to an interview from a fellow member of the
Tribe (including accepting a home interview) than from someone outside the Tribe. In addition,
tribal member interviewers would have a very wide network of relatives and friends within the
tribal community—something that it was thought might prove very helpful in locating sampled
members (sometimes the most difficult step) and gaining their cooperation for an interview. The
tribal leadership and staff also expressed, in advance, the importance of using tribal interviewers,
for cultural, capacity-building and economic reasons ((i.e. providing income and new or
additional job skills for tribal members).). That choice was also made in other Pacific Northwest
fish consumption surveys of Native Americans (CRITFC, 1994, Toy et al., 1996, Suquamish
Tribe, 2000).
Applicants were screened on paper and by telephone. Following a successful initial vetting,
acceptable candidates were interviewed in person, after which non-qualified candidates were
culled and a short list of qualified candidates was provided to the tribal councils for review and
approval. As a professional courtesy, the Tribe had "first right of refusal." Candidates who
passed the screening process, the in-person interview, and tribal approval were offered year-long
positions on the project. Qualified and approved applicants were hired by the survey
administration firm (Pacific Market Research) as part-time employees. An experienced Idaho
project administrator was retained to provide supervision, problem-solving and quality assurance
for interviewers, to act as liaison with tribal officials, and to provide general coordination with
the rest of the contractor and governmental team members. The presence of a local project
administrator was key to coordinating all of the efforts and establishing relationships with tribal
staff and officials.
Nez Perce Tribe
Volume II: Current Fish Consumption Survey
December 2016 Page 27
-------
After hiring, the contractor team conducted an extensive training and mentoring process. The
initial training was a full-day session during which the interviewers were presented with the
background of the survey, its purpose, and the development of the questionnaire. The
interviewers were also taught about the project objectives. The contractors briefed the
interviewers on the history of survey research, the guidelines and principles of in-person and
telephone interviews, and the Belmont Report (a document which explains the importance of
human subject protections). The interviewers were also trained to use the technology (i.e.,
computer tablets and associated software) associated with the survey, as well as the various
display models.
Interviewers were taught how to properly screen respondents, how to conduct in-person
interviews, and how to conduct telephone interviews. It was explained that the first (typically
hour-long) interviews would be conducted in person, while the second (20-minute or less)
follow-up interviews would be administered over the phone. The interviewers were taught to
read all questions verbatim without influencing the respondents' answers. They were also taught
how to record all answers exactly as presented to them. The contractors stressed the importance
of maintaining objectivity throughout the entire process, from respondent recruitment and
screening through the final question of the second interview. There was also instruction and an
emphasis on careful and accurate key entry of interview responses into the correct fields in the
CAPI tablets.
The final part of the training included mock interviews with the interviewers and trainers. The
mock interviews required the use of the tablets, interviewing software, and fish models and
photographs. Interviewers were required to complete a mock hour-long interview as well as a
mock follow-up telephone interview before completion of their training.
After the initial, day-long training session, interviewers were required to conduct practice
interviews, either with family and friends or independently. In this way, they familiarized
themselves with the questionnaire, the computer tablet and the CAPI software. After these
practice interviews, the survey team contacted each interviewer to solicit feedback. The
contractors evaluated the data entered to ensure that the interviewers completed the fields
appropriately. Next, the survey team provided "dummy" responses to the interviewers. This
consisted of providing interviewers paper questionnaires with pre-populated data for them to
enter into CAPI as well as conducting in-person meetings with a member of the survey team who
behaved as a sample respondent, answering with the same dummy data. The pre-populated data
in the paper questionnaires included answers specifically developed to support establishing
personas: high consumers and low consumers of fish. The dummy data from the paper
questionnaires and from the mock interviews were entered into CAPI in May 2014.
In June 2014, the Project Manager at Pacific Market Research checked all dummy data entered
against the master file, a key version of the dummy data. If discrepancies were found between
the key and the data entry by any interviewer, that interviewer was notified and required to
correct the errors. Any interviewers who made such errors were required to conduct additional
data entry exercises prior to receiving authorization to "go live."
All of the dummy data output was double-checked to make sure that the values entered in the
CAPI system matched the values produced by the CAPI system. Concurrent with successful
testing, the live interviews with tribal members began. The first live interview was completed on
May 10, 2014 and the last in-person interview included in this report was completed on April 24,
Nez Perce Tribe
Volume II: Current Fish Consumption Survey
December 2016 Page 28
-------
2015. Telephone interviews continued through May 4, 2015 to complete the second 24-hour
dietary recall.
Numbers of completed interviews were tabulated weekly against expected completion rates and
hours expended. At the beginning of the study, interview numbers were low as the survey
administration contractor and interviewers worked through difficulties in obtaining contact
information for selected sample respondents. Weekly calls between all contractors, tribal staff
and EPA staff were initiated to resolve in-the-field challenges in obtaining interviews.
5.10 Calculation of FFQ Consumption Rates
Annual FCRs, which included consumption at special events and gatherings, were computed
based on responses to the FFQ portion of the first interview. Rates were also computed from the
24-hour recalls using the NCI method, described later in Section 5.23. Respondents described
their consumption using portion models to indicate portion size (converted to grams as described
in Section 5.7) and portion frequency (e.g., once per week or two times per month). For each
separate species, respondents were permitted to describe their consumption in two ways: over the
whole year using a single portion size and frequency (constant throughout the year) or over two
different periods of higher and lower fish consumption, which may or may not correspond to
when the specific species was in or out of season. In the case of consumption varying between a
high and low season, respondents would provide portion size and frequency for each of the two
periods separately, as well as the duration of the higher consumption period in days, weeks, or
months. The low consumption season was then calculated as one year minus the fraction-of-a-
year duration of the high consumption season. Stated again for clarity, the duration of high and
low seasons (or designation of only one regimen of portion size and frequency throughout the
entire year) was reported for each individual species consumed.
Note that the higher consumption period duration was entirely up to the respondent to provide
for each species as he or she wished. It was also optional for the respondent to a) mentally
average over the whole year rather than using two periods; or, b) use a single (full-year) period,
if the respondent felt that that was a better approximation to the respondent's consumption
pattern than two periods. For the two-period responses, the duration of the higher consumption
period provided by the respondent may have been shorter than the biological season of the
species or the period may have been longer, for example by preserving fish caught in season and
consuming it over an extended period or a different period based on cultural events. We have not
compared the respondent-reported and the biological season lengths in this report. This
difference may be evaluated in the future. Most responses (80% of the 2,810 per-species
responses from all respondents combined) were provided using a single, one-year period rather
than a pair of higher and lower consumption periods.
The FFQ asked separately about consumption at and outside of special events and gatherings.
The notation for rates in this section is descriptive of the quantity entering into or the result of a
calculation. The total consumption rate in grams/day {Rate Total in the equations here) was
calculated as the sum of the rate which excluded special events and gatherings (Rate Nonevents)
and the rate for special events and gatherings only (Rate Events). Rate Nonevents was
calculated either based on consumption information provided to represent an entire year as a
single period, (Rate Nonevents Whole) or by combining annualized rates of consumption during
a higher consumption period (Rate Nonevents Higher) and the consumption rate in the
Nez Perce Tribe
Volume II: Current Fish Consumption Survey
December 2016 Page 29
-------
remaining lower period {Rate NoneventsLower). Each of these rates were calculated per
species first, then species-specific rates were summed together to produce species-group rates
(see Section 5.11 for definitions of species groups).
If the respondent reported consumption over the whole year as a single period (rather than
varying during the year), the FCR (g/day), excluding consumption at special events, was
determined by the following equation:
Rate_Nonevents_Whole = SIZE_Nonevents x FREQ_Nonevents, (1)
where:
SIZE Nonevents = total portion size in grams (determined based on the portion
model used by the respondent, the portion-to-mass conversion factor for the
combination of the portion model and species, and the number of portion units
consumed; see Q19 in the questionnaire in Appendix A)
and:
FREQ Nonevents = number of portions consumed per day, which may be
converted to a daily amount from the number of portions reported per week, per
month or per year (Q18 in the questionnaire).
Any frequency per week was converted to frequency per day using 7 days/week. Any frequency
per month was converted to frequency per day by dividing by the factor 365/12 days/month. Any
frequency per year was converted to frequency per day by dividing by the factor 365 days/year.
Of note, the year preceding any interview in the survey did not overlap a leap year.
If the respondent reported consumption over two periods (higher and lower consumption), the
rates (non-annualized) for each period were computed in the same way as equation (1), above.
The two rates were then annualized and combined using the following equation:
Rate_Nonevents =
%HIGH x Rate_Nonevents_Higher + %LOW x Rate_Nonevents_Lower, (2)
where:
%HIGH = the length of the higher consumption period expressed as a
proportion of the year (Q22 in the questionnaire);
%LOW= the length of the lower consumption period expressed as a proportion
of the year (%HIGH + %LOW= 1);
Rate Nonevents Higher = consumption rate in g/day during the higher
consumption period (portion frequency and size came from Q20 and Q21,
respectively);
and,
Rate Nonevents Lower = consumption rate in g/day during the lower
consumption period (portion frequency and size came from Q23 and Q24,
respectively).
Nez Perce Tribe
Volume II: Current Fish Consumption Survey
December 2016 Page 30
-------
The higher-period duration was reported in either weeks or months. Weeks' duration of a high-
consumption season were converted to a proportion of a year by multiplying by the factor 7/365.
Months' duration of a season were converted to a proportion of a year by multiplying by the
factor 1/12.
For special events and gatherings, respondents were asked only about suckers and whitefish (as a
single group), salmon and steelhead (all species combined), resident trout (all species combined)
and sturgeon. This selection of species and groups was done through consultation with both the
Nez Perce and Shoshone-Bannock Tribes, who noted that a more limited set of species were
consumed at special events, and was further motivated by the desire to reduce respondent
burden. For each of these four species/groups, the corresponding FCR (g/day) was computed as
Rate_Events = EFREQ x % EVENTS x SIZE_Events, (3)
where:
EFREQ = number of events per day (converted from the number of events per
week, month, or year; Q31 in the questionnaire in Appendix A);
%EVENTS = proportion of events where the given species is consumed (Q34);
and,
SIZE Events = total portion size in grams (based on the model and units chosen in
Q33 and the standard portion-to-mass conversion routine described in Section 5.7).
The final individual FCR (g/day), which also includes consumption both at and outside of special
events and gatherings, is determined using the following equation:
Rate_Total = Rate_Nonevents + Rate_Events. (4)
As Rate Nonevents was calculated for each individual species (e.g. chinook, coho or sockeye
salmon) while Rate Events was calculated at the group level (e.g. all salmon and steelhead
combined), Rate Nonevents in equation (4) was first aggregated to the group level by summing
individual species rates as appropriate before the summation with Rate Events.
5.11 Species Groups
The species groups included in this report (Table 2) were determined jointly by the Nez Perce
Tribe10 and EPA staff, with the Tribe making the final decision. EPA staff provided guidance on
EPA policy as to what species might be included in developing FCRs that are relevant for
ambient water quality criteria to protect human health.
The Nez Perce Tribe decided that from a water quality standard development perspective, the
appropriate grouping of fish to focus on in this report should include near coastal, estuarine,
freshwater and, in particular, anadromous species (Group 2). Inclusion of anadromous species in
the FCR used to develop AWQC is a policy option that the EPA has made available to states and
tribes (US EPA, 2013). In Oregon, anadromous species are included in the FCR used for that
state's AWQC (Oregon DEQ, 2011). Anadromous species are also currently included in the FCR
10 Email from Joe Oatman to Nayak Polissar (and others) on June 26, 2015, conveying an email from Marlene Trumbo documenting
the Nez Perce Tribal Executive Committee (NPTEC) decision on species groups to be reported.
Nez Perce Tribe
Volume II: Current Fish Consumption Survey
December 2016 Page 31
-------
used for Washington's proposed AWQC (Washington Department of Ecology, 2015). The Nez
Perce Tribe wished to report on total fish consumption (Group 1).
The species included in the groups used for reporting FCRs are described in detail in Table 2.
Group 2 contains Groups 3-5 and part of Group 6. Groups 3-7 are mutually exclusive groups
which completely cover Group 1. During interviews, individual species consumed were named
by the respondent based on their personal knowledge, species photographs (Appendix B) and
discussion with the interviewer; the respondent's final identification was accepted. In particular,
respondents differentiated between freshwater clams and mussels and marine clams and mussels.
In the case of freshwater clams and mussels, some respondents harvested the shellfish
themselves or knew the difference based on appearance. Only 4% of respondents reported
consuming freshwater clams or mussels while 31% reported consuming marine clams and
mussels. Of note, Groups 1 and 2 contain all shellfish species, so this distinction between
freshwater and marine does not affect those groups.
Table 2. Nez Perce Tribe. Species groups.
Species
Group
Description
Species and Species Groups Included
Group 1
All finfish and shellfish
All species in groups 3-7 (these groups are mutually exclusive)
Group 2
Near coastal, estuarine,
freshwater and
anadromous finfish and
shell fish
All species in groups 3, 4 and 5; lobster, crab, shrimp, octopus, oysters,
geoduck, razor clam, bay mussel, scallops, and other marine clams or
mussels
Group 3
Salmon and steelhead
Chinook, coho, sockeye, kokanee, steelhead, chum, pink, Atlantic and any
unspecified salmon species
Group 4
Resident trout
Rainbow, cutthroat, cutbow, bull, brook, lake, brown, bottoms, golden and
any unspecified trout species.
Group 5
Other freshwater finfish
and shellfish
Lamprey, sturgeon, whitefish, sucker, bass, bluegill, carp, catfish, crappie,
sunfish, tilapia, walleye, yellow perch, crayfish, freshwater clams or mussels
and any unspecified freshwater species
Group 6
Marine finfish and
shellfish
Marine finfish (cod, halibut, pollock, tuna, herring, sardines, mackerel, mahi,
orange roughy, red snapper, seabass, kipper, wahoo, yellowtail and shark),
marine shellfish (lobster, crab, shrimp, octopus, squid, oysters, geoduck,
razor clam, bay mussel, scallops, and other marine clams or mussels) and
any unspecified marine finfish or shellfish
Group 7
Unspecified finfish and
shellfish
Any response where the species was not specified sufficiently to be placed
into groups 3, 4, 5 or 6
Note: There is overlap between the species in Group 2 and Groups 3-6. Group 2 used in this report has been revised
from the Group 2 species list presented in a draft interim report of this survey. Species selection for group 2, as
presented in this report, was informed in part by the habitat proportions listed per species in U.S. EPA, 2014, Table
1. In particular, the marine species in Group 2 were considered likely to be near coastal or estuarine.
Nez Perce Tribe
Volume II: Current Fish Consumption Survey
December 2016 Page 32
-------
5.12 Demographic Groups
Group 1 (all fish) consumption rates were computed by population demographic groups defined
by variables available from the enrollment file and the survey questionnaire The enrollment file
was used to define groups based on gender, age, and whether or not the respondent was a
documented fisher as determined from the Nez Perce Tribe fishers list (see Section 4.4). The
questionnaire was used to define groups based on whether the respondent lived on- or off-
reservation, the number of persons resident in the respondent's household, and the respondent's
education and income levels.
5.13 Response Rates
Response rates were calculated according to standard definitions of response rate (AAPOR,
2011). The following specific form of the response rate was calculated:
RR1 = I / [ (I + P) + (R + NC + O) + U ]
where:
I = The number of complete interviews
P = The number of partial interviews
R = The number of refusals and break-offs
NC = The number of eligible sampled members not contacted
O = The number of other eligible non-respondents
U = The number of non-respondents with unknown eligibility
For this survey the use of the RR1 equation is equivalent to the following formulation:
RR1 = I / (N - X)
where N = the size of the originally selected sample and X = the number of members found to be
ineligible after contacting or attempting contact. A completed interview, which contributes to the
numerator of the response rate calculation, was defined as one where the respondent either: 1)
responded to the screening interview or the FFQ items sufficiently to be classified as a non-
consumer (Q3-Q6 of the questionnaire), or 2) completed the full first interview (after the
screening interview) with the FFQ items completed and provided enough information to support
calculation of an FFQ consumption rate. To satisfy the second condition, a respondent did not
need to answer every question but needed to reach the end of the questionnaire. Note that this
definition allows for respondents who sufficiently answered the screening interview to be
classified as consumers (Q3-Q6) but who did not go on to complete the full interview. This
means that the number of known consumers in the survey is higher than the number of
respondents with known FFQ consumption rates.
An ineligible member, who reduces the denominator of the response rate calculation, was
defined as a sampled member who was: 1) found to live outside of the eligible ZIP codes,
2) found to be employed as a tribal interviewer involved in the survey, or 3) deceased,
institutionalized or impaired. The term "institutionalized" included prospective adult respondents
who, at the time of the survey, lived in a setting where they had little or no control over their
Nez Perce Tribe
Volume II: Current Fish Consumption Survey
December 2016 Page 3 3
-------
diets. For example, residents of long-term care facilities, hospice (not in-home), and prison
would be classified as institutionalized.
Not all sampled members were contacted, and therefore the eligibility or ineligibility of every
sample member could not be determined. This measure of response rate is thus conservative (too
low) in the sense that its value is reduced by the presence of sampled members who are ineligible
but presently unknown to be ineligible. Ineligible members whose ineligibility was unknown to
the survey team would include, for example, deceased members whose enrollment records had
not yet been updated or members who recently moved out of the eligible ZIP code area and
whose residence address differed from the address of record at the time the enrollment files were
used to draw the sample. A count that is unknown to the survey staff is the number of sampled
tribal members who were ineligible but were not known to be ineligible. If this number was
known, it could be included in the response rate calculations, and the response rate would be
higher than that reported here.
5.14 Design Changes
As the survey progressed, a number of issues became evident. It was found that the contact
information found in tribal enrollment records was not as accurate as had been hoped, requiring
sometimes significant and creative research to locate potential respondents. The time required for
interviewers to travel to respondents' homes and conduct interviews was also much greater than
expected, and some interviewers encountered difficulty in conducting interviews at tribal
members' homes. Finally, the fraction of individuals agreeing to be interviewed was also lower
than expected. All of these factors led to a lower-than-expected rate of interview acquisition and
concerns about attaining an appropriate number of interviews to assure a credible result.
To address these issues, several design changes were adopted partway through the interviewing
period to increase the number of interviews completed and improve the chances of meeting the
sample size goals for the NCI method. The first of these design changes was to permit the
interviewers to attend special events11 (e.g., tribal meetings and powwows) and recruit and
interview attendees for interviews during the events—still drawing potential respondents only
from the list of tribal members selected into the sample. As part of this design change,
interviewers were permitted to draw respondents from any of the four sample waves of members.
Ultimately, this did lead to the interviews not being spread evenly across the year, as originally
designed. However, during the analysis phase of this project, the wave structure was considered
to be less important, because seasonality was not found to be a factor that was present in reported
consumption rates in a manner that could be included in the analysis. (See the analyses of
seasonality presented in Section 5.23.2.1.) The ability to detect seasonality may have been
limited by the small number of interviews conducted during the peak harvesting period. Due to
the limited number of interviews during the peak season, the finding of no compelling
seasonality during the data analysis does not constitute a demonstration that there is no seasonal
variation in fish consumption.
11 The Nez Perce culture and traditional practices involve regular and time-specific ceremonies. Today, these "special events" are a
continuation of tribal customs and cultural practices, and provide an opportunity to maintain those traditional values and teachings.
These gatherings often have a large attendance, and such occasions served as an opportunity to make contacts and complete
interviews.
Nez Perce Tribe
Volume II: Current Fish Consumption Survey
December 2016 Page 34
-------
As part of their activity at these events, the interviewers were also permitted to schedule interviews
at a later time (after the special event). Thus, the special events provided an opportunity not only
for on-site interviewing, but also to arrange additional interviews later on. After the special event,
the criteria reverted to respondent recruitment only from the wave of members assigned to the
specific calendar period. However, interviewers were also permitted to conduct interviews of
members from the sample list (any wave) whom they might encounter by chance.
As expected, the design change noted above greatly increased the acquisition of completed
interviews. Any adult willing to be interviewed at a special event was likely part of the sample
roster, as approximately 80% of the eligible adults in the Tribe were included in the sample.
A second design change occurred after the EPA and the contractor team received limited,
conditional approval from the Tribe to expand the interview team to include non-Nez Perce
interviewers, which allowed interviewers from the EPA and other tribal organizations to assist
the Nez Perce interview team. These individuals received the same training and instructions that
the Nez Perce interview team received. The non-Nez Perce interviewers were also permitted to
draw respondents from any wave of members. Non-tribal interviewers visited the Tribe in
December, 2014 and March-April, 2015 and interviewed eligible members from the sample list
that were tribal government employees. During the December 2014 visit, non-tribal interviewers
also attended a tribal holiday event and interviewed additional sampled members (not necessarily
Tribal employees). The interviews by the non-tribal interviewers were conducted in the
respondents' offices or other Tribal or commercial venues, rather than respondents' homes.
Tribal employees were offered paid time off, by the Tribe, to participate in the survey. In rare
cases, the interviewing supervisor, a non-tribal member, scheduled or completed interviews with
tribal members who were not also tribal employees, but these instances were rare, and all were
pre-approved by the Nez Perce Tribe. Nearly all first contacts with prospective respondents were
made by tribal interviewers or the interviewers' supervisor, who had developed exceptional
rapport with tribal members. While non-tribal interviewers assisted in interviewing some of the
easier-to-contact tribal members (i.e. employees whose work phones and work addresses were
known), tribal interviewers were freed up to pursue many more of the hard-to-reach respondents,
people who a non-tribal interviewer would have difficulty finding, contacting and convincing to
complete the interview. The scope of work for the non-tribal interviewers was limited, but their
overall efforts were crucial to the success of the project.
An additional change—though not a design change—was increased coordination in scheduling
of interviews. The interviewers' supervisor (from the contractor team) worked more closely with
the interviewers to assist them in arranging interviews. In addition, tribal staff and tribal leaders
played a more prominent role in establishing the framework to coordinate between interviewers,
supervisor, and tribal staff to schedule interviews at special events and on tribal property.
Nez Perce Tribe
Volume II: Current Fish Consumption Survey
December 2016 Page 3 5
-------
5.15 Reinterviews
A sample of respondents who completed the first interview were sampled to be re-interviewed
using a short list of questions related to fish consumption. The goal of the reinterview was to
compare the original and reinterview responses to assess reproducibility.
The reinterview questionnaire is contained in Appendix A. The questions cover the frequency of
consumption of Chinook salmon, the species with the largest number of consumers among the
survey respondents. Additional species were not specified to limit the total burden on
respondents and the duration of the reinterviews. Additional questions in the reinterview cover
changes in overall fish consumption and the number of people living in the respondent's
household. Responses to corresponding questions in the original and reinterview were compared
descriptively using means, standard deviations and Spearman's correlation coefficient.
The reinterviews were conducted from April 3 to June 12, 2015 by the Pacific Market Research
interview supervisor, a non-tribal interviewer. The survey statistical team provided the
interviewer with a list of respondents who were originally interviewed within the last 2 months
to help select respondents. The list was refreshed every two or three weeks with recent
interviews. To help ensure a balanced sample, the list was partitioned into 6 groups, defined by
gender and Chinook consumption. For each gender, Chinook consumption was divided into three
equal-sized groups using tertiles. The target was 30 reinterviews total, with 5 from each group.
The interviewer was aware of the groups but was not aware that the groups were defined by
previously reported consumption levels. The interviewer was instructed to carry out reinterviews
from each group (e.g., high-consumption females) until five reinterviews in the group were
completed.
Over the course of 2 months, 81 respondents were identified for possible contact for a
reinterview, of which 67 (83%) had at least one contact attempt. (There was no requirement to
contact or attempt to contact all respondents on the list.) Thirty-one reinterviews were
completed. The target was 30 but an extra interview was performed (and used in the reinterview
analysis), because—on the first interview—one respondent did not provide a complete response
regarding Chinook consumption (the duration of the high consumption period was missing),
necessitating imputation. The imputed value was retained for comparison to the reinterview
value, since such imputations have been used to present the results of this survey. Of the 36
respondents with a reinterview attempt but no completion, there were no (zero) refusals, 22
respondents did not have a valid phone number recorded, five did not respond after the
maximum number of contact attempts was reached and 9 had 1-6 contact attempts (median: 2)
before the reinterview quota was reached and no further attempts to contact these respondents
were needed.
5.16 Reviews and Approvals
The survey team developed a Survey Design Report in 2014 in extensive discussions and
collaboration with the Nez Perce Tribe and the EPA that outlined the approach and procedures
for implementing the fish consumption survey. The Coeur d'Alene, Kootenai and Shoshone-
Bannock Tribes of Idaho, the Columbia River Inter-Tribal Fish Commission (CRITFC) and the
Upper Snake River Tribes Foundation (USRTF) also reviewed and provided input to the survey
design based on similar design reports that were submitted to them. Staff from the State of Idaho
Nez Perce Tribe
Volume II: Current Fish Consumption Survey
December 2016 Page 36
-------
Department of Environmental Quality also participated in design discussions that paralleled their
own survey development efforts. The coordination with DEQ was implemented in order to
ensure that data collected by the tribal surveys would be of utility in Idaho's efforts to revise
State ambient water quality criteria. Progress on the survey was reported on a regular basis to the
Negotiated Rulemaking process run by the State of Idaho.
In order to meet accepted standards of protection for survey respondents, the Survey Design
Report was submitted for review and approval to two Institutional Review Boards (IRBs) and the
EPA Human Subjects Research Review Official (HSRRO), the latter of which has the final
authority for all human subjects research supported by the EPA.
First, the Northwest Indian College (NWIC) IRB reviewed the design protocol, suggested
modifications to the survey questionnaire to ensure protection of tribal respondents, and gave
"consultative approval" for the survey to proceed on March 14, 2014. The design team felt that it
was important to include an IRB associated with Native American tribes in order to fully assess
any issues the research might pose for unique Native American cultures. Subsequently, Quorum
Review IRB reviewed the design protocol, including revisions made according to the NWIC IRB
recommendations, and issued a "notice of exemption determination" on March 26, 2014
acknowledging that the survey met the criteria for protection of human subjects' personally
identifiable information and did not require further review or restrictions. Quorum IRB was the
official IRB on record for the survey, since the NWIC IRB played a consultative role. Finally,
the EPA HSRRO reviewed the design protocol and supporting documentation, including the IRB
letters, and approved the survey design. Ultimately, the Nez Perce Tribe gave final approval for
the survey to proceed.
The survey was implemented largely consistent with the methods as described in the final survey
design document. Some modifications to the design—in a manner that would not bias the
survey—were implemented during the field work to increase the response rate, as described
above in Section 5.14.
A version of this report was submitted to a four-person peer review committee on July 30, 2015
for a letter peer review. The four reviewers included: a statistician who co-developed the NCI
method and who had extensive experience in dietary surveys; a dietician and nutritionist
involved in monitoring and assessing food consumption and related behavior of the U.S.
population; a professor of nutrition involved in designing and evaluating dietary surveys; and a
researcher in food sciences involved in methodological aspects of dietary intake assessment. The
four reviewers each evaluated the report independently and submitted their reviews to the peer
review contractor, who summarized the reviews and also included them verbatim in the peer
review report. The charge to the reviewers asked them to consider all major aspects of the
design, analysis and reporting of the survey. The peer reviewers' comments were returned at the
end of August, 2015. The current version of the report includes the contractors' revisions in
response to the peer reviews and in response to additional internal reviews from the EPA, from
the two tribes participating in the current fish consumption survey, and from two tribal consortia,
(see Section 5.17.1).
Nez Perce Tribe
Volume II: Current Fish Consumption Survey
December 2016 Page 37
-------
5.17 Internal Reviews
5.17.1 Review by the Tribe and Other Organizations
A design report containing planned procedures was prepared for review by the Tribe, as well as
by two affiliated tribal organizations (Columbia River Inter-Tribal Fish Commission—
CRITFC—and the Upper Snake River Tribes Foundation—USRTF), the EPA, SRA (the
contracting organization managing multiple related contracts for the EPA), and Ross Strategic.
These Tribe and organizations provided feedback or approval, and their suggestions were
addressed or considered in preparation of a final design document.
A draft interim report was provided on April 27, 2015 to and was reviewed by the two Tribes
participating in the current fish consumption survey—the Nez Perce and Shoshone-Bannock
Tribes. The draft interim report included analysis only from the FFQ data collected during part of
the survey year. The report was also provided to and reviewed by the CRITFC and USRTF tribal
organizations, as well as the EPA and two organizations closely involved in the work effort:
SRA and Ross Strategic. The feedback from these reviews played a role in the version of the
draft interim report, dated May 12, 2015, and the benefits of those reviews have carried forward
into the current analysis and report. The May 12, 2015 report was submitted specifically to aid
the State of Idaho in its rulemaking effort.
A revised draft report was issued on July 15, 2015 for internal review by the Tribes, tribal
organizations, EPA and the contractors. The July 15, 2015 report included analysis of both FFQ
data and data from the 24-hour dietary recalls—analyzed by the NCI method. The various parties
offered comments, which the contractors used to prepare the next major version of the report.
That version was submitted to a peer review committee on July 30, 2015 (see Section 5.16), and
the same version was reviewed by the Tribes, tribal organizations, EPA and the contractors, who
also reviewed versions issued on September 21, 2015, and September 25, 2015. The contractors
considered the feedback from each wave of reviews in producing each subsequent version of the
report, including the present version.
5.17.2 Review of Statistical Computing
Two statisticians separately implemented the calculation of the FCRs per respondent, for all
species combined (total consumption rate), all reported species groups (see Section 5.11) and
also for each of the 45 pre-specified species and species group used in the survey questionnaire.
The calculations include the consumption rate formulas described in Section 5.10 and the
imputation of missing values as described in Section 5.28. All of these consumption rate values
were compared between the two statisticians' implementations of the rate calculation
methodology. Any differences found were discussed (without comparing codes), after which
each statistician modified their code independently until there was complete agreement for all
respondents and all species.
5.18 Overview of Statistical Analysis
The description of the statistical analysis methods in the following sections is extensive and
covers a number of topics, including:
• definition of fish consumers vs. non-consumers (which may vary across the more
frequently to less frequently consumed species groups);
Nez Perce Tribe
Volume II: Current Fish Consumption Survey
December 2016 Page 3 8
-------
• handling of missing values in the FFQ responses about consumption—a methodology
which avoided excluding some respondents' consumption records, which were nearly but
not entirely complete;
• sampling probabilities and their adjustment for non-response for use in statistical
weighting with the intent of providing estimates for the target tribal population;
• evaluation of the impact of design changes, including interviewing at special events and
non-tribal interviewers, as well as home vs. non-home interviews;
• confidence interval calculations based on the non-parametric bootstrap using replicate
weights, which provided robust estimates of the precision of consumption rate means and
percentiles; and
• the NCI method, a complex and flexible modeling approach that was applied to the 24-
hour recall responses to estimate consumption rate distributions—in addition to those
provided from the FFQ data on estimated consumption over the preceding year
Consumption rates in this report are generally presented to one decimal place, e.g., 70.1 g/day.
While the true level of precision of a particular rate may not warrant the one decimal place, that
format has been used for four reasons. First, in some cases, for very low consumption rates, e.g.,
1.6 g/day, rounding to an integer (which would be 2 g/day, in the example) would sometimes be
an unacceptable loss of information. Second, users of this report may sometimes carry out
calculations based on the rates reported here, and the one decimal place may sometimes improve
the precision of those derived calculations. Third, stylistically, tables with internally varying
numerical formats are more difficult for some users to read and scan than a table with a
consistent numeric format. Finally, if the format of the rates is intended to truly and consistently
represent precision for every rate presented, then, onerously, each and every rate would need to
be considered separately for possible rounding, and that rounding could extend to the unit, tens
or hundreds digits, as well as being differential rounding for each individual rate. For example, in
one case 43.6 g/day might need to be rounded to 40, while in another case it might be rounded to
44 g/day, and in yet another case, it might need to be preserved in all its specific digits: 43.6
g/day. Thus, though the format of a particular rate might be more precise than warranted in some
cases, the magnitude of the rate is apparent and meaningful, and it would be rare in this study to
have the numeric format interfere with any comparison among rates.
5.19 Sampling Probabilities
The sampling probabilities (or sampling fraction) for each stratum were calculated as the number
of the sampled tribal members in a stratum divided by the number of tribal members in the same
stratum. Section 5.20 describes how the sampling probabilities were modified to produce
statistical weights used in calculating most results presented in this report.
5.20 Non-Response Adjustments to Weights
Completed interviews with useable responses for consumption rate calculations (or with a
determination that the respondents never consumed fish) were not available for all sampled tribal
members. If it could be assumed that non-response to the survey was completely random—for
example, not dependent on sampled members' gender, age or other characteristic—then the
original sampling weights (based on strata only, and calculate as the inverse of the sampling
fraction per stratum) could be used without leading to any bias. However, that assumption is
Nez Perce Tribe
Volume II: Current Fish Consumption Survey
December 2016 Page 3 9
-------
often not valid and was not made here. The sampling weights were therefore adjusted for non-
response using characteristics available from the enrollment file and fisher list.
The terms "responder" and "non-responder" are used in this section and at other locations in this
report. Responders were defined as sampled tribal members who were interviewed and were
determined to be either fish consumers or fish non-consumers. In contrast, sampled tribal
members that were either not interviewed or were interviewed but could not be determined to be
either fish consumers or fish non-consumers, were designated as non-responders. Both terms
"responder" or "non-responder" are not to be confused with the generic term "respondent" that
simply means a survey participant who may be referred to in the particular topic being discussed
or whose data were used in the analysis being presented.
The non-response adjustment is used to adjust the probability of being sampled from the tribal
population—i.e., to adjust the "sampling probability." The sampling probability (Section 5.19) is
the starting point—a quantity used in creating appropriate statistical weights. It is adjusted by
taking account of the probability of a sampled tribal member actually becoming a responder to
the survey. That probability of survey response, in turn, is calculated in relation to demographics
of the sampled tribal members. The goal is to adjust for potential bias due to differences among
responders and non-responders and to yield better (usually less biased) estimates of the
population value of a statistic, such as a mean. A respondent's sampling weight W(used for
statistical analysis) was calculated as the inverse of the product of: (a) the sampling fraction in
the respondent's stratum and (b) the estimated probability Pr of being a respondent
("response probability") for a tribal member with the respondent's specific characteristics (e.g.,
age, gender, etc.):
W = 1/(FS *Pr)
Response probabilities (Pr) were calculated using multivariate logistic regression (Hosmer and
Lemeshow, 2000) for survey response among sampled tribal members, using available
demographic characteristics. The response probabilities are, thus, a multivariate function of a
number of demographic characteristics. Available demographic characteristics from the
enrollment files used to draw the sample or from other sources included:
age group, gender, ZIP code group (83540, 83536, 83501, Other ZIP codes), and fisher indicator
(on vs. not on the fisher list).
Logistic regression models for response were selected using the Hosmer-Lemeshow goodness of
fit statistic (Hosmer and Lemeshow, 2000). The selected models included:
age group, ZIP code group (83540, 83536, 83501, Other), and fisher indicator.
The same weights that were developed per respondent were applied to all weighted analyses
(including the analysis of the FFQ and 24-hour recall consumption data).
Replicate weights from bootstrap re-sampling (1,000 resamples) were used to calculate the
variance estimators (standard errors, confidence intervals, p-values). See the section on replicate
weight calculations, below, for more detail.
Nez Perce Tribe
Volume II: Current Fish Consumption Survey
December 2016 Page 40
-------
5.21 Consumer/Non-Consumer Determination (Overall and per Species)
The analysis included a determination of whether respondents were either fish consumers or fish
non-consumers using screening questions in the CAPI (screening interview questions 3-6, see
Appendix A). These questions asked the respondents sequentially whether they consumed fish
yesterday, last week, last month, or in the past year. Only respondents who reported consuming
fish in response to the screening questions were further interviewed using the FFQ. Any
respondent who did not report consuming fish on the FFQ, despite reporting consumption in
response to the screening questions, was categorized as a non-consumer. Consumers of any other
designated species group (see Section 5.11) were identified using only the FFQ responses;
respondents were considered consumers of the species group if they reported consuming any of
the applicable species during the preceding year, including consumption at special events and
gatherings. All analyses (FFQ analysis, naive and NCI methods for the 24-hour recalls) were
limited to the consumers of the relevant species group according to this designation.
5.22 Mean, Variance and Percentile Methods for non-NCI analyses
Estimates of means, variances and percentiles were carried out using standard survey estimate
methods implemented in the R survey package (Lumley, 2014 and Lumley, 2004). For the
estimates of the percentiles, the package uses a method described in Francisco and Fuller's 1986
(Iowa State University) technical report, Estimation of the Distribution Function With a Complex
Survey. The survey package also enables inference (estimation of means, variances, percentiles,
percentages) in specific groups. When estimating quantities in sub-populations the methodology
accounts for the uncertainty in the weights derived for a specific sub-population. The
methodology is further described in Lumley, 2010.
The survey estimate method applied to the 24-hour recall data is referred to as the "naive"
method. For each respondent providing data for a naive method calculation, the respondent's one
or two 24-hour recall consumption rates were averaged and the naive method was applied to the
per-respondent averages. (For a respondent with only one 24-hour recall, the "average" is the
single consumption rate itself—for the species or species group considered.) The method is
"naive" in that it does not account for the variability of recalls within a respondent or other
complexities of the 24-hour recall data (such as the weekend effect, the effect of the interview
number—first vs. second interview—or the impact of other variables that may cause a difference
between fish consumption during the first vs. second 24-hour period). The naive method was
utilized primarily for a methodologic comparison of the differences between the FFQ and 24-
hour recall consumption rates and it was limited to the estimation of means.12 The percentile
estimates for the upper and lower tails of the distribution of fish consumption, if they are
calculated from the naive method, do not account for the within-person, day-to-day variation in
fish consumption. Those tail percentile estimates tend to be biased, with overestimated
percentiles in the upper tail and underestimated percentiles in the lower tail (see Dodd, 2006).
The NCI method, which is based on the 24-hour recall data, could not be used for the analysis of
species Groups 3-7 due to the smaller number consumers of each of these species groups (than
for Groups 1 and 2) and the associated insufficient number of "double-hits" needed for the NCI
12 A more extensive comparison of FFQ and 24-hour recall data was carried out and the methods and results are described in
Section 6.11.
Nez Perce Tribe
Volume II: Current Fish Consumption Survey
December 2016 Page 41
-------
method. Thus, the naive method was carried out to estimate mean fish consumption rates for
species Groups 3-7—to be compared the means calculated by the FFQ method.
5.23 NCI Method
5.23.1 Overview
The NCI method (Dodd, KW, et al. 2006; Tooze, JA, et al. 2006; Kipnis V, et al. 2009) was used
to estimate the distribution of usual fish consumption from the 24-hour recall data. Compared to
the consumption reported on the FFQ, 24-hour consumption would be expected to have a smaller
recall bias. The 24-hour assessment refers to consumption "yesterday" while the FFQ asks about
typical values of consumption for the preceding year. For this survey, the grams consumed
"yesterday" were calculated from the responses to Q10 from the questionnaire (the question
number is the same for both recalls; see Appendix A) using the standardized portion-to-mass
conversion described in Section 5.7. The analysis of reported 24-hour consumption, however,
presents analytical challenges. The main analytical features of the NCI method for analysis of
fish consumption are described in Polissar et al., 2014. Points (1) to (8), below are adapted (and
extended for application in the present context) from that document.
The NCI method involves fitting a model for usual intake (grams/day) of a commodity, such as
fish, based on data from a survey with reported consumption on two or more days. The mean and
percentiles of consumption are estimated from a derived distribution of usual intake, which is
part of the fitted model. The model assumes:
(1) There is an underlying distribution of true usual intake for the population being
studied. The true intake for a given person might be thought of as their average
daily intake—averaged over the course of a year, often reported as grams per day.
The usual intake for a person does not have the ups and downs that occur with
intake for any given day; the usual intake is a single number for each person. This
usual, average or "true" intake would typically vary from person to person in the
population. The set of values of usual intake would typically have relatively few
people at very low or very high values of intake and relatively more people in
between.
The set of usual intake values for a population do not have to form a "bell-shaped
curve," but the true distribution, it is assumed in the NCI methodology, can be
transformed to the normal (bell curve) distribution in a fairly flexible manner,
specified by the methodology. (It is noted that fish consumption distributions tend
to be skewed toward large consumption values and can often be approximated by
the lognormal distribution; this phenomenon is consistent with the
"transformation-to-the-bell-shape" assumption here.)
(2) There is day-to-day variation in how much a person consumes of a commodity—
on days when they do consume. The daily consumption varies around their usual
intake.
The estimate of the day-to-day variation is a critical part of the NCI model and
requires a substantial number of respondents that report consumption on two days
("double-hits"). The ability to run the NCI model is directly impacted by the
Nez Perce Tribe
Volume II: Current Fish Consumption Survey
December 2016 Page 42
-------
number of available double-hits, with considerations for this study noted as
follows.
The numbers of double-hits for species Group 1 (all finfish and shellfish
species) and for species Group 2 (near coastal, estuarine, freshwater and
anadromous species) were small in the two tribes involved in the fish
consumption survey: 43 double-hits for the Nez Perce Tribe and 8 for the
Shoshone-Bannock Tribes for Group 1 consumption, and 28 for the Nez
Perce Tribe and 3 for the Shoshone-Bannock Tribes for species Group 2
consumption. Thus, an NCI-method model for each species group was fit to
data from both tribes combined13. The NCI method allows the use of
covariates, which are factors (or "variables") influencing consumption—
more specifically, influencing the distribution of usual consumption. (See
items 6-8 below for a more extensive description of the covariates and their
role.) Covariates were introduced into the models in order to capture
differences between the two tribes in the likelihood to consume fish on a
given day and in the amount consumed on a day when fish consumption
occurred. Use of these covariates allowed estimation of tribe-specific
distributions of usual fish consumption. A substantial number of
respondents with Group 1 consumption on at least one of two 24-recall days
were available to enable the inclusion of covariates into the model (179 NPT
respondents and 56 SBT respondents with fish consumption on at least one
of the two 24-hour recall days). The number of respondents was smaller for
Group 2 species: 145 NPT and 31 SBT respondents with at least one fish-
positive 24-hour recall for Group 2.
As a sensitivity analysis to the primary NCI models that used data for the
two tribes together, NCI models were also run for the NPT only. The small
number of double-hits for the SBT did not allow fitting an NCI model for
the SBT only. The combined-tribes model results are presented in this
report, since, under certain assumptions, they are expected to be more
precise than results from a model based on only one of the Tribes.
(3) Returning to an overview of the NCI method, there is a certain probability that a
person will consume on any given day, and this probability can vary from person
to person. For example, there can be frequent and infrequent consumers of fish.
(4) There may be a correlation between the amount consumed on a consumption day
and the frequency of consumption. For many foods, those people who consume
the food more frequently also consume more of it on the actual consumption day
(Tooze et al., 2006).
(5) All survey respondents who are included in the analysis are assumed to be fish
consumers. This includes the possibility that the consumption rate of some
13 This analysis with 31 double hits is an example of the possibility of successfully fitting an NCI model with fewer than 50 double
hits. However, as noted previously, it is wise to plan a sample size that is very likely to yield at least 50 double hits in order to
provide stronger assurance of being able to fit and estimate the parameters for the NCI model.
Nez Perce Tribe
Volume II: Current Fish Consumption Survey
December 2016 Page 43
-------
consumers may be very low. The FFQ data were used to determine if a
respondent was a consumer of fish (or a specific species group) in this study.
(6) The distribution of usual fish consumption may be influenced by factors with
values specific to each individual. In order to accommodate this realistic feature,
the NCI method has the option of including respondent-specific covariates in the
modeling (e.g., FFQ consumption rate, gender, age). The individual-level
covariates can be used to modify the distribution based on the values of the
covariate. For example, respondents with higher FFQ consumption can have a
different distribution of FCRs than respondents with lower FFQ consumption, and
use of gender as a binary covariate can produce a different distribution for each
gender. The selection of covariates into the NCI model is further described in
Section 5.23.2. Another reason for including covariates into the NCI model is to
estimate the distribution for specific groups. Inclusion of a covariate in the model
states that the consumption frequency or amount (or both) vary across the groups
(or values) of the covariate. After the NCI model is fit the estimation of the
distribution in the overall population as well as in specific groups defined by the
model covariates is available.
Consumption may vary depending on the day of the week. Continuing
development of the key points described above, in addition to the respondent-
specific covariates, the NCI method can also adjust for weekday-weekend
differences in consumption and over- or under-representation of weekend or
weekday interviews in the completed pool of 24-hour recall interviews. For the
purpose of this study, the "weekend" was defined as Friday, Saturday and Sunday
and weekdays as Monday through Thursday. Friday has been included in the
definition of the weekend for this analysis, since consumption on Friday has been
found to be more similar to consumption on the traditional two-day weekend than
to consumption on other weekdays (Haines et al., 2003, in a study of the U.S.
general population). The weekday/weekend adjustment accounts for: (a) the
difference in the consumption rate between weekdays and weekends, (b) the
weekday/weekend mix among each respondent's first and second 24-hour recall
interviews, and (c) The noted potential over- or under-representation of weekdays
or weekends in the pool of completed interviews.
(7) The NCI method can also adjust for differences in consumption between the first
and subsequent interviews ("sequence effect"). The sequence effect adjustment in
this study introduces into the model an indicator variable for the second vs. first
interview. In the analysis of this survey's 24-hour recall data by the NCI method,
the fitted model used in calculating the mean and percentiles of the distribution of
usual consumption (the main end product of the NCI method) have keyed the
estimates to the mean consumption rate found in the first interview, though the
data from both interviews are used. In this analysis, both the weekday-weekend
and the sequence effect adjustments have been applied. This choice was
recommended by NCI staff who frequently use the NCI method in dietary
Nez Perce Tribe
Volume II: Current Fish Consumption Survey
December 2016 Page 44
-------
studies14. The NCI staff found these two adjustments to be important in past
application of the NCI method to the NHANES study. Consistent with this
recommendation, the first interview was used as the reference interview. While
there are no formal guidelines dictating this choice, the contractors considered this
to be the most reasonable choice for this survey for two reasons. First, differences
in mean FCRs based on the first and second interviews separately were observed,
indicating that an adjustment for interview sequence was needed (either the first
or the second would be considered as the reference interview). Second, the first
interview was conducted in-person with physical models available in a more
controlled environment than the second interview, which was conducted by phone
using model photos left behind by the interviewer. The contractors also carried
out a sensitivity analysis to assess the impact of these two adjustments on the
estimated distributions. The results of the sensitivity analysis are available in
Appendix E, Section 4.
(8) The model-fitting process leads, in steps, to the estimated distribution of usual
fish consumption. The NCI model is fit by the maximum likelihood method, using
SAS macros available from the following NCI website:
(http://riskfactor.cancer.gov/diet/usualintakes/macros_single.html). All model
parameters, including the Box-Cox transformation parameter (the parameter that
dictates the shape of the distribution of mean consumption per respondent on days
with consumption), are estimated jointly by the likelihood maximization
procedure. The model-fitting by the maximum likelihood method is iterative,
converging on the final parameter estimates. The fitted model describes the daily
fish consumption as a function of covariates and random effects. (The random
effects in the model represent person-to-person differences that are not explained
by the covariates.) The model is used to calculate the distribution of usual fish
consumption. The distribution cannot be determined by a closed form equation,
and it is calculated using simulation.
Specifically, the estimated model parameters are utilized to generate (by
simulation) a population of persons with the same composition of covariates and
between-person variability as has been observed among the respondents. As the
simulation calculates the distribution of usual consumptions rather than
consumptions on specific days, the within-person variation in the amount
consumed day-to-day (also estimated by the model) is not included in the
generating process. The usual consumption for each generated individual is the
product of a) the individual's proportion of days with positive consumption and b)
the individual's mean consumption amount on days with positive consumption.
The two parts (the proportion and the mean amount) are generated by the model
from that individual's covariates and the model parameters. The simulation also
includes generation of a random effect for each person that is added to the fixed
effects of the covariates. As the random effects are model-based but
unobservable, the generated data represent "pseudo-persons" drawn from a
14 Personal communication from Kevin Dodd to Moni Neradilek on June 22, 2015 and to Nayak Polissar on September 14, 2015.
Nez Perce Tribe
Volume II: Current Fish Consumption Survey
December 2016 Page 45
-------
population with characteristics derived from the survey's respondents; these
generated pseudo-persons (and their fish consumption) are not specific
respondents in this survey. The random effects for the proportion and the mean
amount consumed on positive days are generated from a bivariate normal
distribution with zero mean and variances estimated from the NCI model.
Because the average amount for a specific pseudo-person generated from the
amount equation in the NCI model is on the Box-Cox transformed scale, it needs
to be back-transformed to the original scale. The back-transformation (the "9-
point approximation" method) adjusts the values to ensure that the mean fish
consumption rate of the estimated usual intake distribution on the original scale is
approximately15 equal to the overall mean of the original 24-hour recall data (see
Tooze, JA, et al. 2010 for more details).
Finally, the probabilities and the average amounts on the original scale are
multiplied for each pseudo-person to yield the usual consumption rate for the
pseudo-person, and the distribution of the usual consumptions is calculated. The
precision of the estimated usual intake distribution is improved by independently
drawing 100 pseudo-persons per each individual in the sample. When the
sequence or the weekend effect(s) is (are) present in the model, the calculations of
the probabilities and the mean consumption amounts are slightly modified. When
the sequence effect is present, the probabilities and the average amounts are
generated with the interview number covariate set to the reference interview. The
first interview is the reference interview in the analysis presented in this report).
When the weekend effect is included, separate probabilities and mean amounts
are generated for the weekdays and for the weekend and are then averaged using a
weighted mean, with weights of 4 and 3, respectively, to yield a single overall
probability and a single overall average amount per pseudo-person.
The simulation method of creating a distribution of usual fish consumption also
applies to the calculation of distributions of usual consumption for specific
subpopulations. The subpopulation calculations are, in fact, a by-product of the
calculation for the entire distribution, when the simulated pseudo-persons are
separated into the desired subpopulations (e.g., the two genders) and
subpopulation-specific distributions are calculated from the pseudo-person data.
In addition to presenting the means and percentiles of usual consumption for
subpopulations of interest, the estimated subpopulation distributions were also
utilized in the process of covariate selection and quality checking of the model
(described in more detail in sections 5.23.2 and 5.23.3, respectively.)
This section and subsequent sections present specific methodology relevant to the analysis by the
NCI method. Readers who are particularly interested in this approach to estimating the
distribution of usual consumption may wish to also review Appendix E, which has important
additional information on the use of the NCI method for this report.
15 The mean based on the distribution of usual intake estimated from the NCI model can differ from the mean estimated by the naive
method (from the input 24-hour recall fish consumption rates) due to options chosen for the model-fitting process, such as the
choice between the first or second interview as the reference interview for the fitting process.
Nez Perce Tribe
Volume II: Current Fish Consumption Survey
December 2016 Page 46
-------
Additional notes on the NCI methodology are available in Tooze et al., 2006. An instructive
webinar series featuring Dr. Tooze and others is available online at:
http://riskfactor.cancer.gov/measurementerror. The SAS statistical programming language code
for carrying out the calculations using the NCI methodology (version 2.1) is also available online
at: http://riskfactor.cancer.gov/diet/usualintakes/macros_single.html.
5.23.2 Covariate Selection and Assessment of Seasonality
The use of covariates, if properly selected, can improve the consistency between the NCI-method
model and the survey's 24-hour recall data and provide better estimates of the mean and
percentiles of consumption for the population or sub-population being considered. The inclusion
of covariates does not change the mean of the overall distribution of usual fish consumption, but
the use of covariates can change the shape of the distribution. If there are differences in
distributions across different subpopulations, the model is able to accommodate these differences
by introducing these characteristics as covariates in the NCI model. The overall distribution
estimated by the NCI model with specific covariates included is then a result of combining the
different distributions across the subpopulations, leading to a potentially different shape of the
overall distribution compared to the NCI model without covariates. As noted, the model is
improved if covariates that affect the distribution of usual fish consumption are included. The
covariates considered for inclusion in the NCI model were:
• FCR per respondent from the FFQ for the same species group for which the distribution
of usual intake was desired (i.e., the Group 1 FFQ consumption rate was used as a
covariate for analysis of the Group 1 24-hour recall consumption data and Group 2 FFQ
rates were used as a covariate for the 24-hour recall data from Group 2)
• presence vs. absence on the fishers list
• gender
• ZIP code groups (83540, 83536, 83501 and combined other ZIP codes for the Nez Perce
Tribe and 83203 and combined other ZIP codes for the Shoshone-Bannock Tribes)
• age (grouped as 18-29, 30-39, 40-49, 50-59 and 60+)
• the respondent's body weight (in pounds)
A dichotomous tribe indicator (NPT or SBT) was included as a covariate in all models.
The FFQ consumption rate is an especially important covariate, as it is highly predictive of the
24-hour recall data. By including the FFQ as a covariate in the NCI method modeling, the
implication is that a distribution of usual consumption derived from the 24-hour recall data of
tribal members with lower FFQ rates would itself be shifted toward lower rates than such a
distribution derived from tribal members with higher FFQ. As there are different ways in which
FFQ rates can be related to the 24-hour recall data, the analysis path in this study explored
several possible relationships between the two set of rates and chose, among them, the best-
fitting one. (More detail on the choice is provided later in this section.)
Among the candidate covariates listed above, the covariates that were selected into the NCI-
method model had a demonstrable impact on the NCI-estimated consumption rate distribution.
The selection of covariates involved a model-building process that started with a simple NCI
model (including tribe as the only covariate) and that subsequently added other covariates that
had an impact on the NCI-model distribution of usual consumption rates. Specifically, the
Nez Perce Tribe
Volume II: Current Fish Consumption Survey
December 2016 Page 47
-------
model-building process added a candidate covariate (and its statistical interaction with the tribe
covariate) into the model, and then there was a visual comparison of the differences in the NCI-
estimated means and percentiles of usual consumption rates within subpopulations defined by
categories of the covariate.
For example, when considering the fishers list covariate, the contractors compared the NCI-
estimated statistics (mean and percentiles) between fishers and non-fishers within each tribe.
Large differences between different levels or categories of the covariate suggested inclusion of
the covariate in the NCI model. To arrive at the best fit for continuous covariates (FFQ rates and
the respondent's body weight), different transformations of the covariate were considered: the
original (untransformed) value, 3rd root, log and ordered decile number (a variable with integer
values from 1 to 10, depending on which decile of the distribution of the covariate included the
untransformed value for a respondent).
The selection of covariates for the NCI model was carried out in two steps: 1.) choosing the best
functional form for the FFQ covariate (choices: no transformation, 3rd root, log or ordered decile
number), and 2.) selecting other covariates. The FFQ consumption rate covariate was considered
first (and was added to the model first, with other candidate covariates considered afterward),
because it was expected that the FFQ rates would be strongly related to the 24-hour recall
consumption rates. Thus, the contractors first considered the FFQ rates as a covariate in the
model and attempted to find the best transformation of FFQ rates that predict the 24-hour recall
rates as analyzed through the NCI method.
When considering a continuous covariate, such as the FFQ rates, for inclusion into the NCI
model, one needs to ensure that the specific form of the continuous covariate correctly reflects
the trend of the 24-hour recall rates in relation to the FFQ rates. As noted, continuous effects of
rd
the FFQ were considered in four forms: the original (untransformed) value, the 3 root value, the
logio value and the numerical decile of FFQ (coded as 1-1016). To choose the best among these
four models the contractors compared them to a fifth NCI model that used the FFQ covariate as a
categorical decile. The overall population was then broken down into ten approximately equal-
sized subpopulations (bins) according to the FFQ decile. The NCI-model estimated means and
percentiles (medians, 90th percentiles and 95th percentiles) in each bin from the four competing
continuous FFQ NCI models were then compared to the means and percentiles from the
categorical NCI model (reference model).
The categorical FFQ model is the most complex one; it uses nine degrees of freedom per tribe,
compared to one degree of freedom per tribe for each of the four continuous FFQ models. The
median and percentiles of the categorical FFQ model may be "noisy" within each decile bin (due
to the small number of respondents in each bin), but the categorical FFQ model is a useful
reference for choosing the best continuous FFQ model. The categorical FFQ model is a useful
reference because it can reveal important features in the possible curvilinear or nonlinear
relationship of FFQ rates to the 24-hour recall rates, after the latter are processed through the
NCI method. A simplistic model-fitting with the various continuous FFQ models can miss such
non-linear relationships.
16 The deciles were defined separately within each tribe.
Nez Perce Tribe
Volume II: Current Fish Consumption Survey
December 2016 Page 48
-------
In choosing among the four continuous FFQ models the contractors sought a model that captured
important features that are present in the categorical FFQ model (see Appendix E, Section 1 and
Figures El and E7 for more detail). On visual inspection, the 3rd root and the logio
transformations best followed the trend in the categorical decile (true for species Group 1 and for
17
species Group 2 models). As the lambda (X) parameter for both species group models was
relatively close to the 3rd root (lambda = 1/3), the 3rd root FFQ was chosen as the primary model
choice. Analysis by the NCI method with logio FFQ was carried out as a sensitivity analysis. The
sensitivity analysis is presented in Appendix E, Section 4 and further details regarding the choice
between FFQ transformations are presented in Appendix E, Section 1. Finally, the contractors
discovered that the 24-hour recall consumption in the 10th FFQ decile among the SBT
respondents was considerably lower than expected by the trend in the continuous FFQ variable
and a binary indicator for this group was added into the model to improve the model fit.
The second step involved considering the inclusion of the remaining covariates into the model.
The candidate variables available included presence/absence on the fishers list, gender, ZIP code
group (83540, 83536, 83501 and combined other ZIP codes for the Nez Perce Tribe, and 83203
and combined other ZIP codes for the Shoshone-Bannock Tribes), and age (grouped as 18-29,
30-39, 40-49, 50-59 and 60+). All of these variables had an impact on the estimated
distribution of usual fish consumption distribution from the NCI method and were included in
the NCI models. Respondents' body weight (tried in the modeling as untransformed, 3rd root,
logio and the decile rank) had no or only a weak relationship with the estimated consumption
distribution and was therefore not included as a covariate. The selected covariates were used in
two model components of the NCI method: the model for the probability of consuming from the
designated species group on a randomly selected day and the model for the amount of the species
group eaten during the day, given that consumption occurred on the specific day.
rd
The 3 root of FFQ was also selected as the covariate for the Group 2 model. However, due to
the small number of single- and double-hits of Group 2 in the SBT, a model with several
covariates was found to be statistically unstable and the remaining covariates (presence on the
fishers list, gender, ZIP code and age) were not included in the final Group 2 model for the
combined Tribes. The final model for Group 2 consumption thus consisted of tribe
(dichotomous), and the 3rd root of FFQ rates and its interaction with the dichotomous tribe
variable. When the distribution of the Group 2 consumption rates was to be estimated within
subgroups (e.g., by gender) the corresponding covariate (e.g., gender) was added into the final
Group 2 model for the specific subgroup analysis only.
Seasonality as a potential factor influencing fish consumption was explored, as described in the
next section. More details on covariate selection can be found Appendix E, Section 1.
17 Lambda (A) is the power exponent used to transform a normal distribution to a distribution appropriate as one component of a
model consistent with the dietary recall data being analyzed.
Nez Perce Tribe
Volume II: Current Fish Consumption Survey
December 2016 Page 49
-------
5.23.2.1 Assessment of Season ality
Prior to selecting the covariates, potential seasonal variation in 24-hour recall consumption rates
was explored for Group 1, Group 2 and salmon. For each tribe, the mean consumption by month
was plotted (see Figures E22, E23 and E24 in Appendix E for the Group 1, Group 2 and salmon
displays, respectively). As the consumption values differed between the 1st and 2nd interviews,
the means per month were calculated separately for the 1st and 2nd interview data for a more
direct comparison across months. While some variability across the months exists, no difference
or pattern was discerned indicating a clear seasonal differences vs. empirical noise; this null
finding may be due to the small sample size for each month. The findings were further
corroborated in the 24-hour recall data by examining seasonal patterns in mean Group 1 FFQ
consumption rates (Appendix E, Figure E25). Also, there might be seasonal variation in access to
fishers for interviews due to their seasonal absence from home. Such absence might affect the
mix of interviewees by month and induce a time pattern of consumption, particularly
consumption of salmon. A plot of the monthly percentage of respondents that were fishers
(Appendix E, Figure E26) shows no clear indication of seasonal differences.
18
May-July 2014 was the peak salmonid harvest period , which coincided with the first three
months of the survey. Further analysis of the Nez Perce respondents was conducted to explore
the possibility that different types of respondents were interviewed during the peak harvest
period compared with the remainder of the survey. For instance, if respondents who fish heavily
(potentially respondents with more seasonality in their consumption patterns) tended to be too
busy or otherwise unavailable for interviewing during the peak harvest period, some true
seasonality may be masked.
The findings of the seasonality analysis did not provide a basis for adjusting consumption rate
estimates for seasonal variation, but the sample sizes used in these analysis and the findings do
not show that there is not a true, underlying seasonal component. Of the 451 Nez Perce
respondents (138 on the fishers list), 30 (11 fishers) were interviewed during the peak harvest
period. The unweighted percentages of fishers did not vary significantly between the peak
harvest period (May-July, 2014) and the remaining period (37% vs. 30%, Chi-squared test p =
0.6). Appendix Table El8 shows mean FCRs calculated using the 24-hour recalls (naive mean)
and the FFQ means for Group 1 (all fish), Group 3 (salmon or steelhead) and Chinook salmon.
There were no significant differences between the early and later respondents in naive mean
FCRs, when considering the early-late comparison among all respondents or among fishers only
(all p > 0.6; see Appendix Table El8 for details on calculations). Mean Group 1 12-month
consumption rates by the FFQ method were significantly higher in respondents interviewed
during the peak harvest period (170 vs. 120 g/day, p = 0.015), indicating that consumers with
relatively high annual consumption were interviewed during the peak period. There were no
other significant differences in mean FFQ rates between periods (Appendix Table El8).
Appendix Table El 9 shows self-reported frequencies of fishing (times per month) from
respondents interviewed during the two periods. There were no significant differences in fishing
rates between periods (p > 0.2 for all comparisons). Taken together, there is no evidence that
fishers, high consumers, or potentially seasonal consumers were under-represented during the
18 Personal communications from Joe Oatman, Nez Perce Department of Fisheries, to Nayak Polissar during August 28-30, 2015.
Nez Perce Tribe
Volume II: Current Fish Consumption Survey
December 2016 Page 50
-------
peak harvest period, though with the small sample size, there may be such an effect that was not
detected.
Appendix Table E20 summarizes how often respondents reported species-specific consumption
as two separate periods (higher and lower consumption periods, presumably related to
seasonality of the species) as opposed to averaging consumption over the whole year
(presumably indicating no seasonality). For respondents interviewed during the peak salmonid
harvest period (May-July, 2014), 45% of responses involving salmon or steelhead were reported
using two periods, compared with 27% of such responses for respondents interviewed during the
remainder of the survey period. This ratio was similar among fishers and non-fishers, as well.
While not conclusive, this suggests that during the peak harvest period, respondents were more
apt (though still <50% of the time) to report consumption of these species in two periods to
explicitly acknowledge the seasonality of consumption. In contrast, during the remaining survey
period, respondents most often mentally averaged consumption over the entire year as one
period. Note that according to Appendix Table El8, this did not seem to have notably impacted
annual salmon and steelhead consumption rates. Again, the small sample size during the peak
harvest period makes detecting seasonal effects, if there are seasonal effects, more difficult.
5.23.3 Quality Checking of the Model
The NCI method is a powerful yet complex method to estimate the distribution of the usual
consumption from the 24-hour recall data. A few simple analyses were therefore conducted to
assess the validity of the NCI model estimates.
In the first quality check the contractors examined the distribution of the consumed amounts. An
important assumption of the NCI method is that the transformed positive consumption amounts
(fish consumption on days when consumption occurred) are normally distributed. To verify this
assumption the contractors examined the (survey-weighted) histograms of the transformed (3rd
root) respondent-specific mean consumption (for the respondents' one or two days which
included fish consumption) and the within-person residuals (for respondents with double-hits)
for the data from the two tribes combined.
The second quality check consisted of comparison of demographic subgroup means between (a)
the NCI method (considering only the consumption amount part of the NCI model), and (b)
means from a "naive" approach: traditional weighted survey means, calculated directly from the
24-hour recall consumption data (including only days with non-zero consumption). The
demographic subgroups considered were defined by the following covariates, each analyzed
separately for this purpose: the fisher indicator, gender, ZIP code group, age group and the FFQ
decile. The two parameters that the contractors compared for each demographic subgroup were
the mean per-respondent probability of consuming fish on a given day and the mean per-
respondent consumption on days with fish consumption. (Note that the mean consumption per
day, on the average, is the product of these two parameters.)
The naive approach was carried out in three alternative forms, depending on which interviews
were used in the calculations: 1) all interviews, 2) interviews for respondents with two interviews
and 3) only first interviews. Choices 1 and 2 are more comparable to the NCI method in that they
also utilize both interviews and allow examination of the covariate effects on the consumption
rates in both interviews. Choices 1 and 2, however, do not account for the sequence effect
(second vs. first interview) and the results could therefore be systematically lower or higher
Nez Perce Tribe
Volume II: Current Fish Consumption Survey
December 2016 Page 51
-------
compared to the results from the NCI model (as the NCI model adjusts for the sequence effect).
The results from choice 3 (first interview only) should be more comparable to the NCI model
estimates with regard to the adjustment for the sequence effect, as the NCI model adjusts for the
sequence number and calculates the consumption rate distribution keyed to the mean of the first
interview. Some differences between all three choices of the naive approach and the NCI model
estimates are still possible because the NCI model adjusts for differences between weekdays and
weekends while the naive approach does not. The estimates that were compared between the
naive and the NCI methods were consumption probabilities and means of positive consumption
days for groups defined by covariates included in the NCI model. The naive and NCI-method
means were compared within categories of the following variables: presence/absence on the
fishers list, gender, ZIP code group, age and the FFQ rate (categorized in deciles). The
comparison of the NCI and naive approaches was carried out for consumption of Group 1
species only.
A final check of the NCI method estimates involved re-computing the estimates by an independent
statistician. The estimates (mean and percentiles) of the Group 1 consumption distribution from the
NCI method were checked by a member of the NCI staff who deals regularly with the NCI method
(personal communication from Kevin Dodd to Moni Neradilek on July 2, 2015). The staff
member's Group 1 means and percentiles were all within 0.4% of the contractors' estimates for the
Nez Perce Tribe and within 0.9% for the Shoshone-Bannock Tribes.
5.23.4 Sensitivity Analyses
While building the NCI model several choices were made. These choices included: 1) using the
third root transformation for the FFQ covariate; 2) using the weekend adjustment and the
sequence effect adjustment; and 3) including a number of other covariates in the final model for
the distribution of usual consumption of Group 1 species. To quantify the impact of these choices
on the estimated distributions, a sensitivity analysis was run with alternative choices. (All
sensitivity analyses were carried out for Groups 1 and 2 species unless otherwise noted.)
Specifically, the log transformation for the FFQ covariate was considered instead of the third
root transformation. A model without the weekend/weekday adjustment was also considered, as
was a model without the sequence effect adjustment. For each of these three alternatives, only
the specific item (e.g., weekend/weekday) was changed or omitted in the model and all other
covariates from the final model were unchanged.
Three additional sensitivity analyses were carried out: (a) a model based on the NPT data only;
(b) a simpler model (for Group 1 species only) than the final model (certain covariates were not
included in the model); and (c) a model assuming zero correlation between the daily probability
of consuming fish and the amount of fish consumed on an actual consumption day.
The model based on the NPT data alone was created to compare the means and percentiles from
the final model—using both Tribes' data—to means and percentiles from a model using just one
Tribe's data (NPT). The relatively small number of single- and double-hits in the SBT data
required that the final models be fitted to data from both Tribes combined, and that covariates be
introduced into the model to capture differences between the Tribes19. As the number of hits in
19 As noted previously, the NCI model based on combined data from the two Tribes was used for the final estimates of means and
percentiles offish consumption for each Tribe. These estimates are expected to be more precise, under certain assumptions, than
estimates based on a model using data from a single Tribe.
Nez Perce Tribe
Volume II: Current Fish Consumption Survey
December 2016 Page 52
-------
NPT was sufficient to run certain models without problems, a sensitivity analysis was carried out
by running the NCI models with the NPT data only and then comparing the results to the final
estimates from the two-Tribe model.
To examine the impact of combining numerous covariates in the NCI model, a sensitivity
analysis was run in which only a single covariate was added to a model that initially included
Tribe (dichotomous), FFQ consumption rate, the Tribe-FFQ interaction and an indicator variable
for the 10th decile of the FFQ consumption rate in the SBT.
Finally, an important methodological feature of the NCI method is that it can include a non-zero
correlation between the probability of consumption on a random day and the consumption
amount on an actual consumption day. In order to investigate the impact of the correlation
assumption, a sensitivity analysis was run forcing the correlation to be zero (no correlation) in
the NCI models.
5.24 Effect of Changes in Study Design on FFQ Rates
An assessment was conducted to determine the impact of two study design changes on overall
fish consumption. The first impact was that of interviews conducted at special events. All
interviews conducted on September 25-27, 2014 and October 17-19, 2014 were considered
interviews at special events. The second consideration was the impact of non-tribal interviewers
compared to tribal interviewers.
Another assessment was also conducted to determine whether interviews conducted at a
respondent's home differed in fish consumption from interviews not conducted at their home.
Although this is not a design change, the comparison was of interest because this variable might
have had an effect on the reported consumption. The results of the home/not home analysis are
presented along with the results on design changes for convenience.
The impact of the design variables on fish consumption was calculated both without and with an
adjustment for respondent characteristics. The unadjusted analysis consisted of the calculation of
FFQ means and medians of fish consumption in the two groups and the estimation of the
difference of the two means. The latter was estimated from linear regression (with the same
statistical weighting of respondents as in the calculation of means and percentiles). Linear
regression was also used in the adjusted analysis and included respondent characteristics in
addition to the tested design variable. The characteristics included ZIP code (categorized as
83536, 83501 or others combined), age category (<30, 30-39, 40-49, 50-59 and 60+), gender,
on/off reservation residence, fishing (questions 35 and 36) and the respondent's body weight (as
a continuous predictor). Including the respondent characteristics in the regression controls for
differences in the fish consumption that may be due to the respondent's personal characteristics
and not to the tested design variables. The results of this analysis are presented in Section 6.7
"Effect of Changes in Study Design on FFQ Rates."
Nez Perce Tribe
Volume II: Current Fish Consumption Survey
December 2016 Page 5 3
-------
5.25 Confidence Intervals
Confidence intervals express the uncertainty of the estimated population means and percentiles
of fish consumption. The confidence intervals in this report were calculated using the bootstrap
replicate weight method (Lumley, 2010), which is a standard statistical methodology for
calculating confidence intervals and incorporates relevant sources of uncertainty. In this method,
1,000 replicate weights (random perturbations of the adjusted sampling weights) are first
calculated (see Section 5.26 for more detail). The replicated weights are then saved for use in all
subsequent confidence interval calculations (see Section 5.26 for more detail). The bootstrap
method for confidence intervals was applied to all weighted analyses (including the analysis of
the FFQ and 24-hour consumption rates). Running the NCI model for 1,000 replicate weight sets
in the bootstrap procedure took over 3 days of computation for species Group 1; therefore, the
confidence intervals were calculated only for the Group 1 mean and percentiles.
These confidence intervals do not account for any clustering of respondents by household. For
example, people who live together may tend to consume more similarly than randomly selected
individuals from different households. This correlation between individuals within the same
cluster would tend to decrease the precision of the mean and percentile estimates (widen the
confidence intervals). The contractors investigated the potential impact of not accounting for
clustering with the help of the Tribe. The Tribe reviewed the list of respondents and their contact
information, as maintained by the tribal enrollment offices at the time the sample was drawn, to
determine which respondents did live together around the time the survey was conducted. The
review was based on address and the reviewer's knowledge of the population.
Based on this review by the Tribe, there were 35 household clusters that comprised 81 members
of the 451 respondents with a completed FFQ interview and calculable consumption rate (see
Appendix D, Table D4 for a complete list of respondents' survey ID codes). Of the 35 clusters,
27 had a pair of respondents, seven had three respondents and one had five respondents.
If, very conservatively, only one respondent per cluster had been included in the analysis, the
effective sample would have been reduced by 46, to a net of 405 respondents, implying that
consumption information from additional respondents within the same household is completely
"redundant"—a highly conservative and unrealistic assumption. This reduction in effective
sample size would lead to only a 5.5% increase in the confidence interval widths of the mean
Group 1 consumption rate, under a simple random sampling scenario. As this impact is quite
small and would only occur under a very extreme and unlikely scenario, the confidence interval
methodology was not modified to account for clustering.
Nez Perce Tribe
Volume II: Current Fish Consumption Survey
December 2016 Page 54
-------
5.26 Replicate Weight Calculations
A total of 1,000 bootstrap replicates was utilized in the calculation of confidence interval and
other measures or uncertainty or inference. In the calculations, each replicate bootstrap
accounted for two sources of uncertainty: the random sampling of members from the population
in each stratum and the non-response model.
The sampling uncertainty was addressed by drawing 1,000 non-parametric bootstrap resamples.
Each non-parametric bootstrap resample consisted of a stratified random sample from the
original sample, sampling with replacement. Specifically, the strata were the strata used in
drawing the random sample for the study and the sample was the sample of the participants
drawn for this study (see Section 5.5). Each random draw was selected from all sampled tribal
members (both non-responders and responders) in each sample stratum. Logistically, the
recorded information from the non-parametric bootstrap procedure was the number of times (N;)
each respondent was drawn in each bootstrap resample i. Note that for observations not being
drawn into a given resample, N; = 0.
The uncertainty in the non-response model was also addressed by the non-parametric bootstrap.
For each of the 1,000 bootstrap resamples the response probabilities predicted by the logistic
response model (described in Section 5.20) were recalculated after the model was refitted to each
bootstrap resample. The response probabilities from bootstrap i are denoted by Pm. The non-
response adjusted replicate weights were then calculated for all responders in the bootstrap
resample. Replicate weights Wt (i denotes the bootstrap index) were calculated as the inverse of
the product of: (a) the sampling fraction per stratum (Fs) and (b) the parametric bootstrap
response probabilities (Pri), and then multiplied by the number of bootstrap resamples for a
given observation:
W^Ni/fFSPn)
The 1,000 sets of bootstrap replicate weights were saved and used for all confidence interval
calculations.
5.27 Confidence Interval Calculations for a Specific Statistic
Calculations for specific statistics were carried out on the subset of responders that were relevant
for that statistic (e.g., consumers of Group 2 fish species would be included for Group 2
calculations of the mean, median and other percentiles).
The statistic of interest (a mean, percentiles or a regression coefficient) were than calculated on
the relevant subset of responders (e.g., Group 2 fish consumers) for each bootstrap realization.
Issues with item-specific missing values in this step were automatically handled by the subset
function in the R software (by excluding the observations with missing values and adjusting the
weights to accommodate the actual number of observations used in the analysis). The 95%
confidence interval limits for a statistic (when a confidence interval was calculated) were defined
as the 2.5th and the 97.5th percentiles of the bootstrap distribution of the specific statistic across
the 1,000 bootstrap realizations.
In a small fraction of the bootstrap replicates, the NCI model did not converge. The NCI model
estimation is a complex iterative procedure for a non-linear mathematical problem that
occasionally does not arrive at a best solution (non-convergence). The fraction of bootstrap
models that did not converge are reported.
Nez Perce Tribe
Volume II: Current Fish Consumption Survey
December 2016 Page 5 5
-------
5.28 Handling Missing Values
As with all surveys, the interviewers strove to obtain complete responses from all respondents
and to avoid any missing values. However, in a survey of this size and complexity, missing
values are unavoidable and a concerted effort was made to handle the missing values in an
appropriate manner.
During an interview, the respondents usually had the option of indicating "don't know or
refused" to avoid responding to a specific question, but could continue on to the subsequent
question. In those situations, missing values were dealt with in multiple ways, depending on the
type of variable with missing data or its importance. If a non-consumption-related response or
variable was missing (e.g., respondent weight in pounds or household income), the respondent
was simply excluded from any analysis involving that variable.
In contrast, if the missing variable was a consumption rate component, then a value was imputed.
The consumption rate components that were imputed in the case of "missingness" were portion
frequency (e.g., portions per week), portion size (based on portion models) and, if the respondent
reported consumption in two periods (e.g., higher/lower or in season/out of season), the length of
the higher consumption period as a percentage of the year (see Section 5.10 on consumption rate
calculations). The imputation procedure was based on the specific rate component missing and
the corresponding species and was always derived from observed, similar responses without
missingness, as described below.
In the sample, respondents reported consuming 6.2 species on average and 13% of respondents
had at least one missing component among any species reported. In total, there were 2,810
species-specific consumption responses (across all combinations of species and respondents), of
which 3.2% had a missing component. This rate of missingness is relatively low, given the large
number of combinations of respondents and species, but the missingness needed to be addressed
due to the total number of respondents with some missingness.
The guiding principle to the imputation procedure was to impute only individual consumption
rate components rather than the final consumption rate itself, which can vary many-fold between
individuals. In general, the value imputed was a mean calculated from similar responses that had
no missing values, where "similar" means that the species or species group was the same as for
the given respondent's record with a missing value. For example, if a respondent reported
consuming Chinook salmon by describing consumption during higher and lower consumption
periods, but did not provide the portion size for the lower-period rate, other responses for
Chinook consumption during the lower consumption period, without missingness, would be
selected for imputation. The mean portion size from those similar responses would then be
calculated and used in place of the missing portion size. If there were less than five other similar
records to use for imputing a missing value, related species were grouped to increase the sample
size. All groupings used are fully specified in Appendix C (Table CI).
Nez Perce Tribe
Volume II: Current Fish Consumption Survey
December 2016 Page 56
-------
Imputation of missing values was performed according to the following rules:
1. Both portion frequency and portion size are missing.
If a respondent provided neither how often he or she consumed a species nor in what
portion size, both frequency and portion size were imputed to 0, which resulted in a
consumption rate of 0 grams/day for that specific species.
2. Portion frequency is missing but portion size is not
If the respondent reported how much he or she consumed per portion but not the
frequency, the frequency was imputed using the mean value computed using records
from the same species and from the same period type, where period type was the whole
year, higher consumption period, or lower consumption period. If fewer than five such
records were available, similar species were grouped together to provide a larger sample
size. Details on how species were grouped is described in Appendix C.
3. Portion size is missing but portion frequency is not
If the respondent reported how frequently he or she consumed but not how much, the
portion size was imputed in an analogous way as Case 2 above, using similar records
without missing values.
4. Higher consumption period length is missing
If the respondent provided consumption detail for higher and lower consumption periods
but did not provide the length of the higher consumption period, this value was imputed
using the mean calculated from similar responses for higher consumption periods. As for
Cases 2 and 3 above, the imputation was species-specific unless the sample size was less
than 5, in which case similar species were grouped. Appendix C describes this process in
more detail.
Once a value was imputed for the missing consumption rate component, the consumption rate
was calculated according to Section 5.10 as if the imputed value was the actual value provided
by the respondent. Appendix C, Tables C2-C7 shows that the final mean and percentiles of
consumption rates were similar under a range of possible imputed values, demonstrating that the
impact of missingness and imputation on the final results was minimal.
There was one exception to the above rules on handling missing values, leading to the exclusion
of a respondent for some analyses. The respondent reported consuming Chinook, cod, and crab
outside of special events and gatherings, and consuming salmon, steelhead, and sturgeon at
special events and gatherings. However, for all species consumed, this respondent did not
provide a portion size or a frequency of consumption. Instead of imputing all of these species as
0 g/day as the above rule #1 would prescribe, the rates were considered incalculable and the
respondent was excluded from the analysis of consumption rates. The reason for treating this
respondent differently than rule #1 states, above, is that the pattern of response strongly indicated
that the respondent was, indeed, a consumer of salmon (included in Group 2), because salmon
was reported as a consumed species both at special events and at gatherings and outside of them.
As a rate of zero for both Group 1 and Group 2 would clearly be incorrect in this case and there
was no basis for imputation, it was deemed best to exclude the respondent.
Nez Perce Tribe
Volume II: Current Fish Consumption Survey
December 2016 Page 57
-------
5.29 Limited Percentiles for Small Sample Sizes
Some percentiles may be quite imprecise due to the small sample size of respondents used for the
percentile calculation. Such percentiles have generally been indicated using a rule of thumb
borrowed from random sampling; a percentile was designated as potentially very imprecise if—
treating the sample as a simple random sample—there would have been two or fewer
respondents with a consumption rate equal to or greater than the noted percentile. Due to the
statistical weighting used in the calculation of percentiles, it is possible that in a specific case
there may actually be more than two respondents (in the sample used to calculate the percentile)
with a rate at or exceeding the noted percentile value. Nevertheless, this approximate method
does provide a helpful flag of caution attached to some percentiles. This rule was applied to
analyses estimated from traditional survey-weighted techniques (Section 5.22), but not to NCI
method analyses (Section 5.23). The latter set of analyses relies on the entire data set, rather than
only on the observations in the tail of the distribution to estimate the percentiles.
Confidence intervals for percentiles (described in Section 5.25) may also become less reliable
(inappropriately wide or narrow) when the sample size is small. Such intervals have been
indicated in cases where there were less than five observations greater than or equal to the
corresponding percentile. This rule was applied only to the analyses estimated from traditional
survey-weighted techniques, but not to the analyses using the NCI method.
5.30 Large Consumption Values
Histograms (Figure 2) were examined of total consumption based on the FFQ, and three
respondents were found with values noticeably higher (1124-1372 g/day) than the other
respondents. The weight and gender of each respondent and the details of the species consumed
were further examined and the consumption rates were determined to be plausible. Accordingly,
the respondents were retained in the analysis without modification of any data.
Nez Perce Tribe
Volume II: Current Fish Consumption Survey
December 2016 Page 5 8
-------
o
CD
O
LO
£
U)
'35
o
"3-
o
co
o
(N
I I I
1
0 200 400 600 800 1000 1200 1400
Consumption Rate (g/day)
Figure 2. Nez Perce Tribe. Histogram of Group 1 (all fish) FCRs (g/day, raw weight, edible
portion).
The bin width is 100 g/day. The percentages (y-axis), corresponding to the frequency of consumers within
each bin, are weighted to correspond to the percentage among consumers in the eligible population. The
sum of all bars equals 100%.
5.31 Software and Software Modules
Calculations were carried out in R (R Core Team, 2015) versions 3.1.1-3.1.3 and SAS 9.4 (for
NCI method analysis only). The weighted survey analyses performed in R used the survey
package for analysis of complex surveys. (Lumley, 2014 and Lumley, 2004). The NCI method
was performed using a SAS macro (version 2.1) that was obtained directly from the NCI team.
Nez Perce Tribe
Volume II: Current Fish Consumption Survey
December 2016 Page 59
-------
6.0 Results
6.1 Response Rates
Table 3 summarizes the overall survey response rate, calculated to be 38.0%. Of the 1250 Nez
Perce tribal members originally sampled, 40 were found to be ineligible during the contact
attempts by interviewers (e.g., the sampled member lived out of the eligible area, were employed
as Tribal interviewers involved in the survey, or were deceased, institutionalized or impaired). Of
these, 8 were classified as impaired. For the purpose of overall response rate calculations, the
remaining 1210 members after excluding the 40 ineligible members were used as the
denominator (using the RR1 standard—see AAPOR, 2011).
Of these 1210 members, 472 members adequately responded to the screening interview
questions used to distinguish between consumers (n=464) and non-consumers (n=8). One
respondent who reported being a consumer on screening reported not consuming on the FFQ, so
this respondent was re-classified as a non-consumer, for a total of 9 non-consumers. Of the
remaining 463 consumers, 452 completed the first interview and 451 had a calculable FFQ
consumption rate. The respondent without a calculable rate is described in Section 5.28 (last
paragraph). The total number of responders with a complete and usable interview was 460,
including the 451 consumers with an FFQ rate plus the 9 non-consumers. The overall RR1
response rate was thus 460 of 1210 (38.0%) (Table 3). The number of responders corresponds to
29% of the original population size of 1574. During the planning phase (see Section 5.13,
"Response Rates") it was anticipated that approximately 60% of sampled members would
provide a first interview and 48% would provide two interviews. It was also anticipated that
these response rates would provide sufficient assurance of reaching the 50 double-hit interviews
(in combination with the double hits from the SBT interviews) needed to support the NCI
method of analysis. While the achieved response rate was lower than anticipated, the required
number of double hits for the two Tribes combined was achieved.
The 451 consumers with calculable FFQ consumption rates form the primary sample for most
tables presented in this report. However, some tables may be based on more or fewer
respondents, depending on analysis-specific inclusion/exclusion criteria.
Table 3. Nez Perce Tribe. Survey response rate.
No. or %
Responders*
460
Total sample size**
1210
Response rate (RR1)
38.0%
*Either was determined to be a non-consumer or completed the first interview and had a calculable FFQ
consumption rate;
**Excludes 40 tribal members found to be ineligible during contact attempts.
Nez Perce Tribe
Volume II: Current Fish Consumption Survey
December 2016 Page 60
-------
6.2 Factors Affecting Response Rates
This section uses a slightly different definition of response to the survey where ineligible
members are not excluded from the denominator; thus the response rate is somewhat
underestimated relative to the primary definition used in Section 6.1. The sample size and
population size are defined and meaningful numerical counts, whereas the number of ineligibles
detected in the survey depends on various survey-specific factors, such as total survey effort. The
contractors did not wish to use a survey-influenced denominator for response rates in this
section; hence, the entire sample or population is used in the denominators here. Due to the small
number of sampled members found to be ineligible to be interviewed, as noted in Section 6.1, the
inclusion of the ineligibles in the denominators of response rates in this section results in a small
20
underestimate of those response rates . That underestimation is unlikely to have much impact on
the difference in response rates between sample or population subgroups.
The response rate did vary quite substantially by demographic characteristics of the selected
sample. Tables 4 and 5 summarize the details. The response rate among males was higher than
among females (41% vs. 33%), those on the fishers list ("documented fisher")21 had a
substantially higher response rate than non-fishers (48% versus 33%), and those in the most
tribally populated ZIP code, 83540, had a substantially higher response rate than those in other
ZIP codes (46% vs. 22-26%).
Age was an important factor in determining response; among females and males, the youngest
members of the selected sample had the lowest response rate (the age range of 18-29 had a
response rate of 16% for females and 28% for males vs. 34—43% for other ages among females
and 29-58%) for other ages among males). Unweighted demographics of the tribal population,
sampled members, and consumers who responded are summarized in Appendix, D, Table Dl.
20 The rate of ineligibility in the entire sample is likely to be between 3% and 9%, based on 40 known ineligibles among those
contacted within a sample size of 1210, from which 460 became respondents. Calculations: 40/1210 = 3%, 40/460 = 9%
21 NPT staff have noted that the fisher list was derived from the Department of Fisheries Resources Management (DFRM)
information on specific individual tribal members who were sampled during their fishing activity at a certain river/area during a
certain fishery season, and is not a comprehensive representation of all "fishers" of the Tribe. They serve as a "fisher indicator" for
purpose of this survey. This will allow comparison their rates to other Tribal members who were not "documented" as fishers through
the Tribe's sampling program and monitoring activities.
Nez Perce Tribe
Volume II: Current Fish Consumption Survey
December 2016 Page 61
-------
Table 4. Nez Perce Tribe. Response rates by sampling strata. Estimates are unweighted.
No. in Total No.
Group Population* Sampled*
Responded**
% of % of
No. Sample Pop.
All
1574
1250
460
36.8%
29.2%
Sampling Strata
Female Age 18-29
191
152
25
16.4%
13.1%
Age 30-39
145
115
40
34.8%
27.6%
Age 40-49
152
121
52
43.0%
34.2%
Age 50-59
153
122
42
34.4%
27.5%
Age 60 or older
175
139
57
41.0%
32.6%
Male Age 18-29
178
141
39
27.7%
21.9%
Age 30-39
160
127
56
44.1%
35.0%
Age 40-49
144
114
66
57.9%
45.8%
Age 50-59
130
103
49
47.6%
37.7%
Age 60 or older
146
116
34
29.3%
23.3%
ineligible members are not excluded; the response rates are thus somewhat underestimated;
**Either was determined to be a non-consumer or completed the first interview and had a calculable FFQ
consumption rate.
Nez Perce Tribe
Volume II: Current Fish Consumption Survey
December 2016 Page 62
-------
Table 5. Nez Perce Tribe. Response rates by demographic factors. Estimates are unweighted.
No. in Total No.
Group Population* Sampled*
Responded **
% of % of
No. Sample Pop.
All
1574
1250
460
36.8%
29.2%
Gender
Male
758
601
244
40.6%
32.2%
Female
816
649
216
33.3%
26.5%
Documented Fisher***
Yes
371
288
139
48.3%
37.5%
No
1203
962
321
33.4%
26.7%
Zip Code
Lapwai - 83540
906
729
336
46.1%
37.1%
Kamiah - 83536
196
151
39
25.8%
19.9%
Lewiston -83501
172
136
30
22.1%
17.4%
Other
300
234
55
23.5%
18.3%
ineligible members are not excluded; the response rates are thus somewhat underestimated;
**Either was determined to be a non-consumer or completed the first interview and had a calculable FFQ
consumption rate;
***Refer to Section 4.4 on Populations for a description of documented fishers. Some respondents who were not
documented fishers did or do fish.
Nez Perce Tribe
Volume II: Current Fish Consumption Survey
December 2016 Page 63
-------
6.3 Consumers, Non-Consumers and Frequency of Consumption
Non-consumption of fish was rare among the Nez Perce Tribe, as shown in Table 6. An
estimated 2.6% of the adult tribal members do not consume fish. The single most common
reason for non-consumption reported was not liking fish at 87% of non-consumers. Other
common reasons included too busy (36%), do not know how to prepare (28%) and allergy or
health concern (25%). The vast majority (87%) of consumers reported eating fish once per week
or less often, while about 10% eat fish 1-2 times per week (Table 6). This frequency information
was determined during the relatively short screening interview and did not involve detailed
probing of consumption frequency, species by species, of the type that occurred later in the
interview.
Of the 463 consumers who responded, 452 completed the first interview which collected detailed
consumption information. One respondent did not provide enough information to calculate an
FFQ consumption rate (described in more detail in Section 5.28, last paragraph), so the
remaining 451 respondents with calculable FFQ rates formed the primary sample for most tables
presented in this report. However, some tables may be based on more or fewer respondents
depending on analysis-specific inclusion/exclusion criteria.
Nez Perce Tribe
Volume II: Current Fish Consumption Survey
December 2016 Page 64
-------
Table 6. Nez Perce Tribe. Frequency of fish consumption based on 472 responders to the screening
questionnaire. Estimates are weighted.
Unweighted %
No.
Weighted %
Consumer*
Yes
98.1%
463
97.4%
No
1.9%
9
2.6%
If consumer, how many days per
week?**
< 1
86.3%
314
87.3%
1-2
10.4%
38
9.6%
2-3
2.8%
10
2.5%
3-4
0.0%
0
0.0%
4-5
0.6%
2
0.6%
5-6
0.0%
0
0.0%
6-7
0.0%
0
0.0%
If non-consumer, why?***
Contamination
0.0%
0
0.0%
(multiple reasons allowed)
Availability
0.0%
0
0.0%
Access to fishing
12.5%
1
7.3%
Do not like fish
75.0%
6
87.0%
Too busy to catch or prepare
25.0%
2
35.6%
Do not know how to prepare
12.5%
1
28.4%
Cannot afford fish
12.5%
1
7.3%
Allergies or health concerns
25.0%
2
34.0%
Vegetarian or vegan
0.0%
0
0.0%
Religious customs
0.0%
0
0.0%
*Consumer status was determined primary from the screening interview. Only respondents who sufficiently
completed the interview to determine consumer status were considered responders. One respondent claimed to be a
consumer during screening but then denied being a consumer during the first interview. This respondent was
classified as a non-consumer;
**364 consumers responded to this question;
***8 non-consumers responded to this question.
Nez Perce Tribe
Volume II: Current Fish Consumption Survey
December 2016 Page 65
-------
6.4 Demographic Characteristics
The tribe is diverse in demographic composition. Table 7 shows that in addition to the expected
diversity of gender and age, most of the respondents live in households with three or more
persons, about a quarter of the population are fishers, almost all of the population has finished
high school or obtained a GED (99%) and nearly half of members have attended some college
(45%). Household income is also diverse but with the majority of Tribal member respondents
falling into the range of $15,000-$45,000 per year annual household income. Of the consumers
included on the fishers list, 87% were male while 38% of non-fishers were male. More than half
of the fishers (56%) were between 18 and 39 years old.
Among female consumers, 82% reported giving birth. Of these women, 75% reported breast-
feeding or providing breast milk to their babies. Of those women who have finished breast-
feeding their youngest child, the median reported age at which they stopped was 6 months
(range: 1 to 46 months). Table D2 in Appendix D summarizes the same demographic variables
as Table 7, but without statistical weighting.
Table 7. Nez Perce Tribe. Demographic characteristics of consumers. Estimates are weighted.
% or Mean ± SD
No. who
Responded
Gender*
Male
49.9%
451
Female
50.1%
Age*
18-29 years
21.5%
451
30-39 years
19.6%
40-49 years
19.2%
50-59 years
17.8%
60 years or older
21.8%
Weight, kgs
89.4 ± 19.9
434
Weight, kgs (males only)
96.6 ± 19.4
239
Weight, kgs (females only)
81.7 ±17.5
195
No. in household
1
8.8%
451
2
19.4%
3-4
42.8%
5 or more
29.0%
Documented fisher*
Yes
24.2%
451
No
75.8%
Live on reservation
Yes
82.7%
449
No
17.3%
Nez Perce Tribe
Volume II: Current Fish Consumption Survey
December 2016 Page 66
-------
% or Mean ± SD
No. who
Responded
Highest education
Middle school
1.2%
448
High School / GED
54.0%
Associates degree
26.4%
Bachelor's degree
12.4%
Master's degree
5.2%
Doctorate
0.8%
Annual household income
<$15K
20.5%
410
$15K - $25K
20.8%
$25K - $35K
20.0%
$35K - $45K
12.6%
$45K - $55K
8.1%
$55K - $65K
5.6%
>$65K
12.3%
*From the Tribal enrollment file or the fishers list; other demographics were determined from the questionnaire.
Refer to Section 4.4 on Populations for a description of documented fishers. Some respondents who were not
documented fishers did or do fish.
6.5 FFQ Rates for Species and Groups of Species
FFQ consumption rate statistics for the Nez Perce Tribe, which include special event
consumption, are shown in Table 8. The Group 1 (all fish) consumption distribution is skewed
toward large values due to a number of consumers with high consumption rates. The mean of
123.4 grams per day among the 451 consumers with a calculable consumption rate is
accompanied by a standard deviation of 159.4, larger than the mean, indicating skewness toward
large values. In addition, the mean (123.4 g/day) is larger than the median (70.5 g/day), another
indication of skewness.
The 90th percentile of consumption, 270.1 grams per day, is more than twice the mean and
approximately four times the median, and the 95th percentile of consumption, 437.4 grams per
day, is approximately triple the mean and over six times as large as the median. The maximum
observed consumption rate was 1,371.9 grams per day.
Confidence intervals are presented for the means and percentiles of consumption. The width of a
confidence interval is a measure of the uncertainty in the specific estimated value. Regardless of
the width of the confidence interval, the estimated rate (statistically referred to as the "point
estimate") is a useful value and is methodologically superior to any other choice within the
confidence interval as an estimate of the percentile, because it has been derived by an unbiased
method. It is wrong to assume for these survey results that the range of a confidence interval—
from lower bound to upper bound—are all equally appropriate consumption rates to use as a
measure of the true population value. The choice of the "point estimate," for example, of 437.4
Nez Perce Tribe
Volume II: Current Fish Consumption Survey
December 2016 Page 67
-------
grams per day for the 95th percentile (FFQ method, Group 1 species), is the only estimate within
the interval that is derived by an unbiased procedure. It is the preferred value to use as the 95th
percentile.
In Group 2, the mean consumption rate is somewhat lower at 104.0 grams per day, and the
median consumption rate for Group 2, 61.3 grams per day, is approximately 85% as large as the
median for Group 1, Once again, this species group's consumption rate has values skewed
toward high consumption rates, weighting to a 90th percentile of 231.4 grams per day and a 95th
percentile of 327.9 grams per day. The maximum Group 2 consumption rate of 1323.8 grams per
day is, again, large but plausible. The consumption rates are presented in a graphic format in
Figures 2 and 3.
Groups 3 through 7 are mutually exclusive and completely subdivide Group 1. Among Groups 3-
7 the most consumed group is Group 3 (salmon and steelhead), with 446 consumers and a mean
consumption rate of 79.0 grams per day, followed by Group 6 (marine finfish and shellfish), with
308 consumers and a mean rate of 51.0 grams per day. Groups 4 (resident trout) and 5 (other
freshwater finfish and shellfish) had similar consumption with 136 and 150 consumers,
respectively, and mean rates of 13.5 grams per day and 14.3 grams per day. There were only 2
consumers of Group 7 (species not specified sufficiently well to place in one of the
aforementioned groups), with a mean rate of 8.1 grams per day.
Nez Perce Tribe
Volume II: Current Fish Consumption Survey
December 2016 Page 68
-------
Table 8. Nez Perce Tribe. Mean, median and selected percentiles of FCRs (g/day, raw weight, edible portion) in the Nez Perce Tribe, based on the FFQ;
consumers only. Estimates are weighted.
No. of
Species Group1 Consumers Mean SD Min
Percentiles
50%
55%
60%
65%
70%
75%
80%
85%
90%
95%
'"99%
Max
Group 1
(all finfish and shellfish)
451
123.4
159.4
0.4
70.5
79.1
91.2
109.5
124.5
137.6
163.9
207.4
270.1
437.4
795.9
1371.9
(95% CI)
(108.7-
146.5)
(63.6-
80.8)
(69.4-
94.8)
(76.8-
109.8)
(88.6-
126.7)
(106.4-
147.4)
(123.9-
166.6)
(143.9-
206.3)
(174.8-
264.7)
(221.0-
340.3)
(309.5-
522.6)
(562.1-
1172.0)
Group 2
(near coastal/estuarine/freshwater/anadromous
finfish and shellfish)
446
104.0
144.2
0.2
61.3
69.0
77.7
91.5
103.6
123.3
145.1
175.2
231.4
327.9
764.5
1323.8
(95% CI)
(92.0-
125.9)
(52.2-
69.5)
(59.7-
80.3)
(66.7-
94.2)
(76.8-
105.6)
(88.9-
128.2)
(104.1-
146.9)
(127.6-
176.3)
(151.1-
222.9)
(195.8-
288.6)
(250.9-
489.9)
(500.9-
1150.2)
Group 3
(salmon and steelhead)
446
79.0
119.7
0.2
45.2
49.5
58.0
65.6
75.7
89.4
107.1
131.7
166.1
247.3
706.7
949.8
(95% CI)
(68.9-
96.0)
(38.4-
55.3)
(45.9-
61.9)
(51.0-
70.0)
(58.8-
79.3)
(67.5-
96.6)
(78.3-
110.9)
(97.7-
135.4)
(114.1-
163.1)
(145.9-
205.5)
(200.7-
438.1)
(431.1-
798.1)
Group 4
(resident trout)
136
13.5
42.5
0.03
3.8
5.3
5.8
7.3
7.9
9.0
13.0
19.9
26.3
56.8
**129.3
544.2
(95% CI)
(8.2-
28.0)
(1.9-
6.2)
(2.8-
7.5)
(3.7-
8.1)
(5.1-
10.4)
(5.7-
13.9)
(7.5-
19.3)
(8.1-
22.0)
(11.0-
32.4)
(18.8-
56.5)
(28.6-
89.9)
(56.3-
428.3)
Group 5
(other freshwater finfish and shellfish)
150
14.3
32.1
0.02
3.7
5.0
6.2
7.5
8.6
11.2
14.9
20.4
34.2
75.9
"109.2
309.5
(95% CI)
(9.4-
21.9)
(2.0-
5.7)
(2.7-
7.2)
(3.4-
8.4)
(4.2-
11.1)
(5.3-
14.6)
(7.2-
20.5)
(8.7-
29.3)
(12.1-
45.1)
(19.0-
75.0)
(34.7-
103.2)
(77.6-
237.5)
Group 6
(marine finfish and shellfish)
308
51.0
77.6
0.1
29.8
33.8
37.9
44.9
52.8
57.7
70.0
74.9
93.3
155.4
363.0
731.8
(95% CI)
(42.3-
63.5)
(25.1-
34.4)
(28.4-
40.6)
(30.6-
46.7)
(34.5-
53.0)
(42.0-
58.1)
(48.6-
70.3)
(56.3-
80.0)
(68.4-
105.5)
(80.1-
151.2)
(124.4-
288.9)
(255.6-
521.6)
Group 7****
(unspecified finfish and shellfish)
2
8.1
4.9
-
-
-
-
-
-
-
-
-
-
-
-
-
*See Table 2 for definitions of species groups;
"Two or fewer expected respondents with rates equal to or greater than the reported percentile (approximately); interpret this percentile more cautiously;
** *Confidence intervals for the 99th percentile are less reliable because there are less than 5 respondents equal to or greater than the reported percentile (approximately); interpret these intervals more
cautiously;
** "There were only 2 consumers of unspecified species so only the mean and SD are presented.
Nez Perce Tribe
Volume II: Current Fish Consumption Survey
December 2016 Page 69
-------
o
o
o
CD
O
CO
o
h-
o
CD
O
LO
o
o
CO
o
CM
o u
{>¦ 8
. X
C 9
:/
C 9
:f
o#
/}
c »
; i
OS
;/
o®
;;
OB
o©
;j
o®
:j
o»
: j
OS
; i
c«
;/
:j
3
;j
3
.7
s
*
o
T
0
Means
i
100 200
Group 1 species (N=451)
Group 2 species (N=446)
—r i "T"" —r i
300 400 500 600 700 800
Consumption rate (g/day)
Figure 3. Nez Perce Tribe. Estimated cumulative distribution of FFQ FCRs (g/day, raw weight,
edible portion). Group 1 includes all fmfish and shellfish. Group 2 includes near coastal, estuarine.
freshwater, and anadromous finlish and shellfish. The percentiles are spaced every 5% on the vertical
axis, with the 1st percentile and 99'" percentiles also included. Estimates are weighted. The points are the
original estimates and the lines (solid and dotted) are linear interpolations between those estimates. The
mean consumption rates for both species groups are indicated with points on the horizontal axis.
Nez Perce Tribe
Volume II: Current Fish Consumption Survey
December 2016 Page 70
-------
6.6 FFQ Consumption Rates by Demographic Groups
As shown in Table 9, FFQ consumption rates for Group 1 (all fish) did vary substantially across
some of the demographic factors. See Table D3 in Appendix D for an expanded set of
consumption rate percentiles in addition to the 50th, 90th, and 95th percentiles. The documented
fishers (based on the fisher list) had a substantially higher consumption rate than the non-fishers
(or those tribal members who were not documented as fishing recently through the Tribe's
sampling program and monitoring activities). The mean of 171.8 g/day for fishers is 60% larger
than the mean for non-fishers at 107.9 g/day. The medians and higher percentiles for fishers are
also substantially higher than the corresponding values for non-fishers. As noted in Section 4.4
(Populations), some active fishers who were not on the fishers list may have been incorrectly
classified as non-fishers. Thus, it is possible that the difference in population consumption rates
between actual fishers and non-fishers is not correctly estimated by the difference between
labeled fishers and non-fishers presented in Table 9.
The survey included questions for respondents on their frequency of fishing (see questions #35
and #36 in Appendix A for question wording). A comparison of responses to these questions and
presence or absence on the fishers list shows that of 93% of those on the fishers list did report
fishing during the preceding 12 months. In the same group, 79% reported fishing more
frequently—at least 12 times in the preceding 12 months (a calculated average of once per month
or more). Among those not on the fishers list, 50% reported fishing during the last year but only
22% reported fishing at least once per month, on the average. Thus, those on the fishers list
include a much higher fraction of people who fish and a much higher fraction of more frequent
fishers than is found among those respondents not on the list. The fishers list contains about two-
thirds of the respondents who fish more frequently, defined as those fishing once per month or
more, on the average. (These calculations are based on 138 respondents on the fishers list and
313 respondents not on the fishers list, limited to those completing questions #35 and #36 of the
questionnaire.)
The mean consumption rate for males was higher than the female rate by 46%: a mean of 146.6
g/day versus 100.2 g/day, respectively.
Age had less of an impact on consumption rates, being relatively consistent (mean and median)
across all age groups except the oldest age group (60 years or older) which had the lowest mean
rate at 105.8 g/day.
Those living on the reservation had a higher mean consumption than those not living on the
reservation; higher percentiles of consumption were also larger for those living on the
reservation.
Household size did not show a consistent relationship with consumption rates. Nor did
education, with those completing high school (or GED) or less having about the same
consumption rate as those who reported some college education. There was also no consistent
pattern of consumption rates in relation to household income.
Nez Perce Tribe
Volume II: Current Fish Consumption Survey
December 2016 Page 71
-------
Table 9: Nez Perce Tribe. Estimated distribution of FFQ consumption rates (g/day, raw weight,
edible portion) of consumers within demographic groups. All rates are for total consumption
(Group 1). Estimates are weighted.
Group
No. of
Consumers*
Mean
SD
Percentiles
50%
90%
95%
Gender**
Male
241
146.6
179.3
87.4
285.1
488.3
Female
210
100.2
133.1
54.7
244.0
341.4
Age**
18-29 years
61
126.7
175.4
74.7
225.2
522.4
30-39 years
94
140.9
161.1
74.0
298.9
448.6
40-49 years
116
115.4
126.1
68.5
241.2
463.3
50-59 years
89
130.3
193.4
67.4
253.8
308.2
60 years or older
91
105.8
136.8
62.3
264.8
332.0
Documented Fisher**
Yes
138
171.8
207.2
98.0
436.8
543.5
No
313
107.9
137.5
65.5
232.9
337.7
Live on reservation
Yes
391
127.3
164.4
70.6
284.6
451.0
No
58
106.5
134.4
65.6
202.8
237.5
Number who live in household
1
37
133.9
179.3
82.0
288.3
***423
2
84
119.0
144.1
57.2
285.3
451.5
3-4
193
119.3
163.7
71.0
224.3
441.0
5 or more
137
129.2
158.0
74.0
284.0
381.1
Highest education
High school / GED or less
242
126.6
176.5
70.4
253.9
492.0
Associates degree or higher
206
120.4
136.5
70.7
275.0
409.0
Annual household income
LO
VI
79
122.9
168.7
69.7
282.4
324.9
LO
I
LO
219
126.6
165.9
71.1
250.8
488.7
>$45K
112
117.7
113.5
72.4
244.8
339.6
*Consumers with unknown or missing subgroup status were excluded for the analysis of that subgroup;
**From the enrollment list or fisher indicator list; other subgroups were determined from the questionnaire;
***Two or fewer expected respondents with rates equal to or greater than the reported percentile (approximately);
interpret this percentile more cautiously.
Nez Perce Tribe
Volume II: Current Fish Consumption Survey
December 2016 Page 72
-------
6.7 Effect of Changes in Study Design on FFQ Rates
The estimated mean and medians of FFQ fish consumption classified by two variables that
reflect mid-survey changes in design are shown in Table 10. The table compares FFQ rates of
consumption of Group 1 (all fish) species for interviewing at special events vs. regular
interviewing and for tribal vs. non-tribal interviewers. The table also compares FFQ rates for
home vs. non-home interviews, which is included here for convenience, though it does not
reflect a design change. The corresponding differences in means (comparing interviews with vs.
without a given characteristic), unadjusted or adjusted for other respondent characteristics, are
shown in Table 11.
The mean consumption for respondents interviewed at special events was 0.3 grams/day lower
compared to respondents not interviewed at special events. This difference reversed and was still
small (5.4 grams/day) once respondent characteristics were adjusted for. These differences were
not statistically significant (p = 0.8-1.0).
The mean FFQ consumption for respondents with tribal interviewers was 31.7 grams/day lower
compared to respondents with non-tribal interviewers. This difference was similar (30.7
grams/day) once respondent characteristics were adjusted for using a multivariate linear
regression model (Table 11). Both the unadjusted and adjusted difference were not statistically
significant (p = 0.3).
Finally, the mean consumption for respondents interviewed at home was 29.1 grams/day lower
compared to respondents interviewed elsewhere. This difference changed very little (23.0
grams/day) once respondent characteristics were adjusted for. Neither the unadjusted nor the
adjusted differences were statistically significant (p = 0.2-0.3).
While there are some small numeric effects of the variables considered, they are not statistically
significant and there is no need to adjust for them in presenting consumption rates for this
population. The effect of these variables on other species groups was not assessed because the
main part of this report focuses on Group 1 species and the assessment for the other groups
would be more limited due to the smaller sample sizes of data sets limited to the consumers of
the other (and more specific) species groups.
Table 10. Nez Perce Tribe. Mean and median of Group 1 (all fish) FFQ consumption rates (g/day,
raw weight, edible portion) by groups according to design variables. Weighted results.
Group
No.
Mean
Median
Not special event
393
123.4
72.2
Special event
67
123.1
60.7
Non-tribal interviewer
93
147.9
78.6
Tribal interviewer
365
116.2
68.9
Non-home interview
380
128.0
72.9
Home interview
77
98.9
65.4
Nez Perce Tribe
Volume II: Current Fish Consumption Survey
December 2016 Page 73
-------
Table 11. Nez Perce Tribe. Unadjusted and adjusted differences in mean Group 1 (all fish) FFQ
FCRs (g/day, raw weight, edible portion) by the design variables. Linear regression. Weighted
results.
Unadjusted
Adjusted
For Respondent
Characteristics*
Difference
Est.
SE
P
Est.
SE
P
Special event
-0.3
27.0
1.0
5.4
27.1
0.8
Tribal interviewer
-31.7
29.8
0.3
-30.6
28.1
0.3
Home interview
-29.1
20.3
0.2
-23.0
20.2
0.3
*Adjusted for ZIP code (83536, 83501 and others), age category (<30, 30-39, 40-49, 50-59 and 60+), gender, on/off
reservation, fishing (questions 35 and 36) and the respondent's physical weight (as a continuous predictor).
6.8 Consumption Rates from the NCI Method
The 24-hour recall data consisted of 850 interviews (single and paired interviews) from 451
respondents. Of the 850 interviews, 29.8% were conducted on the weekend (Friday, Saturday or
Sunday). A total of 399 respondents had two interviews, for which the average interval between
the interviews was 21 days (median: 11 days). The intervals were 21 days or less in 76% of those
with both interviews, between 21 and 90 days in 21%, and between 90 and 180 days in 3.2%.
One respondent had an interval of 205 days. Of the 399 respondents with two interviews, 43 had
two days with Group 1 positive fish consumption and 122 had one day with Group 1 positive
fish consumption. The remaining 52 respondents had only one interview. Of these 52, 14
respondents had Group 1 positive fish consumption.
There were 446 Group 2 consumers, with a total of 840 interviews, among which 29.9% were on
the weekend. Among the respondents in this group, 394 had two interviews. Of the 394
respondents, 33 had two days with Group 2 positive fish consumption and 108 had one day with
Group 2 positive fish consumption. The remaining 52 respondents had only one interview. Of
these 52, 9 respondents had Group 2 positive fish consumption.
The mean and selected percentiles of the distribution of the fish consumption rates calculated
from the 24-hour recall data by the NCI method are presented in Tables 12, 13 and 14 and in
Figure 4.22 Table 12 presents statistics for overall fish consumption (species Group 1) and Table
22 The NCI method as implemented in SAS software provides integer percentiles of usual consumption rates up to the 99th
percentile. However, an analysis of species Group 1 and species Group 2 consumption for the NPT (all respondents) showed a
lower calculated 99th percentile consumption rate for Group 1 (373.2 g/day) than for Group 2 (409.6 g/day), even though the nearby
95th percentile values were in the order expected (232.1 g/day and 221.8 g/day, respectively). The number of respondents in the two
analyses was very similar (though small for the NCI method), and Group 2 is a subset of the species in Group 1 and would be
expected to have a smaller actual 99th percentile in the population. However, it is not an error for these two estimated values of the
99 percentiles to be in an unexpected order. These are both estimates—not population values—for the 99th percentile for each
group of species, and—as indicated by the width of the confidence interval for the 99th percentile for Group 1 (276.2-692.7 g/day)—
there is a range of plausible values for these kinds of estimates. Among the plausible estimates for each of the two 99th percentiles,
some of the plausible choices will have the 99th in the expected order (Group 2 having a smaller 99th percentile than Group 1). In
Nez Perce Tribe
Volume II: Current Fish Consumption Survey
December 2016 Page 74
-------
14 for species Group 2 consumption. Table 13 shows the 95% confidence intervals for the
species Group 1 statistics among all NPT respondents and among NPT respondents on the fishers
list. The bootstrap distributions that were used to derive these distributions are shown in the
Appendix: Figure E20 (all respondents) and Figure E21 (fisher list only). Only 22 out of the
1,000 bootstrap models (2.2%) did not converge. The 22 resamples were excluded from the
confidence interval calculations.
The mean fish consumption in Groups 1 and 2 among all NPT respondents were 75.0 (95% CI,
57.3-104.6) g/day and 66.5 g/day, respectively. The 95th percentile of the distribution of fish
consumption in Groups 1 and 2 among all NPT respondents were 232.1 (95% CI, 165.0-379.7)
g/day and 233.9 g/day, respectively.
Fishers consumed more Group 1 fish than non-fishers (mean 98.2 g/day vs. 67.6 g/day) and men
consumed more than women (mean 87.7 g/day vs. 62.3 g/day). The means in the four ZIP code
groups (83540, 83536, 83501, and "Other" ZIP codes) were between 63.6 and 84.5 g/day. The
means ranged from 58.1 to 92.5 g/day across the five age groups, with the 60+ age group
consuming the least and the 30-39 age group consuming the most. Similar trends were observed
for Group 2 species.
More extensive tables that include lower percentiles of the Group 1 distributions, Group 2
distributions and confidence intervals for Group 1 for the additional percentiles reported are
available in Appendix Tables E1-E3, respectively.
order to avoid confusion in presentation of results, all NCI-method percentiles for Group 1 and Group 2 have been reported only up
to the 95th percentile.
Nez Perce Tribe
Volume II: Current Fish Consumption Survey
December 2016 Page 75
-------
« t? 1° 5 ^ ^ ^ ^ 4°
Consumption rates (g/day)
95
90
85
80
75
70
65
3^60
655
~50
§ 45
B 40
£35
30
25
20
15
10
5
0
Group 1 -* Group 2
Figure 4. Nez Perce Tribe. Distribution of the usual fish consumption (g/day, raw weight, edible
portion) based on the 24-hour recalls. Estimated by the NCI method. Group 1 includes all finfish and
shellfish. Group 2 includes near coastal, estuarine, freshwater, and anadromous finfish and shellfish.
Nez Perce Tribe
Volume II: CurrentFish Consumption Survey
December 2016 Page 76
-------
Table 12. Nez Perce Tribe. Distribution of the usual fish consumption of species Group 1 (g/day, raw weight, edible portion) based on the
24-hour recalls. Estimated by the NCI method.
Group
No. of
Consumers
Mean
Percentiles
50%
55%
60%
65%
70%
75%
80%
85%
90%
95%
Overall
451
75.0
49.5
56.4
64.6
73.9
85.1
98.9
115.7
138.5
173.2
232.1
Documented fisher
Fisher
138
98.2
64.7
74.3
85.2
97.9
113.2
130.4
154.1
184.1
229.2
305.0
Non-fisher
313
67.6
45.6
52.0
59.2
67.9
77.6
90.0
104.9
124.6
155.1
206.0
Gender
Men
241
87.7
58.4
66.7
76.3
87.2
99.8
115.3
134.1
161.9
199.8
268.1
Women
210
62.3
41.8
47.7
54.4
62.4
71.6
82.8
97.7
116.0
145.1
194.4
ZIP Code
83540
329
73.6
48.2
55.1
62.7
72.1
83.2
96.4
113.1
135.5
168.1
227.2
83536
39
84.5
58.1
67.4
77.4
88.9
101.5
117.6
136.2
164.2
197.9
246.9
83501
28
63.6
48.4
54.5
60.8
67.9
75.2
85.6
98.4
115.8
139.4
177.7
Other
55
79.8
49.2
56.8
65.9
76.5
88.8
102.7
120.7
148.8
193.8
264.2
Age
18-29
61
75.3
52.0
58.6
66.1
74.7
85.5
97.8
114.3
137.0
170.1
232.5
30-39
94
92.5
64.5
73.1
83.1
94.9
108.5
124.4
143.7
171.2
207.7
274.2
40-49
116
83.8
56.6
64.0
73.1
83.6
97.4
112.5
129.9
157.0
192.6
256.3
50-59
89
66.8
41.2
46.8
54.0
62.0
71.4
83.3
98.0
118.4
151.4
212.7
60+
91
58.1
37.7
43.0
49.6
57.3
67.6
77.7
92.9
110.5
136.5
182.5
Nez Perce Tribe
Volume II: Current Fish Consumption Survey
December 2016 Page 77
-------
Table 13. Nez Perce Tribe. Distribution of the usual fish consumption of species Group 1 (g/day, raw weight, edible portion) and their
95% confidence intervals based on the 24-hour recalls. Estimated by the NCI method.
No. of
Consumers
Mean
Percentiles
50%
55%
60%
65%
70%
75%
80%
85%
90%
95%
Overall
451
75.0
49.5
56.4
64.6
73.9
85.1
98.9
115.7
138.5
173.2
232.1
(95%
CI)
(57.3-
104.6)
(27.8-
67.8)
(33.8-
76.1)
(41.0-
86.5)
(49.5-
97.5)
(59.0-
111.6)
(69.9-
133.5)
(82.9-
161.2)
(97.8-
200.1)
(120.9-
262.3)
(165.0-
379.7)
Fisher
138
98.2
64.7
74.3
85.2
97.9
113.2
130.4
154.1
184.1
229.2
305
(95%
CI)
(66.3-
158.3)
(32.8-
106.5)
(38.6-
121.0)
(45.9-
137.9)
(54.8-
159.1)
(65.1-
184.2)
(78.2-
218.7)
(91.1-
257.7)
(112.9-
316.1)
(141.4-
401.6)
(196.7-
540.3)
Nez Perce Tribe
Volume II: Current Fish Consumption Survey
December 2016 Page 78
-------
Table 14. Nez Perce Tribe. Distribution of the usual fish consumption of species Group 2 (g/day, raw weight, edible portion) based on the
24-hour recalls. Estimated by the NCI method.
Group
No. of
Consumers
Mean
Percentiles
50%
55%
60%
65%
70%
75%
80%
85%
90%
95%
Overall
446
66.5
36.0
42.1
49.5
58.0
68.7
81.7
98.2
121.8
159.4
233.9
Documented fisher
Fisher
138
98.4
55.2
64.8
75.4
86.3
101.8
121.9
146.9
181.5
238.6
345.0
Non-fisher
308
55.6
32.0
37.0
43.2
50.8
59.4
70.6
84.1
102.2
132.0
189.5
Gender
Men
240
79.4
44.0
51.4
60.1
70.3
81.8
96.4
116.7
144.6
190.4
277.1
Women
206
55.0
29.0
34.0
39.8
47.5
56.3
67.9
82.7
102.8
135.6
198.0
ZIP Code
83540
326
65.5
34.7
40.6
48.2
56.7
67.0
80.2
97.0
120.7
158.4
232.3
83536
38
83.7
46.6
54.8
63.8
74.8
88.9
104.3
129.6
162.4
219.2
301.5
83501
27
64.0
41.6
48.0
54.3
64.6
75.6
87.6
104.8
123.3
150.6
197.4
Other
55
63.0
30.2
36.4
43.0
51.3
60.0
72.2
87.9
112.8
150.0
231.3
Age
18-29
61
76.9
49.4
56.6
64.2
72.5
82.5
93.7
108.4
130.3
167.0
249.4
30-39
94
83.7
53.1
61.0
69.2
79.0
90.4
104.0
122.5
147.6
189.0
262.8
40-49
115
65.1
43.6
48.9
54.9
62.5
71.1
81.7
95.0
114.2
142.8
196.6
50-59
88
55.2
33.8
38.3
43.6
49.9
57.7
67.5
80.4
96.9
122.1
173.0
60+
88
50.4
31.7
36.1
41.0
47.0
54.4
63.4
73.5
89.3
111.6
153.9
Nez Perce Tribe
Volume II: Current Fish Consumption Survey
December 2016 Page 79
-------
6.9 Quality Checking—NCI Method
Some quality checks were carried out to determine if certain assumptions of the NCI method
were met (see Section 5.23.3).
In order to check the NCI model results, certain distributions were examined to determine if they
were similar to a normal ("bell-shaped") distribution—a requirement of the NCI methodology.
The daily consumption rates were raised to an exponent power lambda(X) prior to this particular
assessment. The contractors examined the distribution of person-means (the mean for a
respondent using only their power-transformed consumption on their one or two 24-hour recall
days with non-zero fish consumption—if they had any such days). The contractors also
examined the distribution of within-person residuals. These residuals are the difference of a
respondent's power-transformed consumption on a 24-hour recall day from the mean of the two
power-transformed values for respondents with two non-zero fish consumption days. These
distributions of power-transformed values or residuals should appear approximately normal.
For several demographic subgroups the naive mean (calculated without the NCI method but
using survey weighting) was compared to the mean calculated from the NCI method. The naive
mean was compared to the NCI-method mean of: 1) the probability of consuming on a random
day, and 2) the mean consumption amount, conditional on a day having some fish consumption.
The first quality check examined the distribution of the person-means and within-person
residuals. The NCI models for species Groups 1 and 2 estimated a model lambda of 0.29 and
0.41, respectively, as powers for transformations that result in a distribution closest to the normal
distribution. As both powers are close to the third root (lambda = 0.33), the contractors
transformed the positive amounts of these consumptions of these species groups by taking the
third (cubic) root of the amounts. The distributions of the transformed person-means and the
within-person residuals were then examined. The histograms of these distributions are shown in
Appendix E, Figure E13 (Group 1) and Figure E14 (Group 2) and are, upon visual inspection,
relatively close to the normal distribution.
In the second quality check, naive and NCI method estimated consumption probabilities and
means of positive consumption were compared. The comparisons were carried out within groups
defined by the NCI model covariates are shown in Appendix Figures E15-E19. The covariates
included the presence on the fishers list (Figure El 5), gender (Figure E16), ZIP code (Figure
E17), age (Figure E18) and the FFQ decile (Figure E19).
For all covariates, the naive and NCI approaches revealed similar patterns of the consumption
probability and mean consumption amount across the different groups (e.g., the fishers and male
consumption are estimated to be higher than their complementary population groups by all
approaches). The means and probabilities from the naive approach that utilized both interviews,
however, tended to be higher than the NCI probabilities and means. This difference can be
attributed to the difference between the first and second interview.23 This difference between the
23 See Appendix E, Table E5 for the second interview coefficients in the NCI model, A23_SECINT (a coefficient for the second
vs. first interview mean in the amount portion of the NCI model) and P23_SECINT (a coefficient for the dally probability of
consumption in the probability portion of the NCI model). The positive values of these coefficient indicate that the mean amount
consumed on a consumption day and the probability of consumption on a randomly chosen day are higher in the second
interview than in the first interview.
Nez Perce Tribe
Volume II: Current Fish Consumption Survey
December 2016 Page 80
-------
naive and NCI method means was expected, because the second 24-hour recall mean
consumption (from a naive, survey-weighted analysis) was somewhat higher than the first 24-
hour recall mean (again, naive). This systematic difference was addressed during the NCI
analysis by using the mean from the first 24-hour interview recall as an unbiased estimate of the
population mean of usual intake, as described in Section 5.23.1. Thus, the naive mean that
averaged both the first (lower mean consumption) and second (higher mean consumption)
interviews was higher than the NCI mean, which used the mean from the first interview as an
unbiased estimate of the population mean.
An additional reason that the naive means differed somewhat from the NCI method means is that
the naive approach does not account for the weekday-weekend differences. Specifically, the
consumption amounts tended to be lower on the weekend than the weekdays and the weekend
interviews were under-represented in the sample compared to equal representation of the seven
days of the week (this is not unexpected as the interviewers were not instructed to achieve a
specific ratio of weekday and weekend interviews). About 30% of the 24-hour recall interviews
represented a weekend day versus 43% expected ([3 days]/[7 days] = 43%). The excess of
higher-consumption weekdays in the 24-hour interview data was addressed and adjusted in the
NCI method analysis, yielding a lower NCI mean than the naive mean.
As an additional quality check, the calculations of the estimates of the species Group 1
distribution (mean and percentiles) from the NCI method were also recomputed by NCI staff
(personal communication from Kevin Dodd to Moni Neradilek on July 2, 2015). The recomputed
mean and percentiles for species Group 1 were all within 0.4% of the contractors' estimates for
the Nez Perce Tribe and within 0.9% for the Shoshone-Bannock Tribes.
6.10 Sensitivity Analyses—NCI Model
We carried out a number of sensitivity analyses to understand the impact of various modeling
choices on the estimated means and percentiles. Detailed results of the sensitivity analyses are
presented in Appendix E, Tables E7-E17. All of the analyses in this section refer to comparisons of
means and percentiles when models with different specifications are run using the NCI method.
Model with logw FFQ replacing the 3rd root of the FFQ consumption rate. Compared to the final
model, the change in this one FFQ variable as a covariate in the model had the following effect.
The means for Group 1 species for NPT and SBT were 0.8% higher and 2.6% lower,
respectively, when adjusted for logio FFQ rather than the cube root of FFQ (Table E7). The
corresponding 95th percentiles were 8.3% higher and 0.4% lower, respectively. The differences
in means and the 95th percentiles between the two models were mostly small (<5%) for specific
subgroups. Somewhat larger differences (10—30%) were present for some of the 95th percentiles,
for the SBT mean for males, for the 18-29 age group and for the 60+ age group. Differences in
Group 2 means and 95th percentiles from the two different FFQ specifications were even smaller
than the differences for Group 1. Compared to the final model, the overall Group 2 means for
NPT and SBT were 0.2% and 1.2% higher, respectively, when adjusted for logio FFQ (Table
E8). The corresponding 95th percentiles were 3.3% lower and 1.9% higher, respectively. All
Group 2 differences in mean and percentile estimates for population subgroups were less than
13%) of the estimate from the final model using the cube root of FFQ.
Nez Perce Tribe
Volume II: Current Fish Consumption Survey
December 2016 Page 81
-------
Model with no weekend adjustment. Estimated means and 95th percentiles for Groups 1 and 2
were only slightly affected by presence or absence of the weekend adjustment (Tables E9 and
E10). Most of the estimates tended to increase when the weekend adjustment was not made, but
the differences were small (<7%, except for Group 2 estimates for the SBT age group 50-59,
which had approximately a 10% difference).
Model with no sequence effect adjustment. The final NCI models adjusted the estimated
consumption for the sequence of the interviews, calibrating the second interview consumption
amounts to correspond to the first interview consumption amounts. To investigate the impact of
this adjustment on the estimated distribution of fish consumption NCI models without this
adjustment were considered. Estimated means and 95th percentiles for Groups 1 and 2 increased
by 10-40% when the interview sequence was not addressed (Tables El 1 and E12). Compared to
the final model, the overall Group 1 means for NPT and SBT were 22.5% and 26.1% higher,
respectively. The corresponding 95th percentiles were 13.8% and 22.3% higher, respectively. The
overall Group 2 means for NPT and SBT were 24.4% and 30.1% higher, respectively. The
corresponding 95th percentiles were 19.2% and 25.3% higher, respectively. This increase can be
attributed to the higher mean consumption rate reported on the second interview. Section 5.23.1
further explains the choice to use the first interview as the reference interview.
Model with no correlation between consumption probability and consumed amount. Estimated
means and 95th percentiles for Group 1 and 2 were almost identical when the NCI model ignored
the correlation between the probability of consuming on a random day and consumption amount
(Tables E13 and E14). All estimates of means and 95th percentiles were within 0.2% of the final
model estimates for Group 1 species consumption and within 3.9% for Group 2 consumption.
Modelfit only to the NPT data. Compared to the NPT mean and percentile estimates from the final
model (using both NPT and SBT data), the Group 1 species mean and 95th percentile from the
model using only NPT data were 5.4% lower and 9.6% higher, respectively (Table El 5). In
estimates for population subgroups, species Group 1 means from the NPT-only model were
3.0-8.4%) lower and the 95th percentiles were 3.8-19.3%) higher. The species Group 2 estimated
mean and 95th percentile for the NPT population were 12.1% and 19.3% lower, respectively, when
the model was fitted only to the NPT data (Table El6). In population subgroups, Group 2 means
from the NPT-only model were 9.9-16.8% lower and the 95th percentiles were 5.6-23.6% lower.
Simpler model for Group 1. The simpler model for Group 1 consumption—a model which
included only the covariates for tribe, the 3rd root of the FFQ rate and the tribe by the 3rd root of
the FFQ interaction—had a relatively small effect on the estimated means and 95th percentiles
compared to the final model (Table El7). In most cases the estimates from the simpler model
differed from the final model estimates by <5%, and all of them differed by <15%.
In summary, the different sensitivity analyses showed the impact of the different modeling
choices on the NCI model estimates. For most estimates of mean and the 95th percentile 1.) the
use of log FFQ as covariate, 2.) the absence of the weekend adjustment, 3.) the use of no
correlation between consumption probability and consumed amount and 4.) a simpler model for
Group 1 resulted in <5% difference in the estimates (compared to the final model). The estimated
means and 95th percentiles for NPT changed up to 23.6% when the model was fit only to the
NPT data. When the model did not adjust for the interview sequence the estimates of the mean
and the 95th percentile increased by 10-40% (compared to the final model).
Nez Perce Tribe
Volume II: Current Fish Consumption Survey
December 2016 Page 82
-------
6.11 Comparison of FFQ Rates to 24-Hour and NCI-Method Rates
The estimated distributions of the 24-hour rates from the NCI method were limited to Group 1
and Group 2 species due to the very low number of double-hits for the other species groups
considered. The naive (survey-weighted) means for these two species groups have been
calculated.24 These means can be compared to the corresponding means from the FFQ rate
analysis. Under certain assumptions, the naive means have the same expected value as the FFQ
means. The assumptions include a steady state of consumption rates over time (including the
assumption of a steady state of the probability of consuming fish on a randomly chosen day),
accurate recall by all respondents when reporting fish consumption, and the assumption that the
underlying NCI model used to calculate the distribution of rates of consumption is the correct
model for the population and species groups being considered. Since the various assumptions
would usually be only approximately correct, it is appropriate to look for approximate agreement
of means. The estimates presented in this report also include the means for 24-hour rates for a
larger series of species groups using the standard, survey-weighted, "naive" method. Some
estimated means, 95th percentiles and ratios are presented in Table 15. Because the naive
approach does not adjust for the interview sequence (first vs. second interview) and weekend vs.
weekday effects on consumption, the naive 24-hour means for Groups 1 and 2 were, as expected,
larger than their NCI method counterparts. The higher naive 24-hour means were expected
because of the higher rates for the second interview and, to a smaller extent, because of smaller
mean consumption rates on the three days designated as the "weekend" (Friday-Sunday),
accompanied by fewer than 3/7ths of the 24-hour recall interviews occurring on the three days
designated as the weekend.
The mean for Group 1 (estimated by the NCI method from 24-hour data) was 61% of the
corresponding mean estimated from the FFQ while the 95th percentile estimated from the NCI
method was 53% of the FFQ estimate. The NCI-estimated Group 2 mean and the 95th
percentile were 64% and 71% of the FFQ values, respectively. The naive means from the 24-
hour data were lower than the corresponding FFQ means for all species groups, as shown by
the ratios (the 24-hour value divided by the corresponding FFQ value) in Table 15. Most of the
species had ratios between 0.33 and 0.88 (the mean of the Group 7 species consumption was
0.0, but it was based on only two consumers of this species). It is obvious that the two survey
methodologies are not in agreement in their estimates of the consumption rate distributions.
These findings are addressed with additional analyses in this section and are considered further
in the discussion section.
24 As noted in Section 6.9, the naive mean is calculated from the 24-hour recall data—without using the NCI method but using the
statistical survey weights.
Nez Perce Tribe
Volume II: Current Fish Consumption Survey
December 2016 Page 83
-------
Table 15. Nez Perce Tribe. Estimated means and 95th percentiles of consumption (g/day, raw weight, edible portion) by species group and
estimation method.
No. of
Species group Consumers
Mean
95th percentile
24h
FFQ
Ratio
24h
FFQ
Ratio
Mean
(naive
method)
Mean
(NCI
method)
#>0
#1
hit
#
2
hit
Mean
24h
(naive)
/FFQ
24h
(NCI)
/FFQ
Perc.
(NCI
method)
Perc.
NCI
/FFQ
Group 1: All Finfish and Shellfish
451
94.0
75.0
179
136
43
123.4
0.76
0.61
232.1
437.0
0.53
Group 2: Near Coastal/Estuarine/Freshwater/Anadromous Finfish and Shellfish
446
81.5
66.5
150
117
33
104.0
0.78
0.64
233.9
327.5
0.71
Group 3: All Salmon and Steelhead
446
69.2
126
99
27
79.0
0.88
247.3
Group 4: Resident Trout
136
4.8
2
2
0
13.5
0.36
56.9
Group 5: Other Freshwater Finfish and Shellfish
150
4.7
4
4
0
14.3
0.33
75.9
Group 6: Marine Finfish and Shellfish
308
31.4
65
62
3
51.0
0.61
155.4
Group 7: Unspecified Finfish and Shellfish Species
2
0.0
0
0
0
8.1
0.00
12.2
#>0 = number of consumers with at least one positive 24h recall,
# 1 hit = number of consumers with one positive 24h recall
# 2 hit = number of consumers with two positive 24h recalls
naive method = standard (weighted) sur\>ev estimate methods applied to the per-respondent averages of the 24-hour recalls
Nez Perce Tribe
Volume II: Current Fish Consumption Survey
December 2016 Page 84
-------
In order to better elucidate the difference in consumption rates calculated from the 24-hour recall
data and the rates calculated from the FFQ data, the analyses presented here show the difference
in rates for respondents classified into ten different ordered groups. The ten groups were defined
by deciles of the respondents' FFQ Group 1 consumption rates. Using these groupings of
respondents, this section also compares the FFQ and 24-hour rates for several species groups as a
function of Group 1 deciles. All means were calculated as weighted means using the survey
weights. This section also compares the FFQ-derived and 24-hour recall-derived frequencies of
consumption and typical portion sizes as a function of Group 1 deciles.
Finally, also reported here is an analysis of the relation between a.) the difference (gap) between
a respondent's FFQ and 24-hour recall consumption rates and b.) the respondent's uncertainty in
their FFQ responses. This analysis explores the possibility that the respondents who were less
certain in some of their responses might have larger differences in FFQ vs. 24-hour consumption
than those who were more certain in their responses. The first measure of respondent uncertainty
was the extent to which a respondent reported consumption of non-specific species groups rather
than individual species; for example, the respondent might report generic salmon consumption
(coded as "... salmon and steelhead / species not identified"), an indication of uncertainty, rather
than reporting consumption of specific species, such as coho or Chinook. The second measure of
uncertainty used in this analysis was the extent to which the respondent did not specify certain
aspects of consumption, such as the frequency of consumption of a species or the portion size
typically consumed.
Each respondent's Group 1 FFQ FCR was used to rank order the respondents from lowest to
highest FCR. Respondents in each tribe were then divided into deciles (ten approximately
25
equally sized groups ) according to their FFQ consumption rates from Group 1 species. These
decile groups defined by FFQ consumption of Group 1 species are used for all of the decile
analyses in this section. For each respondent and for each species group, such as Group 1, the
consumption rate from the 24-hour data was the mean consumption of the one or two days of
consumption that were assessed. Days with zero or positive consumption were included in the
calculation of the per-respondent mean. The number of responses with non-zero consumption in
the FFQ data and in the 24-hour recall data are shown in Table 16. These counts of respondents
also help in interpreting the tables that follow Table 16—in particular, Table 19, where these
numbers correspond to the number of consumers of the species group used to calculate the
means and ratios in the table.
Within each decile group, the average across the respondents of their mean daily consumption
(g/day) was calculated from their 24-hour recall responses. Similarly, in the same decile group,
the average daily consumption based on the FFQ responses was calculated. The decile group
averages from the FFQ data and from the 24-hour data were compared in the form of the ratio of
the 24-hour mean consumption rate to the FFQ mean consumption rate. As described later in this
section, similar ratios were calculated comparing 24-hour recall responses and FFQ responses on
frequencies of consumption and on typical portion sizes.26 The deciles were numbered in an
25 Decile groups are of exactly equal size only if the total sample size being divided into groups is a multiple of 10. If the total is not a
multiple of 10, some decile groups will have one additional respondent.
26 It can be easily shown that the 24-hour/FFQ ratios for consumption rates (presented later in Table 17) and frequencies of
consumption (Table 18) are equal to the ratios that would be calculated by including only consumers of the species group in
calculations of the mean FFQ consumption rate (the denominator of each ratio presented) and the mean 24-hour recall consumption
rate (the numerator of the ratio). That equivalence does not hold for the ratios for portion sizes (Table 19).
Nez Perce Tribe
Volume II: Current Fish Consumption Survey
December 2016 Page 85
-------
increasing order, with the first decile corresponding to the 10% of the respondents with the
lowest Group 1 FFQ consumption rate and the 10th decile corresponding to the 10% of the
respondents with the highest Group 1 FFQ consumption rate. The means of Group 1
consumption and Group 2 consumption from the 24-hour recall data and FFQ data for the ten
deciles are shown in Figures 5 and 6, respectively.
The NPT respondents in the lowest seven deciles had relatively similar mean Group 1
consumption rates between the 24-hour recall data and the FFQ data. Starting from the eighth
decile, the 24-hour recall means are progressively smaller compared to the FFQ means. In the
10th decile, the 24-hour recall mean was half the FFQ mean. These patterns were similar for
Group 2 consumption. More numeric details for this comparison can be found in the Appendix F,
Tables F1 (Group 1) and F2 (Group 2).
The analysis of the difference between consumption rates derived from the FFQ and the 24-hour
recall data includes consideration of the contribution of specific species groups to the Group 1
consumption rate. The specific species groups include Group 2 species (near
coastal/estuarine/freshwater/anadromous finfish and shellfish), non-Group 2 species, Group 3
species (salmon and steelhead), Group 4 species (resident trout) and Group 6 species (all marine
species); see Table 2 for the definitions of species groups. For this decile analysis (and only for
the decile analysis), the mean consumption rate for a decile or for all deciles combined has been
calculated including the non-consumer respondents of the species group considered. These non-
consumers of a species group have a zero consumption rate for the species group.
While not presented in the tabular results of this section, the means calculated including non-
consumers can be used to calculate the percentage of the Group 1 (all species) mean
consumption rate that is contributed by a smaller, embedded species group. For example, using
the Group 1 FFQ mean (all deciles combined) of 123.4 g/day and the corresponding Group 2
mean of 102.8 g/day, both from Table 17, the Group 2 species contribute 83% of the total
amount consumed of Group 1 species (all species combined). The analogous percentage based
on the 24-hour recall means was 86%, calculated as 100% x 80.6 g/day (Group 2 mean) / 94.0
g/day (Group 1 mean) using values from Table 17. As another example using the same table, it
can be calculated that the Group 2 species contribute 84% of the consumption of all species in
the 10th decile group, based on either the FFQ or 24-hour recall means. Throughout, the decile-
specific results should be interpreted more cautiously as each decile contains only one tenth of
the total sample size.
The comparison statistic of particular interest is the ratio of the 24-hour mean consumption rate
to the FFQ mean consumption rate—per decile and overall. A value of 1.0 indicates that the FFQ
mean and the 24-hour mean are in perfect agreement. Ratios smaller than 1.0 indicate that
consumption reported in the 24-hour recall interview is smaller than expected compared to
consumption reported in the FFQ interview. Ratios larger than 1.0 indicate larger consumption
reported in the 24-hour interviews than would be expected from the FFQ interviews.
Table 17 shows the mean consumption rates from the FFQ and 24-hour recall and their ratio.
More detailed versions of these tables can be found in Tables F3-F6 in Appendix F. Although
some differences among the species groups were observed in the ratios of 24-hour-to-FFQ means
or percentiles of consumption, the FFQ means were higher than the 24-hour recall means for all
species groups. The 24-hour/FFQ ratio of means ranged from 0.33 (Group 4) to 0.88 (Group 3).
For each species group, the comparison of the FFQ-based and 24-hour-based mean consumption
Nez Perce Tribe
Volume II: Current Fish Consumption Survey
December 2016 Page 86
-------
rates within each decile showed, generally, greater discrepancies at the lowest (1st and 2nd) and
highest (9th and 10th) deciles, with the 24-hour mean being greater than the FFQ mean at the
lower deciles and running the opposite direction at the higher deciles.
The usual daily consumption rate depends on the frequency of consumption and the portion size
typically consumed. Thus, it is important to consider the role of each—frequency and portion
size—as they may affect the observed differences between consumption rates calculated using
the two different sources of data: FFQ and 24-hour recall. The consumption rate estimate for a
respondent and for a particular species is the product of frequency of consumption multiplied by
the portion size. This product calculation, per respondent, then becomes a numerical component
of the consumption rate calculated "downstream" for a group of species and for a group of
respondents. In order to understand whether the differences between the FFQ and 24-hour recall
means were driven by the reported consumption frequency or by the reported portion size (or
both), this section includes a comparison of the FFQ-derived and 24-hour recall-derived mean
frequencies and portion sizes by decile and overall, presented by species group (Tables 18, 19
and Appendix Tables F7 and F8.)
Comparison of frequencies and portion sizes between the FFQ and 24-hour recall data. For each
respondent and for each species group considered, the following four values were calculated,
describing frequency of consumption or portion sizes.
a.) FFQ-based expected frequency of consumption. For a given species group, the expected
frequency of consumption was calculated as the sum of the individual FFQ-reported frequencies
(portions per day) for all species included in the species group. This approximation is most
accurate if no more than one species is eaten per day; the approximation overestimate the daily
frequency of consumption as the incidence of eating multiple species in a day increases. In this
section the frequency for each species is expressed as the probability of that species being
consumed on a randomly selected day. Thus, for example, if a respondent noted eating Chinook
salmon three times per week (interpreted as three days per week), the daily probability would be
3/7 = 0.43. If a respondent reported two periods of consumption for the species (a higher and a
lower consumption period during the year—an option permitted in the questionnaire), the daily
probabilities for each period were combined in a weighted average: the two probabilities were
weighted by the duration of each period. The sum across species of these daily probabilities
would equal the probability of consumption of fish—from the species group considered—on a
randomly selected day of the year.27 (The sum of probabilities was capped at 1.0, a value that
indicates consumption of fish from the species group every day.) As mentioned above,
calculation of this daily probability assumes that, at most, only one species is eaten on any given
day. That assumption appears to be approximately correct. Among the 222 days with fish
consumption reported on the 24-hour recall interviews (counting all respondents and all of their
24-hour recall days), only ten days (4.5%) showed two or more species consumed. The following
is offered in support of the assumption that, approximately, only one species is eaten per day.
Among the survey respondents and among the adult members of the tribal population, it seems
likely that the percentage of consumption days with two or more species consumed is lower than
the percentage value just noted. The reason is that the survey consumption days with "hits" are
27 The probability is readily converted to the more familiar frequency designation by multiplying the probability by a period of time,
such as a week. For example, a probability of 0.25 is the same as 0.25 x 7 days = 1.75 days per week (or 7 days out of 28), on the
average.
Nez Perce Tribe
Volume II: Current Fish Consumption Survey
December 2016 Page 87
-------
more likely to come from the more frequent fish consumers among the respondents. The balance
of the respondents (who had no days with hits in this survey) are likely to consume fish less
frequently. It also seems likely that the more frequent consumers would more often consume two
species or more on one of their consumption days than would be found among the less frequent
consumers. If that is the case, then the days with hits in the survey would find the "two-or-more
species" consumers over-represented relative to the entire sample of respondents or the entire
adult population. Thus, the 4.5% of consumption days with two or more species consumed
would be biased upward relative to what would be found in the long-term experience of the
sample of respondents or the population.
b.) Expected frequency of consumption based on the 24-hour recall data. This empirical
frequency is simply the number of days that a respondent had a "hit" divided by the number of
days for which the respondent provided a 24-hour recall interview for fish consumption. The
possible values of this ratio are very limited: 0 (zero), 0.5 or 1.0, depending on whether the
respondent reported zero hits or one hit on one 24-hour interview, or zero, one or two hits on two
interviews. This very limited selection of frequencies is obviously too coarse to be accurate for
an individual, and therefore these probabilities are used only in aggregate form (by taking a
mean) for groups of respondents.
c.) FFQ-based expected portion size on days of consumption. For each species group, a weighted
mean of the respondent's reported portion size for each of the group's constituent species was
calculated. The per-species statistical weights (used in the weighted mean portion size for a
specific respondent) were calculated as the reported frequency of consumption of that species
(from the FFQ) divided by the sum of the respondent's reported frequencies for all species
28
within the group. This sum—in the denominator of the statistical weight calculations—is the
FFQ-based expected frequency of consumption (of any species in the group) described in a.)
above. Division of the consumption frequency of a single species by this sum then yields a
statistical weight for that species to be used in the calculation of mean portion size. For example,
considering Group 1 (all species), if a respondent reported consuming Chinook salmon six times
a month, tuna three times a month and shrimp once a month (and no other species were
consumed), the sum of the frequencies would be ten. The corresponding statistical weights to be
applied to Chinook salmon, tuna and shrimp typical portion sizes (as offered by the respondent)
would be 6/10, 3/10 and 1/10, respectively. This weighted mean of portion sizes represents the
average amount in grams consumed, averaged over occasions when fish was eaten. In this
example, Chinook was consumed twice as often as tuna, so it would have twice the weight in the
mean calculation.
d.) Expected portion size based on the 24-hour recall data. This quantity was calculated in three
steps: 1.) for each respondent, calculate an unweighted mean of non-zero portion sizes over each
species consumed and across all eating occasions (e.g., lunch, dinner, etc.) reported on the first
24-hour recall interview, if there was any positive fish consumption reported on the first
interview. (See the example, below, for the calculation of the unweighted mean.) 2.) Calculate an
analogous unweighted mean of non-zero portion sizes reported on the second 24-hour recall
interview, if there was any positive fish consumption reported on the second interview. 3.)
28 Note that if a respondent did not consume a particular species within the group, the frequency would be zero, and, thus, all of the
respondent's non-consumed species would have no influence on the statistical weights or the respondent's mean portion size for
the species group.
Nez Perce Tribe
Volume II: Current Fish Consumption Survey
December 2016 Page 88
-------
Determine an unweighted average of the results of steps 1 and 2, if both days had hits. If there
was a hit on only one 24-hour recall interview, then the unweighted mean from the particular
interview was used as the mean 24-hour recall portion size for the respondent. If there were no
hits, then the expected portion size was undefined/unknown for that respondent. Such
respondents were not included and were not intended to be included in the calculation of the
mean portion sizes from 24-hour recall data. The portion size calculations were performed
separately for each species group. The following is an example of the calculation of the
unweighted mean portion size for a given day of consumption. If a respondent reported on one
24-hour recall interview that he or she ate 200 grams of Chinook salmon for lunch, 100 grams of
tuna for a snack and 300 grams of Chinook salmon for dinner (and did not report eating any
other fish that day), then the mean portion size for Group 1 (all species) would be 200 grams
(600 grams total divided by three eating occasions). As in the computation of frequencies, these
per-respondent average portion sizes may not be very precise for each respondent, but they can
be used for calculation of a more precise mean portion size for a group of respondents, such as
the respondents in a decile group.
Survey-weighted means for the frequencies and portions (described in a-d above) were
calculated for each decile's group of respondents, and also for all deciles combined. For the
portion calculations (c and d), a decile's survey-weighted mean portion size from FFQ data (item
c) was calculated including only respondents with positive consumption rates for the particular
species group. Similarly, a decile's survey-weighted mean calculated from 24-hour recall data
(item d) included only respondents who reported positive fish consumption on at least one of the
24-hour recall days.
As shown in Tables 18 (mean frequency) and 19 (mean portion size), the lower consumption rate
from the 24-hour recall relative to the FFQ came from both lower estimated frequency of
consumption and lower estimated portion size. For Group 1 mean consumption (all deciles
combined) the 24-hour recall's lower consumption, relative to the FFQ, was in about the same
proportion for frequency (24-hour/FFQ ratio: 0.85) and portion size (ratio: 0.87). The smallest
ratios (smallest 24-hour frequency or portion size relative to FFQ) occurred primarily at the
higher deciles of consumption. There were similar patterns for Group 2 consumption, with a
frequency ratio of 0.86 (comparing means) and portion size ratio of 0.89 (24-hour/FFQ). More
detailed summaries of the other species groups are summarized in Tables F7 (mean frequency)
and F8 (mean portion size) in Appendix F.
An additional analysis assesses the relation of a respondent's uncertainty in his or her FFQ
responses to the difference between their FFQ and 24-hour recall means. A small proportion of
the respondents (9%) reported some of their fish consumption without designating the specific
species consumed (e.g., a response coded as "All salmon and steelhead/species not identified",
see Figure F1 and Table F9 in Appendix F). Some respondents also had missing data
(frequencies, portion sizes or both) for one or more species (Figure F2). The relationship of
uncertainty and the FFQ-24-hour difference was analyzed using regression analysis. The FFQ-
minus-24-hour difference in consumption rates (per respondent) was the dependent variable and
the number of unspecified species was the independent variable (Table F10). In a second
regression analysis, the number of species with missing data was the independent variable (Table
F11). The analysis showed no compelling evidence to support an impact of these two uncertainty
factors on the FFQ-24-hour difference, but the confidence intervals for the impact of each of the
Nez Perce Tribe
Volume II: Current Fish Consumption Survey
December 2016 Page 89
-------
two uncertainty measures on the FFQ-24-hour consumption rate difference were so wide that the
analysis is inconclusive. The methods and the results are included for methodologic interest in
Appendix F as referenced above (see Figures F1F2 and Tables F9-F11 for more details.)
In summary, the larger reported consumption rates from the FFQ method than from the NCI
method based on 24-hour recall data were present for the several species groups considered.
Underlying this difference is a corresponding difference in the calculated frequency and portion
size of consumption. These differences were most pronounced among the 10%-20% of
respondents with the largest (Group 1, all species) FFQ consumption rates.
Nez Perce Tribe
Volume II: Current Fish Consumption Survey
December 2016 Page 90
-------
Table 16. Nez Perce Tribe. Number of respondents with consumption on the FFQ and 24-hour recall by species group and decile of FFQ
consumption rate. These show the number of non-zero values included in the calculations for Tables 17 and 18, and the sample sizes for
each cell in Table 19.
ALL
DECILE
1
2
3
4
5
6
7
8
9
10
Group 1
Respondents with >0 consumption on the FFQ
451
46
45
45
45
45
45
45
45
45
45
Respondents with >0 consumption on the 24h recall
179
8
15
11
13
16
20
17
24
26
29
Group 2
Respondents with >0 consumption on the FFQ
446
41
45
45
45
45
45
45
45
45
45
Respondents with >0 consumption on the 24h recall
150
6
10
7
11
12
17
16
21
25
25
Group 3
Respondents with >0 consumption on the FFQ
446
41
45
45
45
45
45
45
45
45
45
Respondents with >0 consumption on the 24h recall
126
5
9
5
10
12
13
12
19
21
20
Group 4
Respondents with >0 consumption on the FFQ
136
1
11
6
11
6
16
9
20
24
32
Respondents with >0 consumption on the 24h recall
2
0
0
0
0
0
0
0
0
1
1
Group 5
Respondents with >0 consumption on the FFQ
150
2
11
14
10
10
20
12
20
26
25
Respondents with >0 consumption on the 24h recall
4
0
0
0
0
0
1
0
1
1
1
Group 6
Respondents with >0 consumption on the FFQ
308
29
33
32
28
24
33
28
28
35
38
Respondents with >0 consumption on the 24h recall
65
3
7
8
3
5
9
8
7
6
9
Group 7
Respondents with >0 consumption on the FFQ
2
0
0
0
0
0
0
0
1
0
1
Respondents with >0 consumption on the 24h recall
0
0
0
0
0
0
0
0
0
0
0
Group 1 = all finfish and shellfish; Group 2 = near coastal/estuarine/freshwater/anadromous finfish and shellfish; Group 3 = all salmon and steelhead; Group 4 = resident
trout; Group 5 = other freshwater finfish and shellfish; Group 6 = marine finfish and shell fish; Group 7 = unspecified finfish and shellfish species (see Table 2).
Nez Perce Tribe
Volume II: Current Fish Consumption Survey
December 2016 Page 91
-------
Table 17. Nez Perce Tribe. Weighted mean consumption (g/day) from the 24-hour recall and FFQ interviews for each species group,
overall and by decile of FFQ consumption rate. Deciles are calculated from the group 1 FFQ consumption rate. All rows are based on all
group 1 consumers. Ratios were not calculated when a species group was not consumed by the FFQ.
ALL
DECILE
1
2
3
4
5
6
7
8
9
10
No. of respondents
451
46
45
45
45
45
45
45
45
45
45
Group 1
FFQ mean consumption, g/day
123.4
8.6
24.2
43.2
52.8
64.7
79.8
110.2
144.4
215.8
516.1
24h mean consumption, g/day
94.0
22.2
46.9
37.2
48.3
74.2
96.3
100.7
106.9
162.6
254.4
24h/FFQ consumption
0.76
2.59
1.94
0.86
0.92
1.15
1.21
0.91
0.74
0.75
0.49
Group 2
FFQ mean consumption, g/day
102.8
5.9
19.4
31.5
41.5
51.7
63.9
93.7
120.1
187.2
434.1
24h mean consumption, g/day
80.6
17.4
35.3
26.6
41.6
52.8
86.1
88.7
99.5
152.8
213.4
24h/FFQ consumption
0.78
2.96
1.82
0.84
1.00
1.02
1.35
0.95
0.83
0.82
0.49
Group 3: 24h/FFQ consumption
0.88
3.43
1.50
0.80
1.11
1.18
1.48
1.08
0.90
1.01
0.55
Group 4: 24h/FFQ consumption
0.36
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.39
0.47
Group 5: 24h/FFQ consumption
0.33
0.00
0.00
0.00
0.00
0.00
2.93
0.00
0.66
0.12
0.17
Group 6: 24h/FFQ consumption
0.62
1.76
3.08
1.02
0.52
1.15
0.63
0.70
0.38
0.48
0.41
Group 7: 24h/FFQ consumption
0.00
-
-
-
-
-
-
-
0.00
-
0.00
Group 1 = all finfish and shellfish; Group 2 = near coastal/estuarine/freshwater/anadromous finfish and shellfish; Group 3 = all salmon and steelhead; Group 4 =
resident trout; Group 5 = other freshwater finfish and shellfish; Group 6 = marine finfish and shell fish; Group 7 = unspecified finfish and shellfish species (see
Table 2).
Nez Perce Tribe
Volume II: Current Fish Consumption Survey
December 2016 Page 92
-------
Table 18. Nez Perce Tribe. Weighted mean expected frequency (percentage of days) with fish consumption from the 24-hour recall and
FFQ interviews for each species group, overall and by decile of FFQ consumption rate. Deciles are calculated from the group 1 FFQ
consumption rate. All rows are based on all group 1 consumers. Ratios were not calculated when a species group was not consumed, based
on the FFQ responses.
ALL
DECILE
1
2
3
4
5
6
7
8
9
10
No. of respondents
451
46
45
45
45
45
45
45
45
45
45
Group 1
FFQ mean frequency, %
31%
5%
11%
19%
21%
23%
24%
35%
42%
58%
78%
24h mean frequency, %
26%
10%
22%
14%
16%
22%
31%
30%
32%
39%
50%
24h/FFQ frequency
0.85
1.98
2.02
0.73
0.75
0.94
1.27
0.86
0.77
0.68
0.64
Group 2
FFQ mean frequency, %
25%
4%
8%
12%
15%
17%
19%
27%
34%
47%
68%
24h mean frequency, %
21%
8%
15%
7%
14%
14%
27%
26%
29%
36%
39%
24h/FFQ frequency
0.86
2.20
1.81
0.60
0.89
0.81
1.45
0.97
0.84
0.78
0.57
Group 3: 24h/FFQ frequency
1.00
2.48
1.55
0.60
1.00
1.06
1.45
1.16
1.00
1.05
0.65
Group 4: 24h/FFQ frequency
0.12
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.39
0.12
Group 5: 24h/FFQ frequency
0.45
0.00
0.00
0.00
0.00
0.00
3.08
0.00
0.99
0.13
0.28
Group 6: 24h/FFQ frequency
0.71
1.39
3.16
0.92
0.37
0.87
0.94
0.55
0.59
0.39
0.56
Group 7: 24h/FFQ frequency
0.00
-
-
-
-
-
-
-
0.00
-
0.00
Group 1 = all finfish and shellfish; Group 2 = near coastal/estuarine/freshwater/anadromous finfish and shellfish; Group 3 = all salmon and steelhead; Group 4 =
resident trout; Group 5 = other freshwater finfish and shellfish; Group 6 = marine finfish and shell fish; Group 7 = unspecified finfish and shellfish species (see
Table 2).
Nez Perce Tribe
Volume II: Current Fish Consumption Survey
December 2016 Page 93
-------
Table 19. Nez Perce Tribe. Weighted mean portion size (grams) from the 24-hour recall and FFQ for each species group, overall and by
decile of FFQ consumption rate. Deciles are the deciles of the group 1 FFQ consumption rate. Each individual's portions sizes were
averaged across species with a weight according to the species consumption frequency. All calculations are limited to positive portion
sizes. Ratios were not calculated when a species group was not consumed based on the FFQ or 24-hour recall.
ALL
DECILE
1
2
3
4
5
6
7
8
9
10
No. of respondents
451
46
45
45
45
45
45
45
45
45
45
Group 1
FFQ mean portion size, grams
356
198
259
271
298
335
380
374
404
445
618
24h mean portion size, grams
310
220
232
264
308
329
293
284
302
372
376
24h/FFQ portion size
0.87
1.11
0.90
0.98
1.03
0.98
0.77
0.76
0.75
0.84
0.61
Group 2
FFQ mean portion size, grams
373
190
285
296
293
340
394
393
406
492
638
24h mean portion size, grams
333
225
259
370
313
392
295
284
322
379
405
24h/FFQ portion size
0.89
1.18
0.91
1.25
1.07
1.15
0.75
0.72
0.79
0.77
0.64
Group 3: 24h/FFQ portion size
0.89
1.23
0.82
1.23
1.08
1.13
0.85
0.69
0.77
0.76
0.64
Group 4: 24h/FFQ portion size
2.10
-
-
-
-
-
-
-
-
1.23
2.32
Group 5: 24h/FFQ portion size
0.86
-
-
-
-
-
0.81
-
0.68
1.08
0.67
Group 6: 24h/FFQ portion size
0.90
0.96
0.96
1.01
0.95
1.17
0.61
1.14
0.61
0.95
0.80
Group 7: 24h/FFQ portion size
-
-
-
-
-
-
-
-
-
-
-
Group 1 = all finfish and shellfish; Group 2 = near coastal/estuarine/freshwater/anadromous finfish and shellfish; Group 3 = all salmon and steelhead; Group 4 =
resident trout; Group 5 = other freshwater finfish and shellfish; Group 6 = marine finfish and shell fish; Group 7 = unspecified finfish and shellfish species (see
Table 2).
Nez Perce Tribe
Volume II: Current Fish Consumption Survey
December 2016 Page 94
-------
Figure 5. Nez Perce Tribe. Weighted Group 1 means (g/day) of the 24-hour recall and of the FFQ
consumption rates by Group 1 FFQ deciles. The numerical values for the means are tabulated in
Appendix Table Fl.
Figure 6. Nez Perce Tribe. Weighted Group 2 means (g/day) of the 24-hour recall and of the FFQ
consumption rates by Group 1 FFQ deciles. The numerical values for the means are tabulated in
Appendix Table F2.
Nez Perce Tribe
Volume II: Current Fish Consumption Survey
December 2016 Page 95
-------
6.12 Consumption at Special Events and Gatherings
The FFQ rates presented throughout this report include consumption at special events and
gatherings, while this section summarizes, specifically, annual consumption at special events
only. Consumers reported attending an average ofll.3± 15.1 special events or gatherings per
year (median: 6). Of those who consumed at special events, their consumption at events was, on
average, 11.5 ± 13.8% of their total consumption (median: 6.7%). Table 20 summarizes how
often selected species and groups were consumed at special events and gatherings. Salmon and
steelhead were the most common species group consumed, with 96% of salmon/steelhead
consumers eating from this species group at an average of 10.4 events per year. The large mean
number of events per year where suckers and whitefish are consumed (19.3 events per year) is
due to the fact that the seven members who consumed these two species at special events
attended more than twice the number events per year than the overall average for all consumers
(28.9 vs. 11.3 events per year).
Table 20. Nez Perce Tribe. Frequency of consumption at special events and gatherings for selected
species and groups. Does not include consumption outside of special events and gatherings.
Estimates are weighted29.
Species or Species Group
Salmon and/or
Steelhead
Resident
Trout
Sturgeon
Suckers
and/or
Whitefish
No. of consumers (based on the FFQ)
446
136
51
28
% who consume from the species or species
group at special events
95.6%
17.5%
45.2%
29.4%
Events per year where species or species
group is consumed*
10.4 ± 14.5
6.8 ±9.1
8.1 ±10.7
19.3 ± 17.4
* Values are mean ± SD from those who consume at special events.
6.13 Fish Parts Eaten, Preparation Methods and Sources
The percent of the time skin, eggs and the head, bones and/or other organs were consumed are
summarized in Table 21. The skin was commonly consumed for salmon/steelhead and resident
trout while the other parts were much less frequently consumed for any species group.
29 As described in Section 5.20, unless noted otherwise, "weighted" estimates indicate that the survey statistical weights were used
in calculating the statistics presented.
Nez Perce Tribe
Volume II: Current Fish Consumption Survey
December 2016 Page 96
-------
Table 21. Nez Perce Tribe. Percent of the time other fish parts were consumed for selected species
and species groups. Consumers only*. Estimates are weighted.
Item
Species or Species Group
Salmon and/or
Steelhead
Resident Trout
Sturgeon
Suckers and/or
Whitefish
Skin
44.8 ± 47.7%
(418)
36.3 ±46.1%
(122)
12.1 ±29.8%
(44)
7.8 ± 25.6% (24)
Eggs
2.2 ±12.3%
(309)**
2.2 ±14.4%
(117)
6.9 ±22.1%
(42)
0.4 ±2.1% (20)
Head, bone and/or organs
3.9 ±14.9%
(309)**
6.4 ± 22.2%
(117)
1.8 ±11.9%
(42)
10.0 ±29.2% (20)
Values are mean ± SD; (sample size). Those who did not did not report a percentage value are excluded from
calculation of the statistics in the given cell, e.g., consumption of sturgeon eggs.
Note: Missing values for eggs and head/bones/organs were interpreted as 0% if the respondent did not choose "Not
applicable" or "Don't know or refused."
*Consumer status determined based on annual consumption reported in the FFQ;
**One interviewer frequently entered "Not applicable" to the question about consuming salmon and steelhead eggs,
head, bone and other organs, contributing to a large number of missing values for these cells.
Table 22 shows the percentage of the time different preparation methods were used. Baked or
broiled was a common preparation for all listed species (mean: 45.5-62.8% of the time,
depending on the species). Smoking was also common for salmon/steelhead (mean: 19.6% of the
time) and sturgeon (mean: 28.8% of the time). Dried or use in soups were uncommon (mean
<5% for each listed species).
Table 22. Nez Perce Tribe. Percent of the time different preparation methods were used for selected
species and species groups. Consumers only*. Estimates are weighted.
Method
Species or S
secies Group
Salmon and
Steelhead (N=445)
Resident Trout
(N=131)
Sturgeon (N=50)
Suckers and
Whitefish (N=24)
Baked or broiled
62.8 ± 27.8%
45.5 ± 45.2%
45.6 ± 45.0%
46.2 ± 47.0%
Smoked
19.6 ± 19.8%
4.0 ± 13.5%
28.8 ± 38.8%
0.0 ± 0.0%
Dried
3.9 ± 8.2%
0.2 ±1.8%
0.9 ± 3.9%
0.7 ±4.1%
In a soup
2.0 ± 5.8%
0.5 ± 4.3%
2.2 ± 5.6%
4.5 ± 13.2%
Other
11.7 ±22.7%
49.8 ±47.1%
22.4 ± 38.2%
48.6 ± 46.8%
Values are mean ± SD;
Note: Missing values for any preparation method were interpreted as 0% if the total of non-missing values was
100%;
*Consumer status determined based on annual consumption reported in the FFQ. Those who did not report any
percentage values for a specific species or species group were excluded from the corresponding column;
**Grilled was the most common "Other" preparation method for salmon and steelhead while fried was the most
common method for resident trout, sturgeon, suckers and whitefish.
Nez Perce Tribe
Volume II: Current Fish Consumption Survey
December 2016 Page 97
-------
The percentage of the time consumed fish were obtained from different sources is summarized in
Table 23. Salmon/steelhead and resident trout were most often caught in Idaho waters at 74.4%
and 89.6% of the time on average, respectively.
Table 23. Nez Perce Tribe. Percent of the time selected species and species groups were consumed
from different sources. Consumers only*. Estimates are weighted.
Variable
Species or Species Group
Salmon and/or
Steelhead (N=442)
Resident Trout
(N=130)
Sturgeon
(N=51)
Suckers and/or
Whitefish (N=24)
Bought from a store
(grocery or market)
1.9 ±9.5%
1.0 ±9.7%
0.1 ±0.7%
20.1 ±31.4%
From a restaurant
1.3 ±5.9%
1.7 ±11.2%
4.7 ± 20.6%
18.4 ±29.9%
Caught by you or someone else
(in Idaho waters)
74.4 ±31.3%
89.6 ± 26.8%
25.6 ±40.1%
58.2 ± 46.3%
Caught by you or someone else
(outside of Idaho)
21.8 ±29.2%
6.9 ±21.7%
64.4 ± 44.3%
3.3 ± 12.0%
Other
0.6 ± 5.9%
0.8 ± 9.0%
5.2 ± 22.5%
0.0 ± 0.0%
Values are mean ± SD;
Note: Missing values for any preparation method were interpreted as 0% if the total of non-missing values was
100%;
*Consumer status determined based on annual consumption reported in the FFQ. Those who did not report any
percentage values for a specific species or species group were excluded from the corresponding column.
Nez Perce Tribe
Volume II: Current Fish Consumption Survey
December 2016 Page 98
-------
6.14 Fishing Activities
Based on the questionnaire responses, it is estimated that 61% of consumers took part in fishing
activities over the past year. Figure 7 shows the mean number of times respondents went fishing
each month. June had the highest fishing frequency, followed by July and then May. January and
December had the lowest fishing frequencies. Table 24 summarizes overall fishing frequency
and respondents' access to fishing gear and boats.
CD -i
LD
CO
CM -
Jan Feb Mar Apr May Jun Jul
Month
Aug Sep Oct Nov Dec
Figure 7. Nez Perce Tribe. Mean number of times respondents went fishing each month among the
283 respondents who reported fishing at least once. Estimates are weighted.
Nez Perce Tribe
Volume II: CurrentFish Consumption Survey
December 2016 Page 99
-------
Table 24. Nez Perce Tribe. Fishing activities during the preceding year as reported by the 283
respondents who reported fishing at least once. Estimates are weighted.
Variable
% or
Mean ± SD
No. who
Responded
Number of times went fishing
24.6 ±35.1
283
Percent offish harvested which were-
Kept
60.0 ± 24.7%
277
Given to others
34.5 ± 22.2%
Sold
5.5 ±16.5%
Own or have access to fishing gear
Yes
97.1%
283
No
2.9%
Own or have access to a boat
Yes
34.3%
283
No
65.7%
6.15 Changes in Consumption and Reasons
Table 25 summarizes reported changes in consumption and access to fish and fishing. Nearly all
consumers believe that fish were/are very important in the Tribe's heritage and culture in the past
(97.9%) and present (96.4%).
An estimated 39% of the consumers have experienced a change in fish consumption over time,
and among those who have experienced the change, 49% experienced increased consumption
and 47%) experienced a decrease. A large proportion of the consumers (48%) have experienced
a change in fishing access and, among those experiencing a change, less access to fishing
(7P/o) far outweighed more access (25%). Similarly, 38% of consumers reported a change in
fishing frequency, of which 30% reported an increase and 67% reported a decrease. Nearly all
consumers want to increase consumption (45%) or maintain current levels of consumption of
fish (55%).
Nez Perce Tribe
Volume II: Current Fish Consumption Survey
December 2016 Page 100
-------
Table 25. Nez Perce Tribe. Changes in consumption and access to fishing in the eligible consumer
population. Estimates are weighted.
Variable
%
No. who
responded to the
question
Importance offish in Tribe's heritage and culture, in the
past
Very important
97.9%
451
Somewhat
important
2.0%
Not important
0.1%
Importance offish in Tribe's heritage and culture, in the
present
Very important
96.4%
450
Somewhat
important
3.6%
Not important
0.0%
Change in fish consumption over time
Yes
39.2%
451
No
60.8%
If so, how has consumption changed
Increased
48.9%
171
Decreased
47.4%
Other
3.8%
Change in access to fish and fishing over time
Yes
48.4%
427
No
51.6%
If so, how has access changed
More access
25.4%
201
Less access
70.7%
Other change
3.9%
Change in frequency of fishing
Yes
38.3%
441
No
61.7%
If so, how has fishing frequency changed
Increased
30.4%
164
Decreased
67.1%
Other
2.5%
Desired fish consumption in the future compared to now
Increase amount
45.2%
451
Maintain amount
54.7%
Decrease amount
0.1%
Nez Perce Tribe
Volume II: Current Fish Consumption Survey
December 2016 Page 101
-------
6.16 Reinterviews
Thirty-one reinterviews were conducted between April 3 and June 12, 2015. The time between
the first interview and the reinterview ranged from 28 days to 85 days (median 55 days). There
were 17 female respondents and 14 male respondents. Of the 31 respondents, 29 (94%) reported
consuming Chinook during the reinterview. Of the 2 who did not report consuming Chinook
during the reinterview, one did report consuming Chinook on the first interview (3 days per
year). Of the 29 who reported consuming Chinook on the reinterview, on the first interview 23
also reported Chinook, three reported an unspecified salmon species only and three reported
Coho salmon as the only salmon species. As the respondents were not always sure of the specific
salmon species they consumed, these six instances of unspecified salmon species or Coho
salmon reported on the first interview were assumed to be Chinook salmon for the purposes of
comparing consumption frequencies between the first interview and the reinterview.
Table 26 summarizes the responses to the first interview and reinterview. The mean (± SD)
frequency of Chinook consumption on the first interview and reinterview was 30.7 ± 30.8 and 30.9
± 38.9 portions/year, respectively, with an average difference of 0.2 ± 36 portions/year. The
correlation in the number of portions per day between the first interview and reinterview was r =
0.57 (Spearman's correlation coefficient). The results were little changed when the one respondent
with an imputed duration of their high consumption period was omitted (Spearman's r=0.60).
Respondents were asked in both interviews whether their overall fish consumption had changed.
Of the 31 respondents, 20 (65%) gave the same response on both. Nine others reported a change in
consumption (5 increased and 4 decreased) on the first interview but no change on the reinterview.
Of the 9 respondents who reported a change in consumption on both interviews, 7 (78%) agreed on
the direction of change. The number living in the household of the respondent was reported to be
4.1 ± 2.4 on the first interview and 4.5 ± 2.5 on the second (Spearman's r = 0.92).
Overall, the first and reinterview responses were consistent, particularly in the summary means
and percentages, though there were disagreements at the individual level. These results support
the use of aggregate summaries of consumption. The reinterview questionnaire is in Appendix A.
Table 26. Nez Perce Tribe. Summary of FFQ interview and reinterview responses. All rows are
based on all 31 respondents who completed both interviews. Summaries are unweighted.
Interview
Questionnaire Item
FFQ Interview
Reinterview
Consumed Chinook salmon
96.8%
93.5%
Frequency of Chinook consumption* portions/year
30.7 ± 30.8
30.9 ± 38.9
Overall fish consumption has changed over time
58.1%
35.5%
Overall fish consumption increased
32.3%
22.6%
Overall fish consumption decreased
25.8%
12.9%
Number living in respondent's household
4.1 ±2.4
4.5 ±2.5
Values are percentages or mean ± SD;
*Includes non-consumers as 0.
Nez Perce Tribe
Volume II: Current Fish Consumption Survey
December 2016 Page 102
-------
6.17 Reliability and Cooperation of Respondents—Interviewer's Assessment
Of the 452 completed first interviews, the duration ranged from 15 minutes to 131 minutes (mean
± SD: 59 ± 28 minutes). This excludes two implausible duration values. Sixteen percent were
conducted at the respondent's home and 70% were conducted in private, without others present.
Table 27 shows that the interviewers found only a very small fraction of respondents to be less
than "highly reliable" or "generally reliable." Similarly, the interviewers found only a small
fraction of respondents to be less than "very good" or "good" in their cooperation. No
interviewers thought any respondents had questionable reliability or were unreliable. Thus
overall the interviewers appeared to trust the information they were obtaining.
Table 27. Nez Perce Tribe. Descriptive summary of interviewers' ratings of respondents'
cooperation and reliability during the first interview. Summaries are unweighted.
Variable
%
No.
Respondent's cooperation
Very good
88.2%
398
Good
10.9%
49
Fair
0.9%
4
Poor
0.0%
0
Respondent's reliability
Highly reliable
80.7%
364
Generally reliable
19.3%
87
Questionable
0.0%
0
Unreliable
0.0%
0
Nez Perce Tribe
Volume II: Current Fish Consumption Survey
December 2016 Page 103
-------
7.0 Discussion
7.1 Overview
This fish consumption survey provides unique information about fish consumption and fish
harvesting by a Tribe residing in the Columbia River Basin. Two different sets of FCR estimates
are presented, each developed by quite different methodologies.
One set of rates is based on a food frequency questionnaire (FFQ), through which respondents
provided information on their fish consumption over the past year. The information on frequency
of consumption, portion sizes and the duration of certain consumption seasons has been
combined to yield a consumption rate (g/day) for each respondent for each of the species they
have consumed—the FFQ rates. Means and percentiles of the FFQ rate distribution for seven
groups of species have been presented in this report.
The second method of estimation of rates uses the respondents' answers about fish consumption
during a 24-hour period ("yesterday") along with some plausible modeling assumptions (the NCI
method) to come up with estimates (means and percentiles) that can be directly compared to
30
those provided by the FFQ method . The NCI method does not provide estimates of rates for the
individual respondents encountered in the survey. Rates from the NCI method have also been
presented in this report. NCI rates could only be computed for two of the seven species groups
for which FFQ rates were determined. The other species groups, for which the NCI method
31
could not be used, had an insufficient number of respondents with double hits .
The FFQ and NCI methods' estimates of means and percentiles differ. This issue is discussed in
Section 7.2. Because the NCI and FFQ methods are quite different, a specific summary statistic
from this population, such as a mean or a percentile, should be compared to a statistic computed
with a similar methodology from another population in order to draw a valid comparative
conclusion. For reasons discussed later, the NCI method statistics would usually be preferable
when available (and if the sample size is sufficiently large to support the method). However, the
NCI method analysis may not be possible for consumption of narrowly defined species groups or
small sample sizes, since the planning goal of achieving approximately 50 double hits would
usually not be fulfilled. The FFQ approach is feasible for surveys with a much smaller sample
size than that needed for the NCI method. While larger sample sizes provide more precise
estimates from any method, the minimum size for assurance of feasibility of using the NCI
method would usually start in the hundreds. The data needs and the resources (including
statistical expertise) required for the NCI method's estimates of FCRs are much greater than
those necessary to develop FCRs from FFQ data.
The fish consumption survey of the Nez Perce Tribe, based on a modest response rate (38%) to
the survey—and one that has likely been addressed by use of survey weighting techniques—has
a substantial FCR, with quite high consumption rates for a notable fraction of the population,
30 The NCI model for the distribution of usual consumption can be developed using only the 24-hour recall data. However, the FFQ
rates can be used as a covariate in fitting the NCI model. The FFQ rates provide a covariate in the same sense that gender, age,
and other variables are covariates which may be helpful in improving the NCI model. Only the relative value of the FFQ rates is
important and not their absolute magnitudes. If the FFQ rates were multiplied by or divided by 10,000 or any other non-zero number,
the fitted NCI model using the re-scaled FFQ rates would be unchanged.
31 A double hit refers to the occurrence of a respondent reporting consumption of fish from a given species group for both of the 24-
hour recall interviews. An adequate number of double hits is needed to support the NCI method.
Nez Perce Tribe
Volume II: Current Fish Consumption Survey
December 2016 Page 104
-------
whether the FFQ or NCI method rates are considered. For example, based on the calculated
consumption rates (Tables 8 and 12, all species), one-quarter of the Nez Perce adults consume at
least 99 g/day (NCI method) or 137 g/day, if the FFQ data are used.
Very few non-consumers of fish were encountered in the survey. Only 9 out of 472 respondents
reported non-consumption of fish, for a weighted non-consumption percentage of 2.6%. The
non-consumption percentage is based on respondents who adequately completed questions 3
through 6 of the screening interview (Appendix A, Section 1.0). The subsequent analysis of the
percentage of non-consumers used appropriate statistical survey weights for each of the 9 non-
consuming respondents and 463 consuming respondents.
The Nez Perce Tribe has experienced changes in FCRs and fishing activities, as documented by the
survey. Changes in the distribution and abundance of fish (compared to the Tribe's history) as well
as changes in access to fishing across the treaty territory may all be factors. Among those who
reported a change in access to fishing, many more reported less access (71%) than more access
(25%) compared to an earlier time. It is of interest that more fish are available (compared to the
1990s, see above) and yet the respondents report less access to fishing. One possible explanation,
among others, is an increase in competition for harvesting of fish from non-Indians, especially at
the more productive fisheries where tribal and state fisheries occur at same time and area.
The Tribal members and staff and Nez Perce Tribal Executive Committee contributed very
32
significantly to the execution of this survey. Through advertising, offering of incentives
(entirely at the Tribe's own expense), assisting the supervisor with issues involving survey
implementation (coordination with tribal interviewers and scheduling of interviews with tribal
members), opening special events and powwows to interviewing opportunities, conducting
mailings to tribal members, and other forms of information and advertising, the Nez Perce came
forward to substantially reverse what was a very challenging and difficult slow start to the
survey. Thus, in addition to the quantitative findings in this report, the role of the Tribe and its
governing body and staff should be considered a critical component in the planning of future
tribal surveys. In addition, the development of individual rapport and mutual trust between
individuals from the contractor's staff and those from the tribal staff was a critical component of
the survey. The Tribe is a separate and distinct nation, and collaboration with this unique nation
is something that involved mutual learning, for both the contractor's staff and the Tribe.
32 Of particular note was the Nez Perce Tribe's offering of an attractive t-shirt to interviewees on certain interviewing occasions. The
t-shirts were very popular and undoubtedly helped recruitment of respondents. The Tribe also utilized a raffle to reward individuals
for participating in the survey.
Nez Perce Tribe
Volume II: Current Fish Consumption Survey
December 2016 Page 105
-------
7.2 Comparison of FFQ Rates to NCI-Method Rates
The estimated mean consumption rates (Groups 1 and 2) differed between the FFQ-based rates
and the rates based on the 24-hour recalls, with the 24-hour mean rates being lower (Table 15). A
simple mean was initially used for this comparison: the "naive" mean was calculated as the
survey-weighted mean of the observations. The naive 24-hour mean consumption rates of Group
1 and Group 2 species were 76% and 78% as large as the FFQ means for Groups 1 and 2,
respectively (Table 15). The difference was statistically significant for Group 1 species (p < 0.01,
based on a bootstrap CI for the difference between the FFQ and naive 24-recall mean) and
marginally significant for Group 2 species (p = 0.053). When the 24-hour data were analyzed
(with covariates) using the NCI method, the NCI method's mean and 95th percentile estimates for
Group 1 species consumption were, respectively, 61% and 53% as large as the corresponding
mean and 95th percentile from the FFQ data and methods (Table 15). For Group 2 species, the
NCI method mean and 95th percentile estimates were, respectively, 64% and 71% as large as the
FFQ values. The other species groups assessed (Groups 3-7) also had lower naive 24-hour
means than the FFQ means, but the NCI method could not be used to provide a mean or
percentiles of consumption for these groups due to the smaller sample size of "double hits."
It is likely that—compared with the FFQ approach—the rates based on the NCI method may be
closer to the actual population values because the challenge to a respondent's memory is less
than that involved in collecting the type of data used by the FFQ method. The 24-hour recall data
used by the NCI method are based on the respondent recalling consumption "yesterday," a
memory task that is easier than recalling and averaging consumption during the preceding 12
months, as required by the FFQ portion of the interview. Secondly, a study by Subar et al. (2003)
found that the 24-hour recall method was more accurate than the FFQ method in reproducing
protein and energy intake as measured by accepted biomarker methods.33 Results from the Subar
study suggest a preference for the 24-hour recall data over FFQ data, but extrapolation from
protein and energy intake to fish consumption may be an issue. In addition, the specific format of
the questions used for data collection in any given survey can be expected to have an impact on
the rates calculated from the survey.
The NCI method, however, contains strong assumptions about the shape34 of the distribution of
usual consumption, and the fitted shape used to provide the NCI estimates may or may not fit
well in the tails of the distribution. Specifically, the upper tail of the NCI method distribution
may not fit the actual distribution for high-level consumers very well. Diagnostics and quality
checks suggest that the NCI model fits the tribal data well overall, but there is no definitive
methodology to check segments of the NCI method distributions, such as the upper tails of
FCRs, including the important 90th and 95th percentiles, which are used in making regulatory and
risk assessment determinations.
The NCI method, using 24-hour recall data, and the FFQ method, using respondents' perception
about the past year of consumption, yield a range of estimates, and this range seems likely to
include the actual FCR values. It seems likely that the actual consumption rates are closer to the
NCI estimates, since they are based on memory of consumption "yesterday" rather than memory
33 Protein intake was measured using an indicator chemical while energy production was measured using doubly labeled water.
34 The NCI method assumes a certain family of shapes of the distribution of usual consumption, and the distribution must be derived
from the normal distribution by a Box-Cox power transformation.
Nez Perce Tribe
Volume II: Current Fish Consumption Survey
December 2016 Page 106
-------
of the past year of consumption. Both the FFQ and NCI method approaches are accepted survey
methodologies. Further research is needed to compare usual consumption distributions from the
two methods and determine what gives rise to their differences. The current Idaho tribal fish
consumption surveys are the only surveys known that have collected data using the NCI method
and a comprehensive form of the FFQ method simultaneously. Given the resources required to
conduct surveys supporting NCI data analysis, acquisition of data comparing NCI and FFQ
approaches will likely be slow. Also, it is important to note that an FFQ survey is the only
method—using limited resources—for deriving the distribution of usual consumption (e.g.,
"usual" refers to mean daily consumption over the course of a year) in cases where the survey
results cannot support use of the NCI method. That can happen, for example, when estimation is
needed for species groups that do not have sufficient double hits; generally, the analysis needs 50
or more respondents who report consumption of the fish species group of interest for at least two
24-hour recall periods to provide confidence, in advance of data collection, that the resulting data
can be used with the NCI method.35 The FFQ approach is also the only method available for a
fish consumption survey of limited sample size, for which only a handful of double hits—not
50—may be expected.
Some factors—including those just discussed—that may help to explain the difference between
the FFQ consumption rates and the rates from the NCI method include the following.
• Chance. The FFQ rates per respondent may correctly reflect their consumption over the
past year, but, by chance, the days on which they were interviewed about their
consumption "yesterday" happened to selectively miss their days of actual fish
consumption. Chance may, indeed, explain part of the difference, but the difference in
means and 95th percentiles between the two methodologies is statistically significant (p
< 0.05), so it is very likely that only a part of the difference might be explained this
way. Chance may provide a partial explanation of the differences, but, due to the wide
gap between means and percentiles by the two methods, the role of chance is likely to
be small.
• Memory and interpretation. Both the FFQ and 24-hour recall responses require the
respondents to exercise their memory and interpret their fish consumption behavior. The
24-hour recall is less challenging to memory than the FFQ. The 24-hour recall questions
ask about what happened "yesterday"; the FFQ asks about what happened over the course
of 12 months before the present moment. The fish consumption occasions addressed by
the 24-hour recall can be at most 48 hours old; e.g., consider a Monday 11:55 p.m.
interview response of a person who ate fish at 12:05 a.m. on Sunday.
The FFQ respondent is referring to an average that may not correspond to any events; e.g., a
person who eats fish twice per week during every second week would need to report an average
frequency of once per week, a frequency which never happens during any single week. Whereas
the 24-hour recall asks for an inventory of fish-eating occasions on the preceding day—no
averaging is involved. Similarly, the 24-hour recall asks for the portion size per eating occasion
yesterday rather than for the FFQ's typical portion size during a year. Finally, the FFQ handles
variation in consumption during the course of a year by allowing up to two periods of
35 At the completion of data collection a dataset with fewer than 50 double hits may well be usable with the NCI method. However,
when planning a survey, the 50-double-hit goal is precautionary.
Nez Perce Tribe
Volume II: Current Fish Consumption Survey
December 2016 Page 107
-------
consumption—a high and low consumption period—if needed. The 24-hour recall simply
records what happened throughout a single day.
The 24-hour recall also may include memory error, including error in: a.) determining when
"yesterday" began and ended; b.) forgetting items consumed yesterday; c.) moving consumption
from another day into "yesterday"; and d.) errors in portions sizes or species consumed
"yesterday." There is evidence that the 24-hour recall data may, on the average, be
underreporting fish consumption, which would imply that the NCI-based estimates may
correspondingly underreport fish consumption rates. A relevant study by Moshfegh et al.
compared a.) energy intake (EI) calculated from 24-hour dietary recall interviews to b.) total
energy expenditure (TEE) calculated using the doubly labeled water technique. The analysis was
based on 524 volunteers from the Washington, D.C. area. The ratio of energy intake to
expenditure expressed as a percentage (100*EI/TEE) can be considered a measure of the extent
to which the dietary recall interview captured energy intake. The study found underreporting of
EI by 11%, on the average, and underreporting depended on the BMI 6 (body-mass index) of the
subjects. Using a common BMI classification (WHO, 2015), the underreporting of EI in the
Moshfegh study was as follows: normal weight (BMI less than 25) males had 0%
underreporting, 6% for females; overweight (BMI = 25 to less than 30) males had 14%
underreporting, 15% for females; obese (BMI = 30 or greater) males had 20% underreporting,
21% for females. While energy intake is not equivalent to mass of food items consumed, fish are
a higher source of energy per unit mass than some other foods, such as vegetables. It is likely
that percentage energy underreporting would be relevant in understanding underreporting of
high-energy foods, such as fish. Given the greater underreporting for individuals with greater
values of BMI, the BMI distribution of the surveyed tribal members is relevant.
The 434 Nez Perce respondents' BMI distribution (excluding those with missing height or
weight) was 18% normal weight, 26% overweight and 56% obese (unweighted percentages).
While the Moshfegh findings about energy intake among a largely non-Hispanic white
population cannot be directly applied to this survey of fish consumption among Native
Americans, there is a possibility of underreporting of fish consumption from this survey's 24-
hour interviews, especially given the presence of large BMI values in this surveyed population.
(The Moshfegh study did not collect FFQ consumption rates, and, thus, did not consider
accuracy of respondents' reports on FFQ rates.) A related study by Subar et al. (2003) also found
underreporting of protein and energy intake from both the FFQ and 24-hour recall methods, but
the underreporting was larger for the FFQ method.
Concerning memory, the differential demand on memory of the two approaches is a plausible but
not a proven factor in the observed difference in rates between the two methods. The results
presented in Section 6.11 (Comparison of FFQ Rates to 24-Hour and NCI Method Rates) show
that the frequency of consumption days in the 24-hour data is too low to be consistent with the
frequencies of consumption reported by the FFQ method. It would be tempting to conclude,
therefore, that the respondents' reported 24-hour incidence of hits (a day with fish consumption)
is more accurate than the reported FFQ consumption frequencies, because the 24-hour method
requires less use of memory and interpretation than the FFQ method. It is also possible that the
36 BMI is a commonly used index, based on weight and height, that is used to classify people along a spectrum from normal weight
up to obese. BMI = wt(kg)/ht2 (m).
Nez Perce Tribe
Volume II: Current Fish Consumption Survey
December 2016 Page 108
-------
extensive list of species included in the questionnaire (45 species had explicit consumption
questions for the respondents) may have led to double-counting of some species in the FFQ. A
respondent unsure of a species eaten may have reported it under two or more species. The
analysis in this document of respondent uncertainty in reporting in relation to the difference
between FFQ and 24-hour recall consumption did not show any trend of an increasing FFQ/24-
hour consumption difference with increasing uncertainty. That analysis was not at all definitive
given the relatively small number of respondents showing some aspect of uncertainty (see
Section 6.11 and Appendix F, Figures F1-F2 and Tables F9-F11.) It will take more surveys with
these paired methodologies to definitively address the issue of greater or lesser accuracy of the
24-hour data vs. the FFQ data.
Differences in frequency or portion size reporting. Both frequency and portion size appear to
37
be either overreported in the FFQ data or underreported in the 24-hour recall data, or both. For
all species combined (Group 1), for example, the mean frequency of fish consumption calculated
from the 24-hour recall data (all respondents combined) was 85% as large as the mean frequency
from the FFQ data. The 24-hour recall mean portion size was 87% as large as the FFQ mean
portion size. This pattern was similar for Group 2 species, with the mean frequency of
consumption from the 24-hour recall data being 86% as large as that from the FFQ data and the
mean portion size from the 24-hour recall being 89% of that from the FFQ. These factors are
directly observable and quantitatively appear to explain much of the difference between FFQ and
24-hour recall rates. (See Section 6.11 for a comparison between the 24-hour and FFQ data on
portion sizes and frequencies.)
As an additional methodologic note, the description of portion size is handled differently in the
FFQ and in the 24-hour recall interviews, and the difference may have some effect on the
difference in average portion sizes determined by the two methods (see Section 6.11). In the 24-
hour recall interview's data, the portion size for a species consumed is identified for each eating
occasion during the 24-hour period. In the FFQ interview, a single portion size (or, at most, two
different portion sizes for two different seasons) is identified to describe typical consumption of
a species for an entire year. For a given species, the average across respondents of the FFQ's
typical portion size would agree with the average across respondents of the 24-hour recall's
portion sizes under some specific assumptions, two of which are: a.) the FFQ typical size is a
faithful average, per respondent, of the individual portion sizes occurring during the preceding
year; and b.) the 24-hour recall portion sizes are accurate.
Reference period. The collection of "yesterdays" reported by the pool of respondents in the
survey spans a period of approximately one year (12 months) corresponding to the duration of
interviewing activity in the survey. The reference period for the fish consumption during the
FFQ's preceding year spans almost two years (24 months), corresponding to the beginning of the
preceding year for the first-interviewed respondent to the end of the preceding year (ending on
the interview day) for the last respondent to complete the FFQ segment of the interview. Thus,
collectively for the pool of respondents, the two reference periods do not match. This appears not
to be an important factor in influencing FFQ rates. In the analysis of seasonality described in
Section 5.23.2.1, the calculated mean FFQ consumption rate did not appear to vary
37 The frequency of consumption of a species (e.g., number of times per week) is not directly reported by the respondents during the
24-hour recall interviews, but the average frequency for a group of respondents can be estimated. See Section 6.11 for methods
and results.
Nez Perce Tribe
Volume II: Current Fish Consumption Survey
December 2016 Page 109
-------
systematically month by month across the 12 months during which FFQ interviews occurred,
which is consistent with (but does not prove) a consumption regimen that was not highly variable
during the entire two-year reference period. Thus, the reference period appears not to be a
definite contributor to the difference in consumption rates (24-hour vs. FFQ), based on the lack
of identifiable seasonal variation observed in the FFQ and 24-hour time series for species Groups
38
1 and 2 and, surprisingly, the salmon species. As noted elsewhere, the number of interviews
completed during the peak harvesting period was low, perhaps preventing detection of true
seasonal variation.
Modeling: tails of the distribution. As noted earlier in this section, the rates based on the 24-
hour recall and the NCI method may be more accurate in the middle of the distribution of usual
consumption rates than in the upper or lower tails, including the important 95th percentile of
consumption rates. Currently, there is no way to verify the accuracy of different segments of the
distribution of usual consumption rates provided by the NCI method. It is good to bear in mind
that the NCI model is fitted using all of the 24-hour data to determine one model, and the tails of
the distribution of usual consumption are determined by and consistent with the entire fitted
distribution, including the central hump of the unimodal distribution. Every part of the
distribution is affected by the data from every respondent, including those with low, medium or
high consumption. With the FFQ data, however, the upper and lower tail are determined by those
with very high or very low consumption. Although the NCI method does allow for certain
skewed distributions, the shape of the entire distribution is restricted to a specific family of
distributions. The shape of the distribution derived by the NCI method from the 24-hour recall
data is affected by the data from every respondent. The distribution of usual consumption
derived from the FFQ data has more independence of the tails from the balance of the
distribution. For example, one can have two FFQ distributions with exactly the same shape
(percentile values) up to, say, the 90th percentile, but then one of the two distributions can
continue with a long tail of very high consumption rates and the other distribution can continue
with, say, consumption rates arbitrarily close to the 90th percentile value. That kind of
"independence" of the upper or lower tail cannot happen with the NCI model. The upper tail has
to conform to the functional form determined by the entire dataset. Thus, the important upper tail of
the NCI-modeled distribution may or may not adequately represent the actual upper tail of the
population distribution of consumption. Nevertheless, it is likely that the NCI-based distribution of
consumption is, overall, closer to the actual distribution than the distribution based on the FFQ data.
In summary, the NCI method's rates based on the 24-hour recall interviews are likely to be
closer to the actual rates than the rates from the FFQ analysis, due to the lighter demand on
memory required by the 24-hour recall approach. In this analysis, memory is the primary
candidate to lean on in favor of the NCI method; memory and its imperfections are involved in
producing both the FFQ data and the 24-hour data. However, recall and interpretation of fish
consumption during the 24-hour interviews is less difficult for the respondent than that during
the FFQ segment of the first interview. Given these factors, the NCI method can be favored,
while the FFQ method provides an additional valid estimate of FCRs. In some cases, the FFQ
may be the only viable option to estimate FCRs given the cost of collecting data for and
conducting the analysis for the NCI method. Additionally, the FFQ approach may be the only
38 See the seasonality material at the end of the section on covariate selection (5.23.2), and related material in Appendix E, Section
5.
Nez Perce Tribe
Volume II: Current Fish Consumption Survey
December 2016 Page 110
-------
feasible method for development of FCRs for narrowly defined species groups or for small
surveys. The difficulty in implementing the NCI method in these cases relates to the need to
accrue a sufficient number of respondents who report some fish consumption on two (or more)
24-hour recall interviews—i.e., a sufficient number of double hits. A low probability of fish
consumption may result in too few double hits to estimate the distribution of consumption rates
even for all species combined (total fish consumption). And, even if the NCI model is
successfully developed for total fish consumption, the separate models attempted for groups of
species or for individual species may not succeed due to the limited number of double hits
encountered. The FFQ method can handle these cases where the NCI method does not succeed.
The FFQ is well established as a method to assess food consumption, and Pacific Northwest FFQ
FCRs have been broadly used by EPA and state environmental agencies for regulatory actions
involving assessment of risks posed to Native Americans exposed to contaminants in seafood.
7.3 Comparison of This Survey's Rates to Other Surveys' Rates
Table 28 compares the Nez Perce rates for Group 1 species from the current consumption survey
(based on the FFQ method and the NCI method) to other similarly targeted tribal surveys, and
also presents results of a survey of the U.S. national population (NCI method). Rates can be
validly compared among surveys when the rates have been calculated using the same
methodology—either the NCI method or the FFQ method. The Nez Perce Tribe has a high rate
of fish consumption. Its mean total fish consumption rate for adults (based on the NCI method) is
75.0 g/day and the 95th percentile of consumption is 232.1 g/day. Compared to the NCI method
rates for the U.S.A, the Nez Perce mean, median and 95th percentile rates are from 2.8-fold to
3.4-fold larger. The FFQ method mean, median and 90th and 95th percentiles of consumption for
the Nez Perce Tribe are also high and larger than the corresponding FFQ rates for some other
Pacific Northwest tribes, including the four pooled CRITFC survey tribes. (The NPT was one of
the four tribes include in the CRITFC survey.) In comparison to tribes with access to Puget
Sound fisheries resources, the Nez Perce FFQ rates are also higher than the FFQ rates of the
Tulalip and Squaxin Island Tribes, but lower than those of the Suquamish Tribe. The only other
Pacific Northwest inland tribes with documented fish consumption rates available are the
Shoshone-Bannock Tribes, who participated in the current survey using the same methodology
and survey management as the Nez Perce Tribe. The Nez Perce FFQ rates are lower than but
comparable to the Shoshone-Bannock rates (NPT mean, 123.4 g/day, SBT mean, 158.5 g/day;
NPT 95th percentile, 437.0 g/day, SBT 95th percentile, 603.4 g/day.) The notes under Table 28
provide references for consumption rates of the tribes.
A contributing factor to the high FFQ FCRs as compared to the CRITFC survey may be the
difference in the abundance of anadromous fish particularly, and other fish species more
generally, that were at lower levels in the 1990s and have been increasing to higher levels in the
past decade or more, based on yearly counts of fish passages at Lower Granite Dam from the
website of the Fish Passage Center (see www.fpc.org). The fish runs in recent years are larger,
which would support more harvest opportunities, and therefore would be expected to support
increased current consumption by tribal members compared to the time of the CRITFC survey
Nez Perce Tribe
Volume II: Current Fish Consumption Survey
December 2016 Page 111
-------
(conducted from late 1991 through early 1992). The 2013-2014 counts at Lower Granite Dam of
39
adult Chinook salmon, for example, are several-fold larger than those during 1991-1992.
Differences in survey methodology in assessing total fish consumption may also contribute to the
higher FFQ FCRs for the current survey relative to the CRITFC study. While the CRITFC
survey did question respondents in detail about consumption of the species primarily harvested
in the Columbia River Basin (e.g., salmon, steelhead, lamprey, etc.), its estimates of total fish
consumption (from all sources, not only the Columbia River Basin) were derived from questions
which referred to all species combined, without enumerating species or allowing the respondent
to provide different portion sizes for each species consumed. In contrast, the questionnaire from
this survey enumerated 45 species and gave respondents an opportunity to consider each species
individually, potentially increasing their recall of consumption.
The NCI method rates, which are likely closer to the actual rates than the FFQ rates—for reasons
discussed elsewhere in this report—show greater fish consumption rates for the Nez Perce than
for the Shoshone-Bannock Tribes (all species combined). Likely reasons for the difference in
rates between the two tribes include the greater access of the Nez Perce to fish available for
harvest in its fisheries, and Nez Perce country having more productive fish runs and fisheries
than the other four Idaho tribes (including the SBT). Furthermore, while the NPT fishing areas
are affected by multiple dams that affect survival of anadromous fish or prevent their passage to
spawning grounds during migration, the SBT fishing areas are further upriver and are affected by
additional dams. However, this rationale does not explain why SBT FFQ rates exceed NPT FFQ
rates, in contrast to the NCI method rates, which show the opposite order. Differences in reported
portion size and frequency of consumption between the 24-hour data and the FFQ data for each
tribe seem to be an important factor underlying differences in NCI vs. FFQ rates for each tribe,
and, possibly, for the difference in rates between tribes. (See Section 6.11).
39 Based on data available at www.fpc.orq (accessed September 24, 2015), the passage count for adult Chinook salmon at Lower
Granite Dam was 11,000 and 25,000 (rounded) during 1991 and 1992, respectively; the passage count was 100,000 and 155,000
during 2013 and 2014, respectively. (Table of passages obtained by starting from the web site
http://www.fpc.org/adultsalmon/adultqueries/Adult_Annual_Totals_Query_form.html and selecting "Lower Granite Dam" and
"Chinook".)
Nez Perce Tribe
Volume II: Current Fish Consumption Survey
December 2016 Page 112
-------
Table 28. Nez Perce Tribe. Total FCRs (g/day) of adults in Pacific Northwest Tribes (with
consumption rates available) and the U.S. general population. Consumers only.
Population
No. of
Consumers*
Percentiles
Mean
50%
90%
95%
Nez Perce Tribe, FFQ rates, Group 1
451
123.4
70.5
270.1
437.4
Nez Perce Tribe, NCI method, Group 1
451
75.0
49.5
173.2
232.1
Shoshone-Bannock Tribes, FFQ rates, Group 1
226
158.5
74.6
392.5
603.4
Shoshone-Bannock Tribes, NCI method, Group 1
226
34.9
14.9
94.5
140.9
Tulalip Tribes, FFQ rates
73
82.2
44.5
193.4
267.6
Squaxin Island Tribe, FFQ rates
117
83.7
44.5
205.8
280.2
Suquamish Tribe, FFQ rates
92
213.9
132.1
489.0
796.9
Columbia River Tribes, FFQ rates
464
63.2
40.5
130.0
194.0
USA, NCI method40
*16,363
23.8
17.6
52.8
68.1
* Adults >21 years old; includes both consumers and non-consumers.
Notes. The rates for Columbia River Tribes are from CRITFC, 1994, Table 10. The rates for the Suquamish Tribe
are from Suquamish Tribe, 2000, Table T-3 and Liao, 2002. These rates were converted from g/kg/day to g/day by
multiplying by the mean body weight of 79.0 kg, found in Table T-2 of Suquamish, 2000. The rates for the Tulalip
and Squaxin Island Tribes are from Polissar, 2014, Table 2 and Table 3, respectively. The national rates are from
U.S. EPA, 2014, Appendix E, Table E-l. The rates for the Nez Perce and Shoshone-Bannock Tribes are from this
report and the other report released at the same time as this report with virtually the same format, in Table 8 (FFQ
rates) and Table 12 (NCI method rates).
40 In Table 28, the quoted U.S. national rate includes non-consumers. An analysis of data from an NHANES survey period
(2003-2006) overlapping the reference period (2003-2010) for the NHANES-based rates quoted in Table 28 indicated that only a
small fraction of the U.S. population are non-consumers offish. (See Polissar et al., 2014, Table 8 and text following it.) An analysis
of 7,145 NHANES respondents from the 2003-2006 survey period, including respondents who supplied 24-hour recall data and
completed the FFQ portion of the questionnaire, showed that 680 (9.5%) of the respondents could be labeled as fish "non-
consumers" based on their FFQ responses. Some of these "non-consumers," however, would be "consumers" based on the foods
they reported eating on the 24-hour recalls. Some of the respondents with inconsistent consumer/non-consumer status between the
24-hour recall and FFQ fish consumption reports may have eaten very small, undetected quantities offish in the foods they reported
consuming on the 24-hour recall and then reported no fish consumption in response to the FFQ questions on consumption during
the preceding year. Trace quantities offish, such as that found in Caesar salad and certain cheese spreads, were captured in the
NHANES survey methodology by use of standard recipes applied to foods reported as eaten during the 24-hour recall periods.
Thus, it appears that less than 10% of the U.S. population are non-consumers offish, and a smaller percentage may hold if
undetected, trace quantities offish are excluded.
Nez Perce Tribe
Volume II: Current Fish Consumption Survey
December 2016 Page 113
-------
7.4 Strengths and Limitations
Strengths and limitations of the survey are noted below.
Strengths
Uniqueness. This study is unique in that it conducted both the FFQ (food frequency
questionnaire, including amount consumed) and the 24-hour recall (NCI) method simultaneously
in a survey of tribal consumption of fish. This study is also unique in the length of time over
which it was conducted. Other than a survey of the Colville Tribe (SRC 2015), no other study of
tribal fish consumption in the United States has run for a whole year, covering multiple periods
of fish runs and seasons as well as cultural activities. The span of the survey allowed some
evaluation of seasonal and temporal impacts on FCRs (although the evaluation was limited by a
relatively small number of respondents for some months of the survey).
Collaborative development. Every aspect of this survey was designed in a research-intensive,
time-consuming and transparent collaborative process beginning in the Fall of 2012 and lasting
until the Fall of 2016 between the five tribes in Idaho, the Environmental Protection Agency, two
tribal collaboratives [the Upper Snake River Tribes Foundation (USRTF) and the Columbia
River Inter-Tribal Fish Commission (CRITFC)], the State of Idaho and a highly skilled and
expert cross-disciplinary team of experienced consultants. Efforts were made to incorporate
state-of-the-art survey and analytical methods and tribal cultural and governmental concerns in a
study that was designed to contribute to an understanding of fish consumption by members of the
two tribes surveyed. The survey questionnaire drew extensively on questionnaire content that
had been used previously (for FFQ and 24-hour recall interviews). The approach that was used to
quantify current fish consumption is consistent with the way food consumption surveys at the
population level are currently performed worldwide. (See, for example, the review of food
consumption surveys in De Keyser, et al., 2015.) The intensive collaboration extended over two
years, beginning with design and continuing through the implementation of the study in the field
and the analyses of the data.
The areas of expertise held by the involved parties included tribal culture, fisheries and fishing
practices, environmental issues, survey design (including CAPI), survey administration, statistics
and government policy. Using a team that included considerable prior survey design experience
likely reduced or eliminated bias and increased precision of the resulting estimates. The team's
considerable experience with survey fieldwork was also essential in providing thorough training
for the field staff, conducting the monitoring needed and providing practical and swift solutions
to address the unexpected events that inevitably arise in complex survey efforts. In addition to
the core technical staff working on the project, the project consulted with and utilized outside
experts, through means which included several teleconferences and a number of e-mail
exchanges with experts in dietary surveys from the National Cancer Institute.41 Experts were
involved in both the IRB consultations at the beginning and in the peer review at the conclusion.
The diversity of expertise provided was essential given the broad range of areas and activities
that needed support under each of the areas noted. Lastly, the extensive experience of working
with Native American tribes among this team created an initial rapport with the tribes and
fostered the cooperation that continued to grow as the survey progressed.
41 Drs. Amy Subar and Kevin Dodd of the National Cancer Institute provided valuable input and support.
Nez Perce Tribe
Volume II: Current Fish Consumption Survey
December 2016 Page 114
-------
Tribal contributions. The Tribe made many important contributions to the success of the survey.
Just a few of the many contributions include: the designation of species consumed, the
identification of fishers within the Tribe, the assistance in locating hard-to-find respondents,
regular participation in review and monitoring of progress, tribal governmental encouragement
of participation, publicity to promote participation in the survey and monetary or other incentives
(entirely from tribal resources) to recognize participation.
Tribal enrollment records. One advantage of this collaboration with the tribal government is that
the contractor team was allowed access to a unique frame for drawing the sample: tribal
enrollment records. The use of the enrollment records avoided a costly effort to develop an
alternative frame for sampling. The random sampling (as opposed to, for example, a convenience
sample) conducted from this complete population listing added to the precision of the survey by
using survey resources to increase the sample size rather than using an alternative and costlier
means of identifying respondents and, inevitably, a reduced sample size. In addition, by having
demographic information available in advance of sample selection, the random sample could be
selected from defined demographic strata of the population. This method of stratification almost
always leads to a sample that is more representative of the population than a sample drawn by
other means that are used when a population roster is not available. Finally, the availability of a
population roster from which to draw the sample also allowed an adjustment to reduce or
eliminate bias in the reported results. By comparison of the sample demographic composition to
the population demographic composition (from the enrollment records), each respondent could
be statistically weighted in a manner that reduced (or eliminated) bias due to different success
rates in obtaining interviews among the various demographic groups.
In-person interviews. The use of in-person interviews is a strength of the study because that form
of data collection was expected to generally lead to more accurate and complete responses in this
population, for cultural reasons and because of the use of physical display models that could be
and were used in in-person interviews. Many of the interviews were conducted in respondents'
homes, which may have provided a more comfortable environment to participate in a long,
detailed personal interview. Personal interviews allowed for question clarification. This included
use of non-verbal cues (e.g. facial expressions, etc.) to further determine when some aspect of
the survey was not understood and to clarify as appropriate. Other advantages of a personal
interview approach included ensuring completeness of responses (e.g., ensuring topics and
questions were not skipped) and correction of potentially inconsistent responses. Clarification,
verification, and completeness are much more difficult to address using other interview
approaches (e.g. telephone or mail surveys). In-person interviews also allowed the interviewers
to use portion model displays—which could be picked up and examined closely—and
photographs of multiple species of fish, which added to the ability of the respondents to more
accurately identify the species consumed and characterize the size of their portions. Also,
because some portion models were more closely related to certain preparation methods, the in-
person interviews with portion models aided in identifying the methods used to prepare fish for
consumption; e.g., the fillet model would be commonly linked to methods such as frying or
grilling, and the jerky portion model would be commonly linked to smoking of fish.
It is possible that social desirability bias might enter into a live interview. In this setting, social
desirability is the tendency of an individual to over- or under-report consumption (overall or for
particular species) to avoid anticipated verbal or nonverbal negative feedback related to the
Nez Perce Tribe
Volume II: Current Fish Consumption Survey
December 2016 Page 115
-------
perceived social norms (Herbert, et al., 1995). This type of bias is common in dietary surveys,
including both those based on FFQs or based on 24-hour recalls (Tooze, et al., 2004). This
phenomenon might be more likely with an interviewer than with a privately-offered response.
But, the strengths of interviewer-collected data as described above are likely to outweigh this
potential bias.
The survey contractors found that use of outside interviewers was acceptable in limited
circumstances and effective in increasing the total number of interviews, due to their activities
being shepherded and supported by the Tribe. First contacts for interviews were usually made by
tribal members or by a non-tribal interview supervisor who had developed exceptional rapport
with the Tribe. In addition, the outside interviewers were directed only to potential respondents
who were staff members of the Tribe, except for a single visit to a tribal holiday event where
other sampled members were interviewed. This practice made finding the potential respondents
easier and, also, put the interviewers in touch with a group of tribal members who usually had
substantive interaction with the people outside the Tribe. As described in Section 6.7, the
potential impact of non-tribal interviewers on responses was examined and no significant
differences in overall consumption rates were found between respondents with a tribal or non-
tribal interviewer.
Electronic capture of interviews (CAPI). Another strength of the survey was the use of the CAPI
interview model, which, as noted previously, greatly enhances survey accuracy and
completeness. The interview results were usually available very shortly after the interview based
on synchronizing the CAPI tablets online with the contractor's website.
Survey accuracy and completeness is increased by CAPI, compared to other modes, because:
• There are fewer "touches" on the data. With a paper and pencil questionnaire, the
interviewer records the respondent's answer, and later a data entry clerk enters the data in
a tabulation program. CAPI needs only one data recording source: the interviewer.
• Computer programming and skip logic conditions are automated, allowing the
interviewer to focus on the respondent. A paper questionnaire, whether self-administered
or administered by an interviewer, relies on the sometimes fallible human to check and
administer real-time skip patterns during the interview.
• Out-of-range values and logic checks are evaluated immediately by the computer. Paper
and pencil questionnaires cannot offer this degree of quality assurance.
• Data from the CAPI system is uploaded as soon as an Internet connection is available.
This provides both a backup (in case a computer tablet is lost or stolen) and a means for
statisticians to check the integrity of the data.
• CAPI data collection is transportable. Interviewers can bring the computer tablets to far-
flung areas, even households without landlines or cell phone coverage. Telephone
interviews and online interviews only work where there is phone or Internet access,
respectively.
• CAPI technology requires no technical knowledge or ability from the respondents.
Interviewers are trained to use the computer tablets unobtrusively and without respondent
assistance, other than asking for answers to survey questions. Online surveys dictate that
each respondent has at least basic computer experience and knows how to navigate the
internet.
Nez Perce Tribe
Volume II: Current Fish Consumption Survey
December 2016 Page 116
-------
Detailed inventory of species. An additional strength of the survey was the level of detail
obtained on consumption by species. The consumption of approximately 45 individual species
was specifically inquired about, and additional species could be reported by respondents and
entered into the database using a text field. All such entries were used in preparing this report.
The inquiries on consumption of numerous species may have stimulated memory and
comprehensively evaluated consumption. However, there may have been some double-counting
of consumption if respondents who were unsure of a specific species consumed may have
reported such consumption under more than one species.
Interviews spanned one year. Yet another strength of the survey was the span of time during
which the survey was carried out, covering multiple periods of fish runs and seasons. While this
was a strong design feature, the full strength of this design feature (a full year of interviews) was
not fully realized. The interviews did, indeed, cover one year, but they started during a peak
fishing season and the accrual of completed interviews was slow relative to later periods of the
survey year. The peak fishing season was relatively sparsely covered by interviews.
Nevertheless, all seasons were represented by some interviews. The representation of all seasons
in the survey allowed an assessment of seasonal effect on FFQ consumption responses. Analysis
did not show that a seasonal adjustment was needed to provide valid consumption rates. The
number of interviews per season did vary substantially, but the coverage of seasons during a year
of interviewing is some insurance against bias. While ideally a retrospective FCR covering the
past year and drawn from the respondent's memory (i.e., the food frequency approach) should be
fairly constant over time, in fact the consumption of the preceding year reported during
interviews at the beginning of the survey year could be quite different than the consumption in
the preceding year reported at the end of the survey year. Thus, spreading the surveys over 12
months covered, potentially, the full annual cycle of harvesting and consuming fish. Relative to
42
extant fish consumption surveys in EPA Region 10, this is one of the first to collect FFQ
information during 12 months. Among published reports, the FFQ surveys of the Squaxin Island
and Tulalip Tribes (February 25 through May 15, 1994), Suquamish (July through September,
1998) and the four tribes included in the CRITFC survey (fall and winter of 1991-1992) were all
carried out in less than a year.
NCI method combined with FFQ method. The use of the NCI method to estimate the distribution
of usual fish consumption is another strength. It involves less reliance on memory (but more
reliance on modeling) than the FFQ approach. The results of the NCI method were thoroughly
vetted through additional quality assurance methods, sensitivity analyses and parallel and
independent calculations by two statisticians for many of the consumption rate analyses
presented—both for the FFQ and NCI methods. The use of the two methods in the survey also
provided the opportunity to compare consumption rates between methodologies and explore
potential factors that might explain the differences.
This survey used a quantitative FFQ interview combined with interviews yielding 24-hour recall
of fish consumption to support the NCI method. The FFQ interviews provided data which, by
itself, led to estimates of fish consumption rates. In addition, even though the NCI method could
provide fish consumption estimates from the 24-hour recall data alone, the FFQ data (along with
other covariates) were used in the NCI modeling to provide fish consumption estimates that are
42 EPA Region 10 includes Alaska, Idaho, Oregon, Washington and Native American Tribes in these states.
Nez Perce Tribe
Volume II: Current Fish Consumption Survey
December 2016 Page 117
-------
very likely to be more accurate than estimates that would be derived from the 24-hour data alone.
The use of two distinct methods to assess dietary intake—FFQ and 24-hour recall—combined
with two distinct analyses to estimate usual consumption of fish provided a very comprehensive
study on fish consumption.
Independent replication and verification of key statistics. The calculation of consumption rates (a
rate for each species for each respondent) by two statisticians working independently (and
agreeing on the computed rates) strongly supports an assertion that there are likely to be zero or
very few computational errors in the many calculated quantities presented in this report. The
double computing was an essential measure of quality assurance. In addition, a number of the
summary estimated fish consumption rates (means and percentiles) and other quantities in this
report were also computed twice, independently, by two of the contractors' statisticians, in the
pathway to preparing results for different sections of this report. Lastly, calculations of the
estimates of the species Group 1 distribution (mean and percentiles) from the NCI method were
also recomputed by NCI staff. The recomputed mean and percentiles for species Group 1 were
all within 0.4% of the contractors' estimates.
Reinterviews. The reinterview analysis shows that while individual responses to the same
questions vary over time, the summary means and percentages are reasonably similar to each
other from interview to reinterview. (See Section 6.16.) As this survey is intended to provide
summary consumption statistics such as means and percentiles, the reinterview analysis supports
the achievement of that goal with these interviews, though significant variation by individuals in
their responses (to identical questions) over time is evident. However, reproducibility of
interview results may potentially be affected by the species selected for re-evaluation of
consumption. This analysis selected Chinook salmon, a commonly consumed species of
considerable cultural relevance. Future studies may wish to evaluate consistency using a broader
range of species.
Limitations
Response rates. The response rate for the survey was lower than expected at 38%. The four other
fish consumption surveys of Pacific Northwest Indian tribes have had response rates of over 60%
(i.e., CRITFC, Squaxin Island, Suquamish and Tulalip surveys). It is often difficult to know the
reasons for non-response; typically, these individuals do not divulge rationales for their lack of
participation. To no small effect, limitations on resources and time (to adequately find and
contact some respondents) contributed to a lower response rate. Contributing to the difficulty of
contacting prospective respondents was the incomplete, outdated, incorrect or missing contact
information. Enrollment offices provided membership lists, but sometimes without accurate
phone numbers or addresses. The survey team employed supplemental methods to search for
tribal members, including checking property records, utility records, commercial databases and
online searches. Some tribal members lived "off the grid," in areas without physical mailing
addresses. Others had addresses which were merely "Rural Route." Even tribal interviewers,
who had direct and in-depth knowledge about tribal members, experienced significant difficulty
locating some members. Because of this difficulty, resources intended for the interviewing task
were necessarily diverted to locating contact information for prospective respondents. The team
also experienced challenges with missed appointments. Some tribal members scheduled
interviews in their homes, but then decided not to participate, or postponed them for another time
and location—a postponement which did not always have a successful ending. The challenges of
Nez Perce Tribe
Volume II: Current Fish Consumption Survey
December 2016 Page 118
-------
home interviews that affected response rate included the time and distances travelled to reach
widely dispersed rural residents and difficulties in trying to group willing respondents into
convenient interview trips. This posed both a financial challenge (i.e., time and gas expense of
interviewers) and the resultant reduced numbers of interviews able to be conducted within the
calendar time and budget of the study.
The weighting method used to estimate the population distribution of consumption rates
mitigated some of the potential selection bias stemming from the modest response rate.
Specifically, the non-response adjustment to the weights accounted for differences between
responders and non-responders in their age, gender, ZIP code of residence, fisher indicator
(presence/absence on a list of fishers) and combinations (two-way statistical interactions) of
these characteristics. Biases related to other (unknown) characteristics may potentially persist.
Limited imputation of missing values. A minor limitation of the survey is that some cases had
missing data which had to be imputed to be able to retain the respondent's other related
responses for inclusion in the survey. For example, a respondent might not remember a typical
portion size of consumption for a species but would remember the frequency of consumption of
the species. In this example, the CAPI system would capture the portion size response as a "don't
know" code, and, if there was no intervention, the consumption rate for that species would end
up being missing for the planned statistical analysis. As a result, the respondent's fish
consumption rate would be underreported. Instead, an imputed value of portion size was supplied
for the missing value for the analyses presented in this report. Usually the much less frequently
consumed species had such missing values, though this was not exclusively the case. (See
Section 5.28, "Handling Missing Values," for imputation methods.) A sensitivity analysis
reported in Appendix C suggests that such imputations had little impact on the final results.
7.5 Characterizing Uncertainty
The confidence intervals for percentiles of consumption rates in the study describe the uncertainty
in various FCR statistics—the "margin of error." The width of these confidence intervals should be
taken as advisory, without a specific cutoff of widths considered to be desirable or undesirable
among the confidence intervals presented in this report. Again, the data are valuable and, as a
practice, the estimated means and percentiles are the best choice to use for practical purposes as
opposed to other values in the confidence interval. Based on methodologic principles used to avoid
bias, the point estimate (the estimated value lying within the confidence interval) is the preferred
estimate to use in practice rather than other values in the confidence interval.
The statistical weights were adjusted for non-response to correct for any selection bias. It cannot
be guaranteed that selection bias has been completely addressed, as not all non-response can be
predicted, but all available demographic variables were considered in making the nonresponse
adjustment. Furthermore, the additional uncertainty in consumption rates due to imputation of
missing fields in a limited number of cases is not fully represented in the confidence intervals.
However, the ultimate impact of imputation was found to be small based on a sensitivity analysis
encompassing a wide range of imputation scenarios. In summary, the use of imputation was
important to avoid deletion of a number of respondents' data from the analysis, but the different
choices for imputation, varying around the parameter values chosen, had little effect on means
and percentiles of consumption rates.
Nez Perce Tribe
Volume II: Current Fish Consumption Survey
December 2016 Page 119
-------
The findings on seasonality—actually, a possible lack of seasonality—were unexpected (see
Section 5.23.2.1. This finding was unexpected because fishing activity, as reported in this survey,
did vary by season, as shown in Figure 7. Interviewers also sometimes reported difficulty reaching
sampled members because they were away, fishing. The CRITFC report also showed strong
variation across the 12 calendar months in the percentage of respondents identifying a month as
one of high consumption, and, separately, identifying low consumption months (CRITFC, 1994,
Figures 3 and 4). (The Nez Perce Tribe was one of the four tribes included in the CRITFC survey.)
Analysis of data from the current survey showed no discernible seasonal patterns—that differed
from "noise"—in consumption rates for the species groups analyzed, including salmon (all salmon
and steelhead species combined). The sample sizes were too small to rule out seasonal variation,
but there was no pattern that could be used to create a method for seasonal adjustment of the
consumption rate distributions. It is possible that a large fraction of the Tribal members tend to be
fairly steady over time in their FCR. A fairly steady consumption rate could be managed if tribal
members alternate species according to availability (by harvest or purchase), and, also, draw on
preserved or otherwise stored fish harvested from peak periods of availability.
An additional source of uncertainty about the results of the NCI method of analysis is the role of
the question wording and question sequence used to gather the 24-hour recall data used for the
NCI method (and also used for calculation of mean consumption rates using the naive method,
described in Section 5.22). The 24-hour recall portion of the questionnaire was adapted (and
shortened) from the AMPM method (Automated Multiple Pass Method), a thorough and probing
method to elicit all foods consumed during a 24-hour period (Raper et al., 2004, Moshfegh et al.,
2008). Similar to the AMPM system, the present survey questionnaire included an inventory of
occasions with fish consumption, but, in order to avoid problems from an overly long interview
(e.g., fatigue, dropout, inaccurate answers) there was only one pass through the eating occasions
rather than the multiple passes of the AMPM system. In the current survey a lead-in question
(Appendix A, question #9) could filter out any respondent who reported eating no fish
"yesterday." Such a respondent would be assigned zero fish consumption, would not answer
subsequent questions about specific eating occasions, and would skip to questions on other
topics. It is possible that some of the respondents who may have been recorded as having zero
fish consumption on the 24-hour recall—due to their response on the lead-in question—would
have reported non-zero fish consumption if they had proceeded to a more detailed questioning
about eating occasions. The impact of this phenomenon is unknown but is expected to be small,
since the lead-in question is thorough in asking about potential types and occasions of
consumption, and the interviewers would commonly probe for fish consumption "yesterday."
Nez Perce Tribe
Volume II: Current Fish Consumption Survey
December 2016 Page 120
-------
7.6 Next Steps, Lessons Learned
Many lessons were learned in the process of developing and implementing the survey, analyzing
survey data, and drafting these reports. A "Lessons Learned" memorandum reflecting the
experience of the Tribes, contractors, and EPA will be forthcoming.
7.7 Conclusions
The Nez Perce Tribe is a high fish-consuming population. The mean adult consumption rate for
all fish species (Group 1) is 75.0 g/day and the 95th percentile of consumption is 232.1 g/day—
estimates based on the survey data as analyzed by the NCI method. The consumption rates based
on the data from the food frequency questionnaires (FFQ) in this survey are also high: mean
consumption, 123.4 g/day, 95th percentile, 437.4 g/day.
The Nez Perce NCI method mean fish consumption rate of 75.0 g/day is high relative to the
general U.S. population (NCI-method mean, 23.8 g/day). The FFQ fish consumption rate is also
high relative to some other Pacific Northwest tribes—tribes that can only be compared using an
FFQ rate (Table 28). The Nez Perce mean FFQ consumption rate of 123.4 g/day can be
compared to the pooled CRITFC survey Tribes (FFQ mean, 63.2 g/day), Squaxin Island Tribe
(FFQ mean, 83.7 g/day) and Tulalip Tribes (FFQ mean, 82.2 g/day). The Nez Perce mean FFQ
rate of 123.4 g/day is lower than the FFQ rate for the Suquamish Tribe, 213.9 g/day. The Nez
Perce NCI method rates can be compared to and are higher than those of the Shoshone-Bannock
Tribes, the second tribal population included in the current consumption survey. The NCI
method mean rates for the NPT and SBT are 75.0 g/day and 34.9 g/day, respectively. The NPT
and SBT 95th percentile rates are 232.1 g/day and 140.9 g/day, respectively.
Nez Perce consumption rates were also high when restricted to Group 2 species (near coastal,
estuarine, freshwater, and anadromous finfish and shellfish) with a mean of 66.5 g/day and 95th
percentile of 233.9 g/day by the NCI method. The mean Group 2 consumption rate based on the
FFQ was 104.0 g/day and the 95th percentile was 327.9 g/day.
The population of documented fishers within the Nez Perce Tribe has even higher FCRs than the
overall tribal population (Tables 9, 12). There has been a substantial reported change in access to
fish and fishing according to tribal respondents, and the greatest change is that a much larger
proportion of the population has experienced a decrease in access to fishing than the proportion
of those experiencing an increase in access (Table 25).
Consumption rates obtained via the NCI method are likely closer to the actual rates than rates
obtained using the FFQ method. However, the FFQ approach is a well-documented and accepted
method for conducting dietary intake surveys, and may be used to produce credible results when
sample size or resources cannot support the NCI method. The resources required to collect data
for and implement the NCI method are considerable and are likely not often available to tribes
with limited resources. The current surveys of fish consumption among two tribes in Idaho, the
Nez Perce and the Shoshone-Bannock Tribes, show differing levels of agreement between NCI
and FFQ method FCRs. The source of these differences in rates between NCI and FFQ methods
appears to be associated with reported fish consumption frequency, and, to a lesser degree,
reported portion size for the FFQ vs. the 24-hour recall data. The current Idaho tribal surveys are
the first to conduct both methods simultaneously. Future surveys will be needed to elucidate
Nez Perce Tribe
Volume II: Current Fish Consumption Survey
December 2016 Page 121
-------
differences between the two methods. Given the resources required to conduct these surveys,
acquisition of further data will not occur rapidly.
Multiple studies using different methodologies (e.g., ethnographic observation, caloric intake,
etc.) demonstrate that heritage FCRs exceeded current FCRs, as is shown in Volume I.
A lesson learned from the survey activity is the importance of strong support from the tribal
leadership and staff in order to achieve acceptance of the survey and higher response rates and
the need for significant advance time and preparation prior to field work.
Nez Perce Tribe
Volume II: Current Fish Consumption Survey
December 2016 Page 122
-------
8.0 References for Volume II
AAPOR (American Association for Public Opinion Research). Standard Definitions: Final
Dispositions of Case Codes and Outcome Rates for Surveys. Revised 2011.
Appendix E to U.S. EPA. 2014. Estimated Fish Consumption Rates for the U.S. Population and
Selected Subpopulations (NHANES 2003-2010). U.S. Environmental Protection
Agency, Washington, DC, EPA-820-R-14-002.
CRITFC (Columbia River Inter-Tribal Fish Commission). "A Fish Consumption Survey of the
Umatilla, Nez Perce, Yakama, and Warm Springs Tribes of the Columbia River Basin."
Technical Report 94-3. Portland, Oregon. 1994.
CRITFC, 1994. A common shorthand notation for the full CRITFC reference above.
De Keyser W, Bracke T, McNaughton S, Parnell W, Moshfegh A, Pereira R, Lee H-S, et al.
(2015). "Cross-Continental Comparison of National Food Consumption Survey
Methods—A Narrative Review." Nutrients 7(5). 2015. MDPI AG. Retrieved from
http://dx.doi.org/10.3390/nu7053587.
Dodd KW, Guenther PM, Freedman LS, Subar AF, Kipnis V, Midthune D, Tooze JA, Krebs-
Smith SM. "Statistical Methods for Estimating Usual Intake of Nutrients and Foods: A
Review of the Theory." Journal of the American Dietetic Association 106(10). 2006. pp
1640-50.
Francisco CA, Fuller W. "Estimation of the Distribution Function with a Complex Survey." Iowa
State University, 1986.
Groves, RM; Fowler, FJ; Couper, MP; Lepkowski, JM; Singer, E; Tourangeau, R. Survey
Methodology. New Jersey: John Wiley & Sons, 2009.
Haines, P. S., Hama, M. Y., Guilkey, D. K., and Popkin, B. M. (2003). Weekend Eating in the
United States Is Linked with Greater Energy, Fat, and Alcohol Intake. Obesity Research,
11: 945-949.
Herbert JR, Clemow L, Pbert L, Ockene IS, and Ockene JK (1995). Social Desirability Bias in
Dietary Self-Report May Compromise the Validity of Dietary Intake Measures.
International Journal of Epidemiology, 24(2): 389-397.
Hosmer DW, Lemeshow S. Applied Logistic Regression. New York: Wiley. 2000.
Kipnis V, Midthune D, Buckman DW, Dodd KW, Guenther PM, Krebs-Smith SM, Subar AF,
Tooze JA, Carroll RJ, Freedman LS. "Modeling Data with Excess Zeros and
Measurement Error: Application to Evaluating Relationships Between Episodically
Consumed Foods and Health Outcomes." Biometrics 65. 2009.
Nez Perce Tribe
Volume II: Current Fish Consumption Survey
December 2016 Page 123
-------
Liao S. Excel Spreadsheets of Percentiles of Consumer-Only Rates (g/kg-day) for the Suquamish
Tribe—Various Species Groups. 2002.
Lumley T. "Analysis of Complex Survey Samples." Journal of Statistical Software 9(1). 2004.
Lumley T. "Analysis of Complex Survey Samples." R package, Version 3.30. 2014.
Lumley T. Complex Surveys: A Guide to Analysis Using R. John Wiley and Sons, 2010.
Johnson CL, Paulose-Ram R, Ogden CL, et al. "National Health and Nutrition Examination
Survey: Analytic Guidelines, 1999-2010." National Center for Health Statistics. Vital
Health Statistics 2(161). 2013. Available for download from:
http://www.cdc.gov/nchs/data/series/sr_02/sr02_161.pdf.
Moshfegh AJ, Rhodes DG, Baer DJ, Murayi T, Clemens JC, Rumpler WV, Paul DR, Sebastean
RS, Kiczynski KJ, Ingwersen LA, Staples RC, Cleveland LE. "The U.S. Department of
Agriculture Automated Multiple-Pass Method Reduces Bias in the Collection of Energy
Intakes." The American Journal of Clinical Nutrition 88(2). 2008.
Oregon Department of Environmental Quality. "Human Health Criteria Issue Paper, Toxics
Rulemaking." May 24, 2011.
http://www.deq.state.or.us/wq/standards/docs/toxics/humanhealth/rulemaking/HumanHea
lthToxicCriteriaIssuePaper.pdf.
Polissar NL, Neradilek M, Hippe DS, Aravkin AY, Danaher P, Kalat J. "Statistical Analysis of
National and Washington State Fish Consumption Data." The Mountain-Whisper-Light
Statistics, September 7, 2014.
R Core Team. "R: A Language and Environment for Statistical Computing." Versions
3.1.1-3.1.3. R Foundation for Statistical Computing, Vienna, Austria. URL
http://www.R-project.org/. 2015.
Raper N, Perloff B, Ingwersen L, Steinfeldt L, Anand J. "An Overview of USDA's Dietary
Intake Data System." Journal of Food Composition and Analysis 17. 2004.
Sechena R, Liao S, Lorenzana R, Nakano C, Polissar N, Fenske R. Asian American and Pacific
Islander Seafood Consumption - A Community-Based Study in King County,
Washington. Journal of Exposure Analysis and Environ Epidemiology, (13): 256-266.
2003.
Sechena, R, C Nakano, S Liao, N Polissar, R Lorenzana, S Truong, and R Fenske. "Asian and
Pacific Islander Seafood Consumption Study in King County, Washington." EPA 910/R-
99-003. May 1999.
Singer, E, Van Hoewyk, J, Gebler, N, Raghunathan, T, McGonagle. K. "The Effect of Incentives
on Response Rates in Interviewer-Mediated Surveys." Journal of Official Statistics,
15(2): 217-230. 1999.
Nez Perce Tribe
Volume II: Current Fish Consumption Survey
December 2016 Page 124
-------
Subar, AF, Kipnis V, Troiano, RP, Midthune D, Schoeller DA, Bingham S, Sharbaugh CO,
Trabulsi J, Runswick S, Ballard-Barbash R, Sunshine J, Schatzkin, A. "Using Intake
Biomarkers to Evaluate the Extent of Dietary Misreporting in a Large Sample of Adults:
The OPEN Study." American Journal of Epidemiology. 158(1): 1-13. July 1, 2003.
Syracuse Research Corporation (SRC). Data Analysis Report for the Tribal Consumption and
Resource Use Survey for the Upper Columbia River Site Human Health Risk Assessment
(HHRA) and Remedial Investigation Feasibility Study (RI/FS). February 2015.
The Suquamish Tribe. 2000. "Fish Consumption Survey of the Suquamish Indian Tribe of the
Port Madison Indian Reservation." Puget Sound Region. August 2000. Report writing
group: M. Duncan, N.L. Polissar, S. Liao, D. LaFlamme.
Tooze, JA, et al. "A Mixed-Effects Model Approach for Estimating the Distribution of Usual
Intake of Nutrients: The NCI Method" Statistical Medicine 2010 Nov 30;29(27):
2857-2868.
Tooze JA, Midthune D, Dodd KW, Freedman LS, Krebs-Smith SM, Subar AF, Guenther
PM, Carroll RJ, Kipnis V. "A New Statistical Method for Estimating the Usual Intake of
Episodically Consumed Foods with Application to Their Distribution." Journal of the
American Dietetic Association 106:10. 2006.
Tooze JA, Subar AF, Thompson FE, Troiamo R, Schatzkin A, and Kipnis V (2004).
Psychosocial Predictors of Energy Underreporting in a Large Doubly Labeled Water
Study. The American Journal of Clinical Nutrition, 79: 795-904.
Toy, K.A., N.L. Polissar, S. Liao, G.D. Mittelstaedt. 1996. A Fish Consumption Survey of the
Tulalip and Squaxin Island Tribes of the Puget Sound Region. Tulalip Tribes,
Department of Environment, 7615 Totem Beach Road, Marysville, Washington 98271.
U.S. EPA. Human Health Ambient Water Quality Criteria and Fish Consumption Rates
Frequently Asked Questions.
http://water.epa.gov/scitech/swguidance/standards/criteria/health/
methodology/upload/hhfaqs.pdf. January 18, 2013.
U.S. EPA. 2014. Estimated Fish Consumption Rates for the U.S. Population and Selected
Subpopulations (NHANES 2003-2010). U.S. Environmental Protection Agency,
Washington, DC, EPA-820-R-14-002.
Washington Department of Ecology. Washington State Water Quality Standards: Human Health
Criteria and Implementation Tools, Overview of Key Decisions in Rule Amendment.
Publication No. 14-10-058.
https://fortress.wa.gov/ecy/publications/documents/1410058.pdf. January, 2015.
World Health Organization (WHO), Fact sheet: "Obesity and Overweight," Updated June 2016.
Accessed at http://www.who.int/mediacentre/factsheets/fs311/en/ on 2 July 2016.
Nez Perce Tribe
Volume II: Current Fish Consumption Survey
December 2016 Page 125
-------
List of Appendices
Appendix A:
Idaho Tribes Fish Consumption Survey: Questionnaire
Appendix B:
Portion-to-Mass Conversion
Appendix C:
Additional Detail on Imputations
Appendix D:
Additional Detailed Tables and Methodologic Notes
Appendix E
Expanded Tables and Additional Notes on the NCI Method
Appendix F
Comparison of FFQ Rates to 24-Hour Recall Rates
Appendix G
Geographic Inclusion Criteria—Additional Information
Appendix H:
Design of a Survey on Fish Consumption by the Nez Perce
Tribe—Final Design Document
Nez Perce Tribe
Volume II: Current Fish Consumption Survey
December 2016 Page 126
-------
Volume III:
Appendices to Volume II,
Current Fish Consumption
Survey - Nez Perce Tribe
-------
Contents
Appendix A— Idaho Tribes Fish Consumption Survey: Questionnaire A-i
Appendix B— Portion-to-Mass Conversion B-l
Appendix C— Additional Detail on Imputations C-l
Appendix D—Additional Detailed Tables and Methodologic Notes D-l
Appendix E— Expanded Tables and Additional Notes on the NCI Method E-l
Appendix F— Comparison of FFQ Rates to 24-Hour Recall Rates F-l
Appendix G— Geographic Inclusion Criteria—Additional Information G-l
Appendix H— Survey Design Document— Nez Perce Tribe H-l
Nez Perce Tribe
Volume III
December 2016
-------
List of Tables
Table A-l. Telephone Screening Contact Log
Table A-2. Disposition Codes for Respondent Contact
Table A-3. 24-Hr Recall: Types, Quantities, Methods, and Sources of Fish Eaten
Yesterday
Table A-3a. Portion Size Model Displays: Description and Use
Table A-4. FFQ: Types, Frequency, and Quantity of Species Eaten in Past 12 Months
Table A-5. FFQ: Fish Parts Eaten, Preparation Methods, and Sources
Table A-6. FFQ: Fish Consumption at Gatherings
Table A-7. 24-Hr Recall #2: Types, Quantities, Methods, and Sources of Fish Eaten
Yesterday
Table A-7a. Portion Size Model Displays: Description and Use
Table Bl. Survey Species List
Table B2. Description of Portion Size Model Displays
Table B3. Estimated Fish Moisture Loss Due to Cooking
Table B4. Portion-to-mass (raw weight, edible portion) conversions for the salmon
replica with fillet divided into sections
Table B5. Portion-to-mass (raw weight, edible portion) conversions for other models
Table B6. Portion-to-mass (raw weight, edible portion) conversions for jerky, depending
on the jerky model and species
Table B7. Choice of shellfish model when not specified by the interviewer
Table CI. Species groupings used to impute missing values for uncommon species (less
than 5 non-missing responses).
Table C2. Sensitivity analysis of imputation method on the Group 1 FCRs (g/day, raw
weight, edible portion). Estimates are weighted
Table C3. Sensitivity analysis of imputation method on the Group 2 FCRs (g/day, raw
weight, edible portion). Estimates are weighted.
Nez Perce Tribe
Volume III
December 2016
-------
Table C4. Sensitivity analysis of imputation method on the Group 3 FCRs (g/day, raw
weight, edible portion). Estimates are weighted.
Table C5. Sensitivity analysis of imputation method on the Group 4 FCRs (g/day, raw
weight, edible portion). Estimates are weighted.
Table C6. Sensitivity analysis of imputation method on the Group 5 FCRs (g/day, raw
weight, edible portion). Estimates are weighted
Table C7. Sensitivity analysis of imputation method on the Group 6 FCRs (g/day, raw
weight, edible portion). Estimates are weighted.
Table Dl. Demographics of the population, selected sample and first interview
consumers with known consumption rates. Estimates are unweighted
Table D2. Demographics of the FFQ consumers with known consumption rates.
Estimates are unweighted.
Table D3. Estimated distribution of FCRs (g/day, raw weight, edible portion) of
consumers within demographic groups. All rates are for total consumption (Group 1).
Estimates are weighted. Mean, SD, median ('50%') and percentiles.
Table D4. Enumeration of household clusters. Respondent IDs within each cluster are
comma separated. See section 5.25 on confidence intervals for a discussion on impact.
Table D5. Expected number of "double-hits" for two independent interviews based on
the noted sample size of respondents and two different sampling methods.
Table D6. Number of fish meals consumed by all adult respondents (fish consumers and
non-fish consumers) per week - throughout the year.
Table El. Distribution of the usual fish Group 1 (all fish) consumption (g/day, raw
weight, edible portion) based on the 24 hour recalls. Estimated by the NCI method.
Table E2. Distribution of the usual fish Group 2 consumption (g/day, raw weight, edible
portion) based on the 24 hour recalls. Estimated by the NCI method.
Table E3. Distribution of the usual fish Group 1 (all fish) consumption (g/day, raw
weight, edible portion) and their 95% confidence intervals based on the 24 hour recalls.
Estimated by the NCI method.
Table E4. Coefficients for the NCI models considered in the selection of the FFQ
covariate form. Model for Group 1 species. Only selected coefficients are presented for
the reference model with categorical decile of FFQ ("Cat. FFQ") and for the model with
no FFQ (i.e., model with tribe only).
Table E5. Final model NCI for Group 1.
Nez Perce Tribe
Volume III
December 2016
-------
Table E6. Final model NCI for Group 2.
Table E7. NCI estimates (g/day, raw weight, edible portion) from the final model vs.
model with logio FFQ replacing 3rd root of FFQ. Group 1 consumption.
Table E8. NCI estimates (g/day, raw weight, edible portion) from the final model vs.
model with logio FFQ replacing 3rd root of FFQ. Group 2 consumption.
Table E9. NCI estimates (g/day, raw weight, edible portion) from the final model vs.
final model without the weekend adjustment. Group 1 consumption.
Table E10. NCI estimates (g/day, raw weight, edible portion) from the final model vs.
final model without the weekend adjustment. Group 2 consumption.
Table El 1. NCI estimates (g/day, raw weight, edible portion) from the final model vs.
final model without the sequence effect adjustment. Group 1 consumption.
Table E12. NCI estimates (g/day, raw weight, edible portion) from the final model vs.
final model without the sequence effect adjustment. Group 2 consumption
Table E13. NCI estimates (g/day, raw weight, edible portion) from the final model vs.
final model without correlation between the probability and consumed amount. Group 1
consumption.
Table E14. NCI estimates (g/day, raw weight, edible portion) from the final model vs.
final model without correlation between the probability and consumed amount. Group 2
consumption
Table El 5. NCI estimates (g/day, raw weight, edible portion) for the NPT from the final
model fit to data from NPT + SBT vs. final model fit only to the NPT data. Group 1
consumption.
Table E16. NCI estimates (g/day, raw weight, edible portion) for the NPT from the final
model fit to data from NPT + SBT vs. final model fit only to the NPT data Group 2
consumption.
Table E17. NCI estimates (g/day, raw weight, edible portion) from the final model vs.
simpler model (tribe, 3rd root of FFQ, tribe by 3rd root of FFQ interaction and a single
covariate for groups as needed). Group 1 consumption.
Table E18. Comparison of FCRs (g/day, raw weight, edible portion, based on 24-hour
recall data) between 24-hour recall interviews conducted during the peak salmon harvest
period (May 2014 through July 2014) vs. the remainder of the survey period (August
2014 through May 2015). Nez Perce Tribe. Consumers only*. Estimates are weighted.
Nez Perce Tribe
Volume III
December 2016
-------
Table E19. Comparison of reported fishing rates (mean times per month) between first
interviews conducted during the peak salmon harvest period (May 2014 through July
2014) vs. FFQ interviews conducted during the remainder of the survey period (August
2014 through April 2015). Nez Perce Tribe. Consumers only. Estimates are weighted.
Table E20. Frequencies of two-period FFQ responses (consumption information provided
for higher and lower consumption periods separately) out of all responses*, compared
between FFQ interviews conducted during the peak salmon harvest period (May 2014
through July 2014) vs. the remainder of the survey period (August 2014 through April
2015). Nez Perce Tribe. Estimates are unweighted.
Table Fl. Weighted group 1 means and other statistics from the 24-hour recall and the
FFQ consumption rates (g/day) by Group 1 FFQ consumption rate deciles.
Table F2. Weighted group 2 means and other statistics from the 24-hour recall and the
FFQ consumption rates (g/day) by Group 1 FFQ consumption rate deciles.
Table F3. Weighted means of the 24-hour recall and of the FFQ consumption rates
(g/day) by species group. All Group 1 consumers.
Table F4. Weighted means of the 24-hour recall and of the FFQ consumption rates
(g/day) by species group. Group 1 consumers in the 10th decile.
Table F5. Weighted means of the 24-hour recall and of the FFQ consumption rates
(g/day) by species group. Group 1 consumers in the 9th decile.
Table F6. Weighted mean consumption from the 24-hour recall and FFQ for each species
group, overall and by decile. Deciles are the deciles of the Group 1 FFQ consumption
rate. All rows are based on all Group 1 consumers. Ratios were not calculated when a
species group was not consumed by the FFQ.
Table F7. Weighted mean frequency of positive daily consumption from the 24-hour
recall and FFQ for each species group, overall and by decile. Deciles are the deciles of
the Group 1 FFQ consumption rate. All rows are based on all Group 1 consumers. Ratios
were not calculated when a species group was not consumed by the FFQ.
Table F8. Weighted mean portion size (grams) from the 24-hour recall and FFQ for each
species group, overall and by decile. Deciles are the deciles of the Group 1 FFQ
consumption rate. Each individual's portions sizes were averaged across species with a
weight according to the species frequency. All calculations are limited to positive (non-
zero) portion sizes. Ratios were not calculated when a species group was not consumed,
as noted on the FFQ or 24-hour recall.
Table F9. Number and % respondents with any "not otherwise specified" species
designation (NOS) on the FFQ. Overall and by species group.
Nez Perce Tribe
Volume III
December 2016
-------
Table F10. Unweighted simple linear regressions of the FFQ-24-hour difference on the
number of "not otherwise specified" species (NOS) in the FFQ data per respondent.
Overall and by species. Slope per 1 NOS species. 95% confidence intervals are
approximate (assuming asymptotic normality).
Table F11. Unweighted simple linear regressions of the FFQ-24-hour difference on the
number of species with missing data per respondent. Overall and by species. Slope per 1
missing-data species. 95% confidence intervals are approximate (assuming asymptotic
normality).
Table Gl. Nez Perce reservation ZIP codes, corresponding population centers, and
distances to sampling hubs for the Nez Perce Tribe survey
Nez Perce Tribe
Volume III
December 2016
-------
List of Figures
Figure Bl. Salmon Fillet Replica (24 Servings)
Figure B2. Trout-Like Fillet Replica (Single Serving)
Figure B3. PVC "Lamprey" Pipe (7 Servings)
Figure B4. Package of Real Jerky/Dried Fish ("Salmon Candy")
Figure B5. Measuring Bowls for Fish Soups
Figure B6. Crayfish Photo-Display
Figure B7. Mussels Photo-Display
Figure B8. Shrimp Photo-Display
Figure B9. Species Identification Photographs
Figure El. Comparison of four forms of FFQ adjustment (colored lines) to the
categorical decile FFQ adjustment (black bars). Model for Group 1 species.
DECILENUM2 = the numerical decile of FFQ (coded as 1-10), LIN = the original
(untransformed) FFQ, LOGIO = the logio FFQ, RT3 = the 3rd root FFQ. All models
included an addition adjustment for the 10th decile in the SBT. meanmct = mean,
tpercentile50, 90 and 95 = the 50th, 90th and 95th percentiles, respectively. Estimates are
NCI estimates of daily consumption in g/day (raw weight, edible portion).
Figure E2. NCI-estimated mean and the 50th, 90th and 95th percentiles by the presence on
the fishers list and tribe. Model for Group 1 species. Other covariates include the 3rd root
of FFQ, its interaction with tribe and the indicator for SBT decile 10. Dots are estimates
from 50 bootstrap runs and give some idea of uncertainty around the estimates.
Estimates are NCI estimates of daily consumption in g/day (raw weight, edible portion).
Figure E3. NCI-estimated mean and the 50th, 90th and 95th percentiles by sender and
tribe. Model for Group 1 species. Other covariates include the 3rd root of FFQ, its
interaction with tribe and the indicator for SBT decile 10. Dots are estimates from 50
bootstrap runs and give some idea of uncertainty around the estimates. Estimates are
NCI estimates of daily consumption in g/day (raw weight, edible portion).
Figure E4. NCI-estimated mean and the 50th, 90th and 95th percentiles by ZIP code.
Model for Group 1 species. Other covariates include the 3r root of FFQ, its interaction
with tribe and the indicator for SBT decile 10. Dots are estimates from 50 bootstrap runs
and give some idea of uncertainty around the estimates. Estimates are NCI estimates of
daily consumption in g/day (raw weight, edible portion).
Nez Perce Tribe
Volume III
December 2016
-------
Figure E5. NCI-estimated mean and the 50th, 90th and 95th percentiles by age and tribe.
Model for Group 1 species. Other covariates include the 3r root of FFQ, its interaction
with tribe and the indicator for SBT decile 10. Dots are estimates from 50 bootstrap runs
and give some idea of uncertainty around the estimates. Estimates are NCI estimates of
daily consumption in g/day (raw weight, edible portion).
Figure E6. Comparison of four forms of respondent weight adjustment (color lines) to
the categorical decile respondent weight adjustment (black bars). Model for Group 1
species. DECILENUM2 = the numerical decile of respondent weight (coded as 1-10),
LIN = the original (untransformed) respondent weight, LOG10 = the logio respondent
weight, RT3 = the 3rd root respondent weight. Models include an adjustment for FFQ.
mean_mc_t = mean, tpercentile50, 90 and 95 = the 50th, 90th and 95th percentiles,
respectively. Estimates are NCI estimates of daily consumption in g/day (raw weight,
edible portion).
Figure E7. Comparison of four forms of FFQ adjustment (colored lines) to the
categorical decile FFQ adjustment (black bars). Model for Group 2 species.
DECILENUM2 = the numerical decile of FFQ (coded as 1-10), LIN = linear—the
original (untransformed) FFQ, LOGIO = the logio FFQ, RT3 = the 3rd root FFQ. All
models included an addition adjustment for the 10th decile in SBT. meanmct = mean,
tpercentile50, 90 and 95 = the 50th, 90th and 95th percentiles, respectively. Estimates are
NCI estimates of daily consumption in g/day (raw weight, edible portion).
Figure E8. NCI-estimated mean and the 50th, 90th and 95th percentiles by the presence on
the fishers list and tribe. Model for Group 2 species. Other covariates include the 3rd root
of FFQ, its interaction with tribe and the indicator for the SBT decile 10. Estimates are
NCI estimates of daily consumption in g/day (raw weight, edible portion).
Figure E9. NCI-estimated mean and the 50th, 90th and 95th percentiles by gender and
tribe. Model for Group 2 species. Other covariates include the 3rd root of FFQ, its
interaction with tribe and the indicator for SBT decile 10. Estimates are NCI estimates of
daily consumption in g/day (raw weight, edible portion).
Figure E10. NCI-estimated mean and the 50th, 90th and 95th percentiles by ZIP code.
Model for Group 2 species. Other covariates include the 3rd root of FFQ, its interaction
with tribe and the indicator for the SBT decile 10. Estimates are NCI estimates of daily
consumption in g/day (raw weight, edible portion).
Figure El 1. NCI-estimated mean and the 50th, 90th and 95th percentiles by age and tribe.
Model for Group 2 species. Other covariates include the 3rd root of FFQ, its interaction
with tribe and the indicator for SBT decile 10. Estimates are NCI estimates of daily
consumption in g/day (raw weight, edible portion).
Nez Perce Tribe
Volume III
December 2016
-------
Figure E12. Comparison of four forms of respondent body weight adjustment (colored
lines) to the categorical decile of respondent weight adjustment (black bars). Model for
Group 2 species. DECILENUM2 = the numerical decile of respondent weight (coded as
1-10), LIN = the original (untransformed) respondent weight, LOGIO = the logio
respondent weight, RT3 = the 3rd root respondent weight. Models include an adjustment
for FFQ. meanmct = mean, tpercentile50, 90 and 95 = the 50th, 90th and 95th
percentiles, respectively. Estimates are NCI estimates of daily consumption in g/day
(raw weight, edible portion).
Figure El 3. The (survey-weighted) distribution of the person-means and within-person
residuals of the third root of the positive Group 1 consumption amounts. Both tribes
combined. The units of the original values were g/day (raw weight, edible portion).
Figure E14. The (survey-weighted) distribution of the person-means and within-person
residuals of the third root of the positive Group 2 consumption amounts. Both tribes
combined. The units of the original values were g/day (raw weight, edible portion).
Figure E15. Quality checking of NCI model for Group 1 species. Consumption
probability and mean amount on consumption days by the respondent's presence on the
fishers list. Prob = Probability, Amt = positive consumption amount (in g/day, raw
weight, edible portion). 0 = not on the fishers list. 1= on the fishers list. The y-axis
shows either the consumption probability (between 0 and 1) or the mean amount on
consumption days. Naive 2-hit = naive approach limited to respondents with 2
interviews, naive all = naive approach with all respondents, naive intl = naive approach
limited to 1st interviews, NCI = the NCI model estimate.
Figure E16. Quality checking of NCI model for Group 1 species. Consumption
probability and mean amount on consumption days by the respondent's sender. Prob =
Probability, Amt = positive consumption amount (in g/day, raw weight, edible portion).
0 = men. 1= women. The y-axis shows either the consumption probability (between 0
and 1) or the mean amount on consumption days. Naive 2-hit = naive approach limited
to respondents with 2 interviews, naive all = naive approach with all respondents, naive
intl = naive approach limited to 1st interviews, NCI = the NCI model estimate.
Figure E17. Quality checking of NCI model for Group 1 species. Consumption
probability and mean amount on consumption days by the respondent's ZIP code. Prob
= Probability, Amt = positive consumption amount (in g/day, raw weight, edible
portion). The y-axis shows either the consumption probability (between 0 and 1) or the
mean amount on consumption days. Naive 2-hit = naive approach limited to respondents
with 2 interviews, naive all = naive approach with all respondents, naive intl = naive
approach limited to 1st interviews, NCI = the NCI model estimate.
Nez Perce Tribe
Volume III
December 2016
-------
Figure E18. Quality checking of NCI model for Group 1 species. Consumption
probability and mean amount on consumption days by the respondent's age. Prob =
Probability, Amt = positive consumption amount (in g/day, raw weight, edible portion).
The y-axis shows either the consumption probability (between 0 and 1) or the mean
amount on consumption days. Naive 2-hit = naive approach limited to respondents with
2 interviews, naive all = naive approach with all respondents, naive intl = naive
approach limited to 1st interviews, NCI = the NCI model estimate.
Figure E19. Quality checking of NCI model for Group 1 species. Consumption
probability and mean amount on consumption days by the respondent's decile of group
1 FFQ consumption. Prob = Probability, Amt = positive consumption amount (in g/day,
raw weight, edible portion). The y-axis shows either the consumption probability
(between 0 and 1) or the mean amount on consumption days. Naive 2-hit = naive
approach limited to respondents with 2 interviews, naive all = naive approach with all
respondents, naive intl = naive approach limited to 1st interviews, NCI = the NCI model
estimate.
Figure E20. Bootstrap distribution of the NCI method estimated means and selected
percentiles for all NPT and SBT respondents. N=978 bootstraps (22 of the 1000
bootstraps did not converge). Group 1 consumption (in g/day, raw weight, edible
portion). Red dot shows the point estimate and the red bar around it shows the 95%
confidence interval.
Figure E21. Bootstrap distribution of the NCI method estimated means and selected
percentiles for NPT and SBT respondents on the fishers list. N=978 bootstraps (22 of the
1000 bootstraps did not converge). Group 1 consumption (in g/day, raw weight, edible
portion). Red dot shows the point estimate and the red bar around it shows the 95%
confidence interval.
Figure E22. Seasonality for Group 1 species consumption on the 24-hour recall. Mean
24-hour recall for species Group 1 consumption (g/day, raw weight, edible portion) by
tribe, month and interview number (1st or 2nd 24-hour recall interview). Numbers within
each month's dot are the sample size. One very large data point for a single NPT second
interview during May (5/14) was excluded from this seasonal analysis
Figure E23. Seasonality for Group 2 species consumption on the 24-hour recall. Mean
24-hour recall for species Group 2 consumption (g/day, raw weight, edible portion) by
tribe, month and interview number. Numbers within each month's dot are the sample
size. One outlier data point for a single NPT second interview during May (5/14) was
excluded.
Figure E24. Seasonality for salmon and steelhead consumption on the 24-hour recall.
Mean 24-hour recall consumption rate (g/day, raw weight, edible portion) for all salmon
and steelhead species (combined) by tribe, interview month and interview number (1st
and 2nd interview). Numbers within each month's dot are the sample size. One outlier
data point for a single NPT second interview during May (5/14) was excluded.
Nez Perce Tribe
Volume III
December 2016
-------
Figure E25. Seasonality for Group 1 species. Group 2 species and salmon+steelhead
consumption on the FFQ. Mean Group 1 FFQ consumption rate (g/day, raw weight,
edible portion) by tribe, species group and interview month. Numbers within each
month's dot are the sample size. Salmon: all salmon and steelhead species combined.
Figure E26. Seasonality in the % fisher respondents. Percentages of fishers among
respondents by tribe, interview month and interview number (1st and 2nd interviews).
Numbers within each month's dot are the sample size.
Figure Fl. Distribution of the # "not otherwise specified" species (NOS) on FFQ per
respondent.
Figure F2. Histogram of the number of species with missing data on the FFQ per
respondent.
Figure Gl. Nez Perce reservation and surrounding eligible ZIP codes for inclusion in the
Nez Perce Tribe fish consumption survey.
Nez Perce Tribe
Volume III
December 2016
-------
Appendix A—
Idaho Tribes Fish Consumption Survey:
Questionnaire
Nez Perce Tribe
Volume III - Appendix A
December 2016 Page A-i
-------
Appendix A—Questionnaire1
IDAHO TRIBES FISH CONSUMPTION SURVEY
APPENDIX A
TABLE OF CONTENTS
1.0 TELEPHONE SCREENING 1
2.0 INTERVIEW INTRODUCTION 9
2.1 Administrative Information 9
2.1.1 Interviewer Identification 9
2.1.2 Respondent Identification 9
2.1.3 Interview Date, Time, and Location 9
2.2 Introduction to Interview 10
3.0 24-HOUR DIETARY RECALL 11
3.1 Fish Consumption 11
3.2 Other Dietary Information 19
4.0 FOOD FREQUENCY QUESTIONNAIRE 20
4.1 Fish Consumption 20
4.1.1 Species, Frequency, Quantities 20
4.1.2 Parts of Fish Consumed, Preparation Methods, and Sources 25
4.2 Special Events and Gatherings 27
4.3 Fishing Activities 28
4.4 Changes in Fish Consumption 30
5.0 GENERAL INFORMATION 36
5.1 Respondent Information 36
5.1.1 Demographic Information 36
5.1.2 Breastfeeding History 37
5.2 Interview End 38
1 This hardcopy version of the questionnaire was used from time to time as needed. The vast majority of interviews were carried out
with the questionnaire embedded in a CAPI system (computer-assisted personal interviewing) on a tablet. See the main body of this
report for a description of the CAPI system used in this survey.
Nez Perce Tribe
Volume III - Appendix A
December 2016 Page A-ii
-------
5.3 Post-Interview 40
5.3.1 Interview Quality 40
5.3.2 Interviewer Guarantee of Authenticity 41
6.0 SECOND 24-HOUR DIETARY RECALL 42
6.1 Administrative Information 42
6.1.1 Interviewer Identification 42
6.1.2 Respondent Identification 42
6.1.3 Interview Date, Time, and Location 42
6.2 Introduction 43
6.3 Fish Consumption 43
6.4 Other Dietary Information 51
6.5 Post-Interview 51
Nez Perce Tribe
Volume III - Appendix A
December 2016 Page A-iii
-------
LIST OF TABLES
Table
A-l.
Telephone Screening Contact Log
Table
A-2.
Disposition Codes for Respondent Contact
Table
A-3.
24-Hr Recall: Types, Quantities, Methods, and Sources of Fish Eaten Yesterday
Table
A-3a.
Portion Size Model Displays: Description and Use
Table
A-4.
FFQ: Types, Frequency, and Quantity of Species Eaten in Past 12 Months
Table
A-5.
FFQ: Fish Parts Eaten, Preparation Methods, and Sources
Table
A-6.
FFQ: Fish Consumption at Gatherings
Table
A-7.
24-Hr Recall #2: Types, Quantities, Methods, and Sources of Fish Eaten Yesterday
Table
A-7a.
Portion Size Model Displays: Description and Use
LIST OF ACRONYMS
CDC Center for Disease Control and Prevention
FFQ food frequency questionnaire
NCI National Cancer Institute
(NOTE: The original Preface and Telephone Screen in 'oi-y narrative were repetitive of
the main design document and, therefore, removed fror, ppendix.
Nez Perce Tribe
Volume III - Appendix A
December 2016 Page A-iv
-------
1.0TELEPHONE SCREENING
1. "Hello, I'm calling on behalf of the (name of Tribe and department) . May I please
speak with (name of respondent) ?" (Enter contact information into Table A-l; refer
to Table A-2 for response entry codes)
Yes
No
If YES and respondent is speaking or when the respondent comes to the telephone,
continue to Question #2.
If NO, probe if he/she lives there, and if so, ask "When is the best time to reach
him/her? (Record on log) "Okay, thank you for your time. Good bye."
If NO, not living there, ask "What is the best way to reach him/her? (Record new
number on log) "Okay, thank you for your time. Good bye."
2. "Hello, my name is (your name). Reintroduce Tribe if necessary. We are conducting
a survey to determine the fish consumption rates within our Tribe. The survey is
endorsed and supported by the (name council other). Your information, plus the
information of other Tribal members, will help us protect our environment and
promote the health of our Tribal members and families. You are free to not answer
any of the questions. Today's survey takes about 5 minutes and we would like to
include your input, if now is a good time?"
Yes
No
If YES, "thank you for agreeing to participate," check box below and continue to
Question #3.
~ INTERVIEWER CHECK THIS BOX IF RESPONDENT AGREES TO
PARTICIPATE IN THE TELEPHONE SCREENING.
If NO, ask "When is a good time to call back? (Record on log) "Okay, thank you for
your time. Good bye."
3. "I'd like to ask you about what you ate yesterday. Did you eat any fish yesterday?
This includes ANY amount of fish, shellfish, or seafood eaten for breakfast, lunch,
dinner, or snacks, by itself or within a dish such as soup." (Record on log)
Nez Perce Tribe
Volume III - Appendix A
December 2016 Page A-l
-------
Yes
No
Don't know / Prefer not to answer
If YES, skip to Question #8.
If NO or other, continue to Question #4.
4. "Did you eat any fish in the past week (or if not, in the past month)?" (Record on
log)
Yes
No
Don't know / Prefer not to answer
If YES, skip to Question #7.
If NO or other, continue to Question #5.
5. "Did you eat any fish in the past year?" (Record on log)
Yes
No
Don't know / Prefer not to answer
If YES, skip to Question #7.
If NO or other, continue to Question #6.
6. "Thank you. Just to be thorough, is it possible that during the past year you ate fish
at a restaurant, a friend's house or another place, or someone brought fish to you?"
(Record on log)
Yes
No
Don't know / Prefer not to answer
If YES, continue to Question #7.
If NO or other, skip to Question #9.
Nez Perce Tribe
Volume III - Appendix A
December 2016 Page A-2
-------
7. "How many days did you eat fish in the past week (or month or year - depending on
previous answers)?" (This information will determine applicability of the NCI Method;
Record on log as number per week, month, or year)
7a. "Now considering your eating habits in general, on average how many days do you
eat fish - this can be number of times each week, each month, or each year?"
(Record on log as number per week, month, or year)
8. Thank you. We are also conducting survey interviews that have been endorsed by
(endorsing authority) . The information that you provide will remain strictly
confidential and it will help to protect the health of our Tribe. We will conduct in-
person interviews in a convenient location. Your participation is very important. If
you do agree to participate, you may withdraw at any time and there would be no
consequence for you. May we meet with you for the survey interview? (Record on
log)
Yes
No
If YES, "Great, thank you for your willingness to participate in this important
survey. Let's schedule a time and place. We have Tribal interviewers available to
meet 7 days a week from 8:00 am until 7:00 pm; which day in the next two weeks
is best for you?" If don't know, schedule a call-back time to set interview. Record on
log, skip to #10.
If NO, "I understand. This survey is very important. We don't have to do it
immediately, we have several months to schedule it. I'd like to call you back at a
later date. We want to make sure we represent the whole Tribe."
If ACCEPT or SOFT REFUSAL, schedule re-call and skip to #10.
If HARD REFUSAL, "Okay, thank you for your time today. Good bye."
9. "Can you please tell me the main reasons why you haven't eaten fish?" Allow
respondent to answer question unaided, then state "now I will list some other reasons
people do not eat fish; please let know if any of these apply to you." List the
following items (of those not already noted by the respondent). Check left and right
columns, then continue to #10:
Contamination:
A. "Do you not eat fish because of fish advisories?"
Nez Perce Tribe
Volume III - Appendix A
December 2016 Page A-3
-------
Yes
No
Answered unaided
Answered by prompt
B. "Do you not eat fish because of pollution?"
Yes Answered unaided
No Answered by prompt
C. "Do you not eat fish because of other environmental concerns (for example,
eating fish is not sustainable)?"
Yes Answered unaided
No Answered by prompt
Fish Availability:
D. "Do you not eat fish because there is not enough fish available to catch?"
Yes Answered unaided
No Answered by prompt
E. "Do you not eat fish because it is hard to find fresh fish and seafood"
Yes Answered unaided
No Answered by prompt
Access to Fishing:
F. "Do you not eat fish because of limited access to fishing areas?"
Yes Answered unaided
No Answered by prompt
G. "Do you not eat fish because you used to have access to a boat or fishing
gear, but don't anymore?"
Yes Answered unaided
. No Answered by prompt
Other Reasons:
H. "Do you not eat fish because you do not like fish or you prefer other foods?"
Yes Answered unaided
No Answered by prompt
Nez Perce Tribe
Volume III - Appendix A
December 2016 Page A-4
-------
I. "Do you not eat fish because you are too busy to catch and/or prepare fish?'
Yes Answered unaided
No Answered by prompt
J. "Do you not eat fish because you do not know how to prepare fish?"
Yes Answered unaided
No Answered by prompt
K. "Do you not eat fish because you cannot afford it?"
Yes Answered unaided
No Answered by prompt
L. "Do you not eat fish because of allergies or other health concerns?"
Yes Answered unaided
No Answered by prompt
M. "Do you not eat fish because you are a vegetarian or vegan?"
Yes Answered unaided
No Answered by prompt
N. "Do you not eat fish because you observe religious customs?"
Yes Answered unaided
No Answered by prompt
Nez Perce Tribe
Volume III - Appendix A
December 2016 Page A-5
-------
Table A-l. Telephone Screening Contact Log
Respondent Name:
Respondent ID #:
Respondent Telephone Number (strike-out incorrect numbers, record new):
Scheduled Call-Back Time for Telephone Screen (if necessary to re-schedule):
W hen Called
Who Contacted
Results (of call questions)
Attempt
Date
l)av
lime
( ircfc
Caller Name
Caller II)
C odes
Notes
1
AM PM
2
AM PM
3
AM PM
4
AM PM
5
AM PM
6
AM PM
7
AM PM
8
AM PM
9
AM PM
W hen Called
Who Contacted
Results
Attempt
Date
l)a\
lime
AM/I'M
Caller Name
Caller II)
( ode
Notes
10
AM PM
11
AM PM
12
AM PM
13
AM PM
14
AM PM
15
AM PM
Reported eatine fish vesterdav (circle): YES / NO / No Answer
Reported eating fish during past week (circle): YES / NO / No Answer / Not Applicable
Reported eating fish during Dast month (circle): YES / NO / No Answer / Not Applicable
Reported eating fish during Dast vear (circle): YES / NO / No Answer / Not Applicable
Number of days ate fish (enter number, circle unit): in past Week / Month / Year
Number of days generally eat fish (enter number, circle unit): times per Week / Month / Year
Schedule in-person interview? (circle, enter): YES / NO (If NO, enter call-back time at top of form)
Date: (mm/dd/yyyy) Day: Time: am / pm Location:
Nez Perce Tribe
Volume III - Appendix A
December 2016 Page A-6
-------
Respondent ID:
Table A-2. Disposition Codes for Respondent Contact
01
Completed interview
02
Mid-termination
03
Hard Refusal
04
Invalid number: out of service, disconnected, fast busy
05
No answer
06
Busy signal
07
Answering machine
08
Appointment set
09
Language barrier: non-English
10
Impairment: hearing, mental health, other
11
Deceased respondent
12
Institutionalized
13
Other (Please Specify)
14
Soft Refusal
15
Email attempt
16
Enrollment office lookup
17
Acquaintance / family lookup
18
Online lookup
19
Household visit
Note: Interviewers will be trained on how to respond to telephone inquiries (leaving a message,
handling refusals, calling back, etc.)
Nez Perce Tribe
Volume III - Appendix A
December 2016 Page A-7
-------
Respondent ID:
10. Finally, for the survey, we need to note the general location where you live. The zip
code we have listed for your residence is (zip code from enrollment); is that correct?
(Check)
Yes
No
If NO, "Can you please provide your correct RESIDENCE zip code (or if you don't
know the zip code, community name)? -
Final zip code of residence:
This concludes the interview. Thank you very much for your cooperation. We really
appreciate your time today. That is all. Good bye."
2 NOTE: Individuals may have a different zip code for mail versus residence; be sure to inquire about residence. Prior to an in-
person interview, the supervisor will need to check that the corrected zip code (or community name) supplied by the respondent is
included in the list of eligible zip codes. If the reported residence zip code is not eligible, but the enrollment zip code used to locate
the respondent is eligible, then a call-back may be made to clarify the location of the current residence address. An interview can
still be scheduled pending the final determination. The final residence zip code for the respondent should be noted here.
Nez Perce Tribe
Volume III - Appendix A
December 2016 Page A-8
-------
Respondent ID:
2.0 INTERVIEW INTRODUCTION
l
Basic information about the interview (e.g., location) will be recorded by the interviewer prior to
the in-person interview. The interviewer will then provide a brief introduction to the respondent
about the project. Words to be spoken by the interviewer are identified in bold. Answers are
written, checked, and/or circled, as indicated.
2.1 Administrative Information
General administrative information will be completed by the interviewer at the time of the
interview, but prior to questioning the respondent.
1.
Interviewer Identification
1. Intervi ewer Nam e
2. Interviewer ID:
2. Respondent Identification
3. Respondent ID:
3. Interview Date, Time, and Location
4. Date: / / (mm/dd/yyyy)
5. Day (of the week):
6. Start time: AM / PM (circle)
7. City, State:
8. Location/Venue (check):
Home Central Location
Tribal Office Other (coffee shop, etc.)
Nez Perce Tribe
Volume III - Appendix A
December 2016 Page A-9
-------
Respondent ID:
2.2 Introduction to Interview
To begin the in-person interview, the interviewer will introduce the purpose of the survey and
provide a brief overview of its structure.
"Hello, my name is , and we're conducting a survey on behalf of the . We
appreciate your willingness to participate in our fish consumption survey. The survey is
endorsed by the .
The information you provide as part of this survey will help us understand the rates of fish
consumption, how fish is prepared, and the species or types of fish regularly eaten by
members of the Tribe. Your information, plus the information of other Tribal
members, will help us protect our environment and promote the health of our Tribal
members and families.
We do not intend to collect ANY culturally-sensitive information during this interview. The
information that you provide during this interview is confidential. Your responses to the
questions will be combined with those of others so that your answers cannot be identified.
In the meantime, if you have any questions, here is an information and contact sheet for
you to keep. (Provide Information Sheet)
This interview will take about an hour. The questionnaire has 3 parts. In the first part, I
will ask you to tell me how much fish you ate yesterday. The second part focuses on the
past 12 months: the types of fish you ate, how often you ate it, where you got it, and how it
was prepared, as well as fishing activities and special events. Finally, in the third part, I
will ask you for some general information about yourself.
Your participation in this study is voluntary and you may withdraw at any time without
any consequence to you. If at any time during the interview, you do not know an answer or
do not feel comfortable answering a question, we can skip to the next question. You are free
to not answer any of the questions. May we start the interview now?"
~ INTERVIEWER CHECK THIS BOX IF RESPONDENT AGREES TO
PARTICIPATE IN THE IN-PERSON INTERVIEW.
Nez Perce Tribe
Volume III - Appendix A
December 2016 Page A-10
-------
Respondent ID:
3.024-HOUR DIETARY RECALL
The first part of the in-person interview is a 24-hour dietary recall. Words to be spoken by the
interviewer are identified in bold. Each question will be asked in numeric order. Photographic
and portion model displays will be available for use during questioning.
3.1 Fish Consumption
9. "The first questions are about your fish consumption yesterday. Please consider
what you ate yesterday. I am going to ask you about EACH time you ate. That
would include meals, snacks, eating at home, eating at a friend's or relative's house
or a purchase somewhere. It includes eating fish anywhere or at any time and in any
amount. Did you eat any fish yesterday?"
Yes
No
Don't know / Prefer not to answer
If YES, continue to next Question #9a
If NO or other, skip to next Section (4.0).
9a. "Please think about the first time you ate yesterday Please enter a description
(name, time, or number) for the first occasion where you ate fish yesterday (which
includes finfish, shellfish, and seafood). Consider all meals and snacks, including
fish within dishes such as soups. Include fish bought from a store, from a restaurant,
or caught by you or someone else." (Enter description or occasion number in Table A-
3)
10. "What type of fish did you eat?" (Refer to species display, if needed, enter species type
in Table A-3; see Table A-4 for list of species).
10a. "How much of the (syecies tyye mentioned) did you eat? (See quantity displays
according to species type; enter portion size according to Table A-3a).
10b. "How was the (species type mentioned) prepared or cooked? (Unprompted, check box
in Table A-3).
10c. "Where did the (syecies tyye mentioned) come from? Was it from a market or
store? Was it from a restaurant? Or was it caught by you or someone else (this
includes Tribal distributions)?
lOd. "Was it from Idaho waters or outside of Idaho?" (Check box in Table A-3).
Nez Perce Tribe
Volume III - Appendix A
December 2016 Page A-l 1
-------
Respondent ID:
lOe. "Did you eat this species prepared in any other way or did you eat any other
species of fish for (eating occasion mentioned) ?"
Repeat Question #9a for first/second/third species type or preparation method mentioned
for that eating occasion and complete Table A-3.
Yes
No
If YES, repeat Question #10b above.
If NO, continue to next Question #11.
11. "Please think about the NEXT time you ate yesterday; when was that (name the
eating occasion)? Did you eat fish? (Check)
Yes
No
Did not eat fish rest of day
If YES, repeat Question #9a above for up to 6 eating occasions.
If NO, repeat Question #11 for all eating occasions yesterday.
If "Did not eat fish rest of day," skip ahead to next section, Question #12.
Nez Perce Tribe
Volume III - Appendix A
December 2016 Page A-12
-------
Respondent ID:
Table A-3. 24-Hr Recall: Types, Quantities, Methods, and Sources of Fish Eaten Yesterday
Occasion # &
Description1
Species Type
Portion Size / Quantity
See Displays (enter display #)
Preparation / Cooking Method
Check box
Source
Check box
1
Species 1:
Salmon sections #s
Trout (thin) fillets:
Lamcrev sections:
Jerkv cackaqes:
Soud bowls: cuds
Shellfish (orqanisms):
Q Fried /' Sauteed Q Stew, Soup
Q Baked / Roasted Q Canned, Pickled
Q Broiled / Grilled Q Microwaved
n Poached / Boiled Q Raw / Uncooked
Q Dried, Smoked, Salted Q Other, Unknown
Q Casserole, Mixed Dish
D Market / Store
O Restaurant
Q Caught
Q In Idaho
Q Outside of Idaho
Species 2:
Salmon sections #s
Trout (thin) fillets:
Lamcrev sections:
Jerkv cackaqes:
Soud bowls: cuds
Shellfish (orqanisms):
Q Fried / Sauteed Q Stew, Soup
Q Baked / Roasted Q Canned, Pickled
Q Broiled / Grilled Q Microwaved
Q Poached / Boiled Q Raw / Uncooked
Q Dried, Smoked, Salted Q Other, Unknown
Q Casserole, Mixed Dish
Q Market / Store
O Restaurant
Q Caught
Q In Idaho
Q Outside of Idaho
Species 3:
Salmon sections #s
Trout (thin) fillets:
Lamorev sections:
Jerkv cackaqes:
Soud bowls: cuds
Shellfish (orqanisms):
Q Fried / Sauteed Q Stew, Soup
Q Baked / Roasted Q Canned, Pickled
Q Broiled / Grilled Q Microwaved
Q Poached / Boiled Q Raw / Uncooked
Q Dried, Smoked, Salted Q Other, Unknown
Q Casserole, Mixed Dish
Q Market / Store
O Restaurant
Q Caught
Q In Idaho
Q Outside of Idaho
2
Species 1:
Salmon sections #s
Trout (thin) fillets:
Lamorev sections:
Jerkv oackaqes:
Soud bowls: cuds
Shellfish (orqanisms):
Q Fried / Sauteed Q Stew, Soup
Q Baked / Roasted Q Canned, Pickled
Q Broiled / Grilled Q Microwaved
Q Poached / Boiled Q Raw / Uncooked
Q Dried, Smoked, Salted Q Other, Unknown
Q Casserole, Mixed Dish
Q Market / Store
O Restaurant
Q Caught
Q In Idaho
Q Outside of Idaho
Nez Perce Tribe
Volume III - Appendix A
December 2016 Page A-13
-------
Respondent ID:
Occasion # &
Description1
Species Type
Portion Size / Quantity
See Displays (enter display #)
Preparation / Cooking Method
Check box
Source
Check box
Species 2:
Salmon sections #s
Trout (thin) fillets:
Lamcrev sections:
Jerkv cackaqes:
Soud bowls: cuds
Shellfish (orqanisms):
n Fried / Sauteed Q Stew, Soup
Q Baked / Roasted Q Canned, Pickled
n Broiled / Grilled Q Microwaved
n Poached / Boiled Q Raw / Uncooked
Q Dried, Smoked, Salted Q Other, Unknown
Q Casserole, Mixed Dish
Q Market / Store
O Restaurant
Q Caught
Q In Idaho
Q Outside of Idaho
Species 3:
Salmon sections #s
Trout (thin) fillets:
Lamcrev sections:
Jerkv cackaqes:
Soud bowls: cuds
Shellfish (orqanisms):
Q Fried / Sauteed Q Stew, Soup
Q Baked / Roasted Q Canned, Pickled
n Broiled / Grilled Q Microwaved
Q Poached / Boiled Q Raw / Uncooked
Q Dried, Smoked, Salted Q Other, Unknown
Q Casserole, Mixed Dish
Q Market / Store
Q Restaurant
Q Caught
Q In Idaho
Q Outside of Idaho
3
Species 1:
Salmon sections #s
Trout (thin) fillets:
Lamorev sections:
Jerkv cackaqes:
Soud bowls: cuds
Shellfish (orqanisms):
Q Fried / Sauteed Q Stew, Soup
Q Baked / Roasted Q Canned, Pickled
n Broiled / Grilled Q Microwaved
Q Poached / Boiled Q Raw / Uncooked
Q Dried, Smoked, Salted Q Other, Unknown
Q Casserole, Mixed Dish
Q Market / Store
Q Restaurant
Q Caught
Q In Idaho
Q Outside of Idaho
Species 2:
Salmon sections #s
Trout (thin) fillets:
Lamorev sections:
Jerkv oackaqes:
Soud bowls: cuds
Shellfish (orqanisms):
Q Fried / Sauteed Q Stew, Soup
Q Baked / Roasted Q Canned, Pickled
n Broiled / Grilled Q Microwaved
Q Poached / Boiled Q Raw / Uncooked
Q Dried, Smoked, Salted Q Other, Unknown
Q Casserole, Mixed Dish
Q Market / Store
Q Restaurant
Q Caught
Q In Idaho
Q Outside of Idaho
Nez Perce Tribe
Volume III - Appendix A
December 2016 Page A-14
-------
Respondent ID:
Occasion # &
Description1
Species Type
Portion Size / Quantity
See Displays (enter display #)
Preparation / Cooking Method
Check box
Source
Check box
Species 3:
Salmon sections #s
Trout (thin) fillets:
Lamcrev sections:
Jerkv cackaqes:
Soud bowls: cuds
Shellfish (orqanisms):
n Fried / Sauteed Q Stew, Soup
Q Baked / Roasted Q Canned, Pickled
n Broiled / Grilled Q Microwaved
n Poached / Boiled Q Raw / Uncooked
Q Dried, Smoked, Salted Q Other, Unknown
Q Casserole, Mixed Dish
Q Market / Store
O Restaurant
Q Caught
Q In Idaho
Q Outside of Idaho
4
Species 1:
Salmon sections #s
Trout (thin) fillets:
Lamcrev sections:
Jerkv cackaqes:
Soud bowls: cuds
Shellfish (orqanisms):
Q Fried / Sauteed Q Stew, Soup
Q Baked / Roasted Q Canned, Pickled
n Broiled / Grilled Q Microwaved
Q Poached / Boiled Q Raw / Uncooked
Q Dried, Smoked, Salted Q Other, Unknown
Q Casserole, Mixed Dish
Q Market / Store
O Restaurant
Q Caught
Q In Idaho
Q Outside of Idaho
Species 2:
Salmon sections #s
Trout (thin) fillets:
Lamorev sections:
Jerkv cackaqes:
Soud bowls: cuds
Shellfish (orqanisms):
Q Fried / Sauteed Q Stew, Soup
Q Baked / Roasted Q Canned, Pickled
n Broiled / Grilled Q Microwaved
Q Poached / Boiled Q Raw / Uncooked
Q Dried, Smoked, Salted Q Other, Unknown
Q Casserole, Mixed Dish
Q Market / Store
O Restaurant
Q Caught
Q In Idaho
Q Outside of Idaho
Species 3:
Salmon sections #s
Trout (thin) fillets:
Lamorev sections:
Jerkv oackaqes:
Soud bowls: cuds
Shellfish (orqanisms):
Q Fried / Sauteed Q Stew, Soup
Q Baked / Roasted Q Canned, Pickled
n Broiled / Grilled Q Microwaved
Q Poached / Boiled Q Raw / Uncooked
Q Dried, Smoked, Salted Q Other, Unknown
Q Casserole, Mixed Dish
Q Market / Store
O Restaurant
Q Caught
Q In Idaho
Q Outside of Idaho
Nez Perce Tribe
Volume III - Appendix A
December 2016 Page A-15
-------
Respondent ID:
Occasion # &
Description1
Species Type
Portion Size / Quantity
See Displays (enter display #)
Preparation / Cooking Method
Check box
Source
Check box
5
Species 1:
Salmon sections #s
Trout (thin) fillets:
Lamcrev sections:
Jerkv cackaqes:
Soud bowls: cuds
Shellfish (orqanisms):
n Fried / Sauteed Q Stew, Soup
Q Baked / Roasted Q Canned, Pickled
n Broiled / Grilled Q Microwaved
n Poached / Boiled Q Raw / Uncooked
Q Dried, Smoked, Salted Q Other, Unknown
Q Casserole, Mixed Dish
Q Market / Store
O Restaurant
Q Caught
Q In Idaho
Q Outside of Idaho
Species 2:
Salmon sections #s
Trout (thin) fillets:
Lamcrev sections:
Jerkv cackaqes:
Soud bowls: cuds
Shellfish (orqanisms):
Q Fried / Sauteed Q Stew, Soup
Q Baked / Roasted Q Canned, Pickled
n Broiled / Grilled Q Microwaved
Q Poached / Boiled Q Raw / Uncooked
Q Dried, Smoked, Salted Q Other, Unknown
Q Casserole, Mixed Dish
Q Market / Store
O Restaurant
Q Caught
Q In Idaho
Q Outside of Idaho
Species 3:
Salmon sections #s
Trout (thin) fillets:
Lamorev sections:
Jerkv cackaqes:
Soud bowls: cuds
Shellfish (orqanisms):
Q Fried / Sauteed Q Stew, Soup
Q Baked / Roasted Q Canned, Pickled
n Broiled / Grilled Q Microwaved
Q Poached / Boiled Q Raw / Uncooked
Q Dried, Smoked, Salted Q Other, Unknown
Q Casserole, Mixed Dish
Q Market / Store
O Restaurant
Q Caught
Q In Idaho
Q Outside of Idaho
6
Species 1:
Salmon sections #s
Trout (thin) fillets:
Lamorev sections:
Jerkv oackaqes:
Soud bowls: cuds
Shellfish (orqanisms):
Q Fried / Sauteed Q Stew, Soup
Q Baked / Roasted Q Canned, Pickled
n Broiled / Grilled Q Microwaved
Q Poached / Boiled Q Raw / Uncooked
Q Dried, Smoked, Salted Q Other, Unknown
Q Casserole, Mixed Dish
Q Market / Store
O Restaurant
Q Caught
Q In Idaho
Q Outside of Idaho
Nez Perce Tribe
Volume III - Appendix A
December 2016 Page A-16
-------
Respondent ID:
Occasion # &
Description1
Species Type
Portion Size / Quantity
See Displays (enter display #)
Preparation / Cooking Method
Check box
Source
Check box
Species 2:
Salmon sections #s
Trout (thin) fillets:
Lamcrev sections:
Jerkv cackaqes:
Soud bowls: cuds
Shellfish (orqanisms):
n Fried / Sauteed Q Stew, Soup
Q Baked / Roasted Q Canned, Pickled
n Broiled / Grilled Q Microwaved
n Poached / Boiled Q Raw / Uncooked
Q Dried, Smoked, Salted Q Other, Unknown
Q Casserole, Mixed Dish
Q Market / Store
O Restaurant
Q Caught
Q In Idaho
Q Outside of Idaho
Species 3:
Salmon sections #s
Trout (thin) fillets:
Lamcrev sections:
Jerkv cackaqes:
Soud bowls: cuds
Shellfish (orqanisms):
Q Fried / Sauteed Q Stew, Soup
Q Baked / Roasted Q Canned, Pickled
n Broiled / Grilled Q Microwaved
Q Poached / Boiled Q Raw / Uncooked
Q Dried, Smoked, Salted Q Other, Unknown
Q Casserole, Mixed Dish
Q Market / Store
O Restaurant
Q Caught
Q In Idaho
Q Outside of Idaho
1. "Description" refers to a distinct fish-eating occasion defined by the respondent (breakfast, lunch, dinner, snack, or a time or number).
2. See Table A-4 for species list; will be coded later as anadromous, freshwater resident, or marine fish and shellfish.
Nez Perce Tribe
Volume III - Appendix A
December 2016 Page A-17
-------
Respondent ID:
Table A-3a. Portion Size Model Displays: Description and Use
Display
Type1
Display
Numbers2
Display
Description
What Display
Represents
How Respondents
Report Portion Size
Associated Mass of Real
Fish
Salmon
S1 to S9
Large rubber
salmon fillet, cut
into 24 servings
Cooked salmon and
other fish species
with thick fillets
Identify multiples and/or
fractions for sections 1
to 24 in 0.25 increments
Serving sections range from
1.5 oz. (42 g) to 6.8 oz. (192
g) of uncooked fish
Trout
T1 to T9
Small plastic trout
fillet, single serving
Cooked trout and
other fish species
with thin fillets
Identify multiples and/or
fractions of the fillet in
0.25 increments
One fillet is 3.0 oz. (85 g) of
baked fish, or 4.0 oz. (113 g)
of uncooked fish
Lamprey
L1 to L9
Gray PVC pipe, 2"
diameter, 14" long,
notched every 2"
for 7 servings
Cooked adult
lamprey (eel)
Identify multiples and/or
fractions of the 2"
servings in 0.25
increments
Each 2" serving is calculated
to be 4.0 ounces (113
grams) of uncooked fish
Jerky
J1 to J9
Package of real
"salmon candy"
(dried fish pieces)
Dried pieces of
salmon and other
fish species
Identify multiples and/or
fractions of the package
in 0.25 increments
Packages range from 2.4 oz.
(68 g) to 3.0 oz. (84 g) of
dried fish, or 5.6 oz. (159 g)
to 6.5 oz. (187 g) raw fish
Bowls
B1 to B9
(each is
set of 5)
Empty plastic bowls
(1/4, 1/2, 1, 11/2, and 2
cups) of different
colors
Containers to hold
fish soup, composite
dishes
Identify multiples and/or
fractions of a cup in
0.25 increments
1 cup offish soup is
estimated to include 0.25
cup of cooked fish (2 oz. or
57 g) or 2.5 oz. (72 g) raw
fish
Crayfish
C1 to C9
Color photograph
(laminated) of
whole crayfish
Cooked crayfish
Identify number of
organisms
1 crayfish contains 0.26 oz.
(7.2 g) of uncooked edible
meat
Mussels
M1 to M9
Color photograph
(laminated) of plate
with 6 half-shell
mussels
Cooked mussels
and other bivalve
shellfish
Identify number of
organisms
1 mussel contains 0.4 oz.
(10 g) of uncooked edible
tissue
Shrimp
S1 to S9
Color photograph
(laminated) of plate
with 6 shrimp
Cooked shrimp
Identify number of
organisms
1 shrimp contains 1.6 oz. (44
g) of uncooked edible tissue
Other
N/A
Can or jar offish
(no display
provided)
Fish (tuna, salmon)
in a can or jar
Identify multiples and/or
fractions of cans or jars
in 0.25 increments
Standard tuna can is 5 oz.
(142 g); mason jar is 8 oz
(227 g)
Notes
1. A total of nine identical copies of each model display type will be available for use during interviews (five
for NPT and four for SBT).
2. Display numbers are written in permanent marker on every model display, as well as contact information
for Kristin Callahan, RIDOLFI, 206-436-2774, in the event there are questions or need for replacements.
" = inches
g = grams
oz. = ounces
Nez Perce Tribe
Volume III - Appendix A
December 2016 Page A-18
-------
Respondent ID:
3.2 Other Dietary Information
"Now I will ask you general questions about your diet."
12. "Was the amount of fish you ate yesterday more, less, or about the same as usual?"
(Check)
More than usual
Less than usual
About the same as usual
13. "Are you currently on any kind of diet, either to lose weight or for some other
reason?" (Check)
Yes
No
Prefer not to answer
Nez Perce Tribe
Volume III - Appendix A
December 2016 Page A-19
-------
Respondent ID:
4.0 FOOD FREQUENCY QUESTIONNAIRE
The second part of the in-person interview is a food frequency questionnaire (FFQ) based on the
past year (12 months), and includes questions on dietary patterns and related activities that may
affect fish consumption.
4.1 Fish Consumption
"Thank you for the information about fish you may have eaten yesterday. The next
questions are about your fish consumption (and activities involving fish) over the past
year."
1. Species, Frequency, Quantities
14. "Did you eat fish in the past 12 months? That includes finfish, shellfish, and seafood.
Consider all meals and snacks, including fish within dishes such as soups. Include
fish bought from a store, from a restaurant, or caught by you or someone else. Did
you eat fish in the past 12 months?" (Check)
Yes
No
If YES, continue to Question #15.
If NO, ask "Please consider ANY amount of fish you may have eaten in the past
year." If still NO, terminate interview (skip to Section 5.2, Interview End).
15. "Please tell me which types of fish you ate in the past 12 months (including the fillet
and any parts). For each fish type you say you have eaten, I will ask you how often
you ate it and how much you usually ate. You will be able to respond according to
two periods: when the fish is in-season and the rest of the year. Remember to
consider breakfast, lunch, dinner, and snacks, and include fillets, stews, and other
dishes. Do NOT include special events, such as feasts and ceremonies; I will ask
about that later."
Substitute each species name listed in Table A-4 for each of the questions below, and
complete the table accordingly. Be prepared to show species photographs, if necessary,
and portion size displays. Ask all questions for each species one-by-one, and record
frequency according to "in season" and the rest of the year and record portion sizes
according to Table A-3a.
16. "In the past 12 months, did you eat (Species X) ?"
If YES, check box in Table A-4 and continue to Question #17.
If NO, repeat question for next species on list.
Nez Perce Tribe
Volume III - Appendix A
December 2016 Page A-20
-------
Respondent ID:
17. "Did you eat about the same amount of (Species X) throughout the year or did
you eat more during certain periods and less during other periods of the year?"
If SAME, ask Questions #18-19 and complete Table A-4 for one period; enter length of
period as 12 months. If contradiction occurs (e.g., reports only 3 months), ask "what
about the rest of the year?" (and consider as NOT SAME below).
If NOT SAME, skip to Question #20 and complete Table A-4 for both high and low fish-
eating periods.
18. "In the past 12 months, how often did you eat (Svecies X) in any form (e.g.
cooked or smoked fillets, dried, or soups)?" Enter value and check the units (number of
portions per day, per week, per month, or per year).
19. Please tell me what your typical portion size was when you ate (Species X). You may
only choose ONE type of measurement, either enter the section numbers or one of
the measurements below." Refer to portion displays.
REPEAT Question #16 for each species type listed on Table A-4.
20. "In the past 12 months, how often did you eat (Species X) in any form (e.g.
cooked or smoked fillets, dried, or soups) when it was in season?" Enter value and
check the units (number of portions per day, per week, per month, or per year).
21. Please tell me what your typical portion size was when you ate (Svecies X) when it
was in season. You may only choose ONE type of measurement, either enter the
section numbers or one of the measurements below." Refer to portion displays.
22. "Recognizing that past years may be different, how long was (Svecies X) in
season (total in weeks or months]?" Enter value in weeks or months.
23. "In the past 12 months, how often did you eat (Svecies X) in any form (e.g.
cooked or smoked fillets, dried, or soups) during the rest of the year ? Enter value
and check the units (number of portions per day, per week, per month, or per year).
24. Please tell me what your typical portion size was when you ate (Species X) during the
rest of the year. You may only choose ONE type of measurement, either enter the
section numbers or one of the measurements below" Refer to portion displays.
25. REPEAT Question #16 for each species type listed on Table A-4.
26. "Are there any other fish or shellfish species that you ate in the past 12 months that
we have not mentioned here?"
REPEAT this question and Question #17 (series of questions).
Nez Perce Tribe
Volume III - Appendix A
December 2016 Page A-21
-------
Respondent ID:
Table A-4. FFQ: Types, Frequency, and Quantity of Species Eaten in Past 12 Months
( iinsiiinpiiiiii \\ Ikii I'ish ;iiy lii Si'iisiiir
( iiiisiiiiipiinn Ri'M hI'iIu- War
( Ill-l k
ir
i-ali-ll
Or S;iiik' ('iiiisiiiiipiiiui ^ i-;ir Round
(lilank il'Samc ('iiiiMimpiion ^ i-;ir Round)
I'ish Spi'iiis1
N umber
of
Portions
Portions per day. week,
month, or vear (circle)
Typical Portion
Si/e (& displav
#r
Length of
period
(weeks or
N limber
of
Portions
Portions per day. week,
month, or \ear (circle)
Typical
Portion Si/e
(& display #)'
I .ength of
period
(auto-
months)
calculated)
SALMON AND STEELHEAD
Chinook (King) Salmon
Day
Wk.
Mo
Yr.
Wk. Mo.
Day
Wk.
Mo
Yr.
Wk. Mo.
Coho (Silver) Salmon
Day
Wk.
Mo
Yr.
Wk. Mo.
Day
Wk.
Mo
Yr.
Wk. Mo.
Sockeye (Red) Salmon
Day
Wk.
Mo
Yr.
Wk. Mo.
Day
Wk.
Mo
Yr.
Wk. Mo.
Kokanee (resident form of sockeye)
Day
Wk.
Mo
Yr.
Wk. Mo.
Day
Wk.
Mo
Yr.
Wk. Mo.
Steelhead (migratory form of
rainbow trout)
Day
Wk.
Mo
Yr.
Wk. Mo.
Day
Wk.
Mo
Yr.
Wk. Mo.
Other salmon species (specify, e.g.,
Chum, Pink, Atlantic salmon)
Day
Wk.
Mo
Yr.
Wk. Mo.
Day
Wk.
Mo
Yr.
Wk. Mo.
All salmon and steelhead / species
not identified
Day
Wk.
Mo
Yr.
Wk. Mo.
Day
Wk.
Mo
Yr.
Wk. Mo.
RESIDENT TROUT
Rainbow Trout
Day
Wk.
Mo
Yr.
Wk. Mo.
Day
Wk.
Mo.
Yr.
Wk. Mo.
Cutthroat Trout
Day
Wk.
Mo
Yr.
Wk. Mo.
Day
Wk.
Mo.
Yr.
Wk. Mo.
Cutbow Trout (hybrid of Rainbow
and Cutthroat Trout)
Day
Wk.
Mo
Yr.
Wk. Mo.
Day
Wk.
Mo.
Yr.
Wk. Mo.
Bull Trout (Dolly Varden)
Day
Wk.
Mo
Yr.
Wk. Mo.
Day
Wk.
Mo.
Yr.
Wk. Mo.
Brook Trout
Day
Wk.
Mo
Yr.
Wk. Mo.
Day
Wk.
Mo.
Yr.
Wk. Mo.
Take Trout
Day
Wk.
Mo
Yr.
Wk. Mo.
Day
Wk.
Mo.
Yr.
Wk. Mo.
Brown Trout
Day
Wk.
Mo
Yr.
Wk. Mo.
Day
Wk.
Mo.
Yr.
Wk. Mo.
Other trout species (specify)
Day
Wk.
Mo
Yr.
Wk. Mo.
Day
Wk.
Mo.
Yr.
Wk. Mo.
All resident trout / species not
identified
Day
Wk.
Mo
Yr.
Wk. Mo.
Day
Wk.
Mo.
Yr.
Wk. Mo.
Nez Perce Tribe
Volume III - Appendix A
December 2016 Page A-22
-------
Respondent ID:
( iiiisiimpiiiui \\ Ik-ii I'isli arc In Si-asmr
('iiiiMimplion RiM hI'iIu- War
( Ill-l k
ir
I'llll'll
Or S;iiik' C'linsuinpiiiin ^ i-ar Round
(lilank il'Saim- C onsumpiiim ^ i-ar Round)
I'isli S|K-iii-s'
N limber
ol"
Portions per day. week,
month, or year (circle)
Typical Portion
Si/e (& display
Length of
period
(weeks or
months)
N limber
ol"
Portions per day. week,
month, or year (circle)
1 ypical
Portion Si/e
Length of
period
(auto-
calculated)
Portions
#y
Portions
(& display # j
OTHER FRESHWATER FISH AND SHELLFISH
Sturgeon
Day
Wk.
Mo
Yr.
Wk. Mo.
Day
Wk.
Mo.
Yr.
Wk. Mo.
Lamprey
Day
Wk.
Mo
Yr.
Wk. Mo.
Day
Wk.
Mo.
Yr.
Wk. Mo.
Whitefish
Day
Wk.
Mo
Yr.
Wk. Mo.
Day
Wk.
Mo.
Yr.
Wk. Mo.
Sucker
Day
Wk.
Mo
Yr.
Wk. Mo.
Day
Wk.
Mo.
Yr.
Wk. Mo.
Burbot
Day
Wk.
Mo
Yr.
Wk. Mo.
Day
Wk.
Mo.
Yr.
Wk. Mo.
Northern Pikeminnow (Squawfish)
Day
Wk.
Mo
Yr.
Wk. Mo.
Day
Wk.
Mo.
Yr.
Wk. Mo.
Bass
Day
Wk.
Mo
Yr.
Wk. Mo.
Day
Wk.
Mo.
Yr.
Wk. Mo.
Bluegill
Day
Wk.
Mo
Yr.
Wk. Mo.
Day
Wk.
Mo.
Yr.
Wk. Mo.
Carp
Day
Wk.
Mo
Yr.
Wk. Mo.
Day
Wk.
Mo.
Yr.
Wk. Mo.
Catfish
Day
Wk.
Mo
Yr.
Wk. Mo.
Day
Wk.
Mo.
Yr.
Wk. Mo.
Crappie
Day
Wk.
Mo
Yr.
Wk. Mo.
Day
Wk.
Mo.
Yr.
Wk. Mo.
Sunfish
Day
Wk.
Mo
Yr.
Wk. Mo.
Day
Wk.
Mo.
Yr.
Wk. Mo.
Tilapia
Day
Wk.
Mo
Yr.
Wk. Mo.
Day
Wk.
Mo.
Yr.
Wk. Mo.
Walleye
Day
Wk.
Mo
Yr.
Wk. Mo.
Day
Wk.
Mo.
Yr.
Wk. Mo.
Yellow Perch
Day
Wk.
Mo
Yr.
Wk. Mo.
Day
Wk.
Mo.
Yr.
Wk. Mo.
Other freshwater finfish (specify)
Day
Wk.
Mo
Yr.
Wk. Mo.
Day
Wk.
Mo.
Yr.
Wk. Mo.
Crayfish
Day
Wk.
Mo
Yr.
Wk. Mo.
Day
Wk.
Mo.
Yr.
Wk. Mo.
Freshwater Clams or Mussels
Day
Wk.
Mo
Yr.
Wk. Mo.
Day
Wk.
Mo.
Yr.
Wk. Mo.
Unspecified freshwater fish
Day
Wk.
Mo
Yr.
Wk. Mo.
Day
Wk.
Mo.
Yr.
Wk. Mo.
SEAFOOD / MARINE FISH AND SHELLFISH
Nez Perce Tribe
Volume III - Appendix A
December 2016 Page A-23
-------
Respondent ID:
( iiiisiimpiiiui \\ Ikii l-'isli ;iiv In Si-sisnir
('iiiiMimplion RiM hI'iIu- W;ir
( Ill-l k
it
I'llll'll
Or S;iiik- C'linsuinpiiiin ^ i-;ir Round
(lihink irS;iim- ('oiiMimpiiiiii ^ i-iir Round)
l-'isli Spi-rii's1
N umber
ol"
Portions per day. week,
month, or year (circle)
Typical Portion
Si/e (& display
Length of
period
(weeks or
months)
N limber
ol"
Portions per day. week,
month, or y ear (circle)
Typical
Portion Si/e
Length of
period
(auto-
calculated)
Portions
#y
Portions
(& display # j
Cod
Day
Wk.
Mo
Yr.
Wk. Mo.
Day
Wk.
Mo.
Yr.
Wk. Mo.
Halibut
Day
Wk.
Mo
Yr.
Wk. Mo.
Day
Wk.
Mo.
Yr.
Wk. Mo.
Pollock
Day
Wk.
Mo
Yr.
Wk. Mo.
Day
Wk.
Mo.
Yr.
Wk. Mo.
Tuna
Day
Wk.
Mo
Yr.
Wk. Mo.
Day
Wk.
Mo.
Yr.
Wk. Mo.
Lobster
Day
Wk.
Mo
Yr.
Wk. Mo.
Day
Wk.
Mo.
Yr.
Wk. Mo.
Crab
Day
Wk.
Mo
Yr.
Wk. Mo.
Day
Wk.
Mo.
Yr.
Wk. Mo.
Marine Clams or Mussels
Day
Wk.
Mo
Yr.
Wk. Mo.
Day
Wk.
Mo.
Yr.
Wk. Mo.
Shrimp
Day
Wk.
Mo
Yr.
Wk. Mo.
Day
Wk.
Mo.
Yr.
Wk. Mo.
Other marine fish or shellfish
(Specify)
Day
Wk.
Mo
Yr.
Wk. Mo.
Day
Wk.
Mo.
Yr.
Wk. Mo.
Other marine fish or shellfish
(Specify)
Day
Wk.
Mo
Yr.
Wk. Mo.
Day
Wk.
Mo.
Yr.
Wk. Mo.
Other marine fish or shellfish
(Specify)
Day
Wk.
Mo
Yr.
Wk. Mo.
Day
Wk.
Mo.
Yr.
Wk. Mo.
UNSPECIFIED FISH OR
SHELLFISH SPECIES
Day
Wk.
Mo
Yr.
Wk. Mo.
Day
Wk.
Mo.
Yr.
Wk. Mo.
Notes
1. Species are listed and grouped according to the most commonly eaten types of fish and shellfish.
2. Fish consumption "in season" is based on respondents perception or experience related to harvest and assumed higher consumption (compared to
the rest of the year); biological seasons (e.g., fish runs) will be evaluated during data analysis and do not have to correspond to the duration of
seasons noted by the respondent.
3. See 24-hour dietary recall (Table A-3) for examples of portion size data to enter according to species type (e.g., salmon, trout, lamprey, shellfish) or
preparation method (jerky, bowls of soup). A description of the portion displays is provided in Table A-3a above.
Nez Perce Tribe
Volume III - Appendix A
December 2016 Page A-24
-------
Respondent ID:
2. Parts of Fish Consumed, Preparation Methods, and Sources
The next questions are about the parts of fish you eat, methods of preparation, and sources
(where acquired) according to species groups. Those groups are 1) salmon and steelhead, 2)
trout species, 3) sturgeon, and 4) suckers and whitefish." Complete Table A-5 for the
following questions.
27. "When you eat a fish fillet, what percent of the time do you eat the following species
of fish with skin?"
ASK question for 1) salmon and steelhead, 2) trout, 3) sturgeon, and 4) suckers and
whitefish. Record answers in percent (including zero) or leave blank if that species type
is not consumed at all. Complete Table A-5.
28. "When you eat (syecies group), what percent of the time do you eat the eggs and
what percent of the time do you eat other organs (including head and bones)?"
ASK question for 1) salmon and steelhead, 2) trout, 3) sturgeon, and 4) suckers and
whitefish. Record answers in percent (including zero) or select "Not Applicable" if that
species type is not consumed at all. Complete Table A-5.
29. "Thinking about how the fish that you eat is prepared, what percent of the time
that you eat (syecies group) is it: baked or broiled? smoked? dried? in a soup? or
other method (specify)? Your answers should total 100%."
ASK question for 1) salmon and steelhead, 2) trout, 3) sturgeon, and 4) suckers and
whitefish. Complete Table A-5.
30. "Thinking about where the fish comes from that you eat, what percent of the time
do you get (syecies tyye) from the following sources? Your answers should total
100%."
• Bought from a store (grocery or market)?
• From a restaurant?
• Caught by you or someone else in Idaho waters, including Tribal
distributions?
• Caught by you or someone else outside of Idaho waters, including Tribal
distributions?
ASK question for 1) salmon and steelhead, 2) trout, 3) sturgeon, and 4) suckers and
whitefish. Complete Table A-5.
Nez Perce Tribe
Volume III - Appendix A
December 2016 Page A-25
-------
Respondent ID:
Table A-5. FFQ: Fish Parts Eaten, Preparation Methods, and Sources
Species Croup:
Salmon and
Steel head
Trout
Sturgeon
Suckers and
W hitcrish
Percent ol' l inio Typically Kat:
Skin
Eggs
Head, bone, and/or
organs
Percent ol' l inie Typically Prepare (total 100%):
Baked or broiled
Smoked
Dried
In a soup
Other:
Don't know
Percent ol' Time Typically Obtained (total 100%):
Bought from a store
(grocery or market)
From a restaurant
Caught by you or
someone else (in
Idaho waters)
Caught by you or
someone else (outside
of Idaho)
Other:
Don't know
Nez Perce Tribe
Volume III - Appendix A
December 2016 Page A-26
-------
Respondent ID:
4.2 Special Events and Gatherings
"I will now ask questions related to your fish consumption during special events and
gatherings, including ceremonies or other community events." Complete Table A-6 for the
following questions.
31. "In the past 12 months, how many special events and gatherings did you attend
(either per week, month or year)?" (Enter number and circle one unit)
Events per Week / Month / Year
If zero, skip to next section (4.3), Question #35.
32. "Did you eat fish in any form (e.g. cooked or smoked fillets, dried, or soups) at these
special events and gatherings, such as 1) salmon and steelhead, 2) trout, 3) sturgeon,
4) suckers or whitefish?" (Circle answer in Table A-6)
Yes
No
Don't know / Prefer not to answer
If YES continue to next question
If NO or other, skip to next section (4.3), Question #35.
33. "What was your typical portion size for the following species at the special events
and gatherings? You may only choose ONE type of measurement, either enter the
section numbers or one of the measurements below."
ASK question for 1) salmon and steelhead, 2) trout, 3) sturgeon, and 4) suckers and
whitefish. Complete Table A-6. (See portion models.)
34. "At what percent of the special events and gatherings did you eat (species group) ?"
ASK question for 1) salmon and steelhead, 2) trout, 3) sturgeon, and 4) suckers and
whitefish. Complete Table A-6.
Nez Perce Tribe
Volume III - Appendix A
December 2016 Page A-27
-------
Respondent ID:
Table A-6. FFQ: Fish Consumption at Gatherings
Species (.roup
C onsumed (circle)
Typical Port ion Size
(enter sections, fillets,
packages, cups see Table
. [-4a for model list)
Percent of time eat
fish at gatherings
Salmon and
Steelhead
YES NO
%
Trout
YES NO
%
Sturgeon
YES NO
%
Suckers and
Whitefish
YES NO
%
4.3 Fishing Activities
"I am now going to ask you some questions about fishing."
35. "Over the past 12 months, did you take part in any fishing-related activities?"
(Check)
Yes
No
Prefer not to answer
If YES, continue to next question.
35a. If NO, ask "Why not"? (Check and skip to next section)
If prefer not to answer, skip to next section.
Fish advisories
Pollution
Other environmental concerns
Not enough fish available to catch
Limited access to fishing areas
Used to access to boat/fishing gear, not anymore
Too far from fishing areas
Too busy, no time
Nez Perce Tribe
Volume III - Appendix A
December 2016 Page A-28
-------
Respondent ID:
No longer custom, prefer other activities
Prefer other foods
Don't know how to fish
Prefer not to answer
Other
36. "Now I'm going to ask you the approximate number of times you went fishing (for
fish and shellfish) each month. How many times did you go fishing during each of
the following months?" (List and enter value for each)
Times
in
January
Times
in
February
Times
in
March
Times
in
April
Times
in
May
Times
in
June
Times
in
July
Times
in
August
Times
in
September
Times
in
October
Times
in
November
Times
in
December
37. "What percent of the fish that you harvest do you keep for you and your household,
what percent do you give/distribute to others outside your household, and what
percent do you sell (your answers should total 100%)?" (Enter)
Percent Keep
Percent Give to others
Percent Sell
100% Total
Nez Perce Tribe
Volume III - Appendix A
December 2016 Page A-29
-------
Respondent ID:
38. "Do you own or have access to fishing gear?" (Check)
Yes
No
Prefer not to answer
39. "Do you own or have access to a boat?" (Check)
Yes
No
Prefer not to answer
4.4 Changes in Fish Consumption
"I am now going to ask you questions about changes in fish consumption and availability.
Some of these may be open-ended questions. We do not intend to collect ANY culturally-
sensitive information."
40. "Has there been a change over time in your fish consumption?" (Check)
Yes
No
Don't know / Prefer not to answer
If YES, continue to next question.
If NO or other, skip to Question #41.
40a. "How has it changed most recently?" (Check)
Increased consumption
Decreased consumption
Other change (e.g., available species)
Nez Perce Tribe
Volume III - Appendix A
December 2016 Page A-30
-------
Respondent ID:
40b. "When did it change?"
Within past 5 years
In the 2000s (or 5 to 15 years ago)
In the 1990s (or 15 to 25 years ago)
In the 1980s (or 25 to 35 years ago)
In the 1970s (or 35-45 years ago)
In the 1960s or earlier (more than 45 years ago)
40c. "Why did it change?" (Multiple choice options may be developed in Pilot Test)
41. "In the past, how important was fish to your Tribe's heritage and culture?"
Very important
Somewhat important
Not important
Don't know / Prefer not to answer
41a. "Currently, how important is fish to your Tribe's heritage and culture?"
Very important
Somewhat important
Not important
Don't know / Prefer not to answer /
Nez Perce Tribe
Volume III - Appendix A
December 2016 Page A-31
-------
Respondent ID:
42. "Has there been a change in access to fish and fishing (for you or others) over
time?" (Check)
Yes
No
Don't know / Prefer not to answer /
If YES, continue to next question.
If NO or other, skip to Question #43.
42a. "How has it changed?" (Check)
More access to fishing
Less access to fishing
Other change
42b. "When did it change?"
Within past 5 years
In the 2000s (or 5 to 15 years ago)
In the 1990s (or 15 to 25 years ago)
In the 1980s (or 25 to 35 years ago)
In the 1970s (or 35-45 years ago)
In the 1960s or earlier (more than 45 years ago)
42c. "Why did it change?" (Multiple choice options may be developed in Pilot Test)
Nez Perce Tribe
Volume III - Appendix A
December 2016 Page A-32
-------
Respondent ID:
43. "Has there been a change in how often you fish (for you or others)?" (Check)
Yes
No
Don't know / Prefer not to answer
If YES, continue to next question.
If NO or other, skip to Question #44.
43a. "How has it changed most recently?" (Check)
Increased frequency
Decreased frequency
Other change
43b. "When did it change?"
Within past 5 years
In the 2000s (or 5 to 15 years ago)
In the 1990s (or 15 to 25 years ago)
In the 1980s (or 25 to 35 years ago)
In the 1970s (or 35-45 years ago)
In the 1960s or earlier (more than 45 years ago)
43c. "Why did it change?" (Multiple choice options may be developed in Pilot Test)
Nez Perce Tribe
Volume III - Appendix A
December 2016 Page A-33
-------
Respondent ID:
44. "Has there been a change in the way you prepare or use fish?" (Check)
Yes
No
Don't know / Prefer not to answer /
If YES, continue to next question.
If NO or other, skip to Question #45.
44a. "How has it changed most recently?"
Different cooking method
Different use
Don't know / Prefer not to answer /
44b. "When did it change?"
Within past 5 years
In the 2000s (or 5 to 15 years ago)
In the 1990s (or 15 to 25 years ago)
In the 1980s (or 25 to 35 years ago)
In the 1970s (or 35-45 years ago)
In the 1960s or earlier (more than 45 years ago)
44c. "Why did it change?" (Multiple choice options may be developed in Pilot Test)
Nez Perce Tribe
Volume III - Appendix A
December 2016 Page A-34
-------
Respondent ID:
45. "Compared to your fish consumption now, how much/how frequently would you
like to consume fish in the future?" (Check)
Increase consumption
Decrease consumption
Maintain same consumption
Don't know / Prefer not to answer
If INCREASED, continue to next question.
If DECREASED or other, skip to next section.
46. "If you prefer to eat more fish or seafood than you're currently eating, what would
have to occur for you to eat that amount in the future?"
Nez Perce Tribe
Volume III - Appendix A
December 2016 Page A-35
-------
Respondent ID:
5.0 GENERAL INFORMATION
The third and final part of the in-person interview involves collecting general information from
the respondent and recording final administrative data.
5.1 Respondent Information
Respondents will be asked demographic questions as well as (for female respondents) questions
related to breastfeeding history.
1. Demographic Information
"This is the final part of the interview. I have a few general questions and then we will be
done. These include reporting your height and weight, which will help us to calculate and
check fish consumption rates, and reporting education and income ranges, which will help
us determine fish consumption rates for various population groups." (Check or enter - if
respondent prefers not to say, enter 999)
47. Gender (check):
Male
Female
48. "What is your age?" (years)
49. "What is your height?" feet inches
50. "How much do you weigh?" pounds
51. "How many people live in your household, including yourself?"
52. "Do you live on your Tribe's Reservation?" (Check)
Yes
No
Prefer not to answer
53. "What is the highest level of education that you've completed?" (Check)
Elementary School
Middle School
High School / GED
Associates Degree
Nez Perce Tribe
Volume III - Appendix A
December 2016 Page A-36
-------
Respondent ID:
Bachelor's Degree
Master's Degree
Doctorate
Prefer not to answer
54. "What is your approximate household income per year?" (List all options below,
except "prefer not to say" and check)
$15,000 or less
More than $15,000 up to $25,000
More than $25,000 up to $35,000
More than $35,000 up to $45,000
More than $45,000 up to $55,000
More than $55,000 up to $65,000
More than $65,000
Prefer not to answer
2. Breastfeeding History
The following questions are for female respondents only; if male, skip to next section.
55. "Have you ever given birth? (Check)
Yes
No
Prefer not to answer
If YES, continue to next question.
Otherwise, skip to next section.
56. "When did you most recently give birth? _/ (MM, YYYY)
Nez Perce Tribe
Volume III - Appendix A
December 2016 Page A-37
-------
Respondent ID:
57. "Was this baby ever breastfed or fed breast milk? (Check)
Yes
No
Prefer not to answer
If YES, continue to next question.
Otherwise, skip to next section.
58. "If the youngest child is no longer breastfeeding, at what age did you stop feeding
breast milk to this child?" (Provide in months or check other option)
Stopped at (months old)
Still breastfeeding
Prefer not to answer
Not applicable (not biological mother, etc.)
5.2 Interview End
Upon completing the interview, the interviewer will offer appreciation and complete the
remaining administrative information, including signing a form verifying participation.
"This concludes the interview. If any of your answers included culturally-sensitive
information, please tell me.
Yes, included culturally sensitive information
No culturally sensitive information included
Don't know / Prefer not to answer
If YES, this questionnaire will be reviewed by a Tribal official and culturally sensitive
information may be edited or redacted prior to further analysis and review.
Thank you SO very much for your time and cooperation today. Your participation will
contribute significantly to the overall success of this survey and help protect the health of
our Tribe. It would also benefit the survey if you could participate in a second, follow-up
interview over the phone in the next one to four weeks. This second interview will be much
shorter and should only take about 15 minutes."
Nez Perce Tribe
Volume III - Appendix A
December 2016 Page A-38
-------
Respondent ID:
59. "Is it okay if I contact you again for a follow-up call?"
Yes
No
59a. If YES, "what is the best phone number to reach you?'
59b. If YES, "Thank you. I am going to leave photographs of the portion display models
with you so that you will have them for reference when I call." Leave actual-size
photographs of models with the respondent.
59c. If NO, remind respondent of the importance of this study and ask again.
60. "Thank you again for your time today, that is all." Complete information below.
Record interview end time and calculate interview length.
61. End time: AM / PM (circle)
62. Length of interview: (hours and/or minutes)
63. Was the interview conducted in private or were others present? (Check)
In private
Others were present
Nez Perce Tribe
Volume III - Appendix A
December 2016 Page A-39
-------
Respondent ID:
5.3 Post-Interview
Following the interview, the interviewer will assess and record the respondent's level of
participation and the interviewer will acknowledge that he/she recorded the information
truthfully and to the best of his/her ability by signing the following guarantee of authenticity.
1. Interview Quality
64. Respondents cooperation: (Check)
Very good
Good
Fair
Poor
65. Respondent's reliability: (Check)
Highly reliable
Generally reliable
Questionable
Unreliable
Notes / Reasons for opinions:
66. Note any topics or specific questions that appeared confusing or particularly challenging
for the respondent to answer.
Nez Perce Tribe
Volume III - Appendix A
December 2016 Page A-40
-------
Respondent ID:
2. Interviewer Guarantee of Authenticity
67.1, (printed name of interviewer) hereby affirm
that the answers recorded on this questionnaire reflect a complete and accurate
accounting of my interview with the respondent.
Signature of Interviewer
Date
Nez Perce Tribe
Volume III - Appendix A
December 2016 Page A-41
-------
Respondent ID:
6.0SECOND 24-HOUR DIETARY RECALL
Based on the results of the first interview, which includes a 24-hour dietary recall, food
frequency questionnaire, and general demographic information, a subset of individuals will be
selected as "high" fish consumers for participation in a second 24-hour dietary recall by
telephone. Words to be spoken by the interviewer are identified in bold. Questions will be asked
in numeric order.
6.1 Administrative Information
Since this telephone interview will be conducted at a later date, general administrative
information will be completed similar to the first interview (prior to questioning the respondent).
1. Interviewer Identification
1. Interviewer Name
2. Interviewer ID:
2. Respondent Identification
3. Respondent ID:
4. Phone number:
3. Interview Date, Time, and Location
5. Date: / / (MM/DD/YYYY)
6. Day (of the week):
7. Start time: AM / PM (circle)
8. City, State:
Nez Perce Tribe
Volume III - Appendix A
December 2016 Page A-42
-------
Respondent ID:
6.2 Introduction
"Hello, my name is , and I am calling on behalf of the Tribe. We appreciate
your continued willingness to participate in our fish consumption survey.
The information you provide during this follow-up interview, as well as your previous
answers, plus the information of other Tribal members, will help us understand the rates of
fish consumption, how fish is prepared, and the species or types of fish regularly eaten by
members of the Tribe.
The information that you provide during this interview is confidential. Your responses to
the questions will be combined with those of others so that your answers cannot be
identified. If you have any questions, please refer to the information sheet I gave you
previously.
This follow-up survey is much shorter and should only take about 15 minutes. I will ask
you to tell me how much fish you ate in the last 24 hours. Please refer to the photographs I
left with you previously. If you do not know an answer or do not feel comfortable
answering, we can skip that question. You are free to not answer any of the questions. May
we start the interview now?"
~ INTERVIEWER CHECK THIS BOX IF RESPONDENT AGREES TO
PARTICIPA TE IN THE FOLLOW- UP TELEPHONE INTER VIEW.
6.3 Fish Consumption
9. "The first questions are about your fish consumption yesterday. Please consider
what you ate yesterday. I am going to ask you about EACH time you ate. That
would include meals, snacks, eating at home, eating at a friend's or relative's house
or a purchase somewhere. It includes eating fish anywhere or at any time and in any
amount. Did you eat any fish yesterday?"
Yes
No
Don't know / Prefer not to answer
If YES, continue to next Question #9a
If NO or Other, skip to next Section (6.5), Question #14.
Nez Perce Tribe
Volume III - Appendix A
December 2016 Page A-43
-------
Respondent ID:
9a. "Please think about the first time you ate yesterday Please enter a description
(name, time, or number) for the first occasion where you ate fish yesterday (which
includes finfish, shellfish, and seafood). Consider all meals and snacks, including
fish within dishes such as soups. Include fish bought from a store, from a restaurant,
or caught by you or someone else." (Enter description or occasion number in Table A-
7)
10. "What type of fish did you eat?" (Refer to species display, if needed, enter species type
in Table A-7; see Table A-4 above for list of species).
10a. "How much of the (species type mentioned) did you eat? (See quantity displays
according to species type; enter portion size according to Table A-7a).
10b. "How was the (species type mentioned) prepared or cooked? (Unprompted, check box
in Table A-7).
10c. "Where did the (species type mentioned) come from? Was it from a market or
store? Was it from a restaurant? Or was it caught by you or someone else (this
includes Tribal distributions)?
lOd. "Was it from Idaho waters or outside of Idaho?" (Check box in Table A-7).
lOe. "Did you eat this species prepared in any other way or did you eat any other
species of fish for (eating occasion mentioned) ?"
11. "Please think about the NEXT time you ate yesterday; when was that (name the
eating occasion)? Did you eat fish? (Check)
Yes
No
Did not eat fish rest of day
If YES, repeat Question #10 above for up to 6 eating occasions.
If NO, repeat Question #11 for all eating occasions yesterday.
If "Did not eat fish rest of day," skip ahead to next section, Question #12
Nez Perce Tribe
Volume III - Appendix A
December 2016 Page A-44
-------
Respondent ID:
Table A-7. 24-Hr Recall: Types, Quantities, Methods, and Sources of Fish Eaten Yesterday
Occasion # &
Description1
Species Type
Portion Size / Quantity
See Displays (enter display #)
Preparation / Cooking Method
Check box
Source
Check box
1
Species 1:
Salmon sections #s
Trout (thin) fillets:
Lamcrev sections:
Jerkv cackaqes:
Soud bowls: cuds
Shellfish (orqanisms):
Q Fried /' Sauteed Q Stew, Soup
Q Baked / Roasted Q Canned, Pickled
Q Broiled / Grilled Q Microwaved
n Poached / Boiled Q Raw / Uncooked
Q Dried, Smoked, Salted Q Other, Unknown
Q Casserole, Mixed Dish
D Market / Store
O Restaurant
Q Caught
Q In Idaho
Q Outside of Idaho
Species 2:
Salmon sections #s
Trout (thin) fillets:
Lamcrev sections:
Jerkv cackaqes:
Soud bowls: cuds
Shellfish (orqanisms):
Q Fried / Sauteed Q Stew, Soup
Q Baked / Roasted Q Canned, Pickled
Q Broiled / Grilled Q Microwaved
Q Poached / Boiled Q Raw / Uncooked
Q Dried, Smoked, Salted Q Other, Unknown
Q Casserole, Mixed Dish
Q Market / Store
O Restaurant
Q Caught
Q In Idaho
Q Outside of Idaho
Species 3:
Salmon sections #s
Trout (thin) fillets:
Lamorev sections:
Jerkv cackaqes:
Soud bowls: cuds
Shellfish (orqanisms):
Q Fried / Sauteed Q Stew, Soup
Q Baked / Roasted Q Canned, Pickled
Q Broiled / Grilled Q Microwaved
Q Poached / Boiled Q Raw / Uncooked
Q Dried, Smoked, Salted Q Other, Unknown
Q Casserole, Mixed Dish
Q Market / Store
O Restaurant
Q Caught
Q In Idaho
Q Outside of Idaho
2
Species 1:
Salmon sections #s
Trout (thin) fillets:
Lamorev sections:
Jerkv oackaqes:
Soud bowls: cuds
Shellfish (orqanisms):
Q Fried / Sauteed Q Stew, Soup
Q Baked / Roasted Q Canned, Pickled
Q Broiled / Grilled Q Microwaved
Q Poached / Boiled Q Raw / Uncooked
Q Dried, Smoked, Salted Q Other, Unknown
Q Casserole, Mixed Dish
Q Market / Store
O Restaurant
Q Caught
Q In Idaho
Q Outside of Idaho
Nez Perce Tribe
Volume III - Appendix A
December 2016 Page A-45
-------
Respondent ID:
Occasion # &
Description1
Species Type
Portion Size / Quantity
See Displays (enter display #)
Preparation / Cooking Method
Check box
Source
Check box
Species 2:
Salmon sections #s
Trout (thin) fillets:
Lamcrev sections:
Jerkv cackaqes:
Soud bowls: cuds
Shellfish (orqanisms):
Q Fried / Sauteed Q Stew, Soup
Q Baked / Roasted Q Canned, Pickled
n Broiled / Grilled Q Microwaved
n Poached / Boiled Q Raw / Uncooked
Q Dried, Smoked, Salted Q Other, Unknown
Q Casserole, Mixed Dish
Q Market / Store
O Restaurant
Q Caught
Q In Idaho
Q Outside of Idaho
Species 3:
Salmon sections #s
Trout (thin) fillets:
Lamcrev sections:
Jerkv cackaqes:
Soud bowls: cuds
Shellfish (orqanisms):
Q Fried / Sauteed Q Stew, Soup
Q Baked / Roasted Q Canned, Pickled
n Broiled / Grilled Q Microwaved
Q Poached / Boiled Q Raw / Uncooked
Q Dried, Smoked, Salted Q Other, Unknown
Q Casserole, Mixed Dish
Q Market / Store
O Restaurant
Q Caught
Q In Idaho
Q Outside of Idaho
3
Species 1:
Salmon sections #s
Trout (thin) fillets:
Lamorev sections:
Jerkv cackaqes:
Soud bowls: cuds
Shellfish (orqanisms):
Q Fried / Sauteed Q Stew, Soup
Q Baked / Roasted Q Canned, Pickled
n Broiled / Grilled Q Microwaved
Q Poached / Boiled Q Raw / Uncooked
Q Dried, Smoked, Salted Q Other, Unknown
Q Casserole, Mixed Dish
Q Market / Store
O Restaurant
Q Caught
Q In Idaho
Q Outside of Idaho
Species 2:
Salmon sections #s
Trout (thin) fillets:
Lamorev sections:
Jerkv oackaqes:
Soud bowls: cuds
Shellfish (orqanisms):
Q Fried / Sauteed Q Stew, Soup
Q Baked / Roasted Q Canned, Pickled
n Broiled / Grilled Q Microwaved
Q Poached / Boiled Q Raw / Uncooked
Q Dried, Smoked, Salted Q Other, Unknown
Q Casserole, Mixed Dish
Q Market / Store
O Restaurant
Q Caught
Q In Idaho
Q Outside of Idaho
Nez Perce Tribe
Volume III - Appendix A
December 2016 Page A-46
-------
Respondent ID:
Occasion # &
Description1
Species Type
Portion Size / Quantity
See Displays (enter display #)
Preparation / Cooking Method
Check box
Source
Check box
Species 3:
Salmon sections #s
Trout (thin) fillets:
Lamcrev sections:
Jerkv cackaqes:
Soud bowls: cuds
Shellfish (orqanisms):
Q Fried / Sauteed Q Stew, Soup
Q Baked / Roasted Q Canned, Pickled
n Broiled / Grilled Q Microwaved
n Poached / Boiled Q Raw / Uncooked
Q Dried, Smoked, Salted Q Other, Unknown
Q Casserole, Mixed Dish
Q Market / Store
O Restaurant
Q Caught
Q In Idaho
Q Outside of Idaho
4
Species 1:
Salmon sections #s
Trout (thin) fillets:
Lamcrev sections:
Jerkv cackaqes:
Soud bowls: cuds
Shellfish (orqanisms):
Q Fried / Sauteed Q Stew, Soup
Q Baked / Roasted Q Canned, Pickled
n Broiled / Grilled Q Microwaved
Q Poached / Boiled Q Raw / Uncooked
Q Dried, Smoked, Salted Q Other, Unknown
Q Casserole, Mixed Dish
Q Market / Store
O Restaurant
Q Caught
Q In Idaho
Q Outside of Idaho
Species 2:
Salmon sections #s
Trout (thin) fillets:
Lamorev sections:
Jerkv cackaqes:
Soud bowls: cuds
Shellfish (orqanisms):
Q Fried / Sauteed Q Stew, Soup
Q Baked / Roasted Q Canned, Pickled
n Broiled / Grilled Q Microwaved
Q Poached / Boiled Q Raw / Uncooked
Q Dried, Smoked, Salted Q Other, Unknown
Q Casserole, Mixed Dish
Q Market / Store
O Restaurant
Q Caught
Q In Idaho
Q Outside of Idaho
Species 3:
Salmon sections #s
Trout (thin) fillets:
Lamorev sections:
Jerkv oackaqes:
Soud bowls: cuds
Shellfish (orqanisms):
Q Fried / Sauteed Q Stew, Soup
Q Baked / Roasted Q Canned, Pickled
n Broiled / Grilled Q Microwaved
Q Poached / Boiled Q Raw / Uncooked
Q Dried, Smoked, Salted Q Other, Unknown
Q Casserole, Mixed Dish
Q Market / Store
O Restaurant
Q Caught
Q In Idaho
Q Outside of Idaho
Nez Perce Tribe
Volume III - Appendix A
December 2016 Page A-47
-------
Respondent ID:
Occasion # &
Description1
Species Type
Portion Size / Quantity
See Displays (enter display #)
Preparation / Cooking Method
Check box
Source
Check box
5
Species 1:
Salmon sections #s
Trout (thin) fillets:
Lamcrev sections:
Jerkv cackaqes:
Soud bowls: cuds
Shellfish (orqanisms):
Q Fried / Sauteed Q Stew, Soup
Q Baked / Roasted Q Canned, Pickled
n Broiled / Grilled Q Microwaved
n Poached / Boiled Q Raw / Uncooked
Q Dried, Smoked, Salted Q Other, Unknown
Q Casserole, Mixed Dish
Q Market / Store
O Restaurant
Q Caught
Q In Idaho
Q Outside of Idaho
Species 2:
Salmon sections #s
Trout (thin) fillets:
Lamcrev sections:
Jerkv cackaqes:
Soud bowls: cuds
Shellfish (orqanisms):
Q Fried / Sauteed Q Stew, Soup
Q Baked / Roasted Q Canned, Pickled
n Broiled / Grilled Q Microwaved
Q Poached / Boiled Q Raw / Uncooked
Q Dried, Smoked, Salted Q Other, Unknown
Q Casserole, Mixed Dish
Q Market / Store
O Restaurant
Q Caught
Q In Idaho
Q Outside of Idaho
Species 3:
Salmon sections #s
Trout (thin) fillets:
Lamorev sections:
Jerkv cackaqes:
Soud bowls: cuds
Shellfish (orqanisms):
Q Fried / Sauteed Q Stew, Soup
Q Baked / Roasted Q Canned, Pickled
n Broiled / Grilled Q Microwaved
Q Poached / Boiled Q Raw / Uncooked
Q Dried, Smoked, Salted Q Other, Unknown
Q Casserole, Mixed Dish
Q Market / Store
O Restaurant
Q Caught
Q In Idaho
Q Outside of Idaho
6
Species 1:
Salmon sections #s
Trout (thin) fillets:
Lamorev sections:
Jerkv oackaqes:
Soud bowls: cuds
Shellfish (orqanisms):
Q Fried / Sauteed Q Stew, Soup
Q Baked / Roasted Q Canned, Pickled
n Broiled / Grilled Q Microwaved
Q Poached / Boiled Q Raw / Uncooked
Q Dried, Smoked, Salted Q Other, Unknown
Q Casserole, Mixed Dish
Q Market / Store
O Restaurant
Q Caught
Q In Idaho
Q Outside of Idaho
Nez Perce Tribe
Volume III - Appendix A
December 2016 Page A-48
-------
Respondent ID:
Occasion # &
Description1
Species Type
Portion Size / Quantity
See Displays (enter display #)
Preparation / Cooking Method
Check box
Source
Check box
Species 2:
Salmon sections #s
Trout (thin) fillets:
Lamcrev sections:
Jerkv cackaqes:
Soud bowls: cuds
Shellfish (orqanisms):
Q Fried / Sauteed Q Stew, Soup
Q Baked / Roasted Q Canned, Pickled
n Broiled / Grilled Q Microwaved
n Poached / Boiled Q Raw / Uncooked
Q Dried, Smoked, Salted Q Other, Unknown
Q Casserole, Mixed Dish
Q Market / Store
O Restaurant
Q Caught
Q In Idaho
Q Outside of Idaho
Species 3:
Salmon sections #s
Trout (thin) fillets:
Lamcrev sections:
Jerkv cackaqes:
Soud bowls: cuds
Shellfish (orqanisms):
Q Fried / Sauteed Q Stew, Soup
Q Baked / Roasted Q Canned, Pickled
n Broiled / Grilled Q Microwaved
Q Poached / Boiled Q Raw / Uncooked
Q Dried, Smoked, Salted Q Other, Unknown
Q Casserole, Mixed Dish
Q Market / Store
O Restaurant
Q Caught
Q In Idaho
Q Outside of Idaho
1. "Description" refers to a distinct fish-eating occasion defined by the respondent (breakfast, lunch, dinner, snack, or a time or number).
2. See Table A-4 for species list; will be coded later as anadromous, freshwater resident, or marine fish and shellfish.
Nez Perce Tribe
Volume III - Appendix A
December 2016 Page A-49
-------
Respondent ID:
Table A-7a. Portion Size Model Displays: Description and Use
Display
Type1
Display
Numbers2
Display
Description
What Display
Represents
How Respondents
Report Portion Size
Associated Mass of Real
Fish
Salmon
S1 to S9
Large rubber
salmon fillet, cut
into 24 servings
Cooked salmon and
other fish species
with thick fillets
Identify multiples and/or
fractions for sections 1
to 24 in 0.25 increments
Serving sections range from
1.5 oz. (42 g) to 6.8 oz. (192
g) of uncooked fish
Trout
T1 to T9
Small plastic trout
fillet, single serving
Cooked trout and
other fish species
with thin fillets
Identify multiples and/or
fractions of the fillet in
0.25 increments
One fillet is 3.0 oz. (85 g) of
baked fish, or 4.0 oz. (113 g)
of uncooked fish
Lamprey
L1 to L9
Gray PVC pipe, 2"
diameter, 14" long,
notched every 2"
for 7 servings
Cooked adult
lamprey (eel)
Identify multiples and/or
fractions of the 2"
servings in 0.25
increments
Each 2" serving is calculated
to be 4.0 ounces (113
grams) of uncooked fish
Jerky
J1 to J9
Package of real
"salmon candy"
(dried fish pieces)
Dried pieces of
salmon and other
fish species
Identify multiples and/or
fractions of the package
in 0.25 increments
Packages range from 2.4 oz.
(68 g) to 3.0 oz. (84 g) of
dried fish, or 5.6 oz. (159 g)
to 6.5 oz. (187 g) raw fish
Bowls
B1 to B9
(each is
set of 5)
Empty plastic bowls
(1/4, 1/2, 1, 11/2, and 2
cups) of different
colors
Containers to hold
fish soup, composite
dishes
Identify multiples and/or
fractions of a cup in
0.25 increments
1 cup offish soup is
estimated to include 0.25
cup of cooked fish (2 oz. or
57 g) or 2.5 oz. (72 g) raw
fish
Crayfish
C1 to C9
Color photograph
(laminated) of
whole crayfish
Cooked crayfish
Identify number of
organisms
1 crayfish contains 0.26 oz.
(7.2 g) of uncooked edible
meat
Mussels
M1 to M9
Color photograph
(laminated) of plate
with 6 half-shell
mussels
Cooked mussels
and other bivalve
shellfish
Identify number of
organisms
1 mussel contains 0.4 oz.
(10 g) of uncooked edible
tissue
Shrimp
S1 to S9
Color photograph
(laminated) of plate
with 6 shrimp
Cooked shrimp
Identify number of
organisms
1 shrimp contains 1.6 oz. (44
g) of uncooked edible tissue
Other
N/A
Can or jar offish
(no display
provided)
Fish (tuna, salmon)
in a can or jar
Identify multiples and/or
fractions of cans or jars
in 0.25 increments
Standard tuna can is 5 oz.
(142 g); mason jar is 8 oz
(227 g)
Notes
1. A total of nine identical copies of each model display type will be available for use during interviews (five
for NPT and four for SBT).
2. Display numbers are written in permanent marker on every model display, as well as contact information
for Kristin Callahan, RIDOLFI, 206-436-2774, in the event there are questions or need for replacements.
" = inches
g = grams
oz. = ounces
Nez Perce Tribe
Volume III - Appendix A
December 2016 Page A-50
-------
Respondent ID:
6.4 Other Dietary Information
"Now I will ask you general questions about your diet."
12. "Was the amount of fish you ate yesterday more, less, or about the same as usual?"
(Check)
More than usual
Less than usual
About the same as usual
13. "Are you currently on any kind of diet, either to lose weight or for some other
reason?" (Check)
Yes
No
Prefer not to answer
"This concludes the interview. Thank you SO very much for your time and cooperation
today. Your participation will contribute significantly to the overall success of this survey
and help protect the health of our Tribe. We will be calling a few people back just as a
quality control measure. Thanks again for your time; that is all."
6.5 Post-Interview
Following the interview, the interviewer will record the telephone interview end time and length
and acknowledge that he/she recorded the information truthfully and to the best of his/her ability
by signing the following guarantee of authenticity.
Nez Perce Tribe
Volume III - Appendix A
December 2016 Page A-51
-------
Respondent ID:
Record interview end time and calculate interview length.
14. End time: AM / PM (circle)
15. Length of interview: (hours and/or minutes)
16.1, (printed name of interviewer) hereby affirm
that the answers recorded on this questionnaire reflect a complete and accurate
accounting of my interview with the respondent.
Signature of Interviewer
Date
Nez Perce Tribe
Volume III - Appendix A
December 2016 Page A-52
-------
RE-INTERVIEW
QUESTIONNAIRE
Nez Perce Tribe
Volume III - Appendix A
December 2016 Page A-53
-------
Respondent ID:
7.0 INTERVIEW INTRODUCTION
Contact attempts (up to 7 attempts) will be made at varying days of the week and times of day. If
no contact is made before the maximum number of attempts or by the end of the permitted one-
month period (whichever comes first), contact attempts will be terminated. Upon contact by
phone, the interviewer will record answers to re-interview questions.
0. Note outcome of contact attempts here:
No reinterview, maximum no. of attempts reached
No reinterview, respondent refused
Reinterview commenced, responses below.
11. "Hello, I'm calling on behalf of (name of Tribe and department) . May I please
speak with (name of respondent) ?"
Yes
No
If YES and respondent is speaking or when the respondent comes to the telephone,
continue to Question #2.
If NO, probe if he/she lives there, and if so, ask "When is the best time to reach
him/her? (Record on log) "Okay, thank you for your time. Good bye."
If NO, not living there, ask "What is the best way to reach him/her? (Record new
number on log) "Okay, thank you for your time. Good bye."
12. "Hello, my name is (your name)." Reintroduce Tribe if necessary. "I am calling to
thank you for your participation in our fish consumption survey. Can you please
confirm that you participated in the first interview for this survey? (Check)
Yes, did participate
No
Do not remember
If YES, continue to Question #3.
If NO or Do not remember, probe by reminding him/her of the interview date, if he/she
has a relative of the same name, etc.; otherwise, record on log, "Okay, thank you
for your time. Good bye."
13. Great, I am calling to ask just a couple of the same questions for verification
purposes. We do this to make sure we recorded it correctly the first time. The
Nez Perce Tribe
Volume III - Appendix A
December 2016 Page A-54
-------
Respondent ID:
information that you provide is confidential. Today's survey takes less than 5
minutes. May we begin?"
If YES, "Thank you for agreeing to participate," check box below and continue to
Question #4.
Interviewer: check this box if respondent agrees to participate in the telephone
verification interview.
If NO, ask "When is a good time to call back? (Record notes for re-contact as needed)
"Okay, thank you for your time. Good bye."
14. When starting interview, record re-interview call information:
Date: / / (mm/dd/yyyy)
Day (of the week):
Start time: AM / PM (circle)
15. The number of contact attempts needed to reach and re-interview this respondent,
including the successful re-interview, was . (note number)
Nez Perce Tribe
Volume III - Appendix A
December 2016 Page A-55
-------
Questions from the original FFQ will be asked again for quality control purposes. Words to be
spoken by the interviewer are identified in bold. Each question will be asked in numeric order.
No photographic or portion model displays will be necessary.
"Thinking about your fish consumption in the past year,"
8.1 Chinook Salmon Consumption
68. "In the past 12 months, did you eat Chinook salmon?"
If YES, check box in Table 1 and continue to Question #3.
If NO, continue with Question #2.
69. "Thank you. Just to be thorough, is it possible that during the past year you ate
Chinook Salmon at a restaurant, a friend's house or another place, or someone
brought fish to you?"
Yes
No
If YES, continue to QUESTION EXPLANATION below, then Question #3.
If NO, skip to Question #8.
QUESTION EXPLANATION
"Please tell me about how much Chinook salmon you ate in the past 12 months
(including the fillet and any parts). I will ask you how often you ate it. You will be
able to respond according to two periods: when Chinook salmon is in-season and the
rest of the year. Remember to consider breakfast, lunch, dinner, and snacks, and
include fillets, stews, and other dishes. Do NOT include special events, such as feasts
and ceremonies.
70. "Did you eat about the same amount of Chinook salmon.throughout the year, or did
you eat more during certain periods and less during other periods of the year?"
Same
Not same
Don't know.refused
Nez Perce Tribe
Volume III - Appendix A
December 2016 Page A-56
-------
If SAME, ask Question #4 (but not Questions #5, #6 and #7), and complete Table 1 for
one period; enter length of period as 12 months. If contradiction occurs (e.g., reports only
3 months), ask "what about the rest of the year?" (and consider as NOT SAME below).
If NOT SAME, skip to Questions #5, #6 and #7 and complete Table 1 for both high and
low fish-eating periods.
71. "In the past 12 months, how often did you eat Chinook salmon in any form (e.g.,
cooked or smoked fillets, dried, or soups)?" Enter value and check the units (number of
portions per day, per week, per month, or per year).
Skip to Question #8.
72. "In the past 12 months, how often did you eat Chinook salmon in any form (e.g.,
cooked or smoked fillets, dried, or soups) when it was in season?" Enter value and
check the units (number of portions per day, per week, per month, or per year). Record in
Table 1.
73. "Recognizing that past years may be different, how long was Chinook salmon in
season (total in weeks or months]?" Enter value in weeks or months.
74. "In the past 12 months, how often did you eat Chinook salmon in any form (e.g.,
cooked or smoked fillets, dried, or soups) during the rest of the year? Enter value and
check the units (number of portions per day, per week, per month, or per year).
Nez Perce Tribe
Volume III - Appendix A
December 2016 Page A-57
-------
Table 1. FFQ: Frequency and Quantity of Chinook Salmon Eaten in Past 12 Months
Kisli Species
Chock
ilciilcn
( onsumplion When l-'isli ;iiv III Sc.ison'
Or Siiinc ( oilsuinnlion Yc;ir Round
( onsii ill pi ion lies
(liliink il' Siiinc ( onsiini
of (lie Year
)lion Year Round)
Number
i»l"
Portions
Poi'iions per da>. week.
iiioiiiIi. in' \ear (circle)
1 > pical
Kiriioii Si/e
(A: displas )
1 .enulli of
period
(weeks or
IIIOIIlllS)
Niimher
of
I'oi'lloils
l\iilkiiis per da>. week,
iikinili. or \ear (circle)
1 > pical
I'oilioii Si/e
(A: displas )
1 .enulli of
period
(aulo-
calculaled)
Chinook (King) Salmon
Day Wk. Mo. Yr.
NOT ASKED
Wk. Mo.
Day Wk. Mo. Yr.
Wk. Mo.
Notes
1. Fish consumption "in season" is based on respondent's perception or experience related to harvest and assumed higher consumption (compared to
the rest of the year); biological seasons (e.g., fish runs) will be evaluated during data analysis and do not have to correspond to the duration of
seasons noted by the respondent.
Nez Perce Tribe
Volume III - Appendix A
December 2016 Page A-58
-------
8.2 Changes in Fish Consumption.
"The next two questions refer to your consumption of any species of fish, not just Chinook
Salmon." Note, this interviewer's introductory sentence does not appear in the original
questionnaire or in the CAPI software (see section 5.8 of Volume II). It is added here
because the theme just prior to this has been about consumption of Chinook salmon.
75. "Has there been a change over time in your fish consumption?" (Check)
Yes
No
Prefer not to answer / Don't know
If YES, continue to Question #9.
If NO or PREFER NOT TO ANSWER/DON'T KNOW, skip to Question #10.
76. "How has it changed most recently?" (Check)
Increased consumption
Decreased consumption
Other change (simply note if there has been a change that is not either 'increased'
or'decreased')
Technical note: The responses to this question have been modified from the original
question in the full questionnaire by dropping the 'specify' entry for what 'other change'
represents.
8.3 Demographic Information
(Check or enter - if respondent prefers not to say, enter 999)
77. "How many people live in your household, including yourself?"
Nez Perce Tribe
Volume III - Appendix A
December 2016 Page A-59
-------
9.0 INTERVIEW EN
I
Upon completing the interview, the interviewer will offer appreciation and complete the
remaining information, including signing a form verifying participation.
78. "Thank you SO much for your time and cooperation." Complete information below.
Record telephone verification interview end time.
79. End time: AM / PM (circle)
80. Record the circumstances of the re-interview.
81. The interview was conducted (check one)
By phone
In person
Following the interview, the interviewer will acknowledge that he/she recorded the information
truthfully and to the best of his/her ability by signing the following guarantee of authenticity.
I, (printed name of interviewer) hereby affirm
that the answers recorded on this questionnaire reflect a complete and accurate
accounting of my verification interview with the respondent.
Signature of Interviewer
Date
Nez Perce Tribe
Volume III - Appendix A
December 2016 Page A-60
-------
Appendix B—
Portion-to-Mass Conversion
Nez Perce Tribe
Volume III - Appendix B
December 2016 Page B-l
-------
Appendix B—Portion-to-Mass Conversion
Fish Consumption Survey
Portion Model Displays and Mass Calculations
For dietary assessments where food items are not weighed, portion sizes must be used (with
frequency of consumption) to calculate consumption rates (Wrieden, et al., 2003). The U.S.
Department of Agriculture (USDA), in partnership with the Centers for Disease Control and
Prevention (CDC), uses 3-D food models for in-person interviews and 2-D photographs for
follow-up telephone interviews to collect dietary information as part of the National Health and
Nutrition Examination Survey (NHANES) (USDA, 2013). A similar approach has been
successfully used for Tribal fish consumption surveys in California where University of
California Davis researchers use 3-D fish fillet models of varying pre-determined masses to
estimate Tribal fish consumption rates (Shilling, 2014). The USDA recommends that models
represent foods "as consumed" as much as possible (for most accurate reporting); i.e., familiar in
appearance and preparation method (Moshfegh, 2014). Broadly, the models used in this survey
can be grouped into three types: life size depictions of fish portions (e.g. fillets), depictions of
numbers of organisms consumed per serving (e.g. shellfish), or volumes of tissue or composite
dishes consumed (e.g. bowls for fish meat or soup containing fish). The U.S. Environmental
Protection Agency (USEPA) recommends reporting the portions in uncooked weights, however,
since contaminant concentrations are measured in raw fish tissue (Kissinger, 2014). Recognizing
that fish is eaten in various forms, bowls may be used as a measuring guide for fish stews and
other composite dishes; although a standard recipe must be determined in advance to equate the
bowl quantity to fish mass. Some respondents to this survey also reported consumption of fish
tissue in volumetric terms. For example, consumption of crab meat might be reported in terms of
cups of crab meat consumed. Once respondents are familiar with the models, photographs of the
models can be given to respondents for the follow-up telephone interviews (CDC, 2010).
The list of common species used during the interviews to determine fish consumption is provided
in Tii Me li I below. The fish model displays used to determine portion sizes consumed of those
species are described in Table B2, followed by photographs and a discussion of the models and
the mass calculations. There were nine to 11 copies of each display type, depending on the
number of interviewers and whether replacements were necessary during the survey. The model
displays, which represent common species and preparation methods, included the following:
1. Large cooked salmon fillet replica, cut into servings
2. Small cooked trout fillet replica, single serving
3. PVC pipe to represent lamprey
4. Fish jerky pieces (real, packaged) to represent dried fish
5. Measuring bowls for soups and composite dishes
6. Photographs of shellfish, including mussels, crayfish, and shrimp
Nez Perce Tribe
Volume III - Appendix B
December 2016 Page B-2
-------
Table Bl. Survey Species List
SALMON AM) STKKI.II KAI)
Chinook (King) Salmon
Coho (Silver) Salmon
Sockeye (Red) Salmon
Kokanee (resident form of sockeye)
Steelhead (migratory form of rainbow trout)
Oilier salmon species (s|vcif\. c ii . rinini. Pink. Alkinlic salmon)
UKSIDKM I KOl I
Rainbow Trout
Cutthroat Trout
Cutbow Trout (hybrid of Rainbow and Cutthroat Trout)
Bull Trout (Dolly Varden)
Brook Trout
Lake Trout
Brown Trout
Other trout species (specify)
O!IIKU I UKSIIWA I IK II SI I AM) SIIKI.I.I ISM
Sturgeon
Lamprey
Whitefish
Sucker
Burbot
Northern Pikeminnow (Squawfish)
Bass
Bluegill
Carp
Catfish
Crappie
Sunfish
Tilapia
Walleye
Yellow Perch
Other freshwater finfish (specify)
Crayfish
I 'reshwater Clams or Mussels
SKAKOOI) / MAUIMK I ISM AM) SIIKI.I.I ISM
Cod
Halibut
Pollock
Tuna
Lobster
Crab
Marine Clams or Mussels
Shrimp
Other marine fish or shellfish (specify)
Nez Perce Tribe
Volume III - Appendix B
December 2016 Page B-3
-------
Table B2. Description of Portion Size Model Displays
Display
Tjpe"
Display
Numbers2
Display
Description
Wlial Display
Represents
How Respondents
Report Portion
Associated Mass of
I ncooked l isli
Salmon
SI to S9
Large rubber salmon
fillet, cut into 24
servings
Cooked salmon and
other fish species
with thick fillets
Identify multiples
and/or fractions for
sections 1 to 24 in 0.25
increments
Servings range from
1.5 oz. (42 g) to 6.8
oz. (192 g) uncooked
fish
Trout
T1 to T9
Small plastic trout
fillet, single serving
Cooked trout and
other fish species
with thin fillets
Identify multiples
and/or fractions of the
fillet in 0.25
increments
One fillet is 3.0 oz.
(85 g) of baked fish,
or 4.0 oz. (113 g) of
uncooked fish
Lamprey
LI to L10
Gray 14" PVC pipe,
2" diameter notched
every 2" for 7
servings
Cooked adult
lamprey (eel)
Identify multiples
and/or fractions of the
2" servings in 0.25
increments
Each 2" serving is
calculated to be 4.0
oz. (or 113 g) of
uncooked fish
Jerky
Jlto Jll
Package of real
"salmon candy"
(dried fish pieces)
Dried pieces of
salmon and other
fish species; also
crab or similar-
shape tissue
Identify multiples
and/or fractions of the
package in 0.25
increments
Packages range from
2.4 oz. (68 g) to 3.0
oz. (84 g) of dried
fish, or 5.6 oz. (159
g) to 6.5 oz. (187 g)
uncooked fish
Bowls
B1 to B9
(each is set
of 5)
Empty plastic bowls
0/4, !4 1, lVi, and 2
cups) of different
colors
Containers to hold
fish soup, composite
dishes
Identify multiples
and/or fractions of a
cup in 0.25 increments
1 cup of fish soup
includes 0.25 cup of
cooked fish (2 oz. or
57 g) or 2.5 oz. (72 g)
uncooked fish;
If not soup, 1 cup of
fish (8 oz or 227 g) or
10.7 oz (302.4 g)
uncooked fish
Crayfish
CI to CIO
Color laminated
photograph of whole
crayfish
Cooked crayfish
Identify number of
organisms
1 crayfish contains
0.26 oz. (7.2 g) of
uncooked edible
tissue
Mussels
Ml to M10
Color laminated
photograph of plate
with 6 half-shell
mussels
Cooked mussels and
other bivalve
shellfish
Identify number of
organisms
1 mussel contains 0.4
oz. (10 g) of
uncooked edible
tissue
Shrimp
Shi to ShlO
Color laminated
photograph of plate
with 6 shrimp
Cooked shrimp
Identify number of
organisms
1 shrimp contains 1.6
oz. (44 g) of
uncooked edible
tissue
Notes: " = inches, g = grams, oz. = ounces
Nez Perce Tribe
Volume III - Appendix B
December 2016 Page B-4
-------
1.0 Salmon Fillet Model Displa
A 3-D replica of a Chinook salmon fillet was obtained from a local Seattle artist (Figure Bl).
The fillet (with skin and tail) was made of a flexible and durable urethane rubber, which was
poured into a latex mold built based on a fresh (brined) ocean-caught Chinook salmon fillet. The
rubber model was painted the color of cooked salmon muscle (fillet) and other tissues (skin and
tail). The rubber model weighed 6.8 pounds; the fillet part of the model, which was used to
report portion sizes (without skin or tail), had a total length of 29 inches, a width ranging from 3
inches (at the tail end) to 7.5 inches (in the middle), and a depth up to approximately 1 inch.
The salmon replica was used as a model display to indicate portion sizes of all species of baked
or smoked salmon, including Chinook, coho, and sockeye salmon, and also other large fish with
thick fillets, such as sturgeon or halibut, assuming the respondents could associate the model
cross-species. The fillet was cut into 24 servings, each of which was labeled with a number (1
through 24). During the interviews, respondents indicated which serving pieces represented their
average portion size, and the interviewers recorded those numbers for each species type
(translated to mass during data analysis). The display number (SI to S9) of the specific model
used during the interview was also recorded.
Figure Bl. Salmon Fillet Replica (24 Servings)
To equate fish model servings to mass of fresh fish, a Chinook salmon of comparable size was
obtained from the Pike's Place Market in Seattle, Washington. Professional staff at the fish
market filleted and skinned an ocean-caught Chinook salmon and cut it into servings as equal to
the model servings as possible. The whole raw fish (with skin, but no tail) weighed
approximately 7 pounds; 6.8 pounds without the skin. Each serving was later weighed (in ounces
and grams) on a scale (precision of +/- 2 grams), both uncooked and cooked (after oven-baking
for 30 minutes). There was an average 12% loss of mass from the light baking process. Due to
the amorphousness of fresh fish (and, therefore, the model), servings nearest the head and tail
were found to have less mass (about half) than those in the middle of the fillet. Uncooked fish
mass of each of the 24 servings of fresh fish (representing the 24 servings of the portion model)
is presented in Table B4 in section 11.
Nez Perce Tribe
Volume III - Appendix B
December 2016 Page B-5
-------
2.0Trout-Like Fillet Model Displa
A 3-D replica of a baked tilapia fillet from Barnard, Ltd. (made of flexible plastic resin, latex-
and lead-free, 3.5 x 5-inches, and weighing 2.6 ounces), was used as a model display to indicate
portion sizes of baked or smoked trout and other fish species with lighter-colored tissue and
thinner fillets as compared to salmon (Figure B2). The trout-like replica represented a 3-ounce
(or 85-gram) fillet of baked fish, and was versatile enough to represent a variety of freshwater
and marine species. Respondents reported fractions (0.25, 0.5, and 0.75) and/or multiples (1, 2,
3, etc.) of the fillet to indicate their portion size, and interviewers recorded that number
(translated into total mass during data analysis). The display number (T1 through T9) of the
specific model used during the interview was also recorded.
Figure B2. Trout-Like Fillet Replica (Single Serving)
Based on the replica representing a 3-ounce baked fish fillet, and assuming a 25% moisture loss
during the baking process (see Attachment 1; USEPA, 2014), Table B5 in section 11 presents
various portion sizes converted into uncooked fish mass (based on fractions or multiples of 1).
One serving (one whole trout fillet) that is 3 ounces (85 grams) baked equates to 4 ounces (113
grams) uncooked.3 Additional multiples and/or fractions reported by respondents were calculated
during data analysis.
3 Values shown in ounces and grams reflect the direct mass conversions from cooked to uncooked weights (according to the
equation in Attachment 1).
Nez Perce Tribe
Volume III - Appendix B
December 2016 Page B-6
-------
3.0 Lamprey (PVC Pipe) Displa
Lamprey (eel) is a unique anadromous species type consumed by Tribal members. As
recommended by Tribal Representatives, a 14-inch long, 2-inch diameter gray PVC pipe was
used as a model display to indicate portion sizes of lamprey (Figure B3). The length was an
approximate average size of an adult lamprey post-migration, preparing to spawn up-river
(Kostow, 2002). The PVC pipe had section marks notched every 2 inches to indicate servings.
Each 2-inch serving was labeled with a number (1 through 7). Respondents reported fractions
(0.25, 0.5, or 0.75) and/or multiples (1, 2, 3, etc.) of a serving to represent their average portion
size, and the interviewers recorded that number (translated into total mass during data analysis).
The display number (LI to L10) of the specific pipe used during the interview was also recorded.
Figure B3. PVC "Lamprey" Pipe (7 Servings)
Assuming a density as least as great as other fresh (raw) fish muscle, approximately 1.1 g/cm3
(UNFAO, 2014a), and a calculated volume of a cylinder section (102.9 cm ), the mass of each 2-
inch serving was estimated to be 4.0 ounces (113 grams). Table B5 in in section 11 presents
portion sizes as fractions and multiples of one (1) serving. Additional multiples and/or fractions
of these servings reported by respondents were calculated during data analysis.
Nez Perce Tribe
Volume III - Appendix B
December 2016 Page B-7
-------
4.0 Jerky I Dried Fish Displa
In cases where respondents reported eating any species of fish (salrnonid or other) in a dried
form, real fish jerky (known as "salmon candy"), protected in a sealed package, was used to
indicate portion sizes (Figure B4). Respondents reported fractions (0.25, 0.5, or 0.75) and/or
multiples (1, 2, 3, etc.) of the approximately 3-ounce (85-gram) package of dried salmon to
indicate their portion size, and the interviewers recorded that number (translated into total mass
during data analysis). The display number (J1 to J11) of the specific package used during the
interview was also recorded.
In this case, recording the specific display number was particularly important because, although
the label stated that there were 3 ounces (85 grams) in every package, the true mass was found to
vary between packages (and was generally less). Two extra packages were purchased and
opened, and the contents were weighed (in ounces and grams) on a scale (precision of +/- 2
grams). The dried salmon within each of these packages was measured at 2.6 ounces (72 grams),
and the package alone weighed 0.2 ounces (5.7 grams). Without opening the display packages to
be used during the survey (to maintain the integrity of the contents), each whole package was
weighed and, subtracting the weight of the bag (0.2 ounces), total mass of dried fish was
calculated. That mass, without a moisture loss conversion, was used for reporting fresh tissue
such as crab.
Figure B4. Package of Real Jerky/Dried Fish ("Salmon Candy")
Nez Perce Tribe
Volume III - Appendix B
December 2016 Page B-8
-------
To represent dried fish, assuming a 57% moisture loss during the desiccation process
(Attachment 1; USEPA, 2014), Table B6 in section 11 presents the mass of salmon jerky
measured in each display package converted to uncooked mass (based on fractions or multiples
of 1). One serving (one whole package of display Jl) that was 2.5 ounces (70 grams) dried, for
example, converted to 5.8 ounces (163 grams) uncooked. Fractions and/or multiples of one
serving (one package) were calculated based upon one (1) serving of the particular display
package during data analysis.
5.0 Soup Bowl Displa;
For fish soups and composite dishes, portion sizes were determined using empty hard-plastic
bowls of different quantities (and colors) within a '/t-cup (red), V^-cup (yellow), 1-cup (purple),
1 Mi-cup (blue), or 2-cup (green) bowl (Figure B5). Respondents reported the fractions (0.25 or
0.5 cup) or multiples (1, 1.5, 2 cups, etc.) of one cup to indicate their portion size, and the
interviewers recorded that number (translated into mass of fish during data analysis). The display
number (B1 to B9) of the measuring bowl set used during the interview was also recorded.
Figure B5. Measuring Bowls for Fish Soups
As suggested by Tribal representatives (Holt, et al., 2014), it was estimated that 1 cup of soup
contained approximately 0.25 cup (or 2 ounces or 57 grams) of cooked fish (i.e., soup was 25%
fish). Based on the assumption that a one (l)-cup serving of soup contained 2 ounces (57 grams)
of cooked fish, and assuming a moisture loss of 21% from cooking in soup ("wet cooked in
moist heat"), Table B5 in section 11 presents the mass of uncooked fish according to number of
cups (servings) of soup (based on fractions or multiples of 1) (Attachment 1; USEPA, 2014).
Additional multiples and/or fractions that were reported by respondents were calculated during
data analysis. Note that the measuring bowls were intended to represent soups, stews, chowders,
or other composite dishes such as casseroles, applying the same general assumption of 1 cup
composite dish: 0.25 cup cooked fish ratio. As has been noted, some respondents reported
consumption of fish or shellfish tissue in volumetric terms. When the bowls were used to
describe fish volume rather than soup, it was assumed that one cup corresponded to 8 ounces
(227 g) of cooked fish (assumes an overall density of 1) and 10.7 ounces (302.4 g) of uncooked
fish, assuming a 25% moisture loss, as from canning or a dry heat method (Table B3).
Nez Perce Tribe
Volume III - Appendix B
December 2016 Page B-9
-------
6.0Shellfish Photograph Displays
For shellfish, portion sizes were determined using laminated color photograph displays (photo-
displays), printed to 100% scale (actual size). There was a photo-display of a single, whole
crayfish (tail tucked under); a photo-display of mussels (six half shells on a plate) to represent
marine and freshwater bivalves (clams and mussels); and a photo-display of shrimp (six on a
plate), as shown on Figures B6 through B8, respectively. Respondents reported numbers of
organisms (e.g., number of crayfish, mussels, or shrimp) to indicate their portion size, and the
interviewers recorded that number (translated into mass of shellfish during data analysis). The
photo-display number (CI to CIO for crayfish; Ml to M10 for mussels; or SHI to SH10 for
shrimp) of the specific photo-display used during the interview was also recorded.
Figure B6 illustrates a native crayfish, Pacifastacus leniuscu/iis, the most widely distributed
species in the Pacific Northwest (Johnsen and Taugbol, 2010; Larson and Olden, 2011), which
was obtained from the Columbia River watershed and purchased at the Pikes Place Market in
Seattle, Washington. Weight of the whole uncooked organism was measured at 1.3 ounces (36
grams). The primary edible tissue of crayfish is the tail (abdominal muscle), the percent (to
whole body) of which depends on size and maturity. The edible portion of P. leniuscalus has
been estimated to be 15 to 25% of total body weight (Lee and Wickins, 1992, as cited in
Harlioglu, 1996). Assuming that an average 20% of body mass is edible tissue, the mass
consumed per single organism (of a size organism shown in the figure) is 0.26 ounces (7.2
grams). Total numbers of crayfish reported by respondents as the portion size consumed were
recorded and the associated mass was calculated during data analysis.
Figure B6. Crayfish Photo-Display
Figure B7 illustrates a common intertidal zone bivalve, Mytilus edulis or Blue Mussel, which is
found on the Pacific coast of the U.S. and is domestically farmed (NOAA, 2014). Freshwater
mussels are in a different subclass of bivalves than the marine species, but are superficially
similar in appearance. The figure is intended to represent all types of marine and freshwater
bivalves that may be consumed by participants. The shell (half) is included with cooked mussel
meat in the photograph to display a familiar preparation method, but it is the edible soft tissue
Nez Perce Tribe
Volume III - Appendix B
December 2016 Page B-10
-------
that is of interest. Soft tissue can be nearly 50% of total live (wet) weight when the organism is
in best condition (UNFAO, 2014b). One study reported that organisms investing energy in shell
growth may actually limit soft tissue growth (Gimin et al., 2004). For this study, average tissue
weights, which vary by species, age, gender, density, season, food availability, and other
environmental conditions, were used for portion size calculations.
Multiple sources of information were investigated to determine the average mass of soft tissue
consumed per bivalve organism. The mean wet weight of edible soft tissue of a single mussel
consumed by California Indians was reported (in an archeological study) as 1.065 grams, but
with no supporting documentation (Heizer and Whipple, 1971). A more recent study of Mylilus
edulis in Quebec, Canada, collected 4,224 juvenile mussels and measured an average soft tissue
dry weight (ash free) of 0.037 grams (Alunno-Bruscia et al,, 2001), which equates to 0.42 grams
wet weight (likely a juvenile that is too small to be edible). Finally, a reference documenting the
life history of mussels suggested that average large adult mussel soft tissue weighs 1 g dry
weight (Newell and Moran, 1989), which (assuming 10% solids) equates to 10 g. This value was
used to represent the mass of a single bivalve organisms. Total numbers of mussels or clams
reported by respondents as the portion size consumed were recorded, and the associated mass
was calculated during data analysis.
Figure B7. Mussels Photo-Display
• 71"
9!»
hiiiiliiilililiiAtitl
mi
t* Enl 2
,.l In i
Figure B8 illustrates a large shrimp, likely Panda/us borealis, northern prawn or pink shrimp.
Large males commonly reach 170 millimeters (mm) (6.69 inches), which (when including head)
approximates the organism sizes in the photograph. Based on a total length to weight conversion
cited by the U.S. Fish and Wildlife Service (Nichols, 1982 as cited in Bielsa, et al., 1983), a
length of 170 mm equates to 44 grams (1.6 ounces). This value was used to represent the mass of
a single shrimp organism, based upon fractions and multiples of 1. Total numbers of shrimp
Nez Perce Tribe
Volume III - Appendix B
December 2016 Page B-l 1
-------
reported by respondents as the portion size consumed were recorded, and the associated mass
was calculated during data analysis.
Figure B8. Shrimp Photo-Display
I'I'I'I'li]1!! I|I|I|I|||I|I|I I|I|'I<|||||<|I| ; . |i|'l 11 |I|I|I|I||| ||1| ||||| I h| 11'| I'l I'l in m'I'I'i;
4 5] 61 Tan 8 9 lO , 111
• 2 3
UuuW,
7.0 Fish in Cans or Jars
For fish reported as eaten from cans or jars, the following assumptions were made: 1 standard
can of tuna (or other commercially canned fish) contains 5 ounces of cooked fish and 1 standard
Mason jar of salmon (or other fish, home-canned) contains 8 ounces of cooked fish. Based on a
moisture loss of 25% during the canning process (Attachment 1; USEPA, 2014), a single can or
jar equates to 6.7 ounces (189 grams) and 10.7 ounces (302 grams) of uncooked fish,
respectively. Table B5 in section 11 presents the uncooked fish mass associated with fractions
and multiples of 1 can or 1 jar, respectively, of cooked fish.
Nez Perce Tribe
Volume III - Appendix B
December 2016 Page B-12
-------
8.0 References
Alunno-Bruscia, M, E. Bourget, andM. Frechette. 2001. Shell Allometry and Length-Mass-
Density Relationship for Mytilus edulis in an Experimental Food-Regulation Situation.
MarEcol Prog Ser. 219: 177-188. September.
Bielsa, L.M., W.H. Murdich, and R.F. Labisky. 1983. Species Profiles: Life Histories and
Environmental Requirements of Coastal Fishes and Invertebrates (South Florida): Pink
Shrimpl. U.S. Fish and Wildlife Service. Biological Report 82(11.17), TREL-82-4.
Prepared for U.S. Army Corps of Engineers.
Center for Disease Control and Prevention (CDC). 2010. USDA Food Model Booklet. National
Health and Nutrition Examination Survey. Last modified: May 2010. Available:
http://www.cdc.gov/nchs/nhanes/measuring_guides_dri/2002/fmb.htm
Colter, C. and C. Tanaka. 2014. Personal communication with the Shoshone-Bannock Tribes.
Restoration Workshop, March 5-6, and subsequent communication with K. Callahan,
RIDOLFI. Shoshone-Bannock Tribes Fish and Wildlife and Water Resources
Departments. Fort Hall, Idaho.
Dare, P.J. and D.B. Edwards. Seasonal Changes in Flesh Weight and Biochemical Composition
of Mussels (Mytilus edulis L.) in the Conwy Estuary, North Wales. J Exper Mar Biol.
18(2):89-97. Elsevier.
Gimin, R., R. Mohan, L.V. Thinh, and A.D. Griffiths. 2004. The relationship of shell dimensions
and shell volume to live weight and soft tissue weight in the mangrove clam, Plymesoda
erosa (Solander, 1786) fi'om northern Australia. WorldFish Center Quarterly. Vol. 27
No. 3&4. Jul-Dec.
Harlioglu, M.M. 1996. Comparative Biology of the Signal Crayfish, Pacifastacus leniusculus
(Dana), and the narrow-clawed crayfish, Astacus leptodactylus Eschsholtz. Thesis
submitted to University of Nottingham for the degree of Doctor of Philosophy.
Department of Life Science.
Heizer, R.F. andM. A. Whipple (Eds). 1971. The California Indians, A Source Book. Second
Edition. University of California Press. Accessed through Google Books. Available:
http: //b ooks. googl e. com/b ooks? i d=Zj U Om W W y GSMC &print sec=frontcover& source=g
bs_ge_summary_r&cad=0#v=onepage&q&f=false
Holt, J., K. Clark, and J. Oatman. 2014. Personal communication with the Nez Perce Tribe.
Restoration Workshop, March 5-6, and subsequent communication with K. Callahan,
RIDOLFI. Nez Perce Tribe Water Resources and Fisheries Departments. Lapwai, Idaho.
Idaho. 2013. Idaho Fish Consumption Rate Recommended Sample and Questions. Prepared for
the Idaho State Department of Environmental Quality by the Boise State University
Public Policy Center. December 19.
Johnsen, S.I. and T. Taugbol. 2010. NOBANIS - Invasive Alien Species Fact Sheet -
Pacifastacus leniusculus. Online Database of the European Network on Invasive Alien
Species - NOBANIS www.nobanis.org. Accessed 6/3/14.
Kissinger, Lon. 2014. Personal communication with K. Callahan, RIDOLFI, on February 20,
2014. Risk Assessor, U.S. Environmental Protection Agency, Region 10.
Nez Perce Tribe
Volume III - Appendix B
December 2016 Page B-13
-------
Kostow, K. 2002. Oregon Lampreys: Natural History Status and Analysis of Management
Issues. Oregon Department of Fish and Wildlife, Fish Division. Information Reports No.
2002-01. February 25.
Larsen, E.R. and J.D. Olden. 2011. The State of Crayfish in the Pacific Northwest. University of
Washington, School of Aquatic and Fishery Sciences. Fisheries, Vol. 36, No. 2. February.
Moshfegh, Alanna. 2014. Personal communication with K. Callahan, RIDOLFI, on January 13,
2014. Supervisory Nutritionist, U.S. Department of Agriculture.
National Oceanic and Atmospheric Administration (NOAA). 2014. FishWatch, U.S. Seafood
Facts, Blue Mussel. Available:
http://www.fishwatch.gov/seafood_profiles/species/mussels/species_pages/blue_mussel_
farmed.htm. Accessed 6/6/14.
Newel, R.I.E. and D. Moran. 1989. Species Profiles: Life Histories and Environmental
Requirements of Coastal Fishes and Invertebrates (North and Mid-Atlantic): Blue
Mussel. University of Maryland and U.S. Fish and Wildlife Service. Biological Report
82(11.102), TREL-82-4. Prepared for U.S. Army Corps of Engineers. June.
Shilling, Fraser. 2014. Personal communication with K. Callahan, RIDOLFI, on February 6,
2014. University of California at Davis.
United Nations Food and Agriculture Organization (UNFAO). 2014a. Measures, Stowage Rates
and Yields of Fishery Products. Available:
http://www.fao.Org/wairdocs/tan/x5898e/x5898e01.htm#Densities and stowage rates
United Nations Food and Agriculture Organization (UNFAO). 2014b. Fisheries and Aquaculture
Department. United Nations. Accessed June 3, 2014. Available:
http://www.fao.org/fishery/culturedspecies/Mytilus_galloprovincialis/en
United States Department of Agriculture (USD A). 2013. Food Surveys: What We Eat in
America, NHANES. Agriculture Research Service. Last modified: November 2013.
Available: http://www.ars.usda.gov/Services/docs.htm?docid=13793
United States Environmental Protection Agency (USEPA). 2014. Estimated Fish Consumption
Rates for the U.S. Population and Selected Subpopulations (NHANES 2003-2010), Final
Report. EPA-820-R-14-002. April.
Wrieden, W., H. Peace, J. Armstrong, andK. Barton. 2003. A short review of dietary assessment
methods used in National and Scottish Research Studies. Available:
http://www.food.gov.uk/sites/default/files/multimedia/pdfs/scotdietassessmethods.pdf
Nez Perce Tribe
Volume III - Appendix B
December 2016 Page B-14
-------
COOKING LOSS FACTORS
Similar to the Idaho Tribal Fish Consumption Survey, NHANES participants report the amount
of fish consumed "as prepared," which is converted to a raw wet weight in grams. Since the
process of cooking changes the moisture content of fish, a weight conversion based on the
estimated moisture loss due to cooking is required to calculate the grams of raw fish consumed
(USEPA, 2014). Adjustment factors for cooking loss used by NHANES, and reported by EPA,
are provided in Table B3 (with values in bold associated with key preparation methods presented
in this study; notes in italics have been added by the authors).
The following equation is used to convert cooked mass to uncooked (raw) mass:
Weight of raw fish = Weight of cooked fish
1 - (% Moisture Loss/100)
Table B3. Estimated Fish Moisture Loss Due to Cooking
Cooking / Preparation Method
Percent moisture loss
Dried (e.g. jerky)
57
Kippered
46
Smoked, (other than salmon)
36
Salted
33
Canned
25
Cooked, dry heat (e.g., baked)
25
Restructured
25
Cooked, moist heat (e.g., soup)
21
Smoked salmon
17
Pickled
16
Fried
12
Raw
0
Source: USEPA, 2014
Nez Perce Tribe
Volume III - Appendix B
December 2016 Page B-15
-------
Figure B9. Species Identification Photographs
The species identification photographs (image resolution reduced for inclusion into this report)
used by the interviewers to facilitate the administration of the questionnaire (4 pages). Sources:
Columbia River Inter-Tribal Fish Commission, Idaho Fish and Game, Washington Department
of Fish and Wildlife, Montana Field Guide, Freshwater Mollusks Guide, U.S. Fish and Wildlife
Service, U.S. Geological Survey.
[ Salmon and Other Anadromous (Migratory) Fish Native to Idaho
Chinook (King) Salmon {Oncwhynchjs tshowytscbal Coho (Silver) Salmon (Oncorhynchus kisutch) Sockeye (Rod) or Kokanee* [Bluotoack) Salmon
{Oftcarhynehuf oerka) 'Ketones are resident iockeye
Steelhead* Trout (Oncorhyrchus m/kks) U>mpr«y (Lompttro tridentota)
" Steel head are the migratory type of rainbow trout
| Salmon Species Not Native to Idaho
Atlantic Salmon (Satmo salor)
Chum (Dog) Salmon (Or.corhyncbus keta)
Pink (Humpback) Salmon (Oncarlrynchus gorbuscfio)
Resident Fish Species Native to Idaho
Inland Columbia Oa&lii Rcdband
Trout (OncwrltyiULhos mykls-»
galrdnerl)
Trout Species Native and Not Native to Idaho
Bull Trout* {SaUeNrws tanflueittui)
• Also known as Dolly Varden (coastal relatives)
Cutbow Trout (Oncorhynchus clarkii x
Oncorhynchus mykiss hybrid)
Brown Trout (Salmo trutta)
Rainbow Trout* (Oncorhyncnus myhss)
* Includes Golden Trout subspecies
Nez Perce Tribe
Volume III - Appendix B
December 2016 Page B-16
-------
Figure B9. Species Identification Photographs (continued, page 2 of 4)
Sources: Columbia River Inter-Tribal Fish Commission, Idaho Fish and Game, Washington
Department of Fish and Wildlife, Montana Field Guide, Freshwater Mollusks Guide, U.S. Fish
and Wildlife Service, U.S. Geological Survey.
Resident Fish Species Native to Idaho (continued)
Bonneville Whitefish
(Prosopium spi/onotusf
Burbot (Lota lota)
Chkelmaulh (AcrocheBu*
alutaceus)
i f111r1111 r 11111111h 111 * 11111111r1111111111 <•1111 •
Lealherefde Chub (Lcpidomeda
copei)
lake Chub (Cohipmus plumbeeni]
Leopard Dace (Rhinidithys
falcatos)
Dace (Rhiniehttrys
cataractae)
Northern Squawtfch
WhiterisJi {Prosopium (Ptychodheilus oreqonensls)
wMamsonl) 'tfabrve Viiai -round
| Sculpln"]
Peamoutlt ( MyJochellus
cswrinus)
Pygmy Whitefish (Pro*.opium
coulter!)
Bridgelip Sucker (Catostomus
catumbianus)
Largescale sucker (Catostomus
macrocheilusj
Mountain Sucker (Catostomus
piatyrhynchtis)
Utah Sucker (Catostomus
ardsns)
H«"«Kidp Shiner (Richardvoii.iui>
balteatu*)
ijiiii|iiii|iiH|iiii|im]i(ii
Spt'rklod Oner (Rhlnlchthyt
os cuius)
Sand Roller (PwcopsK llmaHlln Dace (Hhlokhthys
transmontana) amatilln)
White Sturgeon (Aclpenser transmontanus)
Sourc*i>:Hver Ifflrr ThtuI frtft Commrtuon, kMwftlft Ml G«"«, Dtpttomm of FWi and WWW*, Wwum F»W 6u«, f nslmtfet Matoiki Gurfc, US. FishWWt Strnw, US. G*atafrM! Sunny
Nez Perce Tribe
Volume III - Appendix B
December 2016 Page B-17
-------
Figure B9. Species Identification Photographs (continued, page 3 of 4)
Sources: Columbia River Inter-Tribal Fish Commission, Idaho Fish and Game, Washington
Department of Fish and Wildlife, Montana Field Guide, Freshwater Mollusks Guide, U.S. Fish
and Wildlife Service, U.S. Geological Survey.
Resident Fish Species Not Native to Idaho
Bass* (Micropterus dotomieu and M. salmoides)
* Includes Smallmouth and Largemouth bass
m
Catfish* (Pylodictis ollvaris, Ictolurus punctatus, I,
furcotus)' Includes flathead, channel, and blue catfish
Green Stinfish (Lepomis
cyonellus)
Nez Perce Tribe
Volume III - Appendix B
December 2016 Page B-18
-------
Figure B9. Species Identification Photographs (continued, page 4 of 4)
Sources: Columbia River Inter-Tribal Fish Commission, Idaho Fish and Game, Washington
Department of Fish and Wildlife, Montana Field Guide, Freshwater Mollusks Guide, U.S. Fish
and Wildlife Service, U.S. Geological Survey.
Resident Fish Species Not Native to Idaho (Continued)
Yellow Perch (Perca flavescens)
Clams, Mussels, and Crayfish
Say Mussel
(Mytilus trossulus)
Western Ridgemusse! (Conidea onguiata)
Asian Clams (Corbicula spp.)
Western Pearl shell (Margaritifero
fatcata)
Floaters (Anodonta spp.)
Zebra/quagga mussels (Dreisserta spp.)
Nez Perce Tribe
Volume III - Appendix B
December 2016 Page B-19
-------
9.0 Portion-to-Mass Calculations
More specific details of the portion-to-mass conversion procedure are described below, including
the specific factors used for each portion model, how write-in species were handled, how can and
jar portion sizes were determined, how shellfish portion sizes were determined, and special-case
exceptions to the overall procedure.
10.0 Portion-to-Mass Conversion Tables
The portion-to-mass conversion factors for each model are shown in Tables A (salmon fillet
sections), B (trout, soup bowl, lamprey, shellfish, can and jar models), and C (jerky models).
Two different conversion factors were determined for bowls, depending on whether the
respondent likely intended the bowl to refer to the total volume of a composite dish of which fish
was only one component or whether the bowl referred to the actual volume of fish. The most
common example of the latter would be canned tuna, as used, for example, in a tuna fish
sandwich. The bowl conversions are described in detail in section 13 of this appendix.
Lastly, two conversion factors were used for each jerky model, with and without adjustment for
moisture loss due to drying. The moisture-loss-adjusted conversion was used for most species.
However, for certain species (noted in Table B6) it was assumed that the respondent utilized the
jerky model to describe consumption due to the visual appearance of the model rather than to
imply it was consumed in a dried form. In those cases, the conversion without moisture loss
adjustment was used.
Table B4. Portion-to-mass (raw weight, edible portion) conversions for the salmon replica
with fillet divided into sections
Fillet Section
Number
Portion-to-Mass
(grams)
Fillet Section
Number
Portion-to-Mass
(grams)
1
50
13
192
2
80
14
180
3
92
15
178
4
112
16
162
5
124
17
170
6
132
18
138
7
176
19
124
8
190
20
110
9
174
21
88
10
170
22
88
11
178
23
66
12
176
24
42
Nez Perce Tribe
Volume III - Appendix B
December 2016 Page B-20
-------
Table B5. Portion-to-mass (raw weight, edible portion) conversions for other models
Model
1 nil
Port ion-lo-Mass («rams):-:
Trout replica
1 fillet
113.4
Measuring bowls (for soup, stew,
etc.)**
1 cup
72.2
Measuring bowls (for fish volume)**
1 cup
302.4
Lamprey
1 serving
113.2
Crayfish
1 organism
7.2
Mussel
1 organism
10.0
Shrimp
1 organism
44.0
Can
1 5 oz can***
302.4
Jar
1 8 oz jar***
189.0
*Values rounded to 1 decimal digit for display although 4 decimal digits were used for calculations to avoid
accumulating rounding errors;
**The 72.2 grams conversion factor was used when the respondent described consumption using the measuring
bowl and either 1) specified the preparation as soup or stew (24-hour recall only) or 2) the species being described
was clams, mussels or lamprey (FFQ only); this factor assumed only a portion of the volume was fish; otherwise, the
302.4 grams factor was used, which assumed the entire volume was fish (see section 13 of this appendix);
***The conversion factor was adjusted proportionally if a non-standard size was specified (i.e., not 5 oz. or 8 oz.) as
described in the Portion-to-mass conversions for cans and jars section below.
Nez Perce Tribe
Volume III - Appendix B
December 2016 Page B-21
-------
Table B6. Portion-to-mass (raw weight, edible portion) conversions for jerky, depending on
the jerky model and species
Portion-to-Mass (grams)*
Jerky
With Moisture Loss Adjustment (Species
Without Moisture Loss Adjustment
Model
Group A)
(Species Group B)
J1
163.5
70.3
J2
172.8
74.3
J3
168.1
72.3
J4
163.5
70.3
J5
163.5
70.3
J6
158.8
68.3
J7
168.1
72.3
J8
163.5
70.3
J9
186.7
80.3
J10
196.0
84.3
Jll
191.4
82.3
Group A contains all salmon, steelhead, freshwater finfish, cod, halibut, pollock, and other marine finfish not in
group B;
Group B contains all freshwater and marine shellfish tuna and sardines;
See Table B3 for moisture loss adjustment factors;
*Values rounded to 1 decimal digit for display although 4 decimal digits were used for calculations to avoid
accumulating rounding errors.
11.0 Write-in Species Corrections and Mapping
In CAP I, several general species categories allowed the respondent to describe consumption of
specific but unlisted species, such as pink salmon or oysters. These species categories include
other salmon, other trout, other freshwater finfish, other marine fish or shellfish, and other fish or
shellfish. In each case, the interviewer was able to write in the name of the specific species.
Because these write-in fields allowed unrestricted free text, there were occasional spelling
variations and instances where a listed species (e.g., tuna) was written in or a write-in species
belonged in a more specific species category. For example, marine clams or mussels would be a
more specific category for a write-in of butter clams rather than "other marine fish and shellfish."
All write-in text instances were examined manually to correct for spelling variation and remap to
a more specific CAPI species category when needed. These changes, which were made in
consultation with Ridolfi staff, facilitated species-specific portion-to-mass conversions and
species grouping for reporting.
Nez Perce Tribe
Volume III - Appendix B
December 2016 Page B-22
-------
12.0 Portion-to-Mass Conversions for Soup Bowls
The soup bowls were originally intended to be used only for specifying soups, stews, or other
composite dishes where the fish was only a component of the total volume; however, during the
course of interviewing it was found that respondents more often used this model to describe the
volume of fish they consumed, not including other non-fish components. This was particularly
common for tuna, crab and lobster meat and small shrimp, the latter being difficult to count
individually, as would be required to utilize the shrimp model. In contrast, clams or mussels were
most often consumed and described as soups.
Whether the respondent intended the soup bowl to refer to A) the total volume of a composite
dish or B) only to the volume of fish contained in the dish was not recorded by the interviewer.
However, through discussions with the interviewer supervisor (who performed and observed a
number of interviews) and some of the interviewers who performed a large number of
interviews, it was determined which species were most commonly described as type A or type B.
The type A species (fish was a component of soup or stew) were determined to be freshwater
clams or mussels, marine clams or mussels and lamprey. All other species were type B.
When performing the mass conversions for the FFQ interviews, where a preparation method was
not recorded, type A species described using bowls were converted using 72.2 grams per 1 cup
bowl (see Figure B5 of this appendix). Type B species were converted using 302.4 grams per 1
cup bowl. This conversion assumed a 25% moisture loss, the same factor assumed for canned
fish or fish cooked with a dry heat (Table B3).
However, when performing the mass conversions for the 24-hour recall, the 72.2 grams per 1 cup
bowl conversion (type A) was used only when the preparation was noted as soup or stew,
regardless of species. The 302.4 grams per 1 cup bowl conversion (type B) was used for all other
preparations, including casserole or mixed dish (a single category). This preparation was most
often used to refer to the final form of the dish rather than how the respondent described the
portion size. For example, a tuna fish sandwich or shrimp salad would be described as a mixed
dish, but the soup bowl model was used to describe the amount of tuna or shrimp included
instead of the total volume of the final dish. This is the only aspect of the portion-to-mass
conversions which differed between the 24-hour recall and FFQ.
13.0 Portion-to-Mass Conversions for Cans and Jars
When respondents provided portion sizes in terms of cans or jars, the interviewer had a text field
in which to capture specific descriptions. Unless otherwise specified, cans were assumed to be 5-
oz. and jars 8-oz. In consultation with Ridolfi, an algorithm was developed which utilizes the
species and text description field to determine the most appropriate portion-to-mass conversion.
The steps of the algorithm are as follows:
1. If an unambiguous container size could be determined from the text field (e.g., 6 oz., 1
qt., 1 cup), this size was used for the conversion.
2. Otherwise, if the text field contained the string "can" and did not contain "jar" (which
would create an ambiguity), then 5 oz. was assumed.
3. If the text field contained the string "jar" but not "can," then 8 oz. was assumed.
4. Finally, if a size could not be determined by steps 1-3, a default was assumed based on
the species. For all freshwater species, cod, halibut, and pollock, 8 oz. was assumed. For
the remaining marine species, 5 oz. was assumed.
Nez Perce Tribe
Volume III - Appendix B
December 2016 Page B-23
-------
14.0 Portion-to-Mass Conversions for Number of Shellfish
When reporting consumption of shellfish, the respondent had the option of specifying the
number of organisms. There were three portion models for this purpose: crayfish, mussels, and
shrimp, each with different portion-to-mass conversion factors. In November 2014, a field was
added to CAPI to allow the interviewer to record which model was used. Due to restrictions in
CAPI, this was implemented as a text field and the interviewer was instructed to use "C" for
crayfish, "M" for mussels, and "S" for shrimp. However, the text field also allowed other text,
and an algorithm was developed in consultation with Ridolfi staff to examine the model text field
and the species field to determine the most appropriate model for mass conversion. The
procedure used is:
1. For any clams or mussels species, "mussels" was chosen regardless of the shellfish model
recorded.
2. For other species, if a valid shellfish model code (C, M, S) could be determined from the
text field, that model was chosen.
3. If a valid shellfish model could not be determined, Table B7 was used to choose the
likely model used:
Table B7. Choice of shellfish model when not specified by the interviewer
Species in CAPI
Chosen
Shellfish Model
Crayfish, lobster, crab
Crayfish
Freshwater clams or mussels, marine clams or mussels, oysters, scallops
Mussels
Shrimp, prawns, squid, octopus
Shrimp
15.0 Exceptions to the Portion-to-Mass Conversion Procedure
Two records that did not follow the expected protocol were manually modified to perform the
mass conversion. These are described below:
1. One respondent reported shark consumption in a higher consumption period and a lower
consumption period. The respondent reported consuming shark once per year in the
higher period and 0 times per year in the lower period, but did not provide the duration of
the higher period. This was manually converted into once per year as a single period
instead of a higher and lower period. The standard portion-to-mass conversion procedure
was then applied to the modified record.
2. One respondent reported consuming alligator as 2 soup bowls per year. This response
was excluded because the alligator is neither a finfish nor a shellfish.
Nez Perce Tribe
Volume III - Appendix B
December 2016 Page B-24
-------
Appendix C—
Additional Detail on Imputations
Nez Perce Tribe
Volume III - Appendix C
December 2016 Page C-l
-------
Appendix C—Additional Detail on Imputations
1.0 Grouping of Species for Imputation of Uncommon Responses
As described in Section 5.28 of the main body of this report, when there was a component
missing which was needed to calculate a species-specific consumption rate (portion frequency,
portion size or higher consumption period percentage of the year), similar non-missing responses
were used to estimate a mean value for imputation. To be considered similar, a response needed
to be for the same species and have the same period type (the types were: whole year, higher
consumption period or lower consumption period). This rule was used when the number of
similar responses was at least 5. When the number was less than 5, species were grouped to
expand the number of similar responses on a case-by-case basis, as described in Table CI. In
general, the choice of groupings was restrictive and based on consultation with Ridolfi staff.
When high-consumption period percentage was being imputed, the grouping was less restrictive
than for size and frequency because the number of available responses was smaller and because
the majority of responses were in the range of 8%—33% (1-4 months) across all species. As the
sensitivity analysis in the next section shows, the final results are similar under a wide range of
imputed values, so the precise value used for the imputation is not critical.
Table CI. Nez Perce Tribe. Species groupings used to impute missing values for uncommon
species (less than 5 non-missing responses)
Species in CAPI
Missing
Field
No.
Imputed
Species group used
for Imputation
Other salmon*
Portion
frequency
1
Other salmon,* Kokanee, Sockeye, which are
less commonly consumed salmon species
Other salmon*
Higher period
percentage
1
Other salmon,* Kokanee, Sockeye, which are
less commonly consumed salmon species
All trout or
unspecified
Higher period
percentage
1
All resident trout species/groups
Freshwater clams
or mussels
Higher period
percentage
2
All freshwater or marine shellfish species
Lobster
Higher period
percentage
3
All freshwater or marine shellfish species
*Other salmon is a species category in CAPI that allowed for a specific salmon species not listed to be written in
most commonly pink or Atlantic salmon.
Nez Perce Tribe
Volume III - Appendix C
December 2016 Page C-2
-------
2.0 Sensitivity Analysis on Imputations
The impact of imputing missing values in calculating consumption rates was explored by
recomputing rates under two extreme approaches: imputing 0 for all missing values, which
would systematically underestimate consumption, and imputing twice the mean value (based on
the same species), which in many cases would overestimate consumption. Consumption rates
based on alternative imputations for Groups 1-6 are shown in Tables C2-C7, respectively. There
was usually little or no difference in the estimates based on the extreme imputation approaches
compared to the imputation approach used in the report (imputing the mean value from the same
species), with the differences ranging from 0.0-2.6% except for the 90th percentile of Group 5
(Table C6) had a difference of 18.4% between the mean approach and the twice the mean
approach. The mean approach is likely to be much more accurate than twice the mean and the
difference of 18.4% is not very large compared to the upper bound of the 95% CI (120% higher
than the point estimate). These results show that imputation of missing values had a minimal
impact on the final consumption rates presented in this report.
Table C2. Nez Perce Tribe. Sensitivity analysis of imputation method on the Group 1 FCRs
(g/day, raw weight, edible portion). Estimates are weighted
Imputation Method
Zero*
Mean**
(used in report)
High***
No. of consumers
451
451
451
Mean
122.1
123.4
123.9
50th percentile
70.2
70.5
71.2
90th percentile
263.8
270.1
270.9
95th percentile
437.3
437.4
437.6
Max
1371.9
1371.9
1371.9
*A11 missing values were assigned the value 0;
**A11 missing values were assigned the mean value from the same species;
***A11 missing values were assigned twice the mean value from the same species.
Nez Perce Tribe
Volume III - Appendix C
December 2016 Page C-3
-------
Table C3. Nez Perce Tribe. Sensitivity analysis of imputation method on the Group 2 FCRs
(g/day, raw weight, edible portion). Estimates are weighted.
Imputation Method
Zero*
Mean**
(used in report)
High***
No. of consumers
446
446
446
Mean
102.8
104.0
104.5
50th percentile
60.1
61.3
62.9
90th percentile
229.5
231.4
233.7
95th percentile
321.8
327.9
326.9
Max
1323.8
1323.8
1323.8
*A11 missing values were assigned the value 0;
**A11 missing values were assigned the mean value from the same species;
***A11 missing values were assigned twice the mean value from the same species.
Table C4. Nez Perce Tribe. Sensitivity analysis of imputation method on the Group 3 FCRs
(g/day, raw weight, edible portion). Estimates are weighted.
Imputation Method
Zero*
Mean:-::-:
(used in report)
Nigh***
No. of consumers
446
446
446
Mean
77.9
79.0
79.4
50th percentile
45.2
45.2
45.8
90th percentile
166.1
166.1
167.1
95th percentile
244.8
247.3
247.3
Max
949.8
949.8
949.8
*A11 missing values were assigned the value 0;
**A11 missing values were assigned the mean value from the same species;
***A11 missing values were assigned twice the mean value from the same species.
Nez Perce Tribe
Volume III - Appendix C
December 2016 Page C-4
-------
Table C5. Nez Perce Tribe. Sensitivity analysis of imputation method on the Group 4 FCRs
(g/day, raw weight, edible portion). Estimates are weighted.
Imputation Method
Zero"
Mean**
(used in report)
High***
No. of consumers
136
136
136
Mean
13.5
13.5
13.5
50th percentile
3.8
3.8
3.8
90th percentile
26.3
26.3
26.3
95th percentile
56.8
56.8
56.8
Max
544.2
544.2
544.2
*A11 missing values were assigned the value 0;
**A11 missing values were assigned the mean value from the same species;
***A11 missing values were assigned twice the mean value from the same species.
Table C6. Nez Perce Tribe. Sensitivity analysis of imputation method on the Group 5 FCRs
(g/day, raw weight, edible portion). Estimates are weighted.
Imputation Method
Zero*
Mean**
(used in report)
Nigh***
No. of consumers
150
150
150
Mean
14.0
14.3
14.6
50th percentile
3.7
3.7
3.7
90th percentile
34.2
34.2
40.5
95th percentile
75.9
75.9
75.9
Max
309.5
309.5
309.5
*A11 missing values were assigned the value 0;
**A11 missing values were assigned the mean value from the same species;
***A11 missing values were assigned twice the mean value from the same species.
Nez Perce Tribe
Volume III - Appendix C
December 2016 Page C-5
-------
Table CI. Nez Perce Tribe. Sensitivity analysis of imputation method on the Group 6 FCRs
(g/day, raw weight, edible portion). Estimates are weighted.
Imputation Method
Zero*
Mean**
(used in report)
Nigh---
No. of consumers
308
308
308
Mean
50.8
51.0
51.1
50th percentile
29.8
29.8
29.8
90th percentile
93.3
93.3
93.3
95th percentile
155.4
155.4
155.4
Max
731.8
731.8
731.8
*A11 missing values were assigned the value 0;
**A11 missing values were assigned the mean value from the same species;
***A11 missing values were assigned twice the mean value from the same species.
Nez Perce Tribe
Volume III - Appendix C
December 2016 Page C-6
-------
Appendix D—
Additional Detailed Tables and
Methodologic Notes
Nez Perce Tribe
Volume III - Appendix D
December 2016 Page D-l
-------
Appendix D—Additional Detailed Tables and
Methodologic Notes
The tables in this appendix supplement tables included in Volume II. Table D1 summarizes
demographics in the original population of eligible Tribal members based on the enrollment
records, in the sample drawn from that population, and the final sample of consumers based on
the responses to the FFQ portion of the questionnaire. All of these estimates are unweighted.
There were some differences in demographic distributions between the original population, the
list of tribal members designated to be included in the sample and the consumers about whom
various analyses are presented in the report tables. This illustrates why the survey weights were
used throughout the analyses presented in this report, as the weights are designed to account for
these differences and produce estimates which are representative of the tribal population from
which the sample was drawn. Weighting is discussed.in Section 5.20 of Volume II.
Table Dl. Nez Perce Tribe. Demographics of the population, selected sample and first
interview consumers with known consumption rates. Estimates are unweighted.
Population
(N=I574)
Sample
(N=l 250)
HQ
Consumer"
(M=45l)
Variable
At
No.
At
Mo.
At
Mo.
Gender
Male
48.2%
758
48.1%
601
53.4%
241
Female
51.8%
816
51.9%
649
46.6%
210
Age
18-29 years
23.4%
369
23.4%
293
13.5%
61
30-39 years
19.4%
305
19.4%
242
20.8%
94
40-49 years
18.8%
296
18.8%
235
25.7%
116
50-59 years
18.0%
283
18.0%
225
19.7%
89
60 years or older
20.4%
321
20.4%
255
20.2%
91
Documented fisher
Yes
23.6%
371
23.0%
288
30.6%
138
No
76.4%
1203
77.0%
962
69.4%
313
Zip code
83540
57.6%
906
58.3%
729
73.0%
329
83536
12.4%
196
12.1%
151
8.6%
39
83501
10.9%
172
10.9%
136
6.2%
28
Other
19.1%
300
18.7%
234
12.2%
55
*Includes those who completed the first interview and have a calculable non-zero FFQ consumption rate.
Nez Perce Tribe
Volume III - Appendix D
December 2016 Page D-2
-------
Table D2 presents unweighted estimates of demographics among consumers, analogous to Table
7 in Volume II, which presents weighted estimates.
Table D2. Nez Perce Tribe. Demographics of the FFQ consumers with known consumption
rates. Estimates are unweighted.
V-ii or
Moan ± SI)
No. w ho
Responded
Gender*
Male
53.4%
451
Female
46.6%
Age*
18-29 years
13.5%
451
30-39 years
20.8%
40-49 years
25.7%
50-59 years
19.7%
60 years or older
20.2%
Weight, kgs
89.9 ± 19.5
434
Weight, kgs (males only)
95.9 ± 18.8
239
Weight, kgs (females only)
82.6 ± 17.8
195
No. in household
1
8.2%
451
2
18.6%
3-4
42.8%
5 or more
30.4%
Documented fisher*
Yes
30.6%
451
No
69.4%
Live on reservation
Yes
87.1%
449
No
12.9%
Highest education
Middle school
1.6%
448
High School / GED
52.5%
Associates degree
25.7%
Bachelor's degree
14.1%
Master's degree
5.6%
Doctorate
0.7%
Annual household income
< $15K
19.3%
410
$15K - $25K
20.7%
$25K - $35K
19.8%
$35K - $45K
12.9%
$45K - $55K
8.3%
$55K - $65K
5.6%
>$65K
13.4%
*From the enrollment list or fisher indicator list; other demographics were determined from the questionnaire.
Annual fish consumption rates based on the FFQ for various demographic groups are
summarized in detail in Table D3. This expanded version of Table 9 in Volume II shows more
percentiles.
Nez Perce Tribe
Volume III - Appendix D
December 2016 Page D-3
-------
Table D3. Nez Perce Tribe. Estimated distribution of FCRs (g/day, raw weight, edible
portion) of consumers within demographic groups. All rates are for total consumption
(Group 1). Estimates are weighted. Mean, SD, median ('50%') and percentiles.
Nil. ill
lYrcenliles
(¦niiip
('IIIISIIIIKTS'-
Mean
SI)
50",,
55",,
r.o"„
f,5"„
70",,
75%
SO",,
N5"„
•>0"„
•>5"„
(Jender'- '¦
Male
241
146.6
179.3
87.4
101.0
1 19.7
133.3
148.8
168.7
191.3
223.6
285.1
488.3
Female
210
100.2
133.1
54.7
64.1
70.9
81.8
105.3
116.3
133.1
155.1
244.0
341.4
Age**
18-29 years
61
126.7
175.4
74.7
84.2
102.0
109.5
123.1
126.2
148.3
190.7
225.2
522.4
30-39 years
94
140.9
161.1
74.0
84.9
112.8
132.7
148.1
164.8
185.9
243.2
298.9
448.6
40-49 years
116
115.4
126.1
68.5
78.2
85.9
95.5
114.8
130.8
150.4
194.5
241.2
463.3
50-59 years
89
130.3
193.4
67.4
76.6
88.6
107.2
136.1
150.0
184.6
212.8
253.8
308.2
60 years or
older
91
105.8
136.8
62.3
71.0
76.4
101.9
113.2
129.4
143.9
207.5
264.8
332.0
Documented
Fisher**
Yes
138
171.8
207.2
98.0
108.6
125.7
156.1
177.5
203.2
228.5
283.9
436.8
543.5
No
313
107.9
137.5
65.5
70.6
81.1
98.8
113.4
132.5
147.7
178.1
232.9
337.7
Live on
reservation
Yes
391
127.3
164.4
70.6
77.9
88.9
112.4
131.4
148.4
177.6
222.5
284.6
451.0
No
58
106.5
134.4
65.6
95.0
106.2
108.9
110.8
123.2
128.1
152.7
202.8
237.5
Number who
live in
household
1
37
133.9
179.3
82.0
93.0
108.9
113.8
131.8
135.9
147.9
243.1
288.3
***423.0
2
84
119.0
144.1
57.2
65.5
82.1
103.5
136.3
179.1
202.4
240.0
285.3
451.5
3-4
193
119.3
163.7
71.0
78.4
88.0
105.7
117.1
125.9
144.9
175.8
224.3
441.0
5 or more
137
129.2
158.0
74.0
83.1
100.0
113.7
133.1
155.8
176.3
201.1
284.0
381.1
Highest
education
High school /
GED or less
242
126.6
176.5
70.4
79.5
96.9
109.1
123.0
134.8
156.2
190.8
253.9
492.0
Associates
degree or
higher
206
120.4
136.5
70.7
78.6
89.8
113.3
134.3
151.9
185.3
211.0
275.0
409.0
Annual
household
income
< $15K
79
122.9
168.7
69.7
74.0
97.2
105.8
125.8
135.4
159.9
204.0
282.4
324.9
$15K- $45K
219
126.6
165.9
71.1
79.3
89.4
107.1
121.2
136.4
156.7
208.5
250.8
488.7
>$45K
112
117.7
113.5
72.4
78.9
97.1
122.8
135.9
155.7
174.1
215.5
244.8
339.6
*Consumers with unknown or missing subgroup status were excluded for the analysis of that subgroup;
**From the enrollment list or fisher indicator list; other subgroups were determined from the questionnaire;
***Two or fewer expected respondents with rates equal or greater than the reported percentile (approximately);
interpret this percentile more cautiously.
Nez Perce Tribe
Volume III - Appendix D
December 2016 Page D-4
-------
Some sampled respondents lived together in the same household (a statistical cluster). As
described in Section 5.25 of Volume II on confidence interval calculations, this clustering was
ignored in the calculations, as the number of clusters was small and likely to have minimal
impact on estimates and the precision of estimates. All known clusters are enumerated in Table
D4 to facilitate future analyses which may utilize the clustering information.
Table D4. Nez Perce Tribe. Enumeration of household clusters. Respondent IDs within
each cluster are comma separated. See section 5.25 on confidence intervals for a discussion
on impact.
( luster II)
P.MU IDs
1
E1AIO, EAIT1
2
E1P63, ESFBV
3
E33P9, EM176
4
E3P73, E063E
5
E3XBE, EJ9K1
6
E4NEO, EREES
7
E58XO, EEMQQ
8
E5RHK, EQ8BI
9
E65M, EB452
10
E6CQ2, E6P1W
11
E6PAI, ET8FX
12
E6YG0, ECODT
13
E7EJ6, ECOUR
14
E8AMB, ESM4S
15
E8HEK, EXY46
16
E8RLC, EDPQA
17
EA4VL, EIOXT
18
EB478, EVD86
19
EBT5B, EGRJP
20
EC8V1, EFQQ4
21
EESW7, EYSWS
22
EFE4A, EWQB2
23
EH21Q, ESDK7
24
EHAKO, EMWSN
Nez Perce Tribe
Volume III - Appendix D
December 2016 Page D-5
-------
( luster II)
P.MU IDs
25
EOIID, EV2MI
26
EPULA, EZRSR
27
ETTSY, EWQ7T
28
El 1X9, E6HY0, EOL5S
29
E1Q8I, ETWDT, EX2ND
30
E20JH, E5LMF, EJ2V7
31
E40TM, E6URJ, EQBA2
32
EB3TX, EQ3Y5, EZE8V
33
EDKUW, EDVWP, EIDIO
34
EQGKA, ER9Y3, EY40I
35
E389E, E4WM7, ELXWI, EPIWP, EU22B
Sample size and expected number of double hits. A planning exercise to support the NCI
method.
In this section, the expected counts of fish consumption in two 24-hour recall periods ("double
hits") are calculated using various assumptions on the frequency of fish consumption. Of
particular interest is the expected number of individuals who consume fish in each of two 24-
hour recall interviews. The fish consumption rates from the CRITFC report are used (see
reference below the second table, below), which gives the fraction of the population that
consumes various numbers of fish meals per week.
Table 5, on page 77 of the CRITFC report, gives the estimated number of fish meals per week.
However, the probability of fish consumption on a randomly chosen day is required in order to
calculate the expected number of double hits. To account for the possibility of multiple meals
being consumed on the same day (e.g., a person who consumes two fish meals in one week may
consume both on the same day), several alternative methods were used to calculate the
probability of fish consumption:
1) Method 1: Assume each meal was consumed on a separate day. That is, estimate the
probability of fish consumption as "number of fish meals per week'77. Those who
consumed 7 or more meals per week were assumed to consume fish every day.
2) Method 2: Divide the number of meals per week by 2, for those who eat 1 or more fish
meals per week, and then implement Method 1 on the modified (weighted) percentages.
Using this method, someone who consumes 2 fish meals per week would have a 1 in 7
chance of consuming fish on a particular day, while someone who consumes fish once
every 2 weeks (i.e., less than one fish meal per week) would still have a 1/14 chance of
fish consumption on a randomly chosen day, as in Method 1.
3) Method 3: Divide the number of meals for those who eat 2 or more fish meals per week
by two, and then implement Method 1 on the modified counts.
4) Method 4: Divide the number of meals for those who eat 4 or more fish meals per week
by two, and then implement Method 1 on the modified counts.
Nez Perce Tribe
Volume III - Appendix D
December 2016 Page D-6
-------
For a given consumption category (e.g., those who consume 1 meal per week), the probability of
fish consumption on two separate days can be calculated, assuming consumption is independent
between the days. If this probability is labeled pj, the probability that a randomly sampled person
from the population consumes fish in each of two independent 24-hour recall periods is then a
weighted average of these pj, where the pj is weighted by the fraction of the population which
they represent.
Two methods of sampling individuals were considered:
a) No over-sampling: Take a random sample of fish consumers.
b) Over-sampling: Sample those who consume fish 2 or more times per week at twice the
rate of the rest of the population.
Over-sampling is intended to increase the number of respondents who report eating fish during
each of two 24-hour recall periods.
In summary, four methods are presented for estimating the probability of fish consumption on a
particular day for individuals in the population, and two ways of sampling individuals from the
population are presented. For a given sample size, this gives us 8 estimates of the expected
number of individuals who eat fish in both 24-hour recall periods ("double hits"). These
estimates are given in the following table, along with a 95% lower bound on the expected
number in parentheses.
Nez Perce Tribe
Volume III - Appendix D
December 2016 Page D-7
-------
Table D5. Expected number of "double-hits" for two independent interviews based on the
noted sample size of respondents and two different sampling methods.
Methodl
Method2
Method3
Method 4
Sample
Size
random
sample
over
sample
random
sample
over
sample
random
sample
over
sample
random
sample
over
sample
100
10(4)
13(6)
4(0)
5(1)
6(1)
7(2)
7(2)
9(3)
200
20(11)
27 (17)
7(2)
10 (4)
11(5)
15(7)
13(6)
17(9)
300
30(19)
40 (28)
11(4)
15(7)
17(9)
22(13)
20(11)
26 (16)
400
40 (27)
54 (40)
14(7)
20 (11)
23 (13)
30(19)
26(16)
34 (23)
500
49 (36)
67 (51)
18(9)
24(15)
28(18)
37 (25)
33 (21)
43 (30)
600
59 (44)
81(63)
21 (12)
29 (19)
34 (23)
45 (32)
39 (27)
52 (38)
700
69 (53)
94 (75)
25(15)
34 (23)
40 (28)
52 (38)
46 (32)
60 (45)
800
79 (62)
108 (87)
28(18)
39 (27)
46 (32)
60 (44)
52 (38)
69 (53)
900
89 (70)
121 (100)
32(21)
44 (31)
51 (37)
67 (51)
59 (44)
77 (60)
1000
99 (79)
135 (112)
35 (24)
49 (35)
57 (42)
75 (58)
65 (49)
86 (68)
1100
109 (88)
148 (124)
39 (27)
54 (39)
63 (47)
82 (64)
72 (55)
95 (76)
1200
119(97)
162 (137)
42 (30)
59 (44)
68 (52)
89 (71)
78 (61)
103 (83)
1300
128 (106)
175 (149)
46 (33)
63 (48)
74 (57)
97 (78)
85 (67)
112(91)
1400
138(115)
189 (162)
49 (36)
68 (52)
80 (62)
104 (84)
91 (72)
121 (99)
1500
148 (124)
202 (174)
53 (39)
73 (56)
85 (67)
112(91)
98 (78)
129 (107)
1600
158 (134)
216 (187)
57 (42)
78 (61)
91(72)
119(98)
104 (84)
138 (115)
1700
168 (143)
229 (199)
60 (45)
83 (65)
97 (78)
127 (105)
111 (90)
146 (123)
1800
178 (152)
243 (212)
64 (48)
88 (69)
103 (83)
134(111)
117(96)
155 (131)
1900
188(161)
256 (225)
67 (51)
93 (74)
108 (88)
142(118)
124 (102)
164 (138)
2000
198 (170)
270 (237)
71 (54)
98 (78)
114 (93)
149 (125)
130 (108)
172 (146)
Technical Notes
In this report, self-reported survey data collected in 1994 were used from the Yakama, Warm
Springs, Umatilla or Nez Perce Tribes. It is implicitly assumed that: i.) the fish consumption
rates in this historical population are similar to those in our target population; and ii.) the
respondents accurately reported consumption frequencies. Fish consumption patterns may vary
both by population and over time. Also, the survey suggests significant recall bias. For example,
consumption once every week was much more common than once every 6 days or once every 8
days. It is also possible that fish consumption varies widely by season, and that the rates in the
CRITFC report may be averaged over several seasons.
In obtaining the lower bound for counts of "double-hits", it was assumed that the counts were
Poisson-distributed. With this approximation, the standard deviation (SD) of a count is the
square-root of the count. The 95% lower confidence bound was then estimated, using a normal
approximation, as "count - 1.96*SD." In reality, heterogeneity in the fish consumption
categories may make this assumption unrealistic, making the reported lower bound approximate
to some degree.
Nez Perce Tribe
Volume III - Appendix D
December 2016 Page D-8
-------
Table D6. Number of fish meals consumed by all adult respondents (fish consumers and
non-fish consumers) per week - throughout the year.
.NumlKTiif
Ml-.lls |KT
\\ ll'k
l nwii^likil
l"lV(|lll-|HA
IVrii-nl
CiiiiiuhiliM'
IVlTUll
Number nl"
Mwils |KT
\\ ilk
l'lV(|lll'IH\
IVnviil
CiiiiiuhiliM'
IVrivnl
0.0
46
8.9%
8.9%
4
16
4.8%
95.5%
0.1
5
0.5%
9.4%
5
4
0.8%
96.2%
0.2
24
3.0%
12.4%
6
3
0.5%
96.7%
0.3
3
0.3%
12.7%
7
2
0.8%
97.6%
0.4
24
2.6%
15.3%
8
2
0.2%
97.8%
0.5
28
3.9%
19.2%
9
1
0.1%
97.9%
0.6
9
1.0%
20.2%
10
4
0.9%
98.8%
0.8
1
0.1%
20.3%
12
2
0.3%
99.1%
1.0
203
43.8%
64.1%
15
3
0.4%
99.6%
1.2
1
0.1%
64.2%
20
1
0.1%
99.7%
1.9
1
0.1%
64.3%
24
1
0.1%
99.9%
2.0
90
21.0%
85.4%
30
1
0.1%
100%
3.0
25
5.3%
90.7%
Total
500
100%
From Table 5, page 77, CRITFC report (Columbia River Inter-Tribal Fish Commission, "A Fish Consumption
Survey of the Umatilla, Nez Perce, Yakama, and Warm Springs Tribes of the Columbia River Basin." Technical
Report 94-3. Portland, Oregon. 1994). Used with permission.
Nez Perce Tribe
Volume III - Appendix D
December 2016 Page D-9
-------
Appendix E—
Expanded Tables and Additional Notes
on the NCI Method
Nez Perce Tribe
Volume III - Appendix E
December 2016 Page E-l
-------
Appendix E—Expanded Tables and Additional Notes on the NCI Method
The tables in this section provide additional percentiles and other statistics of fish consumption rates based on the NCI method.
Selected values in these tables have been presented in the Results section of Volume II, in particular Sections 6.8-6.10.
Table El. Nez Perce Tribe. Distribution of the usual fish Group 1 (all fish) consumption (g/day, raw weight, edible portion)
based on the 24 hour recalls. Estimated by the NCI method.
No. of
Consumers
Mean
Percentiles
5%
10%
15%
20%
25%
30%
35%
40%
45%
50%
55%
60%
65%
70%
75%
80%
85%
90%
95%
Overall
451
75.0
7.3
11.2
15.1
19.2
23.5
27.8
32.5
37.7
43.3
49.5
56.4
64.6
73.9
85.1
98.9
115.7
138.5
173.2
232.1
Documented fisher
Fisher
138
98.2
9.4
14.8
20.1
25.2
30.9
36.7
42.7
49.3
56.3
64.7
74.3
85.2
97.9
113.2
130.4
154.1
184.1
229.2
305.0
Non-fisher
313
67.6
6.8
10.5
14.0
17.7
21.7
25.8
30.2
34.6
39.9
45.6
52.0
59.2
67.9
77.6
90.0
104.9
124.6
155.1
206.0
Gender
Men
241
87.7
9.1
14.0
18.8
23.6
28.4
33.4
39.1
44.8
51.3
58.4
66.7
76.3
87.2
99.8
115.3
134.1
161.9
199.8
268.1
Women
210
62.3
6.1
9.5
12.5
15.9
19.5
23.5
27.3
31.7
36.5
41.8
47.7
54.4
62.4
71.6
82.8
97.7
116.0
145.1
194.4
ZIP Code
83540
329
73.6
7.0
10.9
14.7
18.7
22.8
27.2
31.8
36.9
42.3
48.2
55.1
62.7
72.1
83.2
96.4
113.1
135.5
168.1
227.2
83536
39
84.5
8.7
13.1
17.6
23.0
27.8
32.8
38.5
44.2
50.8
58.1
67.4
77.4
88.9
101.5
117.6
136.2
164.2
197.9
246.9
83501
28
63.6
7.4
11.2
14.8
19.3
23.7
27.7
32.4
37.0
42.3
48.4
54.5
60.8
67.9
75.2
85.6
98.4
115.8
139.4
177.7
NP Other
55
79.8
7.2
11.0
15.0
19.0
23.2
26.9
31.7
36.8
42.6
49.2
56.8
65.9
76.5
88.8
102.7
120.7
148.8
193.8
264.2
Age
18-29
61
75.3
8.4
12.4
17.0
21.4
25.8
30.7
35.1
40.5
46.5
52.0
58.6
66.1
74.7
85.5
97.8
114.3
137.0
170.1
232.5
30-39
94
92.5
10.8
16.5
21.8
27.2
31.8
37.2
43.0
49.4
56.2
64.5
73.1
83.1
94.9
108.5
124.4
143.7
171.2
207.7
274.2
40-49
116
83.8
9.3
13.7
18.1
22.9
27.2
32.2
37.9
43.5
49.9
56.6
64.0
73.1
83.6
97.4
112.5
129.9
157.0
192.6
256.3
50-59
89
66.8
5.8
9.1
12.3
15.4
19.0
22.6
26.5
30.8
35.8
41.2
46.8
54.0
62.0
71.4
83.3
98.0
118.4
151.4
212.7
60+
91
58.1
5.4
8.2
11.0
13.8
16.9
20.5
24.1
28.3
33.0
37.7
43.0
49.6
57.3
67.6
77.7
92.9
110.5
136.5
182.5
Nez Perce Tribe
Volume III - Appendix E
December 2016 Page E-2
-------
Table E2. Nez Perce Tribe. Distribution of the usual fish Group 2 consumption (g/day, raw weight, edible portion) based on
the 24 hour recalls. Estimated by the NCI method.
No. of
Consumers
Mean
Porconlik's
5"i.
iir;.
15"-;.
20V »
25':;.
30"';.
35V--;.
4ov;.
45";.
5ov--;.
55"-;.
(.0"-;.
(.5v;.
",ov;.
75%
so1:;.
S5V-;.
«>0V'i.
<)5"'i.
Overall
446
66.5
4.1
6.8
9.4
12.2
15.1
18.3
21.9
26.1
30.8
36.0
42.1
49.5
58.0
68.7
81.7
98.2
121.8
159.4
233.9
Documented fisher
Fisher
138
98.4
7.3
11.6
15.8
20.0
24.6
29.8
35.1
40.8
47.7
55.2
64.8
75.4
86.3
101.8
121.9
146.9
181.5
238.6
345.0
Non-fisher
308
55.6
3.9
6.4
OO
OO
11.2
13.8
16.6
19.7
23.2
27.4
32.0
37.0
43.2
50.8
59.4
70.6
84.1
102.2
132.0
189.5
Gender
Men
240
79.4
5.3
8.7
11.8
15.2
18.7
22.7
27.2
32.2
37.5
44.0
51.4
60.1
70.3
81.8
96.4
116.7
144.6
190.4
277.1
Women
206
55.0
3.0
5.1
7.2
9.4
11.8
14.4
17.4
20.5
24.5
29.0
34.0
39.8
47.5
56.3
67.9
82.7
102.8
135.6
198.0
ZIP Code
83540
326
65.5
3.8
6.3
OO
OO
11.4
14.3
17.4
20.8
24.8
29.6
34.7
40.6
48.2
56.7
67.0
80.2
97.0
120.7
158.4
232.3
83536
38
83.7
4.8
8.0
11.0
14.6
18.2
22.7
27.9
33.3
39.7
46.6
54.8
63.8
74.8
88.9
104.3
129.6
162.4
219.2
301.5
83501
27
64.0
5.1
8.4
11.7
15.0
18.6
22.5
26.5
31.0
36.0
41.6
48.0
54.3
64.6
75.6
87.6
104.8
123.3
150.6
197.4
NP Other
55
63.0
3.8
6.3
8.5
10.8
13.1
15.9
19.2
22.4
26.1
30.2
36.4
43.0
51.3
60.0
72.2
87.9
112.8
150.0
231.3
Age
18-29
61
76.9
9.1
13.4
17.6
21.2
25.1
29.4
33.2
38.5
43.4
49.4
56.6
64.2
72.5
82.5
93.7
108.4
130.3
167.0
249.4
30-39
94
83.7
10.5
15.1
19.5
23.2
27.4
31.7
36.6
41.6
46.9
53.1
61.0
69.2
79.0
90.4
104.0
122.5
147.6
189.0
262.8
40-49
115
65.1
OO
OO
12.8
16.2
19.7
23.1
26.6
30.2
34.7
38.8
43.6
48.9
54.9
62.5
71.1
81.7
95.0
114.2
142.8
196.6
50-59
88
55.2
5.3
8.0
10.5
13.0
15.5
18.5
21.8
25.3
29.4
33.8
38.3
43.6
49.9
57.7
67.5
80.4
96.9
122.1
173.0
60+
88
50.4
5.5
8.2
10.6
13.1
15.6
18.3
21.0
24.4
28.0
31.7
36.1
41.0
47.0
54.4
63.4
73.5
89.3
111.6
153.9
Nez Perce Tribe
Volume III - Appendix E
December 2016 Page E-3
-------
Table E3. Nez Perce Tribe. Distribution of the usual fish Group 1 (all fish) consumption (g/day, raw weight, edible portion)
and their 95% confidence intervals based on the 24 hour recalls. Estimated by the NCI method.
No. or
Consumers
Mean
Percentiles
5"-;.
lov;.
15"';.
2ov--;.
25";.
30"-;.
35"'i.
40"n
45";.
Overall
451
75.0
7.3
11.2
15.1
19.2
23.5
27.8
32.5
37.7
43.3
(95% CI)
(57.3-104.6)
(1.5-18.5)
(3.0-24.0)
(4.7-29.3)
(6.7-34.4)
(9.1-38.8)
(11.8-44.0)
(14.8-48.6)
(18.5-54.0)
(22.6-60.7)
Fisher
138
98.2
9.4
14.8
20.1
25.2
30.9
36.7
42.7
49.3
56.3
(95% CI)
(66.3-158.3)
(1.8-32.2)
(3.8-39.9)
(6.1-47.9)
(8.4-55.8)
(11.1-62.6)
(14.4-69.9)
(18.5-77.6)
(23.1-86.4)
(28.0-96.0)
-continued
Pereenliles
50",,
55"-;.
(.OV'i.
(.5"-;.
¦'0"-;.
75%
so1:;.
S5v;.
<>ov;.
95",.
Overall
49.5
56.4
64.6
73.9
85.1
98.9
115.7
138.5
173.2
232.1
(95%
CI)
(27.8-67.8)
(33.8-76.1)
(41.0-86.5)
(49.5-97.5)
(59.0-
111.6)
(69.9-
133.5)
(82.9-
161.2)
(97.8-200.1)
(120.9-
262.3)
(165.0-
379.7)
Fisher
64.7
74.3
85.2
97.9
113.2
130.4
154.1
184.1
229.2
305
(95%
CI)
(32.8-
106.5)
(38.6-
121.0)
(45.9-
137.9)
(54.8-
159.1)
(65.1-
184.2)
(78.2-
218.7)
(91.1-
257.7)
(112.9-
316.1)
(141.4-
401.6)
(196.7-
540.3)
Nez Perce Tribe
Volume III - Appendix E
December 2016 Page E-4
-------
1.0 NCI Method—Covariate Selection
This section expands on the selection of covariates into the NCI models described in section
5.23.2 "The NCI Method—Covariate Selection." That section described two steps for selecting
the covariates into the NCI models: (1) the choice of the FFQ covariate adjustment; and (2) the
inclusion of other covariates. The other candidate covariates included: presence on the fishers
list, gender, ZIP code groups (83540, 83536, 83501 and Other for the Nez Perce Tribe; 83203
and Other for the Shoshone-Bannock Tribes), age (grouped as 18-29, 30-39, 40-49, 50-59 and
60+) and the responder's weight (in pounds). Prior to these two steps we also assessed potential
seasonality in the 24-hour recall data.
We first present covariate selection for the species Group 1 NCI model. We first considered four
forms of continuous FFQ covariate adjustment: the original (untransformed) FFQ rate value, the
3rd root value, the logio value and the numerical decile of FFQ (coded as 1-104). Each of these
forms was accompanied in the model by its interaction with the tribe to allow different effects in
the two tribes. The goodness-of-fit of the four FFQ forms was compared to the model with the
categorical FFQ decile by calculating statistics for respondents divided into the ten decile groups
per tribe5. Specifically, the mean, median, 90th percentile and 95th percentile of consumption
were calculated by the NCI method within each decile of FFQ for each of the four forms, and
were compared to the same statistics (means and percentiles) calculated by a fifth NCI model
that used the FFQ decile as a categorical variable. The NCI model with the categorical FFQ
decile regresses the likelihood of consuming fish on a given day and the amount consumed on
days with positive consumption on the indicators of the FFQ deciles. The model estimates one
average probability and one average amount for each FFQ decile. As a result, the estimated
relationship between the FFQ and the 24-hour recall from this model is a step function (step =
one estimated value per decile). The model allows for any shape of the FFQ-24-hour-recall trend
line across the ten FFQ deciles (but constant values within each decile). The four forms of
continuous FFQ covariate adjustment, in contrast, assume specific curve-linear trends,
constraining the estimated trends to specific shapes. Although the categorical decile model need
not necessarily reveal the "best" relationship between FFQ and 24-hour recalls (due to noise in
the data and other possible relationships), the categorical model is a useful reference because it
can reveal potential non-linear trends in the relationship. In choosing between the four
continuous FFQ adjustments we sought to find a transformation of FFQ that would reasonably
follow the trend suggested by the categorical decile model and lead to a good, simple
characterization of the relationship between FFQ and the 24-hour recalls. The categorical decile
model also suggested another adjustment that we previously did not expect. We discovered that
the 24h recall consumption in the 10th FFQ decile among the SBT respondents was considerably
4 The decile cut points were defined separately within each tribe.
5 The categorical FFQ model, representing 10 decile categories, is defined by indicator variables (often called 'dummy' variables).
These indicator variables are dichotomous and are usually coded as either 1 (one) or 0 (zero) for each particular observation being
used in the analysis (e.g., a consumption rate for a particular respondent). A value of '1' for an indicator variable (e.g., the indicator
variable for the 3rd decile) indicates that the observation falls in the particular decile group represented by that variable (e.g., the 3rd
decile). For the 10 deciles, nine indicator variables are needed to define the categorical variable. One of the decile groups serves as
a reference group and is not represented by an indicator variable. Thus, if the 10 decile is the reference group, an observation in
the 3rd decile would be represented by the following values for the nine indicator variables: 0,0,1,0,0,0,0,0,0. An observation in the
10th decile group would be represented by the following values for the nine indicator variables: 0,0,0,0,0,0,0,0,0. That collection of
nine zeroes tells us that the observation is not in any of the first nine deciles, so it must be in the 10th decile. In the context of this
study, the categorical FFQ variable probably is the best representation of the FFQ data for a regression analysis. However, it uses
nine variables and is not a parsimonious representation of the data; thus, to avoid problems in fitting the NCI models, a more
parsimonious form of the FFQ consumption rate is sought, which is the topic considered here.
Nez Perce Tribe
Volume III - Appendix E
December 2016 Page E-5
-------
lower than expected by the trend in any of the four forms of FFQ. We therefore added an
indicator for this group into each model, which greatly improved the fit. The impact of the 10th
SBT decile is further described in the following paragraph.
The comparison of the four FFQ forms of covariate adjustments to the categorical FFQ
adjustment is shown in Figure E2. The eight panels of the figure show the fit for the two tribes
(the first four panels for NPT and the second four panels for SBT), all calculated from an NCI
model based on data combined form the two Tribes. The four panels for each tribe show the
estimated mean, the 50th, 90th and 95th percentiles (in that order). The estimates from the
reference categorical decile model are shown as black bars and the estimates from the four
considered FFQ forms are superimposed as colored lines. The categorical estimates show that in
the NPT, the NCI-estimated usual intake estimated from the 24-hour recalls increased with
higher FFQ deciles. This, however, was not the case in the SBT, where the estimated intake
decreased after the 8th decile. While the decrease from the 8th decile to the 9th decile was relative
moderate, the decrease from the 8th decile to the 10th decile was pronounced. We therefore
introduced an indicator for the 10th SBT decile (but not for the 9th SBT decile) into the model.
The impact of this indicator is also illustrated in Table E4, which shows the NCI model
coefficients for 10 different models: (1) the four continuous forms of FFQ with the indicator for
SBT decile 10; (2) the four continuous forms of FFQ without the indicator for SBT decile 10; (3)
the model with the categorical FFQ decile; and (4) the model without FFQ. The coefficient
A VAR U2 shows the between-person variance, in the transformed positive amount, not
explained by the covariates. The similar values of the coefficients lambda (A LAMBDA) across
the models suggests that the transformations of the amount consumed are similar across the 10
models (ranging from 0.25 to 0.32) and, thus, the variances are approximately comparable
(larger differences would suggest different amount scales and a lack of comparability of the other
model coefficients). The model without FFQ (the last column) has AVARU2 equal to 6.09. As
this model has no FFQ adjustment, the unexplained between-person variance is large.
Importantly, the models with the SBT decile 10 indicator variable have AVARU2 values
between 0.91 and 2.55 whereas the models without it have much larger A_VAR_U2 values
(ranging between 2.78 and 6.12). The difference in A_VAR_U2 shows the ability of the SBT
decile 10 to explain differences in the amount variation across respondents.
Figure El and Table E4 help us to choose between the four forms of continuous FFQ adjustment.
The untransformed FFQ and numerical FFQ decile models have much larger AVARU2 than
the 3rd root and logio FFQ models. Visually, the untransformed FFQ model tends to overestimate
the intake for the bottom two FFQ deciles and the 10th decile, and to underestimate the intake for
the FFQ deciles 5-9 in SBT (with the exception of decile 10). The model with numerical FFQ
deciles tends to overestimate the intake for FFQ deciles 7 and 8 in NPT. The fits for the 3rd root
and logio FFQ models are similar visually as well as in terms of their AVARU2 values. The
choice between these two models was therefore arbitrary. We used the 3rd root of FFQ as our
primary choice because the 3rd root transformation is numerically very close to the
transformation of the positive 24-hour recalls in this model (lambda of 0.33 corresponds to the
third root). With the 3rd root of FFQ, the FFQ predictor and the transformed 24-hour recall values
are approximately on the same scale. To investigate the impact of this choice, we ran a
sensitivity analysis with logio FFQ as the form for the FFQ variable and compared the results to
the primary choice of the 3rd root of FFQ. The results of this sensitivity analysis are presented in
this appendix.
Nez Perce Tribe
Volume III - Appendix E
December 2016 Page E-6
-------
By -tribe deciles
123456789 10 123456739 10
FFQ Decile
model — DECILENUM2 — LIN — LOG10 — RT3
Figure El. Comparison of four forms of FFQ adjustment (colored lines) to the categorical decile FFQ adjustment (black bars).
Model for Group 1 species. DECILENUM2 = the numerical decile of FFQ (coded as 1-10), LIN = the original (untransformed)
FFQ, LOGIO = the logio FFQ, RT3 = the 3rd root FFQ. All models included an addition adjustment for the 10H! decile in the
SBT. mean mc t = mean, tpercentileSO, 90 and 95 = the 50th, 90th and 95th percentiles, respectively. Estimates are NCI
estimates of daily consumption in g/day (raw weight, edible portion).
Nez Perce Tribe
Volume III - Appendix E
December 2016 Page E-7
-------
Table E4. Coefficients for the NCI models considered in the selection of the FFQ covariate form. Model for Group 1 species.
Only selected coefficients are presented for the reference model with categorical decile of FFQ ("Cat. FFQ") and for the model
with no FFQ (i.e., model with tribe only).
Muik-ls wiili imlk.iliir fur IO1I1 ik-iik- in SIS 1"
Mmk-ls williuiil iiuliiiilor fur lOili ik-iik- in SIJT
IIO model as linear Junction of
IIO model as linear function of
Orix. IIO
3rd root
of IIO
/.«.!; IIO
IIO Decile
Oris. I I O
3rd root
of no
l.og IIO
IIO Decile
( at.
IIO
\o
IIO
AO 1 _INTERCEPT
13.9559
10.3166
8.0985
10.7239
13.0141
10.2516
8.0091
11.1414
A02_TRIBE
-1.5858
-3.7307
-3.3414
-0.2963
-0.485
-0.0059
-1.0845
-0.5927
0.006336
0.6543
0.8374
0.5618
0.007474
0.8504
1.1147
0.5113
0.007179
0.6377
0.6002
-0.02219
-0.00503
-0.286
-0.03819
-0.05807
A05_SBT_DEC 10
-9.0943
-6.6204
-4.1483
-4.0528
A06_WEEKEND
-0.9247
-0.7346
-0.4761
-0.9493
-1.2819
-1.2208
-0.8656
-1.0534
A07_SECINT
0.8183
0.846
0.5661
1.0871
1.2293
1.3213
1.0724
1.2909
A_LAMBDA
0.3117
0.283
0.2467
0.3
0.3163
0.3156
0.2864
0.3074
0.2504
0.2956
A_LOGSDE
1.3783
1.2269
1.006
1.3037
1.3682
1.3839
1.2245
1.3473
A_LOGSDU2
0.407
0.02313
-0.04887
0.4687
0.9056
0.7576
0.5107
0.6819
P01_INTERCEPT
-1.9953
-3.4115
-4.2844
-3.0236
-1.9964
-3.4485
-4.3217
-2.7742
P02_TRIBE
-0.8803
-1.2198
-1.0185
-0.615
-0.6906
-0.2404
-0.155
-0.77
0.003719
0.4265
0.6466
0.2804
0.003724
0.4326
0.6516
0.2413
0.000153
0.08232
0.03917
-0.01308
-0.0024
-0.1727
-0.1923
-0.01529
P05_ SBT _DEC10
-2.1493
-2.0507
-1.3541
-1.1575
P06_WEEKEND
-0.1348
-0.07827
-0.04341
-0.04868
-0.1743
-0.1089
-0.09914
-0.1101
P07_SECINT
0.5072
0.4915
0.4825
0.4907
0.5132
0.484
0.4936
0.4897
P_LOGSDUl
0.179
0.07796
0.03015
0.07674
0.1934
0.1392
0.1122
0.1205
z_u
0.5427
0.5503
0.5118
0.5889
1.1695
1.1138
1.02
1.1021
P_VAR_U1
1.4304
1.1687
1.0622
1.1659
1.4721
1.3211
1.2515
1.2726
1.0642
1.625
A_VAR_U2
2.2571
1.0473
0.9069
2.5533
6.1181
4.5502
2.7772
3.9107
1.8615
6.0925
Nez Perce Tribe
Volume III - Appendix E
December 2016 Page E-8
-------
Modi-Is willi iiulii';iliir 1'nr Will ik-iik- in SIJT
Modi-Is wilhiiul iii(lii';iliir lor llllli ik-iik- in Sli'l
IIO model as linear Junction oj
IIO model as linear function of
Oris. IIO
Jrtl root
of IIO
IIO
IIO Decile
Oris. IIO
3rd root
oj IIO
/.«.!; I'I'O
IIO Decile
< at.
IIO
\o
IIO
A_VAR_E
15.7464
11.6335
7.4788
13.565
15.4315
15.9229
11.5756
14.8004
6.7362
12.0332
cov_ulu2
0.8895
0.554
0.4626
0.9129
2.4733
1.9746
1.4353
1.7875
1.3851
2.7027
RHO
0.4951
0.5008
0.4713
0.5291
0.8241
0.8054
0.7699
0.8012
0.9841
0.859
Estimated parameters: Parameters starting with the letters "A" and "P" refer to the amount and probability models, respectively.
AO 1 INTERCEPT and P01_INTERCEPT= intercept;
A02 TRIBE and P02 TRIBE = tribe (NPT=0, SBT=1);
and = the (untransformed or transformed) FFQ;
and = the tribe-FFQ interaction;
A05SBTDEC10 and P05_ SBTDEC10 = indicator of 10th decile in SBT (0=no,l= yes);
A06WEEKEND and P06WEEKEND = weekend indicator (0=no,l= yes);
A07SECINT and P07_SECINT= 2nd interview (0=no,l= yes);
A LAMBD A = lambda for the Box-Cox transformation of the consumed amount;
ALOGSDE = log SD of the residual variance;
A LOGSDU2 and P_L0GSDU1= log SD of the between-subject variance;
Z_U = the Fisher's transformation of the correlation parameter;
P VAR U1 = the between-subject variance for the probability model (Ul);
A_VAR_U2 = the between-subject variance for the amount model (Ul);
A VAR E = the residual variance for the amount model;
cov_ulu2 = covariance between U1 and U2;
RHO = the correlation parameter between U1 and U2.
Nez Perce Tribe
Volume III - Appendix E
December 2016 Page E-9
-------
rd
After adding the 3 root of FFQ and its interaction with the dichotomous tribe variable and the
indicator for SBT decile 10 into the model, the next step considered inclusion of the remaining
covariates into the model. These candidate covariates included the presence on the fishers list,
gender, ZIP code groups (83540, 83536, 83501 and Other for the Nez Perce Tribe and 83203
and Other for the Shoshone-Bannock Tribes), age (grouped as 18-29, 30-39, 40-49, 50-59
and 60+) and the responders' weight (attempted as untransformed, 3rd root, logio and the
numerical decile, coded 1-10). These covariates were included in the model along with their
interactions with the tribe.
For the categorical covariates (all covariates except the responders' weight), we calculated the
NCI-estimated mean and percentiles and compared them across the groups of the covariate. The
results are shown in Figures E2-E5. All four covariates showed an impact on the Group 1
consumption. Specifically, fishers tended to consume more (Figure E2), women less (Figure E3),
and respondents in the other SBT ZIP codes more than in the ZIP code 83203 and respondents in
the NPT ZIP code 83501 less than in the remaining three NPT ZIP codes (Figure E4). We also
observed differences in age for both tribes. Going from younger age groups (left) to older groups
(right), consumption first increased and then decreased (Figure E5).
rd
Respondents' weight (attempted as untransformed, 3 root, logio and the numerical decile) was
analyzed in a fashion similar to the FFQ covariate (Figure E6). There seems to be no or, at best, a
weak relationship between the respondents' weight and the 24-hour recall. Respondents' weight
was therefore not included in the final model.
The selected covariates were used as covariates in both the probability and the amount equations
of the NCI model. The coefficients for the final model for Group 1 are presented in Table E5. In
addition to the coefficients for the selected covariates, the output shows coefficients for the
weekend adjustment, the sequence effect adjustment and the variance components.
Documentation of the parameters can be found in the user's guide for the NCI model macros
(Ruth Parsons, Stella S. Munuo, Dennis W. Buckman, Janet A. Tooze, Kevin W. Dodd. User's
Guide for Analysis of Usual Intakes. 2009.
http://appliedresearch.cancer.gov/diet/usualintakes/Users_Guide_Mixtran_Distrib_Indivint_l. 1 .pdf)
Nez Perce Tribe
Volume III - Appendix E
December 2016 Page E-10
-------
150
aJ
F
LU
100 -
50
Mean
100
75
50-
25
p50
500-
400-
300
200 -i
100-
p90
p95
600 -
400 -
200-
<1)
Li_
I
£=
O
lQ
Qj ¦
f— '£>
n
o
-------
100 -
aJ
LU
50 -
0 -
Mean
~1 1 1 1
I-
LL CL £ lQ 03 CD £
~Z. IE ~Z. go jE 00
LL LL
75-
50-
25-
p50
n i 1 1
I— J3! l-ffl I- ® l— le
CL (S CLE iOcC L1JP
Z .> Z!oj (f)^ CO
400 -
300-
200 -
100 -
0-
p90
p95
400-
200-
0-
~i I 1 1—
<1) <1)
h- ® Hlc h ® Hlc
CL ZI CO
I- \-~a I- I— 75
Cloj £ CO cc CO £
21 !> 21
-------
150 -
ce
100 -
IU
50
Mean
~i 1 1 r
co
o
c\]
CO
o
LO
co
to
LO
CO
O
'T
LO
co
CO CO CO CO
QJ
00
"1 1—
Q_ lQ
Z CO
OO CO 00 -f
o o
lO
(A
Figure E4. NCI-estimated mean and the 50th, 90th and 95th percentiles by ZIP code. Model for Group 1 species. Other
covariates include the 3td root of FFQ, its interaction with tribe and the indicator for SBT decile 10. Dots are estimates from 50
bootstrap runs and give some idea of uncertainty around the estimates. Estimates are NCI estimates of daily consumption in
g/day (raw weight, edible portion).
Nez Perce Tribe
Volume III - Appendix E
December 2016 Page E-13
-------
150 J
£ 100
or
E
LU
50
Mean
~l 1 1 1—
O) O) o> o>
7 o> O) o>
150
100
l l
o o
LO
50-
+
o
I— I— I— Q_l— I— I— I— CD
ClCLCLZlQCQCQCDc/)
Z Z Z co co co co
p50
"1 1 1 1—
0)0)0)0)
7
00
LO
I I
O O
LO
+
o
"1—I—I—I—
O) O) O) O)
CM
_ i-
LO
~_ I— I— I— I— iTi
Z lD lD QJ lQ (/)
CO CO CO CO
p90
~i—i—i—r
O) O) O)
7 7
o o
7
CO
LO
I
o
1—I I—I—I—
Ch O-/ 0>
600-
400-
200-
c\]
?
co lo
to oo o o o
,_ T— co ^ lo
+
o
CO
I— I— I— I— CL I— I— I— I— iTi
ClClClCl^GjCDlDlQoO
ZZZZ CO CO CO co
p95
1 1 1 1
O) O) O) O)
7 7
oo o
+
o
~l—I—I—I—
O) O) O) O)
C_5 C_5 CO CO
I
o
LL LL Q_ LL
+
o
to
K
QJ
I— I— I— ., .
DJ Hi CD lD (/)
CO CO CO CO
Figure E5. NCI-estimated mean and the 50th, 90th and 95th percentiles by ase and tribe. Model for Group 1 species. Other
covariates include the 3td root of FFQ, its interaction with tribe and the indicator for SBT decile 10. Dots are estimates from 50
bootstrap runs and give some idea of uncertainty around the estimates. Estimates are NCI estimates of daily consumption in
g/day (raw weight, edible portion).
Nez Perce Tribe
Volume III - Appendix E
December 2016 Page E-14
-------
Overall deciles
NPT, mean mc t
I
~i 1 1 1 1 1 1 1 1 r~
NPT, tpercentile50
ifiiiiiiii
7 8 9 10
—i 1 1 1 1 1 1 1 1 1—
1 23456769 10
400 J
300
200
NPT tpercentileSO
iiiin
i 1 1 1—
12 3 4
-| 1 1 1 r~
6 7 8 9 10
NPTr tpercentile95
~i 1 1 1 1 1 1 1 1 r~
SBT, mean_mc_t
60-
40-
20-
0-
9 10
-1 1 1 1 1 1 1 1 1 1—
1 23456789 10
40-
20-
oH
SBT, tpercentileSO
¦i ii_i3
~1 1 1 1 1 1 1 T"
12 3 4
8 9 10
SBT. tpercentileSO
~l 1 1 1 1 1 1 1 1 T"
1 23456789 10
SBT, tpercentile95
250-
'ill IT! Lll!
~i 1 1 1 1 1 1 r
1 23456739 10
WT Decile
model — DECILENUM1 — LIN — LOQ10 — RT3
Figure E6. Comparison of four forms of respondent weight adjustment (color lines) to the categorical decile respondent weight
adjustment (black bars). Model for Group 1 species. DECILENUM2 = the numerical decile of respondent weight (coded as 1-
10), LIN = the original (untransformed) respondent weight, LOG10 = the logio respondent weight, RT3 = the 3rd root
respondent weight. Models include an adjustment for FFQ. mean mc t = mean, tpercentileSO, 90 and 95 = the 50th, 90"' and
95l 1 percentiles, respectively. Estimates are NCI estimates of daily consumption in g/day (raw weight, edible portion).
Nez Perce Tribe
Volume III - Appendix E
December 2016 Page E-15
-------
Table E5. Final model NCI for Group 1.
Term
llsliiiiiilo
Term
llsliiiiiilo
\u 1 INTERCEPT
11.3909
P01 INTERCEPT
-3.3335
A02TRIBE
-3.76
P02TRIBE
-2.2826
A03ROOT3FFQ
0.5626
P03ROOT3FFQ
0.4529
A04TRIBEROOT3FFQ
0.8751
P04TRIBEROOT3FFQ
0.07145
AO5 TRIBEFFQ GROUP ALL GPD D
ECX10
-7.9413
P05 TRIBEFFQ GROUP ALL GPD D
ECX10
-2.1986
A06FISHER
0.4883
P06FISHER
-0.2079
A07FISHERTRIBE
0.7557
P07FISHERTRIBE
0.2321
A08FEMALE
-1.5451
P08FEMALE
0.2951
A09FEMALETRIBE
1.5025
P09FEMALETRIBE
-0.08841
A10ZIPGROUP83536
-0.2356
P10ZIP GROUP83536
0.2814
A11ZIPGROUP83501
0.01798
P11ZIPGROUP83 501
0.06362
A12ZIPGROUPNPOTHER
0.04987
P12ZIPGROUPNPOTHER
-0.3446
A13ZIPGROUPSBOTHER
1.6268
P13ZIPGROUPSBOTHER
0.7921
A14AGEGROUP1
1.185
P14_ AGEGROUP1
-0.138
A15AGEGROUP2
1.9248
P15AGEGROUP2
-0.3214
A16AGEGROUP3
0.7249
P16AGEGROUP3
-0.4385
A17AGEGROUP4
0.3805
P17AGEGROUP4
-0.3371
A18AGEGROUP1 TRIBE
-3.4037
P18AGEGROUP1 TRIBE
1.3651
A19AGEGROUP2TRIBE
-2.0021
P19AGEGROUP2TRIBE
1.0734
A20AGEGROUP3 TRIBE
-2.8827
P20AGEGROUP3TRIBE
0.8447
A21AGEGROUP4TRIBE
-1.9345
P21AGEGROUP4TRIBE
1.3002
A22WEEKEND
-0.9696
P22WEEKEND
-0.05227
A23_SECINT
0.7675
P23SECINT
0.48
ALAMBDA
0.289
PLOGSDU1
-0.03087
ALOGSDE
1.2507
z_u
0.5493
ALOGSDU2
-4.669
PVARUl
0.9401
A_VAR_U2
0.000088
AVARE
12.1995
cov_ulu2
0.004549
RHO
0.5
Estimated parameters: Parameters starting with the letters "A" and "P" refer to the amount and
probability models, respectively.
AO 1 INTERCEPT and P01_INTERCEPT= intercept;
A02 TRIBE and P02TRIBE = tribe (NPT=0, SBT=1);
Nez Perce Tribe
Volume III - Appendix E
December 2016 Page E-16
-------
A03ROOT3FFQ and P03ROOT3FFQ = the (untransformed or transformed) FFQ;
A04 TRIBEROOT3FFQ and P04 TRIBEROOT3FFQ = the tribe-FFQ interaction;
A05_TRffiEFFQ_GROUP_ALL_GPD_DECXl 0 and
P05TRIBEFFQ GROUP ALL GPD DECX10 = indicator of 10th decile in SBT (0=no,l=
yes);
A06 FISHER and P06 FISHER = on the fishers list (0=no,l= yes);
A07 FISHERTRIBE and P07 FISHERTRIBE = on the fishers list and SBT (0=no,l= yes);
A08 FEMALE and P08 FEMALE = female (0=no,l= yes);
A09 FEMALETRIBE and P09 FEMALETRIBE = SBT female (0=no,l= yes);
A10 ZIPGROUP83536 and P10 ZIPGROUP83536 = ZIP = 83538 (0=no,l= yes);
A11ZIPGROUP83501 and PI 1ZIPGROUP83501 = ZIP = 83501 (0=no,l= yes);
A12 ZIPGROUPNPOTHER and P12 ZIPGROUPNPOTHER = NPT but not ZIP 83538 or
83501 (0=no,l= yes);
A13 ZIPGROUPSBOTHER and P13 ZIPGROUPSBOTHER = SBT but not ZIP 83203
(0=no,l= yes);
A14 AGEGROUP1 and P14 AGEGROUP1 = age 30-39 (0=no,l= yes);
A15 AGEGROUP2 and P15 AGEGROUP2 = age 40-49(0=no,l= yes);
A16 AGEGROUP3 and P16 AGEGROUP3 = age 50-59 (0=no,l= yes);
A17 AGEGROUP4 and P17 AGEGROUP4 = age 60+ (0=no,l= yes);
A18 AGEGROUP1 TRIBE and P18 AGEGROUP1 TRIBE = age 30-39 and SBT (0=no,l= yes);
A19 AGEGROUP2TRIBE and P19 AGEGROUP2TRIBE = age 40-49 and SBT(0=no,l= yes);
A20 AGEGROUP3TRIBE and P20 AGEGROUP3TRIBE = age 50-59 and SBT (0=no,l= yes);
A21AGEGROUP4TRIBE and P21 AGEGROUP4TRIBE = age 60+ and SBT (0=no,l= yes);
A22_WEEKEND and P22_WEEKEND = weekend indicator (0=no,l= yes);
A23 SECINT and P23_SECINT= 2nd interview (0=no,l= yes);
A LAMBDA = lambda for the Box-Cox transformation of the consumed amount;
A LOGSDE = log SD of the residual variance;
ALOGSDU2 and P_LOGSDUl= log SD of the between-subject variance;
Z_U = the Fisher's transformation of the correlation parameter;
P VAR Ul = the between-subject variance for the probability model (Ul);
A_VAR_U2 = the between-subject variance for the amount model (Ul);
A VAR E = the residual variance for the amount model; cov_ulu2 = covariance between Ul
and U2;
RHO = the correlation parameter between Ul and U2.
Nez Perce Tribe
Volume III - Appendix E
December 2016 Page E-17
-------
We ran a similar covariate selection for the Group 2 NCI model.
Figure E7 shows comparison of the four forms of FFQ adjustment (the original (untransformed)
value, the 3rd root value, the logio value and the numerical decile of FFQ). In this case, the FFQ
was the FFQ for the Group 2 species to correspond to the Group 2 outcome. As in the group 1
model addition of the indicator for the SBT decile 10 improved the model greatly and the 3rd root
and logio transformations lead to the best fit among the four forms of continuous FFQ. The 3rd
root transformation more closely corresponded to the lambda from the NCI model and was thus
used as the primary choice while the logio transformation was used in the sensitivity analysis.
Similar to group 1, the presence on the fishers list (Figure E8), gender (Figure E9), ZIP code
(Figure E10) and age (Figure El 1) had an important impact on the group 2 consumption while
the impact of the respondents' weight was weak (Figure E12). We attempted to add all of the
important covariates into the final NCI model for group 2 consumption. However, the model
coefficients were unstable. The instability was a consequence of a small number of "hits" in the
SBT data, and the model could not clearly separate the independent effects of some of the
covariates. To obtain a more stable model we used the model FFQ and tribe adjustments only as
the final NCI model for group 2 (Table E6). The additional covariates (such as the presence on
the fishers list) were introduced into the model only when needed (i.e. when specific subgroup
estimates of consumption were needed). For example, the gender covariate was added when
gender-specific distributions were estimated.
Nez Perce Tribe
Volume III - Appendix E
December 2016 Page E-18
-------
By-tribe deciles
i i i i i l l i i i —i 1 1 1 1 1 1 1 1 r~
123456789 10 1 23456789 10
FFQ Decile
model — DECILENUM2 — LIN — LOG10 — RT3
Figure E7. Comparison of four forms of FFQ adjustment (colored lines) to the categorical decile FFQ adjustment (black bars).
Model for Group 2 species. DECILENUM2 = the numerical decile of FFQ (coded as 1-10), LIN = linear—the original
(untransformed) FFQ, LQG10 = the logio FFQ, RT3 = the 3 ' root FFQ. All models included an addition adjustment for the
10th decile in SBT. mean mc t = mean, tpercentileSO, 90 and 95 = the 50,h, 90lh and 95th percentiles, respectively. Estimates are
NCI estimates of daily consumption in g/day (raw weight, edible portion).
Nez Perce Tribe
Volume III - Appendix E
December 2016 Page E-19
-------
100 -
75
0>
"E5
E 50
"K
LU
25
Mean
40'
20 ¦
p50
250'
200
150-1
100
50-
p90
p95
300 ¦
200
100 ¦
Q
i- .sa
Cl 11.
h- «
T 2 .<2
£Z LL
o
q>
I- « I- o
£2 LJ- CO -£
co 1 y> --
C LL
o
LU
I--2
LL. Li-
H *
T 2 .EG
C LL
o
CD LL [H -£¦
-
C id-
's
{1>
J=
I-.e
o
m
JZ
o
I- J
SI
I— I—
~_ LL ~_
W
a?
£Z
o
£Q LL u2 -rri
a
o
Figure E8. NCI-estimated mean and the 50th, 90th and 95th percentiles by the presence on the fishers list and tribe. Model for
Group 2 species. Other covariates include the 3rd root of FFQ, its interaction with tribe and the indicator for the SBT decile 10.
Estimates are NCI estimates of daily consumption in g/day (raw weight, edible portion).
Nez Perce Tribe
Volume III - Appendix E
December 2016 Page E-20
-------
Mean
80-
60-
E 40 -
111
20 -
"T T T T
|- I nj \- & I— nj
CLffl lQA3 t£ £
ZS co ^ cafe
p50
40-
30 ¦
20-
10 ¦
T T T r
a
I—^ h— tfl I—^ I— nj
CLffi Q-f 0Q «5 HI L
~z. ® m s
150-
100-
50-
0-
p90
"i 1 1 r
CLffi ~_ p m CD £
p95
200 -
100
Uj CO 0)
"I 1 1 1
Oi
\- M He? I- _® I- as
Q_ CT3 £L C i£(T3 LQ?
^ 00 ^ UJiti
Figure E9. NCI-estimated mean and the 50th, 90th and 95th percentiles by sender and tribe. Model for Group 2 species. Other
covariates include the 3rd root of FFQ, its interaction with tribe and the indicator for SBT decile 10. Estimates are NCI
estimates of daily consumption in g/day (raw weight, edible portion).
Nez Perce Tribe
Volume III - Appendix E
December 2016 Page E-21
-------
Mean
TO
o
c\i
o
'TO
TO
U">
rt-
o
lO
TO
CQ
CO
CO CO CO ® += 5
£2
p50
40-
3D-
20 ¦
10-
TO
o
Cs|
o
LO
TO
TO
U7
CQ
V>
CO CO CO
I—- I—- I—
GQ Q- D.
a> ~z. z
o
^r
U7
TO
co
200-
150-
100-
50-
flj
CO
m
_c _c
O O
m
p95
300 -
200-
100-
TO
o
{M
Q
LTJ
CO
<0
TO
Lti
TO
o
^r
LTj
TO
$ (O CD CD ^ £
lD
tfl
CD
03
5>
Al
5
m
to
Figure E10. NCI-estimated mean and the 50th, 90th and 95th percentiles by ZIP code. Model for Group 2 species. Other
covariates include the 3td root of FFQ, its interaction with tribe and the indicator for the SBT decile 10. Estimates are NCI
estimates of daily consumption in g/day (raw weight, edible portion).
Nez Perce Tribe
Volume III - Appendix E
December 2016 Page E-22
-------
80 -
60 ¦
£
"S
~ 40 ¦
in
HI
20-
Mean
~l 1 1 1—
0> O? <3?
p50
40-
20 -
"i—r
~T~
"i—i—i—r
CM «
I I
CO Q
*- CO
+
©
(£> CO
K ~~
a.
Ol CF3 CJJ £J1
CM CO
I
o
CD Q2i GQ CIH
<3> Oi
CM CO
I
o
PI
I
CO
T
Q
tJ-
+
©
CD CO
0_ Q_ CL LL
tr cl f
<3> <35 -en
CM ^
i I i
o
o
150
100-
50-
m CD CD CD Oj
(a co co o
o o
cn ^
I
CO
+
o
~1 1 1 1—
O) O) (JJ ffl
o ©
o ^r
LD
I
©
LQ
Cl CL CL
p95
200-
100-
~i—r
~i—i—i—r
n—i—i—r
o
77T
"" o o
CO -=J-
00
I- Cl HH \— \- CD
CL 2 £D m CO GO tfl
z: <0 co co co
h-
CL
h- I—
Q_ Cl
+
O
CD OCl
en ra W'
-------
Overall deciles
150-
100
50 i
1 23456789 10
1 23456769 10
123456789 10
500
400-
o>
"j? 300 -
200
100-
0 -
NPTr tperceritile95
SBT, mean mc t
iiikliii _ii , i
20-
10
0-
SBT, tpercentile50
1 2 3 4 5 6 7
9 10
1 23456789 10
1 2 3 4 5 6 7
9 10
150 J
50-
SBT. tpercentile90
t 1 r
n 1 1 1—
1 2 3 4 5 6 7
SBT, tpercentile95
100-
0-
lis! .9
9 10
1 23456739 10
WT Decile
model — DECILENUM1 — LIN LOG10 — RT3
Figure E12. Comparison of four forms of respondent body weight adjustment (colored lines) to the categorical decile of respondent
weight adjustment (black bars). Model for Group 2 species. DECILENUM2 = the numerical decile of respondent weight (coded
as 1-10), LIN = the original (untransformed) respondent weight, LOG 10 = the logio respondent weight, RT3 = the 3rd root
respondent weight. Models include an adjustment for FFQ. inean mc t = mean, tpercentile50, 90 and 95 = the 50th, 90th and 95"
percentiles, respectively. Estimates are NCI estimates of daily consumption in g/day (raw weight, edible portion).
Nez Perce Tribe
Volume III - Appendix E
December 2016 Page E-24
-------
Table E6. Final model NCI for Group 2.
Term
Kstimale
Term
Kslimale
AO 1 INTERCEPT
16.2626
P01 INTERCEPT
-3.6988
A02TRIBE
8.6578
P02TRIBE
-2.6738
A03ROOT3FFQ
1.5434
P03ROOT3FFQ
0.4562
A04TRIBEROOT3FFQ
-1.8424
P04TRIBEROOT3FFQ
0.3336
A05SBTDEC10
0.546
P05_ SBT DEC10
-6.0168
A06WEEKEND
-2.0663
P06WEEKEND
-0.1213
A07SECINT
1.2819
P07SECINT
0.5122
ALAMBDA
0.4074
PLOGSDUl
-0.01034
ALOGSDE
1.6965
Z U
-0.09476
ALOGSDU2
1.663
PVARUl
0.9795
AVARU2
27.8251
AVARE
29.7566
cov_ulu2
-0.4932
RHO
-0.09448
Estimated parameters: Parameters starting with the letters "A" and "P" refer to the amount and probability models, respectively.
AO 1 INTERCEPT and P01_INTERCEPT= intercept;
A02 TRIBE and P02TRIBE = tribe (NPT=0, SBT=1);
A03ROOT3FFQ and P03 ROOT3FFQ = the (untransformed or transformed) FFQ;
A04 TRIBEROOT3FFQ and P04_ TRIBEROOT3FFQ = the tribe-FFQ interaction;
Nez Perce Tribe
Volume III - Appendix E
December 2016 Page E-25
-------
A05SBTDEC10 and P05_ SBTDEC10 = indicator of 10th decile in SBT (0=no,l= yes);
A06WEEKEND and P06WEEKEND = weekend indicator (0=no,l= yes);
A07SECINT and P07_SECINT= 2nd interview (0=no,l= yes);
A LAMBDA = lambda for the Box-Cox transformation of the consumed amount;
A LOGSDE = log SD of the residual variance;
ALOGSDU2 and P_LOGSDUl= log SD of the between-subject variance;
Z_U = the Fisher's transformation of the correlation parameter;
P VAR Ul = the between-subject variance for the probability model (Ul);
AVARU2 = the between-subject variance for the amount model (Ul);
A VAR E = the residual variance for the amount model;
cov_ulu2 = covariance between Ul and U2;
RHO = the correlation parameter between Ul and U2
Nez Perce Tribe
Volume III - Appendix E
December 2016 Page E-26
-------
2.0 NCI Method—Quality Checkin
This appendix section contains displays concerning various quality checks for the NCI model. These displays are discussed and
referenced in section 6.9 "Quality Checking—NCI Method" in Volume II.
Person means - Weighted histogram
Residuals - Weighted histogram
ID
CNI
O
<=>
CNI
C>
ID
O
CD
d
o
o
o
c=>
ID
<=>
<=>
^ co
-m-' ^ .
-------
Person means - Weighted histogram
Residuals - Weighted histogram
ID
OI _
ID
<=>
CD
cz
o
10
Per-person mean of 3rd root 24h {pos. values only)
2-101
Residuals of 3rd root 24h
(pos. values only, 2 hit respondents only)
Figure E14. The (survey-weighted) distribution of the person-means and within-person residuals of the third root of the
positive Group 2 consumption amounts. Both tribes combined. The units of the original values were g/day (raw weight, edible
portion).
Nez Perce Tribe
Volume III - Appendix E
December 2016 Page E-28
-------
NPT, Prob
0.2-
o.o-
0.15-
0.10-
0.05-
~J
nn
NPT, Amt
400 -
200 -
100-
o-
SBT, Amt
300-
100 -
o-
1
fisher
Method Naive 2-hit ¦ Naive All I
MCI
Figure E15. Quality checking of NCI model for Group 1 species. Consumption probability and mean amount on consumption
days by the respondent's presence on the fishers list. Prob = Probability, Amt = positive consumption amount (in g/day, raw
weight, edible portion). 0 = not on the fishers list. 1= on the fishers list. The y-axis shows either the consumption probability
(between 0 and 1) or the mean amount on consumption days. Naive 2-hit = naive approach limited to respondents with 2
interviews, naive all = naive approach with all respondents, naive intl = naive approach limited to 1st interviews, NCI = the
NCI model estimate.
Nez Perce Tribe
Volume III - Appendix E
December 2016 Page E-29
-------
female
Method Naive 2—hit | Naive All | Naive Int1 NCI
Figure E16. Quality checking of NCI model for Group 1 species. Consumption probability and mean amount on consumption
days by the respondent's sender. Prob = Probability, Amt = positive consumption amount (in g/day, raw weight, edible
portion). 0 = men. 1= women. The y-axis shows either the consumption probability (between 0 and 1) or the mean amount on
consumption days. Naive 2-hit = naive approach limited to respondents with 2 interviews, naive all = naive approach with all
respondents, naive intl = naive approach limited to 1st interviews, NCI = the NCI model estimate.
Nez Perce Tribe
Volume III - Appendix E
December 2016 Page E-30
-------
83203
SB Oilier
83203
SB Other
zipgroupc
Method Naive 2—hit V Naive All | Naive Int! MCI
Figure E17. Quality checking of NCI model for Group 1 species. Consumption probability and mean amount on consumption
days by the respondent's ZIP code. Prob = Probability, Amt = positive consumption amount (in g/day, raw weight, edible
portion). The y-axis shows either the consumption probability (between 0 and 1) or the mean amount on consumption days.
Naive 2-hit = naive approach limited to respondents with 2 interviews, naive all = naive approach with all respondents, naive
intl = naive approach limited to 1st interviews, NCI = the NCI model estimate.
Nez Perce Tribe
Volume III - Appendix E
December 2016 Page E-31
-------
(17.9.29] (29.39] {39,49] {49,59] (59,95] (17.9,29] {29,39] (39,49] (49,59] (59;95]
1 1 1 1 1 1 1 1 1 1
(17.9.29] (29.39] {39,49] {49,59] (59,95] (17.9,29] {29,39] (39,49] (49,59] (59.95]
agegroup
Method Naive 2—hit Naive All ^ Naive Int1 NCI
Figure E18. Quality checking of NCI model for Group 1 species. Consumption probability and mean amount on consumption
days by the respondent's age. Prob = Probability, Amt = positive consumption amount (in g/day, raw weight, edible portion).
The y-axis shows either the consumption probability (between 0 and 1) or the mean amount on consumption days. Naive 2-hit
= naive approach limited to respondents with 2 interviews, naive all = naive approach with all respondents, naive intl = naive
approach limited to 1st interviews, NCI = the NCI model estimate.
Nez Perce Tribe
Volume III - Appendix E
December 2016 Page E-32
-------
NPT, Prob
0.5-
0.4 -
0.3-
0.2-
0.1 -
0.0-
400
200-
0-
JMPT^Amt
10
9 10
SBT, Prob
0.4 -
0.3-
0.1 -
0.0-
600-
400-
200-
0-
SBT, Amt
"i 1 1 r~
8 9 10
1
9 10
decilexnum
Method Naive 2-hit ¦ Naive All Naive Int1 NCI
Figure E19. Quality checking of NCI model for Group 1 species. Consumption probability and mean amount on consumption
days by the respondent's decile of group 1 FFQ consumption. Prob = Probability, Amt = positive consumption amount (in
g/day, raw weight, edible portion). The y-axis shows either the consumption probability (between 0 and 1) or the mean amount
on consumption days. Naive 2-hit = naive approach limited to respondents with 2 interviews, naive all = naive approach with
all respondents, naive intl = naive approach limited to 1st interviews, NCI = the NCI model estimate.
Nez Perce Tribe
Volume III - Appendix E
December 2016 Page E-33
-------
3.0 NCI Method—Confidence Intervals
The bootstrap distributions which were used to compute confidence intervals are shown in Figures E20 and E21 below. These are
discussed in Section 6.8 of Volume II.
75
50
25
0
75
50
25
0
100
75
50
25
0
100
50
0
100
75
50
25
0
U k
NPT p5
NPT, plO
NPT pi5
50 75 100 125
0 10 20 30
0 10 20 30 40
t—r
0 10 20 30 40
NPX p20
NPT, p25
kj 75- J|, 75- lJ, 75- uj
s;JL "A : jL
NPT; p30
0 10 2030 40 50
0 20 40 60 0 20 40 60
NPT p35 | NPT p40 NPT, p45 NPT p50 | NPT. p55 NPT, p60 NPT, p65
75- 75- A 1°5- J 75- JJL 75- t 75- J|
J | 50" 50 - K 50- 50- 50- 50- jHlj
I 25- I 25 - 25- 25- 25- 25-
J o-
20 40 60
20 40 60 SO
25 50 75 100 20 40 60 80 100
t r
50 75 100125
MPT; p70
NPT? p75
NPT, p60
NPT; p85
J»- 4 k H k 1 7'- IL ™1 ii,
L 50- jHJ 50- 50- ML 50- 50- A 50-
L 25- 25- ¦ | 25- 25- 25- 25-
NPT p90
100-
75-
50-
25-
NPT, p95
NPT: p99
60 90 120 150 50 75100125150
100 150 200
100150200250
10a 5CPO(25C80ffi50 100200300400500
250 500 7501000
SBT, Mean SBT, p5 SBT.pIO SBT, p15 5BT, p20 SBT. p25 SBT, p30
1 I 250 " I i 150" i 150 "J
Jl.C_C_-Lr-:1l1 =£_"~i1
o- ^7- , , o- , J o-L"^—- ,_J o--PP»p^l ,J o- , - oJ
25 50 75
SBT, p35
0 3 6 9 12
SBT p40
10 15
0 5 10 15
O 5 10 15 20
10 20
0 10 20 30
SBT, p45
$BT, p50
L100- A 90- J[ 90-1 75- il 75- |j 75- Ik
, „ l£ t£ , t Jk. : £ -Jk
SBT p55
SBT; p60
SBT, p65
0 10 20 30
0 10 20 30
0 10 20 30 40
0 10 20 30 40
0 10 20 30 40 50
0 20 40 60
0 20 40 60
SBT p70 | SBT, p75 SBT p8Q SBT, p65 | SBT, p90 | SBT p9S SBT, p99
100- ii 120- j( 125- | . 100- u
£ ii£ ~£ iJl *£ i-£ i£
0 25 50 75
0 30 60 90
0 50 100150
0 50100150200
Replicate value
100 200 300 10C20Q30G1-OC500
250 5007501000
Figure E20. Bootstrap distribution of the NCI method estimated means and selected percentiles for all NPT and SBT
respondents. N=978 bootstraps (22 of the 1000 bootstraps did not converge). Group 1 consumption (in g/day, raw weight, edible
portion). Red dot shows the point estimate and the red bar around it shows the 95% confidence interval.
Nez Perce Tribe
Volume III - Appendix E
December 2016 Page E-34
-------
50 100 150
0 50 100150200
0 100 200
0 100 200 300
Replicate value
0 lOffiOCBOOUHEOO o 200400600800
~i r~
0 500100ft 5 OQOOO
Figure E21. Bootstrap distribution of the NCI method estimated means and selected percentiles for NPT and SBT respondents
on the fishers list. N=978 bootstraps (22 of the 1000 bootstraps did not converge). Group 1 consumption (in g/day, raw weight,
edible portion). Red dot shows the point estimate and the red bar around it shows the 95% confidence interval.
Nez Perce Tribe
Volume III - Appendix E
December 2016 Page E-35
-------
This section of the appendix shows the numerical results of the sensitivity analyses described in
Sections 5.23.4 and 6.10 of Volume II (Sensitivity analyses). Each table in this section compares
the Group 1 and Group 2 consumption results from two different models: a.) the final model
(used to derive the means and percentiles of consumption presented in Volume II) vs. b.) a
variation on the final model, as noted in the table title. The variations considered are 3rd root vs.
logio transformation of FFQ consumption (Tables E7 and E8), with and without weekend
adjustment (Tables E9 and E10), with and without interview sequence effect adjustment (Tables
El 1 and E12), with and without correlation between probability of consumption and consumed
amount (Tables E13 and E14), NPT and SBT data combined vs. NPT data only (Tables E15 and
E16), and final model vs. simplified model with three covariates (Table E17). The mean
consumption rate and the 95th percentile of consumption are compared between the final model
and the alternative model in each table.
Nez Perce Tribe
Volume III - Appendix E
December 2016 Page E-36
-------
Table E7. NCI estimates (g/day, raw weight, edible portion) from the final model vs. model
with logio FFQ replacing 3rd root of FFQ. Group 1 consumption.
Tribe
(il'Olipill^
\;iri;ihlc
Croup
No. <»r
Consumers
(A)
l-'iiiiil model
(lit
l.ojilO I'l-'Q model
"ii dilTcivncc
•loo':;.
Mosul
.)5"'
IVrmililc
M Oil n
.)5"'
Poivciililo
M o;i ii
.)5"'
Poivonlilo
NPT
Overall
Overall
451
75.0
232.1
75.6
251.4
0.8%
8.3%
NPT
Fisher
Fisher
138
98.2
305.0
95.5
304.5
-2.7%
-0.2%
NPT
Fisher
Non-fisher
313
67.6
206.0
69.3
232.4
2.5%
12.8%
NPT
Gender
Male
241
87.7
268.1
88.0
283.8
0.3%
5.9%
NPT
Gender
Female
210
62.3
194.4
63.3
216.1
1.6%
11.2%
NPT
ZIP
83501
28
63.6
177.7
66.4
222.1
4.4%
25.0%
NPT
ZIP
83536
39
84.5
246.9
86.4
267.6
2.2%
8.4%
NPT
ZIP
83540
329
73.6
227.2
74.9
251.2
1.7%
10.6%
NPT
ZIP
Other
55
79.8
264.2
76.4
257.6
-4.2%
-2.5%
NPT
Age
18-29
61
75.3
232.5
75.2
241.7
-0.1%
4.0%
NPT
Age
30-39
94
92.5
274.2
92.8
293.9
0.4%
7.2%
NPT
Age
40-49
116
83.8
256.3
84.8
279.2
1.3%
8.9%
NPT
Age
50-59
89
66.8
212.7
68.1
236.0
1.9%
11.0%
NPT
Age
60+
91
58.1
182.5
58.7
204.6
1.1%
12.1%
SBT
Overall
Overall
226
34.9
140.9
34.0
140.3
-2.6%
-0.4%
SBT
Fisher
Fisher
134
42.4
163.6
40.4
158.1
-4.6%
-3.4%
SBT
Fisher
Non-fisher
92
33.9
138.3
33.2
138.1
-2.3%
-0.2%
SBT
Gender
Male
143
38.1
158.3
33.9
144.3
-11.0%
-8.8%
SBT
Gender
Female
83
32.2
126.8
34.1
138.4
5.7%
9.1%
SBT
ZIP
83203
207
29.9
121.1
29.1
120.1
-2.5%
-0.8%
SBT
ZIP
Other
19
59.2
209.7
57.5
217.3
-2.9%
3.6%
SBT
Age
18-29
36
24.3
110.2
21.1
89.2
-13.1%
-19.1%
SBT
Age
30-39
39
44.6
159.0
41.6
155.4
-6.8%
-2.2%
SBT
Age
40-49
51
51.7
202.5
51.0
203.3
-1.2%
0.4%
SBT
Age
50-59
48
31.8
125.8
31.3
126.3
-1.7%
0.4%
SBT
Age
60+
52
26.8
90.7
31.4
116.6
17.1%
28.4%
Nez Perce Tribe
Volume III - Appendix E
December 2016 Page E-37
-------
Table E8. NCI estimates (g/day, raw weight, edible portion) from the final model vs. model
with logio FFQ replacing 3rd root of FFQ. Group 2 consumption.
Tribe
(il'Olipill^
\;ui;il)lc
(>l'(lll|)
No. <»r
Consumers
(A)
l-'iiiiil model
(lit
l.njilO I I Q model
difference
•lOOVi.
Mean
Pereenlile
Me;in
«)5"'
Percentile
Mean
«)5"'
Pereenlile
NPT
Overall
Overall
446
66.5
233.9
66.6
226.2
0.2%
-3.3°.,
NPT
Fisher
Fisher
138
98.4
345.0
95.1
302.0
-3.4%
-12.5%
NPT
Fisher
Non-fisher
308
55.6
189.5
56.7
189.0
1.9%
-0.2%
NPT
Gender
Male
240
79.4
277.1
79.0
261.9
-0.6%
-5.5%
NPT
Gender
Female
206
55.0
198.0
55.3
196.5
0.7%
-0.7%
NPT
ZIP
83501
27
64.0
197.4
66.6
204.4
4.0%
3.5%
NPT
ZIP
83536
38
83.7
301.5
84.1
282.9
0.4%
-6.2%
NPT
ZIP
83540
326
65.5
232.3
65.1
224.8
-0.7%
-3.2%
NPT
ZIP
Other
55
63.0
231.3
61.1
208.0
-2.9%
-10.1%
NPT
Age
18-29
61
76.9
249.4
74.8
222.4
-2.7%
-10.8%
NPT
Age
30-39
94
83.7
262.8
82.1
241.5
-1.9%
-8.1%
NPT
Age
40-49
115
65.1
196.6
65.0
193.8
-0.1%
-1.4%
NPT
Age
50-59
88
55.2
173.0
54.0
169.6
-2.2%
-2.0%
NPT
Age
60+
88
50.4
153.9
51.9
162.8
3.0%
5.8%
SBT
Overall
Overall
225
18.6
80.0
18.9
81.5
1.2%
1.9%
SBT
Fisher
Fisher
134
23.3
92.6
23.4
91.3
0.2%
-1.4%
SBT
Fisher
Non-fisher
91
17.8
76.8
18.1
78.6
1.6%
2.2%
SBT
Gender
Male
143
18.0
79.4
18.1
82.0
0.8%
3.3%
SBT
Gender
Female
82
19.5
84.3
19.6
85.2
0.9%
1.1%
SBT
ZIP
83203
206
15.8
67.2
16.0
68.4
1.3%
1.8%
SBT
ZIP
Other
19
34.1
130.7
34.0
127.5
-0.4%
-2.4%
SBT
Age
18-29
36
1.3
5.4
1.4
5.8
7.1%
8.9%
SBT
Age
30-39
39
36.5
136.3
36.5
138.1
0.0%
1.4%
SBT
Age
40-49
51
50.9
203.0
51.0
197.9
0.1%
-2.5%
SBT
Age
50-59
48
12.6
55.2
12.8
55.6
1.6%
0.8%
SBT
Age
60+
51
13.1
45.1
12.8
45.2
-2.8%
0.3%
Nez Perce Tribe
Volume III - Appendix E
December 2016 Page E-38
-------
Table E9. NCI estimates (g/day, raw weight, edible portion) from the final model vs. final
model without the weekend adjustment. Group 1 consumption.
Tribe
(il'Olipill^
\;iri;il>le
(>l'(lll|)
No. <»r
Consumers
(A)
l-'iiiiil model
(lit
No weekend
iid.jusl meiil
difference
•lOOVi.
Mean
Pereenlile
Mesin
«)5"'
Pereenlile
Mean
«)5"'
Pereenlile
NPT
Overall
Overall
451
75.0
232.1
78.0
240.2
4.0%
3.5°.,
NPT
Fisher
Fisher
138
98.2
305.0
100.0
309.3
1.8%
1.4%
NPT
Fisher
Non-fisher
313
67.6
206.0
71.0
215.3
5.1%
4.5%
NPT
Gender
Male
241
87.7
268.1
90.8
276.9
3.5%
3.3%
NPT
Gender
Female
210
62.3
194.4
65.4
203.4
4.9%
4.6%
NPT
ZIP
83501
28
63.6
177.7
67.3
188.9
5.8%
6.3%
NPT
ZIP
83536
39
84.5
246.9
87.4
254.2
3.4%
3.0%
NPT
ZIP
83540
329
73.6
227.2
77.0
237.3
4.6%
4.5%
NPT
ZIP
Other
55
79.8
264.2
81.4
268.6
2.1%
1.7%
NPT
Age
18-29
61
75.3
232.5
77.2
236.8
2.6%
1.8%
NPT
Age
30-39
94
92.5
274.2
97.2
286.7
5.1%
4.6%
NPT
Age
40-49
116
83.8
256.3
86.7
262.4
3.5%
2.4%
NPT
Age
50-59
89
66.8
212.7
69.2
219.8
3.5%
3.4%
NPT
Age
60+
91
58.1
182.5
61.3
192.4
5.5%
5.4%
SBT
Overall
Overall
226
34.9
140.9
35.0
142.2
0.3%
0.9%
SBT
Fisher
Fisher
134
42.4
163.6
44.5
170.9
5.1%
4.5%
SBT
Fisher
Non-fisher
92
33.9
138.3
33.8
138.0
-0.4%
-0.3%
SBT
Gender
Male
143
38.1
158.3
38.8
160.6
1.9%
1.5%
SBT
Gender
Female
83
32.2
126.8
31.8
124.6
-1.2%
-1.8%
SBT
ZIP
83203
207
29.9
121.1
30.3
123.6
1.4%
2.1%
SBT
ZIP
Other
19
59.2
209.7
57.9
205.7
-2.2%
-1.9%
SBT
Age
18-29
36
24.3
110.2
23.8
108.0
-2.1%
-2.0%
SBT
Age
30-39
39
44.6
159.0
46.7
166.0
4.6%
4.4%
SBT
Age
40-49
51
51.7
202.5
50.1
195.0
-3.1%
-3.7%
SBT
Age
50-59
48
31.8
125.8
33.4
133.1
4.8%
5.8%
SBT
Age
60+
52
26.8
90.7
25.9
88.0
-3.3%
-3.1%
Nez Perce Tribe
Volume III - Appendix E
December 2016 Page E-39
-------
Table E10. NCI estimates (g/day, raw weight, edible portion) from the final model vs. final
model without the weekend adjustment. Group 2 consumption.
Tribe
(il'Olipill^
\;iri;il>le
(>l'(lll|)
No. <»r
Consumers
(A)
l-'iiiiil model
(lit
No weekend
iid.jusl meiil
difference
•lOOVi.
Mean
Pereenlile
Mesin
«)5"'
Pereenlile
Mean
«)5"'
Pereenlile
NPT
Overall
Overall
446
66.5
233.9
68.9
243.1
3.5%
3.9°.,
NPT
Fisher
Fisher
138
98.4
345.0
99.7
350.8
1.3%
1.7%
NPT
Fisher
Non-fisher
308
55.6
189.5
58.4
200.6
5.0%
5.9%
NPT
Gender
Male
240
79.4
277.1
81.9
288.8
3.1%
4.2%
NPT
Gender
Female
206
55.0
198.0
57.5
209.3
4.6%
5.7%
NPT
ZIP
83501
27
64.0
197.4
67.2
209.8
4.9%
6.3%
NPT
ZIP
83536
38
83.7
301.5
86.3
313.7
3.1%
4.1%
NPT
ZIP
83540
326
65.5
232.3
68.4
244.9
4.4%
5.4%
NPT
ZIP
Other
55
63.0
231.3
64.0
238.0
1.6%
2.9%
NPT
Age
18-29
61
76.9
249.4
77.2
254.9
0.5%
2.2%
NPT
Age
30-39
94
83.7
262.8
86.9
272.7
3.8%
3.7%
NPT
Age
40-49
115
65.1
196.6
66.6
201.2
2.3%
2.4%
NPT
Age
50-59
88
55.2
173.0
55.7
175.3
0.9%
1.3%
NPT
Age
60+
88
50.4
153.9
52.0
159.2
3.2%
3.5%
SBT
Overall
Overall
225
18.6
80.0
18.8
81.5
1.0%
1.9%
SBT
Fisher
Fisher
134
23.3
92.6
23.8
95.7
1.9%
3.3%
SBT
Fisher
Non-fisher
91
17.8
76.8
17.9
77.9
0.4%
1.3%
SBT
Gender
Male
143
18.0
79.4
18.0
80.2
0.5%
1.0%
SBT
Gender
Female
82
19.5
84.3
20.1
88.1
3.2%
4.6%
SBT
ZIP
83203
206
15.8
67.2
15.4
67.0
-2.2%
-0.4%
SBT
ZIP
Other
19
34.1
130.7
35.9
140.2
5.4%
7.3%
SBT
Age
18-29
36
1.3
5.4
1.3
5.5
4.0%
2.6%
SBT
Age
30-39
39
36.5
136.3
37.7
139.4
3.0%
2.3%
SBT
Age
40-49
51
50.9
203.0
50.7
199.8
-0.4%
-1.5%
SBT
Age
50-59
48
12.6
55.2
13.8
60.1
9.6%
8.9%
SBT
Age
60+
51
13.1
45.1
12.8
43.1
-2.6%
-4.4%
Nez Perce Tribe
Volume III - Appendix E
December 2016 Page E-40
-------
Table Ell. NCI estimates (g/day, raw weight, edible portion) from the final model vs. final
model without the sequence effect adjustment. Group 1 consumption
Tribe
Grouping
\;iri;il)lc
(•roup
No. or
Consumers
(A)
l-'iiiiil model
(lit
No SO(|IK'IICO
ol'l'ocl
iidjiislmonl
"/« dilToivnco
¦¦•loir;.
M oil n
<)5"'
IVi venl ile
M oil n
«)5"'
IVivcnlilo
M Oil n
1)5"'
IVivoiililo
\\>\
U\ era II
U\ era II
451
" 0
i
|
2(4 1
:: 5%
1 vS"„
NPT
Fisher
Fisher
138
98.2
305.0
119.4
343.2
21.6%
12.5%
NPT
Fisher
Non-fisher
313
67.6
206.0
83.1
236.2
22.9%
14.6%
NPT
Gender
Male
241
87.7
268.1
107.9
306.7
23.0%
14.4%
NPT
Gender
Female
210
62.3
194.4
75.9
219.2
21.7%
12.7%
NPT
ZIP
83501
28
63.6
177.7
80.3
209.4
26.2%
17.8%
NPT
ZIP
83536
39
84.5
246.9
102.6
277.1
21.4%
12.2%
NPT
ZIP
83540
329
73.6
227.2
90.0
258.9
22.3%
14.0%
NPT
ZIP
Other
55
79.8
264.2
97.3
302.1
22.0%
14.3%
NPT
Age
18-29
61
75.3
232.5
92.9
265.4
23.5%
14.1%
NPT
Age
30-39
94
92.5
274.2
112.1
305.5
21.3%
11.4%
NPT
Age
40-49
116
83.8
256.3
102.8
290.4
22.7%
13.3%
NPT
Age
50-59
89
66.8
212.7
83.4
250.7
24.7%
17.9%
NPT
Age
60+
91
58.1
182.5
70.0
205.4
20.5%
12.5%
SBT
Overall
Overall
226
34.9
140.9
44.0
172.3
26.1%
22.3%
SBT
Fisher
Fisher
134
42.4
163.6
54.3
199.2
28.1%
21.7%
SBT
Fisher
Non-fisher
92
33.9
138.3
42.7
168.2
25.8%
21.6%
SBT
Gender
Male
143
38.1
158.3
47.0
187.8
23.4%
18.6%
SBT
Gender
Female
83
32.2
126.8
41.5
153.7
28.8%
21.2%
SBT
ZIP
83203
207
29.9
121.1
38.1
148.7
27.6%
22.8%
SBT
ZIP
Other
19
59.2
209.7
72.5
246.1
22.4%
17.4%
SBT
Age
18-29
36
24.3
110.2
29.6
134.3
21.9%
21.8%
SBT
Age
30-39
39
44.6
159.0
56.2
190.0
25.9%
19.5%
SBT
Age
40-49
51
51.7
202.5
66.9
250.0
29.5%
23.5%
SBT
Age
50-59
48
31.8
125.8
38.8
144.5
21.9%
14.9%
SBT
Age
60+
52
26.8
90.7
35.1
113.5
31.1%
25.0%
Nez Perce Tribe
Volume III - Appendix E
December 2016 Page E-41
-------
Table E12. NCI estimates (g/day, raw weight, edible portion) from the final model vs. final
model without the sequence effect adjustment. Group 2 consumption.
Tribe
Grouping
\;iri;il)lc
(•roup
No. or
Consumers
(A)
l-'iiiiil model
(lit
No SO(|IK'IICO
ol'l'ocl
iid.juslmonl
"/« dilToivnco
¦¦•loir;.
M oil n
<)5"'
IVi venl ile
M oil n
.)5"'
IVrccnlilo
M Oil n
1)5"'
Poivonlik'
\\>\
U\ era II
U\ era II
44<>
(i(i 5
¦>;; >i
x: ~
:~x x
24.4" u
i<>:%
NPT
Fisher
Fisher
138
98.4
345.0
122.0
396.6
23.9%
15.0%
NPT
Fisher
Non-fisher
308
55.6
189.5
69.8
221.8
25.5%
17.0%
NPT
Gender
Male
240
79.4
277.1
98.6
323.8
24.1%
16.9%
NPT
Gender
Female
206
55.0
198.0
67.3
231.2
22.5%
16.8%
NPT
ZIP
83501
27
64.0
197.4
79.6
232.5
24.4%
17.8%
NPT
ZIP
83536
38
83.7
301.5
100.7
343.6
20.2%
14.0%
NPT
ZIP
83540
326
65.5
232.3
80.9
275.3
23.5%
18.5%
NPT
ZIP
Other
55
63.0
231.3
78.4
278.6
24.4%
20.4%
NPT
Age
18-29
61
76.9
249.4
92.0
283.3
19.7%
13.6%
NPT
Age
30-39
94
83.7
262.8
100.2
297.6
19.7%
13.2%
NPT
Age
40-49
115
65.1
196.6
78.9
227.4
21.2%
15.7%
NPT
Age
50-59
88
55.2
173.0
67.3
202.6
21.9%
17.1%
NPT
Age
60+
88
50.4
153.9
61.4
179.7
21.8%
16.8%
SBT
Overall
Overall
225
18.6
80.0
24.2
100.1
30.1%
25.3%
SBT
Fisher
Fisher
134
23.3
92.6
29.5
110.8
26.4%
19.6%
SBT
Fisher
Non-fisher
91
17.8
76.8
23.4
96.5
31.0%
25.6%
SBT
Gender
Male
143
18.0
79.4
23.3
98.5
29.9%
24.0%
SBT
Gender
Female
82
19.5
84.3
25.4
106.3
30.3%
26.2%
SBT
ZIP
83203
206
15.8
67.2
20.7
86.5
31.2%
28.7%
SBT
ZIP
Other
19
34.1
130.7
42.5
157.6
24.7%
20.6%
SBT
Age
18-29
36
1.3
5.4
1.7
7.2
36.5%
33.6%
SBT
Age
30-39
39
36.5
136.3
45.9
161.2
25.6%
18.3%
SBT
Age
40-49
51
50.9
203.0
63.0
240.9
23.7%
18.7%
SBT
Age
50-59
48
12.6
55.2
16.2
69.2
29.0%
25.4%
SBT
Age
60+
51
13.1
45.1
16.6
54.1
26.5%
20.0%
Nez Perce Tribe
Volume III - Appendix E
December 2016 Page E-42
-------
Table E13. NCI estimates (g/day, raw weight, edible portion) from the final model vs. final
model without correlation between the probability and consumed amount. Group 1
consumption.
Tribe
Grouping
\iiriiihlo
(•roup
No. or
Consumers
(A)
l-'iiiiil model
(lit
\\ iihoui
I'roh-iiml.
(orroliilion
'."ii (NIToronoo
¦ 1 Oil"..
M oil n
<)5"'
Poreonlilo
M on n
<)5"'
Poreonlilo
M Oil ii
«)5"'
Poreonlilo
NPT
Overall
Overall
451
75.0
232.1
75.0
232.1
0.0%
0.0%
NPT
Fisher
Fisher
138
98.2
305.0
98.3
305.0
0.0%
0.0%
NPT
Fisher
Non-fisher
313
67.6
206.0
67.6
205.9
0.0%
-0.1%
NPT
Gender
Male
241
87.7
268.1
87.7
268.1
0.0%
0.0%
NPT
Gender
Female
210
62.3
194.4
62.3
194.4
0.0%
0.0%
NPT
ZIP
83501
28
63.6
177.7
63.6
177.6
0.0%
-0.1%
NPT
ZIP
83536
39
84.5
246.9
84.5
246.9
0.0%
0.0%
NPT
ZIP
83540
329
73.6
227.2
73.6
227.1
0.0%
0.0%
NPT
ZIP
Other
55
79.8
264.2
79.8
264.4
0.0%
0.1%
NPT
Age
18-29
61
75.3
232.5
75.3
232.5
0.0%
0.0%
NPT
Age
30-39
94
92.5
274.2
92.5
274.2
0.0%
0.0%
NPT
Age
40-49
116
83.8
256.3
83.8
256.4
0.0%
0.0%
NPT
Age
50-59
89
66.8
212.7
66.9
212.9
0.0%
0.1%
NPT
Age
60+
91
58.1
182.5
58.1
182.3
0.0%
-0.1%
SBT
Overall
Overall
226
34.9
140.9
34.9
140.9
0.1%
0.0%
SBT
Fisher
Fisher
134
42.4
163.6
42.4
163.6
0.1%
0.0%
SBT
Fisher
Non-fisher
92
33.9
138.3
34.0
138.4
0.1%
0.0%
SBT
Gender
Male
143
38.1
158.3
38.1
158.5
0.1%
0.1%
SBT
Gender
Female
83
32.2
126.8
32.2
126.7
0.1%
-0.1%
SBT
ZIP
83203
207
29.9
121.1
29.9
121.2
0.1%
0.1%
SBT
ZIP
Other
19
59.2
209.7
59.3
209.6
0.1%
0.0%
SBT
Age
18-29
36
24.3
110.2
24.3
110.4
0.1%
0.1%
SBT
Age
30-39
39
44.6
159.0
44.6
158.7
0.1%
-0.1%
SBT
Age
40-49
51
51.7
202.5
51.7
202.7
0.1%
0.1%
SBT
Age
50-59
48
31.8
125.8
31.9
125.9
0.1%
0.1%
SBT
Age
60+
52
26.8
90.7
26.8
90.8
0.0%
0.1%
Nez Perce Tribe
Volume III - Appendix E
December 2016 Page E-43
-------
Table E14. NCI estimates (g/day, raw weight, edible portion) from the final model vs. final
model without correlation between the probability and consumed amount. Group 2
consumption.
Tribe
Grouping
\;iri;il)lc
(•roup
No. or
Consumers
(A)
l-'iiiiil model
(lit
\\ illioul
Proh-iiml.
( orivliilion
'."ii (lilTcivnco
¦ 1 Oil"..
M oil n
<)5"'
Peivenlile
M o;i n
«)5"'
Pe real (ilc
M o;i ii
«)5"'
Permit ill*
NPT
Overall
Overall
446
66.5
233.9
66.9
238.8
0.6%
2.1%
NPT
Fisher
Fisher
138
98.4
345.0
97.9
347.5
-0.5%
0.7%
NPT
Fisher
Non-fisher
308
55.6
189.5
56.4
196.9
1.4%
3.9%
NPT
Gender
Male
240
79.4
277.1
79.3
274.0
-0.1%
-1.1%
NPT
Gender
Female
206
55.0
198.0
54.8
196.5
-0.4%
-0.8%
NPT
ZIP
83501
27
64.0
197.4
63.6
193.6
-0.7%
-1.9%
NPT
ZIP
83536
38
83.7
301.5
83.5
300.0
-0.3%
-0.5%
NPT
ZIP
83540
326
65.5
232.3
65.2
229.5
-0.4%
-1.2%
NPT
ZIP
Other
55
63.0
231.3
62.9
230.5
-0.1%
-0.4%
NPT
Age
18-29
61
76.9
249.4
76.7
251.8
-0.2%
1.0%
NPT
Age
30-39
94
83.7
262.8
83.9
264.9
0.3%
0.8%
NPT
Age
40-49
115
65.1
196.6
64.0
195.9
-1.6%
-0.3%
NPT
Age
50-59
88
55.2
173.0
54.6
173.9
-1.0%
0.5%
NPT
Age
60+
88
50.4
153.9
50.7
156.5
0.6%
1.7%
SBT
Overall
Overall
225
18.6
80.0
18.8
81.6
0.9%
2.0%
SBT
Fisher
Fisher
134
23.3
92.6
23.5
95.8
0.9%
3.5%
SBT
Fisher
Non-fisher
91
17.8
76.8
18.1
79.5
1.5%
3.5%
SBT
Gender
Male
143
18.0
79.4
17.9
78.9
-0.3%
-0.6%
SBT
Gender
Female
82
19.5
84.3
19.4
83.5
-0.2%
-0.9%
SBT
ZIP
83203
206
15.8
67.2
15.7
66.4
-0.5%
-1.2%
SBT
ZIP
Other
19
34.1
130.7
33.7
128.1
-1.1%
-2.0%
SBT
Age
18-29
36
1.3
5.4
1.2
5.2
-2.2%
-2.6%
SBT
Age
30-39
39
36.5
136.3
36.3
137.3
-0.7%
0.8%
SBT
Age
40-49
51
50.9
203.0
50.5
206.8
-0.7%
1.9%
SBT
Age
50-59
48
12.6
55.2
12.5
55.4
-0.6%
0.4%
SBT
Age
60+
51
13.1
45.1
12.9
45.0
-1.5%
-0.2%
Nez Perce Tribe
Volume III - Appendix E
December 2016 Page E-44
-------
Table E15. NCI estimates (g/day, raw weight, edible portion) for the NPT from the final
model fit to data from NPT + SBT vs. final model fit only to the NPT data. Group 1
consumption.
Tribe
(ii'oupiii^
\;iri;il>lc
(¦roup
No. <»r
Consumers
(A)
I'iiiiil model
(lit
NPT diilii onl\
difference'
•lOOVi.
Mean
Peivenlile
Mi'sin
•,5"'
IVrconlilo
Mciin
«)5"'
Pcrmilik*
NPT
Overall
Overall
451
75.0
232.1
70.9
254.3
-5.4%
9.6%
NPT
Fisher
Fisher
138
98.2
305.0
92.0
327.2
-6.3%
7.3%
NPT
Fisher
Non-fisher
313
67.6
206.0
64.2
231.5
-5.0%
12.4%
NPT
Gender
Male
241
87.7
268.1
84.0
300.9
-4.2%
12.3%
NPT
Gender
Female
210
62.3
194.4
57.9
212.5
-7.0%
9.3%
NPT
ZIP
83501
28
63.6
177.7
61.7
212.1
-3.0%
19.3%
NPT
ZIP
83536
39
84.5
246.9
79.8
265.9
-5.6%
7.7%
NPT
ZIP
83540
329
73.6
227.2
70.1
253.5
-4.7%
11.6%
NPT
ZIP
Other
55
79.8
264.2
73.1
274.3
-8.4%
3.8%
NPT
Age
18-29
61
75.3
232.5
71.7
247.0
-4.8%
6.2%
NPT
Age
30-39
94
92.5
274.2
88.6
305.5
-4.2%
11.4%
NPT
Age
40-49
116
83.8
256.3
78.6
280.1
-6.2%
9.3%
NPT
Age
50-59
89
66.8
212.7
62.8
238.3
-6.1%
12.1%
NPT
Age
60+
91
58.1
182.5
54.4
202.7
-6.4%
11.0%
Nez Perce Tribe
Volume III - Appendix E
December 2016 Page E-45
-------
Table E16. NCI estimates (g/day, raw weight, edible portion) for the NPT from the final
model fit to data from NPT + SBT vs. final model fit only to the NPT data Group 2
consumption
Tribe
(ii'oupiii^
\;iri;il>le
(¦roup
No. <»r
Consumers
(A)
I'iiiiil model
(lit
NP I I (hilii onl\
"ii (lilTciviicc
•lOOVi.
Mciin
Peivenlile
Mi'iin
«)5"'
Poivonlilo
M Oil n
.)5"'
Poivonlilo
NPT
Overall
Overall
446
66.5
233.9
58.1
188.9
-12.7%
-19.3%
NPT
Fisher
Fisher
138
98.4
345.0
88.5
296.9
-10.0%
-13.9%
NPT
Fisher
Non-fisher
308
55.6
189.5
48.0
147.5
-13.7%
-22.1%
NPT
Gender
Male
240
79.4
277.1
71.6
233.8
-9.9%
-15.6%
NPT
Gender
Female
206
55.0
198.0
46.7
158.2
-15.1%
-20.1%
NPT
ZIP
83501
27
64.0
197.4
55.5
150.9
-13.3%
-23.6%
NPT
ZIP
83536
38
83.7
301.5
74.7
268.1
-10.8%
-11.1%
NPT
ZIP
83540
326
65.5
232.3
56.0
184.9
-14.5%
-20.4%
NPT
ZIP
Other
55
63.0
231.3
54.9
202.2
-12.8%
-12.6%
NPT
Age
18-29
61
76.9
249.4
67.0
235.4
-12.9%
-5.6%
NPT
Age
30-39
94
83.7
262.8
73.5
242.9
-12.2%
-7.6%
NPT
Age
40-49
115
65.1
196.6
54.8
174.6
-15.9%
-11.2%
NPT
Age
50-59
88
55.2
173.0
45.9
149.7
-16.8%
-13.5%
NPT
Age
60+
88
50.4
153.9
43.1
137.7
-14.4%
-10.5%
Nez Perce Tribe
Volume III - Appendix E
December 2016 Page E-46
-------
Table E17. NCI estimates (g/day, raw weight, edible portion) from the final model vs.
simpler model (tribe, 3rd root of FFQ, tribe by 3rd root of FFQ interaction and a single
covariate for groups as needed). Group 1 consumption.
Tribe
(ii'oupiii^
\;iri;il>le
(¦roup
No. <»r
Consumers
(A)
I'iiiiil model
(lit
Simpler model
"ii dilTcivncc
•lOOVi.
Mciin
Peivenlile
M oil n
.)5"'
IVrceiUik'
M Oil n
.)5"'
IVivonlilo
NPT
Overall
Overall
451
75.0
232.1
75.2
252.3
0.3%
8.7%
NPT
Fisher
Fisher
138
98.2
305.0
101.4
333.7
3.2%
9.4%
NPT
Fisher
Non-fisher
313
67.6
206.0
68.3
226.8
1.1%
10.1%
NPT
Gender
Male
241
87.7
268.1
89.8
286.3
2.4%
6.8%
NPT
Gender
Female
210
62.3
194.4
62.3
198.7
-0.1%
2.2%
NPT
ZIP
83501
28
63.6
177.7
57.2
182.7
-10.1%
2.8%
NPT
ZIP
83536
39
84.5
246.9
84.0
276.2
-0.6%
11.8%
NPT
ZIP
83540
329
73.6
227.2
74.3
256.6
1.0%
13.0%
NPT
ZIP
Other
55
79.8
264.2
80.9
287.9
1.4%
9.0%
NPT
Age
18-29
61
75.3
232.5
74.2
224.2
-1.5%
-3.6%
NPT
Age
30-39
94
92.5
274.2
92.8
278.8
0.4%
1.7%
NPT
Age
40-49
116
83.8
256.3
84.8
258.5
1.2%
0.8%
NPT
Age
50-59
89
66.8
212.7
65.5
215.3
-2.1%
1.2%
NPT
Age
60+
91
58.1
182.5
58.1
182.6
0.0%
0.1%
SBT
Overall
Overall
226
34.9
140.9
34.5
142.8
-1.1%
1.3%
SBT
Fisher
Fisher
134
42.4
163.6
42.1
161.9
-0.8%
-1.0%
SBT
Fisher
Non-fisher
92
33.9
138.3
33.5
138.6
-1.4%
0.2%
SBT
Gender
Male
143
38.1
158.3
38.7
161.7
1.7%
2.2%
SBT
Gender
Female
83
32.2
126.8
31.3
123.3
-3.0%
-2.8%
SBT
ZIP
83203
207
29.9
121.1
29.3
126.9
-1.8%
4.8%
SBT
ZIP
Other
19
59.2
209.7
56.8
212.6
-4.1%
1.4%
SBT
Age
18-29
36
24.3
110.2
21.0
94.3
-13.7%
-14.4%
SBT
Age
30-39
39
44.6
159.0
45.9
169.2
2.9%
6.4%
SBT
Age
40-49
51
51.7
202.5
52.3
196.2
1.3%
-3.1%
SBT
Age
50-59
48
31.8
125.8
33.5
131.1
5.2%
4.2%
SBT
Age
60+
52
26.8
90.7
27.2
97.1
1.6%
7.0%
Nez Perce Tribe
Volume III - Appendix E
December 2016 Page E-47
-------
5.0 NCI Method—Covariate Selection: Assessment of Seasonalit
Figure E22 shows the survey-weighted mean6 of the 24-hour recall of the Group 1 species consumption by tribe, month and interview
number (1st vs. 2nd interview). The 1st and 2nd interviews are separated because we found important differences between them (the 2nd
interview tended to be higher, on average, than those in the first interview). Means for some of the months have very small sample
sizes (the sample size is shown within each dot). The sample sizes are limited and there is large variability of the 24-hour recall data
across time: no clear seasonal trend is apparent. We do not claim that such a trend does not exist, but that a trend was not empirically
evident from the data. With fewer single and double hits than the NPT, the trend lines for the SBT do not suggest a trend. Although
some of the months appear to have lower consumption rates, on the average (e.g., July and August 2014 for NPT), this could be an
artifact of the small sample size. And, while other months seem to be high in a specific group (e.g., November for 1st interviews in
NPT), these trends are not strongly supported by the other interviews (e.g., the 2nd interview for the NPT November mean) or across
tribes. Because of the lack of empirical evidence for seasonal differences in the 24-hour recalls for Group 1, species seasonality was
ignored in the NCI models for Group 1.
Figure E23 shows the survey-weighted mean of the 24-hour recall of the Group 2 species consumption by tribe, month and interview
number (1st vs. 2nd interview). The conclusions for the seasonal effects in Group 2 consumption are similar to those for Group 1
(Figure El) in that no clear seasonal trends were identified.
The remaining figures (E24-E26) and tables (E18-E20) presented in this section provide additional summaries and analysis of the
data regarding possible seasonality in consumption. These materials are described and interpreted in section 5.23.2.1 of Volume II of
this report.
6 The means were calculated standard survey estimate methods described in section 5.22 using the same weights as in all other analyses (see in sections 5.19 and 5.20).
Nez Perce Tribe
Volume III - Appendix E
December 2016 Page E-48
-------
All respondents
^ ^ ^ ^ ^ ^ ^ ^ ^ ^ ^ ^ ^
i ~—i-; \~—l-:—ft—r
Interview
1
Month
Figure E22. Seasonality for Group 1 species consumption on the 24-hour recall. Mean 24-hour recall for species Group 1
consumption (g/day, raw weight, edible portion) by tribe, month and interview number (1st or 2,M| 24-hour recall interview).
Numbers within each month's dot are the sample size. One very large data point for a single N PT second interview during
May (5/14) was excluded from this seasonal analysis
Nez Perce Tribe
Volume III - Appendix E
December 2016 Page E-49
-------
All respondents
Interview
1
^ ^ &*• ^ ^ ^ 6'S^ ^ ^ 4^ ^
Month
Figure E23. Seasonality for Group 2 species consumption on the 24-hour recall Mean 24-hour recall for species Group 2
consumption (g/day, raw weight, edible portion) by tribe, month and interview number. Numbers within each month's dot are
the sample size. One outlier data point for a single NPT second interview during May (5/14) was excluded.
i : i : i : i . rr i k i . i z i „ rr i - rr: i
Nez Perce Tribe
Volume III - Appendix E
December 2016 Page E-50
-------
All respondents
Figure E24. Seasonality for salmon and steelhead consumption on the 24-hour recall Mean 24-hour recall consumption rate
(g/day, raw weight, edible portion) for all salmon and steelhead species (combined) by tribe, interview month and interview
number (1st and 2nd interview). Numbers within each month's dot are the sample size. One outlier data point for a single NPT
second interview during May (5/14) was excluded.
Nez Perce Tribe
Volume III - Appendix E
December 2016 Page E-51
-------
FFQ
350
300
250
200
150
C 100
fl
-------
Nez Perce Tribe
80
70
60 -
50 -
-------
Table E18. Comparison of FCRs (g/day, raw weight, edible portion, based on 24-hour recall data) between 24-hour recall
interviews conducted during the peak salmon harvest period (May 2014 through July 2014) vs. the remainder of the survey
period (August 2014 through May 2015). Nez Perce Tribe. Consumers only*. Estimates are weighted.
All Respondents
(451 consumers)
l-ishei's
(I3S consumers)
lnlcr\ ic'\\ s
During Peak lliincsl
lnlcr\ iews
During Po;ik Ihmcsl
Yes
No
P-\ illllC"""
Yes
No
P-\ illllC"" "
Naive 24-hour mean*
Group 1 (all fish)
108.3 (40.7)
93.6 (8.4)
0.81
124.7 (56.0)
129.0(18.9)
0.96
Group 3 (Salmon or steelhead)
64.9 (22.7)
70.2 (7.8)
0.80
113.8(56.3)
108.9(18.0)
0.93
Chinook salmon
56.3 (21.7)
46.7 (7.2)
0.65
82.2 (49.7)
61.4(13.9)
0.61
FFQ Mean
Group 1 (all fish)
170.0(31.6)
119.8(8.7)
0.015
304.4 (91.1)
161.2(18.7)
0.041
Group 3 (Salmon or steelhead)
82.5 (19.7)
78.7 (6.9)
0.68
189.2(62.1)
121.9(15.1)
0.31
Chinook salmon
46.3 (14.0)
48.2 (5.4)
0.61
119.3 (43.3)
73.9(12.5)
0.24
Note: see Section 5.23.2.1 (Assessment of Seasonality) in Volume II for a more detailed explanation and interpretation of this table.
Values are mean (standard error) unless otherwise specified;
*The number of consumers (based on the FFQ) were 451, 446 and 389 (138, 138 and 128 for fishers only) for Group 1, Group 2 and Chinook salmon,
respectively; within the peak harvest period, the number of consumers were 30, 30 and 29 (11, 11 and 11 for fishers only) for Group 1, Group 2 and Chinook
salmon, respectively;
**The naive mean was calculated in two steps: 1) for each respondent, the mean of the consumption on up to two 24 hour recalls and 2) mean of these means. In
this table only, this calculation was adjusted to exclude the second 24-hour recall if the first recall occurred during the peak harvest period and the second
occurred after the peak harvest period;
***Survey weighted t-test of the cube root of the FCR values.
Nez Perce Tribe
Volume III - Appendix E
December 2016 Page E-54
-------
Table E19. Comparison of reported fishing rates (mean times per month) between first interviews conducted during the peak
salmon harvest period (May 2014 through July 2014) vs. FFQ interviews conducted during the remainder of the survey period
(August 2014 through April 2015). Nez Perce Tribe. Consumers only. Estimates are weighted.
All Respondents
(451 consumers)
l-ishers
(I3S consumers)
lnlcr\ iews
During Peak llam-sl
lnlcr\ icw s
During Peak llancsl
Yes
No
P-\ aluc"
Yes
No
P-\ aluc"
Went fishing at least once (%)
Over the whole year
73%
61%
0.22
92%
91%
0.88
In May, June and July
71%
59%
0.26
92%
91%
0.88
No. of times fishing, everyone (times/month)
Over the whole year
1.1 (0.3)
1.3 (0.1)
0.51
2.7 (0.8)
2.3 (0.2)
0.65
In May, June and July
2.5 (0.5)
2.8 (0.3)
0.48
5.3 (1.6)
5.3 (0.5)
0.94
No. of times fishing, if > 0 times** (times/month)
Over the whole year
1.5 (0.4)
2.1 (0.2)
0.20
2.9 (0.8)
2.6 (0.2)
0.65
In May, June and July
3.5 (0.7)
4.7 (0.4)
0.22
5.7 (1.7)
5.8 (0.5)
0.81
Values are percentages or mean (standard error) unless otherwise specified;
* Survey weighted chi-squared test for went fishing at least once and t-test of the cube root of the fishing rate values;
**Only including those who went fishing at least once.
Nez Perce Tribe
Volume III - Appendix E
December 2016 Page E-55
-------
Table E20. Frequencies of two-period FFQ responses (consumption information provided for higher and lower consumption
periods separately) out of all responses*, compared between FFQ interviews conducted during the peak salmon harvest period
(May 2014 through July 2014) vs. the remainder of the survey period (August 2014 through April 2015). Nez Perce Tribe.
Estimates are unweighted.
All Respondents
(451 consumers)
l-'ishers
(I3S consumers
Interviews
During Peak Harvest
Interviews
During Peak Harvest
Yes
No
Ualio of %*s
Yes
No
Ualio of %*s
Group 1 (all fish)
30% (80/267)
19% (475/2543)
1.6
20% (18/90)
22% (171/761)
0.9
Group 3 (Salmon or steelhead)
45% (32/71)
27% (238/893)
1.7
39% (9/23)
24% (71/298)
1.6
Chinook salmon
48% (14/29)
27% (98/361)
1.8
36% (4/11)
24% (28/117)
1.5
Values are percentages (numerator / denominator) unless otherwise specified;
*For the purposes of this table, a "response" is a record of the consumption of an individual species on the FFQ. That is, if a respondent reports eating Chinook,
rainbow trout and sturgeon, this counts as three responses. For each response, the respondent may report consumption for a higher and lower period separately (a
two-period response). This counts as a single response. Therefore, the total number of responses is the total number of individual species mentioned by all
respondents on the FFQ. For simplicity, this analysis includes all responses, without making any exclusions for missing values.
Nez Perce Tribe
Volume III - Appendix E
December 2016 Page E-56
-------
Appendix F—
Comparison of FFQ Rates to 24-Hour
Recall Rates
Nez Perce Tribe
Volume III - Appendix F
December 2016 Page F-l
-------
Appendix F—Comparison of FFQ Rates to 24-Hour
Recall Rates
This section presents additional description of the differences between the FFQ fish consumption
rates and the 24-hour recall rates. It examines the differences in the consumption (g/day),
frequency of consumption and portion size by deciles of the Group 1 FFQ rate and for other
species groups. It also examines the relationship of these rate differences to two indices that
describe the level of uncertainty of the respondents in their answers to FFQ questions on fish
consumption. These tables supplement the material in Section 6.11 of Volume II, where more
background and definitions are provided. Note that these tables are based on Group 1 consumers,
even when consumption rates of other species groups are examined. Non-consumers of a given
species group contribute a consumption rate of zero to the calculations.
Tables F1 and F2 shows the mean of FFQ and 24-hour recall rates for Group 1 and Group 2,
respectively, grouped by the decile of the Group 1 FFQ rate. The identical partitioning of the
respondents into ten deciles by the Group 1 FFQ rate is also used in the subsequent tables.
Nez Perce Tribe
Volume III - Appendix F
December 2016 Page F-2
-------
Table Fl. Nez Perce Tribe. Weighted group 1 means and other statistics from the 24-hour recall and the FFQ consumption
rates (g/day) by Group 1 FFQ consumption rate deciles.
(iroup 1
HQ Decile
l-T'Q range
in decile
\
Sum of'
weights
# respondents
with a 24h hit
Mean HQ
(Ml )
Mean, naive 24h
(M24)
M24-MI
M24/MI
1
0.41-16.98
46
156
8
8.6
22.2
13.6
2.59
2
17.03-33.53
45
167
15
24.2
46.9
22.8
1.94
3
33.84-48.65
45
146
11
43.2
37.2
-6.0
0.86
4
48.82-57.52
45
129
13
52.8
48.3
-4.5
0.92
5
57.53-70.60
45
147
16
64.7
74.2
9.5
1.15
6
70.79-89.94
45
140
20
79.8
96.3
16.6
1.21
7
91.06-124.94
45
159
17
110.2
100.7
-9.5
0.91
8
126.54-166.77
45
150
24
144.4
106.9
-37.5
0.74
9
167.39-281.18
45
150
26
215.8
162.6
-53.2
0.75
10
281.75-1372
45
140
29
516.1
254.4
-261.7
0.49
All
451
1485
179
123.4
94.0
-29.4
0.76
Nez Perce Tribe
Volume III - Appendix F
December 2016 Page F-3
-------
Table F2. Nez Perce Tribe. Weighted group 2 means and other statistics from the 24-hour recall and the FFQ consumption
rates (g/day) by Group 1 FFQ consumption rate deciles.
(iroup 1
HQ Decile
I'T'Q range
in decile
\
Sum of
weights
# respondents
with a 24h hit
Mean HQ
(Ml )
Mean, naive 24h
(M24)
M24-MI
M24/MI
1
0.00-16.39
46
156
6
5.9
17.4
11.5
2.96
2
5.52-33.53
45
167
10
19.4
35.3
15.9
1.82
3
5.45-48.44
45
146
7
31.5
26.6
-5.0
0.84
4
0.79-57.21
45
129
11
41.5
41.6
0.1
1.00
5
18.94-70.53
45
147
12
51.7
52.8
1.1
1.02
6
3.55-88.61
45
140
17
63.9
86.1
22.2
1.35
7
52.50-124.41
45
159
16
93.7
88.7
-5.0
0.95
8
54.78-162.81
45
150
21
120.1
99.5
-20.5
0.83
9
126.85-272.81
45
150
25
187.2
152.8
-34.4
0.82
10
104.13-1324
45
140
25
434.1
213.4
-220.8
0.49
All
451
1485
150
102.8
80.6
-22.2
0.78
Nez Perce Tribe
Volume III - Appendix F
December 2016 Page F-4
-------
Table F3 shows the differences in the mean FFQ and 24-hour recall rates for all species and for
other species groups. The table illustrates that the differences between the FFQ and 24-hour rates
were present in all species groups.
Table F3. Nez Perce Tribe. Weighted means of the 24-hour recall and of the FFQ
consumption rates (g/day) by species group. All Group 1 consumers.
\
# respondents
with a 24h hit
Mean HQ
(Ml )
Mean, naive 24h
(M24)
M24-MI
M24/MI
Group 1
451
179
123.4
94.0
-29.4
0.76
Group 2
451
150
102.8
80.6
-22.2
0.78
Non-Group 2
451
41
20.6
13.4
-7.2
0.65
Group 3
451
126
78.1
68.4
-9.7
0.88
Group 4
451
2
4.3
1.5
-2.7
0.36
Group 6
451
65
36.2
22.5
-13.7
0.62
Group 1 = all finfish and shellfish; Group 2 = near coastal/estuarine/freshwater/anadromous
finfish and shellfish; Group 3 = all salmon and steelhead; Group 4 = resident trout; Group 6 =
marine finfish and shell fish (see Table 2).
Tables F4 and F5 are analogous to Table F3 but are limited only to the respondents in the 10th
decile and the 9th decile, respectively. The table illustrates that the differences between the FFQ
and 24-hour recall rates were largely driven by differences for the respondents in these two
deciles and these differences were present for all species groups.
Table F4. Nez Perce Tribe. Weighted means of the 24-hour recall and of the FFQ
consumption rates (g/day) by species group. Group 1 consumers in the 10th decile.
\
# respondents
with a 24h hit
Mean II-Q
(Ml)
Mean, naive 24h
(M24)
M24-MI
M24/MI
Group 1
45
29
516.1
254.4
-261.7
0.49
Group 2
45
25
434.1
213.4
-220.8
0.49
Non-Group 2
45
5
82.0
41.0
-40.9
0.50
Group 3
45
20
331.7
181.3
-150.4
0.55
Group 4
45
1
21A
12.9
-14.5
0.47
Group 6
45
9
137.1
56.7
-80.4
0.41
Group 1 = all finfish and shellfish; Group 2 = near coastal/estuarine/freshwater/anadromous
finfish and shellfish; Group 3 = all salmon and steelhead; Group 4 = resident trout; Group 6 =
marine finfish and shell fish (see Table 2).
Nez Perce Tribe
Volume III - Appendix F
December 2016 Page F-5
-------
Table F5. Nez Perce Tribe. Weighted means of the 24-hour recall and of the FFQ
consumption rates (g/day) by species group. Group 1 consumers in the 9th decile.
\
# respondents
with a 24h hit
Mean HQ
(Ml )
Mean, naive 24h
(M24)
M24-MI
M24/MI
Group 1
45
26
215.8
162.6
-53.2
0.75
Group 2
45
25
187.2
152.8
-34.4
0.82
Non-Group 2
45
3
28.6
9.8
-18.8
0.34
Group 3
45
21
125.4
126.1
0.7
1.01
Group 4
45
1
7.6
3.0
-4.6
0.39
Group 6
45
6
64.4
31.2
-33.2
0.48
Group 1 = all finfish and shellfish; Group 2 = near coastal/estuarine/freshwater/anadromous
finfish and shellfish; Group 3 = all salmon and steelhead; Group 4 = resident trout; Group 6 =
marine finfish and shell fish (see Table 2).
Table F6 shows the consumption rates from the FFQ and 24-hour recall by species group and
decile. This table is formatted in the same way as Tables F7 and F8, which summarized mean
frequencies and portion size. The consumption rate is a product of frequency and portion size, so
by examining these components separately more insight into the source of the disagreement
between the FFQ and 24-hour recall can be gained (also see Section 6.11 of Volume II).
Nez Perce Tribe
Volume III - Appendix F
December 2016 Page F-6
-------
Table F6. Nez Perce Tribe. Weighted mean consumption from the 24-hour recall and FFQ
for each species group, overall and by decile. Deciles are the deciles of the Group 1 FFQ
consumption rate. All rows are based on all Group 1 consumers. Ratios were not calculated
when a species group was not consumed by the FFQ.
All.
1)1 (III.
1
:
4
5
(.
7
s
ID
No. of respondents
451
4<>
45
45
45
45
45
45
45
45
45
Group 1
FFQ mean consumption, g/day
123.4
8.6
24.2
43.2
52.8
64.7
79.8
110.2
144.4
215.8
516.1
24h mean consumption, g/day
94.0
22.2
46.9
37.2
48.3
74.2
96.3
100.7
106.9
162.6
254.4
24h/FFQ mean consumption
0.76
2.59
1.94
0.86
0.92
1.15
1.21
0.91
0.74
0.75
0.49
Group 2
FFQ mean consumption, g/day
102.8
5.9
19.4
31.5
41.5
51.7
63.9
93.7
120.1
187.2
434.1
24h mean consumption, g/day
80.6
17.4
35.3
26.6
41.6
52.8
86.1
88.7
99.5
152.8
213.4
24h/FFQ mean consumption
0.78
2.96
1.82
0.84
1.00
1.02
1.35
0.95
0.83
0.82
0.49
Group 3
FFQ mean consumption, g/day
78.1
4.3
15.0
22.9
36.3
44.8
47.9
69.6
99.7
125.4
331.7
24h mean consumption, g/day
68.4
14.8
22.6
18.4
40.4
52.8
70.7
75.2
89.9
126.1
181.3
24h/FFQ mean consumption
0.88
3.43
1.50
0.80
1.11
1.18
1.48
1.08
0.90
1.01
0.55
Group 4
FFQ mean consumption, g/day
4.3
0.0
0.8
0.3
0.7
0.2
2.4
2.1
2.5
7.6
27.4
24h mean consumption, g/day
1.5
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
3.0
12.9
24h/FFQ mean consumption
0.36
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.39
0.47
Group 5
FFQ mean consumption, g/day
4.8
0.0
0.5
1.4
0.5
1.1
3.1
2.0
2.8
18.3
19.7
24h mean consumption, g/day
1.6
0.0
0.0
0.0
0.0
0.0
9.0
0.0
1.8
2.3
3.4
24h/FFQ mean consumption
0.33
0.00
0.00
0.00
0.00
0.00
2.93
0.00
0.66
0.12
0.17
Group 6
FFQ mean consumption, g/day
36.2
4.2
7.9
18.6
15.3
18.6
26.4
36.5
39.4
64.4
137.1
24h mean consumption, g/day
22.5
7.3
24.3
18.9
7.9
21.4
16.6
25.5
15.2
31.2
56.7
24h/FFQ mean consumption
0.62
1.76
3.08
1.02
0.52
1.15
0.63
0.70
0.38
0.48
0.41
Group 7
FFQ mean consumption, g/day
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.1
0.0
0.2
24h mean consumption, g/day
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
24h/FFQ mean consumption
0.00
-
-
-
-
-
-
-
0.00
-
0.00
Nez Perce Tribe
Volume III - Appendix F
December 2016 Page F-7
-------
Group 1 = all finfish and shellfish; Group 2 = near coastal/estuarine/freshwater/anadromous
finfish and shellfish; Group 3 = all salmon and steelhead; Group 4 = resident trout; Group 5 =
other freshwater finfish and shellfish; Group 6 = marine finfish and shell fish; Group 7 =
unspecified finfish and shellfish species (see Table 2).
Table F7 examines the differences in the FFQ and 24-hour recall consumption frequencies.
Frequencies are defined as the expected percentage of days of the year with consumption (see
Section 6.11 of Volume II). The 9th and 10th deciles feature the biggest differences, and these
differences are found for the majority of species groups.
Nez Perce Tribe
Volume III - Appendix F
December 2016 Page F-8
-------
Table F7. Nez Perce Tribe. Weighted mean frequency of positive daily consumption from
the 24-hour recall and FFQ for each species group, overall and by decile. Deciles are the
deciles of the Group 1 FFQ consumption rate. All rows are based on all Group 1
consumers. Ratios were not calculated when a species group was not consumed by the FFQ.
All.
di :c
i.i:
i
:
J)
4
5
6
7
s
In
No. of respondents
451
4h
45
45
45
45
45
45
45
45
45
Group 1
FFQ mean frequency, %
31%
5%
11%
19%
21%
23%
24%
35%
42%
58%
78%
24h mean frequency, %
26%
10%
22%
14%
16%
22%
31%
30%
32%
39%
50%
24h/FFQ mean frequency
0.85
1.98
2.02
0.73
0.75
0.94
1.27
0.86
0.77
0.68
0.64
Group 2
FFQ mean frequency, %
25%
4%
8%
12%
15%
17%
19%
27%
34%
47%
68%
24h mean frequency, %
21%
8%
15%
7%
14%
14%
27%
26%
29%
36%
39%
24h/FFQ mean frequency
0.86
2.20
1.81
0.60
0.89
0.81
1.45
0.97
0.84
0.78
0.57
Group 3
FFQ mean frequency, %
18%
3%
6%
8%
13%
13%
13%
19%
27%
29%
50%
24h mean frequency, %
18%
7%
10%
5%
13%
14%
18%
22%
27%
31%
32%
24h/FFQ mean frequency
1.00
2.48
1.55
0.60
1.00
1.06
1.45
1.16
1.00
1.05
0.65
Group 4
FFQ mean frequency, %
2%
0%
1%
0%
0%
0%
1%
1%
1%
2%
9%
24h mean frequency, %
0%
0%
0%
0%
0%
0%
0%
0%
0%
1%
1%
24h/FFQ mean frequency
0.12
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.39
0.12
Group 5
FFQ mean frequency, %
2%
0%
0%
1%
0%
1%
1%
1%
1%
5%
5%
24h mean frequency, %
1%
0%
0%
0%
0%
0%
4%
0%
1%
1%
1%
24h/FFQ mean frequency
0.45
0.00
0.00
0.00
0.00
0.00
3.08
0.00
0.99
0.13
0.28
Group 6
FFQ mean frequency, %
12%
2%
4%
10%
7%
9%
9%
14%
12%
23%
29%
24h mean frequency, %
8%
3%
12%
9%
3%
7%
9%
8%
7%
9%
16%
24h/FFQ mean frequency
0.71
1.39
3.16
0.92
0.37
0.87
0.94
0.55
0.59
0.39
0.56
Group 7
FFQ mean frequency, %
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
24h mean frequency, %
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
24h/FFQ mean frequency
0.00
-
-
-
-
-
-
-
0.00
-
0.00
Group 1 = all finfish and shellfish; Group 2 = near coastal/estuarine/freshwater/anadromous
finfish and shellfish; Group 3 = all salmon and steelhead; Group 4 = resident trout; Group 5 =
other freshwater finfish and shellfish; Group 6 = marine finfish and shell fish; Group 7 =
unspecified finfish and shellfish species (see Table 2).
Table F8 examines the differences in the FFQ and 24-hour recall portion sizes. The 9th and 10th
decile feature the biggest differences. These differences are found for Group 2 and salmon
species but not for trout, other freshwater species and all marine species.
Nez Perce Tribe
Volume III - Appendix F
December 2016 Page F-9
-------
Table F8. Nez Perce Tribe. Weighted mean portion size (grams) from the 24-hour recall
and FFQ for each species group, overall and by decile. Deciles are the deciles of the Group
1 FFQ consumption rate. Each individual's portions sizes were averaged across species
with a weight according to the species frequency. All calculations are limited to positive
(non-zero) portion sizes. Ratios were not calculated when a species group was not
consumed, as noted on the FFQ or 24-hour recall.
ALL
1)1.(11.1.
1
2
4
>
6
7
8
9
10
Nil. "I' rvspiuuk-nls
451
46
45
45
45
45
45
45
45
45
45
Group 1
FFQ mean portion size, grams
356
198
259
271
298
335
380
374
404
445
618
24h mean portion size, grams
310
220
232
264
308
329
293
284
302
372
376
24h/FFQ mean portion size
0.87
1.11
0.90
0.98
1.03
0.98
0.77
0.76
0.75
0.84
0.61
Group 2
FFQ mean portion size, grams
373
190
285
296
293
340
394
393
406
492
638
24h mean portion size, grams
333
225
259
370
313
392
295
284
322
379
405
24h/FFQ mean portion size
0.89
1.18
0.91
1.25
1.07
1.15
0.75
0.72
0.79
0.77
0.64
Group 3
FFQ mean portion size, grams
383
175
288
314
300
348
408
402
423
501
673
24h mean portion size, grams
342
215
235
387
325
392
346
278
326
383
431
24h/FFQ mean portion size
0.89
1.23
0.82
1.23
1.08
1.13
0.85
0.69
0.77
0.76
0.64
Group 4
FFQ mean portion size, grams
229
227
143
123
175
145
306
237
195
277
261
24h mean portion size, grams
481
NA
NA
NA
NA
NA
NA
NA
NA
340
605
24h/FFQ mean portion size
2.10
-
-
-
-
-
-
-
-
1.23
2.32
Group 5
FFQ mean portion size, grams
254
233
281
168
248
204
281
192
191
309
339
24h mean portion size, grams
218
NA
NA
NA
NA
NA
227
NA
130
334
227
24h/FFQ mean portion size
0.86
-
-
-
-
-
0.81
-
0.68
1.08
0.67
Group 6
FFQ mean portion size, grams
288
240
221
209
248
248
294
290
355
325
426
24h mean portion size, grams
259
231
213
211
236
290
179
329
217
310
342
24h/FFQ mean portion size
0.90
0.96
0.96
1.01
0.95
1.17
0.61
1.14
0.61
0.95
0.80
Group 7
FFQ mean portion size, grams
1297
NA
NA
NA
NA
NA
NA
NA
340
NA
2419
24h mean portion size, grams
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
24h/FFQ mean portion size
-
-
-
-
-
-
-
-
-
-
-
Group 1 = all finfish and shellfish; Group 2 = near coastal/estuarine/freshwater/anadromous
finfish and shellfish; Group 3 = all salmon and steelhead; Group 4 = resident trout; Group 5 =
Nez Perce Tribe
Volume III - Appendix F
December 2016 Page F-10
-------
other freshwater finfish and shellfish; Group 6 = marine finfish and shell fish; Group 7 =
unspecified finfish and shellfish species (see Table 2).
Figure F1 and Table F9 describe the distribution of the number of "not otherwise specified''
species (NOS) on the FFQ, per respondent, for different species groups.
group_all
group
trout
group
_epa2
group_
salmon
group fresh nonsalmori nontrout
group marine all
group other
U not otherwise species per respondent
Figure Fl. Nez Perce Tribe. Distribution of the # "not otherwise specified" species (NOS)
on FFQ per respondent.
group all = Group 1 (all finfish and shellfish); group_epa2 = Group 2 (near
coastal/estuarine/freshwater/anadromous finfish and shellfish); groupsalmon = Group 3 (all
salmon and steelhead), group trout = Group 4 (resident trout); groupfreshnonsalmonnontrout
= Group 5 (other freshwater finfish and shellfish); groupmarine = Group 6 (marine finfish and
shell fish); group other = Group 7 (unspecified finfish and shellfish species).
Nez Perce Tribe
Volume III - Appendix F
December 2016 Page F-l 1
-------
Table F9. Nez Perce Tribe. Number and % respondents with any "not otherwise specified"
species designation (NOS) on the FFQ. Overall and by species group.
Species Croup
\
/»
Group 1
39
9%
Group 2
36
8%
Group 3
35
8%
Group 4
7
2%
Group 5
0
0%
Group 6
0
0%
Group 7
3
1%
Group 1 = all finfish and shellfish; Group 2 = near coastal/estuarine/freshwater/anadromous
finfish and shellfish; Group 3 = all salmon and steelhead; Group 4 = resident trout; Group 5 =
other freshwater finfish and shellfish; Group 6 = marine finfish and shell fish; Group 7 =
unspecified finfish and shellfish species (see Table 2).
Table F10 presents linear regression results that analyze the relationship between # "not
otherwise specified" species (NOS) and the FFQ-24-hour difference in consumption rates.
Table F10. Nez Perce Tribe. Unweighted simple linear regressions of the FFQ-24-hour
difference on the number of "not otherwise specified" species (NOS) in the FFQ data per
respondent. Overall and by species. Slope per 1 NOS species. 95% confidence intervals are
approximate (assuming asymptotic normality).
Species (iroup
Intercept
Slope
95% CI
Group 1
26.8
-7.6
-56.1
41.0
Group 2
21.9
-17.8
-63.7
28.2
Group 3
7.5
-7.2
-60.1
45.6
Group 4
2.9
2.7
-21.9
27.4
Group 5
-
-
-
-
Group 6
-
-
-
-
Group 7
0.0
5.7
5.1
6.2
Nez Perce Tribe
Volume III - Appendix F
December 2016 Page F-12
-------
Group 1 = all finfish and shellfish; Group 2 = near coastal/estuarine/freshwater/anadromous
finfish and shellfish; Group 3 = all salmon and steelhead; Group 4 = resident trout; Group 5 :
other freshwater finfish and shellfish; Group 6 = marine finfish and shell fish; Group 7 =
unspecified finfish and shellfish species (see Table 2).
Figure F2 describes the distribution of the number of species with missing data on FFQ for
different species groups.
300-
i
! 200-
100-
0-
group_all
group trout
grojp_epa2
group fresh nonsalmori nontrout
group_salrnan
group marine all
group other
4 species with missing data per respondent
Figure F2. Nez Perce Tribe. Histogram of the number of species with missing data on the
FFQ per respondent.
group all = Group 1 (all finfish and shellfish); group_epa2 = Group 2 (near
coastal/estuarine/freshwater/anadromous finfish and shellfish); groupsalmon = Group 3 (all
salmon and steelhead), group trout = Group 4 (resident trout); groupfreshnonsalmonnontrout
= Group 5 (other freshwater finfish and shellfish); groupmarine = Group 6 (marine finfish and
shell fish); group other = Group 7 (unspecified finfish and shellfish species).
Nez Perce Tribe
Volume III - Appendix F
December 2016 Page F-13
-------
Table F11 presents linear regression results that analyze the relationship between the number of
species with missing data and the FFQ-24-hour difference in consumption rates.
Table Fll. Nez Perce Tribe. Unweighted simple linear regressions of the FFQ-24-hour
difference on the number of species with missing data per respondent. Overall and by
species. Slope per 1 missing-data species. 95% confidence intervals are approximate
(assuming asymptotic normality).
Species Croup
Intercept
Slope
95% CI
Group 1
24.5
7.3
-21.3
35.8
Group 2
20.0
1.0
-28.3
30.3
Group 3
6.9
-0.6
-44.5
43.4
Group 4
2.9
4.0
-20.7
28.6
Group 5
3.5
8.0
-3.0
18.9
Group 6
12.9
-7.5
-30.6
15.6
Group 7
0.0
0.0
-1.3
1.2
Group 1 = all finfish and shellfish; Group 2 = near coastal/estuarine/freshwater/anadromous
finfish and shellfish; Group 3 = all salmon and steelhead; Group 4 = resident trout; Group 5 =
other freshwater finfish and shellfish; Group 6 = marine finfish and shell fish; Group 7 =
unspecified finfish and shellfish species (see Table 2).
Nez Perce Tribe
Volume III - Appendix F
December 2016 Page F-14
-------
Appendix G—
Geographic Inclusion Criteria
Additional Information
Nez Perce Tribe
Volume III - Appendix G
December 2016 Page G-l
-------
Appendix G—Geographic Inclusion Criteria—
Additional Information
The process for selecting a geographic area for sampling members of the Nez Perce Tribe was
based on ZIP code boundaries for zip codes in and around the Nez Perce reservation. The Zip
code boundaries were delineated using a Geographic Information System (GIS)—specifically,
the ArcGIS software program. ZIP code boundaries were downloaded from the U.S. Census
Bureau, circa 2010. To subset the ZIP codes from national to local scale, buffers of 25 and 50
miles (called sampling "hubs") were created around the primary population centers of Lapwai
and Kamiah using ArcGIS. Any ZIP code boundary that included any portion of the land area
within either buffer was then selected for inclusion in the first iteration of the ZIP code subset.
Using this ZIP code subset, a population center for each ZIP code was identified using the U.S.
Postal Service ZIP code lookup tool. These population centers were then selected in ArcGIS
from the "Cities and Towns" dataset available from the National Atlas of the United States
(NAUS). If the population center was not present in the NAUS dataset, it was instead digitized in
ArcGIS through aerial interpretation of high-resolution basemaps. Once the population centers
were assigned to every ZIP code, a second iteration of the ZIP code subset was created. For this
second iteration, any ZIP code whose population center was not included within the 25- or 50-
mile buffer from either sampling hub was removed from the ZIP code subset.
Using this second iteration of the ZIP code subset, each code was first assigned to a sampling
hub (either Lapwai or Kamiah) based on the closest aerial distance of the ZIP code population
center to the sampling hub. Once each ZIP code was assigned to a sampling hub, it was then
assigned to a buffer zone of either 25 or 50 miles (depending on the distance from the ZIP code's
population center to the sampling hub). The ZIP codes were then plotted on a map, symbolizing
each ZIP code as either 25 or 50 miles from either sampling hub, as shown in Figure Gl.
The distances between each ZIP code population center and the sampling hubs were calculated in
ArcGIS using an automatic straight-line distance-calculation tool. Since the geographical
coordinates of the population centers were provided in feet according to the Idaho State Plane
Coordinate System, the distances were measured in feet and then converted to miles. The
distances calculated from each population center to Lapwai and Kamiah, according to ZIP code,
are provided in Table Gl.
Nez Perce Tribe
Volume III - Appendix G
December 2016 Page G-2
-------
Figure Gl. Nez Perce reservation and surrounding eligible ZIP codes for inclusion in the
Nez Perce Tribe fish consumption survey.
99174
SteptoiA
^ )
'M
armingron
99128
•Far'mii
99130
Gyffii'UI
' Pfntanii'
pr
83855
83834
HcrvanJ-
83312
99111
Alhlirji #99102
Piillrruui jVuBw
f!3®^2 83857
83306
Jji x il
Ijmy
83823
•Jftivcr
*83827
«• K O nU tu i
83832 83535
Kunurui^ e3537
99113
r(s<)'U'n*' I
llniiljitunn" I
,RBH 99175
.... ..U-iviifStn "835W 83M1
95347 6lark>":" I t T 8f24 63545
59403 • S35GT L jl - ' 63545
.inn,,--
"TjRol'.n
•Xiutur
83536
y—I
FfTilitMml*
83528 S3533
IS™*} . -
So«itt«K rSRl 2014 US CefiHjs 2010; US Netifial Atlaj 2010
ym,, _j
83530 °JV'"
83522
^ v *tF
\ '?¦> • A
, White J}§ J-
v. I Bird "" ' '
83554
83542 Luc lie ^
Legend
ZIP Codes within 25 Miles of Lapwai or Kamiah
ZIP Codes within 50 Miles of Lapwai or Kamiah
Nez Perce Reservation Boundary
Population Centers
83546
Elk City
83525
Note: Moscow. ID and Pullman, WA
arc the population centers for two ZIP code areas
April 7015 | Figui~
RID0LFI
Environmental
Fish Consumption Survey of the Nei Perce Tribe
Prepared for the Nez Perce Tribe
Nez Perce ZIP Code Map
for Sample Selection
Nez Perce Tribe
Volume III - Appendix G
December 2016 Page G-3
-------
Table Gl. Nez Perce reservation ZIP codes, corresponding population centers, and
distances to sampling hubs for the Nez Perce Tribe survey.
ZIP Code
Population
Center
Distance to
l.:i|)\\;ii
(Miles)
Distance to
Kiimiiih
(Miles)
liulTer
Distiince
Closest
S;ini|)lin<>
11 ii h
83501
Lewiston
10.21
49.14
25
Lapwai
83520
Ahsaka
23.93
23.91
25
Kamiah
83522
Cottonwood
32.94
19.74
25
Kamiah
83523
Craigmont
19.75
21.03
25
Lapwai
83524
Culdesac
6.64
32.50
25
Lapwai
83525
Elk City
76.90
39.69
50
Kamiah
83526
Ferdinand
26.50
18.04
25
Kamiah
83530
Grangeville
46.58
21.26
25
Kamiah
83533
Green Creek
33.15
13.88
25
Kamiah
83535
Juliaetta
12.92
40.49
25
Lapwai
83536
Kamiah
39.15
0.00
25
Kamiah
83537
Kendrick
16.33
39.84
25
Lapwai
83539
Kooskia
43.54
6.20
25
Kamiah
83540
Lapwai
0.00
39.14
25
Lapwai
83541
Lenore
14.01
31.71
25
Lapwai
83542
Lucile
64.69
49.77
50
Kamiah
83543
Nezperce
29.48
10.16
25
Kamiah
83544
Orofino
26.78
20.52
25
Kamiah
83545
Peck
18.84
25.53
25
Lapwai
83546
Headquarters
50.03
29.80
50
Kamiah
83548
Reubens
13.80
25.48
25
Lapwai
83552
Stites
45.28
9.71
25
Kamiah
83553
Weippe
41.52
11.18
25
Kamiah
83554
White Bird
50.68
34.75
50
Kamiah
83555
Winchester
14.32
28.57
25
Lapwai
83806
Bovill
37.01
47.01
50
Lapwai
83812
Clarkia
49.39
55.32
50
Lapwai
83823
Deary
29.75
46.88
50
Lapwai
83827
Elk River
39.67
39.14
50
Kamiah
83832
Genesee
11.62
48.37
25
Lapwai
83834
Harvard
35.61
58.43
50
Lapwai
83843
Moscow
24.50
58.08
25
Lapwai
83844
Moscow
24.50
58.08
25
Lapwai
83855
Potlatch
36.02
63.44
50
Lapwai
83857
Princeton
35.24
61.21
50
Lapwai
83871
Troy
23.02
49.93
25
Lapwai
83872
Viola
32.06
63.84
50
Lapwai
99102
Albion
34.13
70.16
50
Lapwai
99111
Colfax
42.33
78.09
50
Lapwai
99174
Steptoe
49.21
83.14
50
Lapwai
99113
Colton
19.14
57.64
25
Lapwai
99128
Farmington
48.70
76.76
50
Lapwai
99130
Garfield
44.68
75.66
50
Lapwai
Nez Perce Tribe
Volume III - Appendix G
December 2016 Page G-4
-------
ZIP Code
Population
Center
Distance to
l.:i|)\\;ii
(Miles)
Distance to
Kiimiiih
(Miles)
liulTer
Distiince
Closest
S;ini|)lin<>
11 ii h
99161
Palouse
37.26
68.73
50
Lapwai
99163
Pullman
28.80
65.09
50
Lapwai
99164
Pullman
28.80
65.09
50
Lapwai
99179
Uniontown
16.41
55.07
25
Lapwai
99347
Pomeroy
38.47
77.29
50
Lapwai
99401
Anatone
24.47
53.46
25
Lapwai
99402
Asotin
12.50
49.47
25
Lapwai
99403
Clarkston
11.52
50.40
25
Lapwai
Nez Perce Tribe
Volume III - Appendix G
December 2016 Page G-5
-------
Appendix H—
Survey Design Document
Nez Perce Tribe
Nez Perce Tribe
Volume III - Appendix H
December 2016 Page H-l
-------
Appendix H—Survey Design Document,
Nez Perce Tribe
DESIGN OF A SURVEY
ON FISH CONSUMPTION
BY THE NEZ PERCE TRIBE
Prepared for
The Nez Perce Tribe
The U.S. Environmental Protection Agency
SRA International, Inc.
Prepared by
The Mountain-Whisper-Light Statistics and RIDOLFI Inc.
February 2014
Nez Perce Tribe
Volume III - Appendix H
December 2016 Page H-2
-------
Design of a Survey
on Fish Consumption
by the Nez Perce Tribe
Prepared for
The Nez Perce Tribe
The U.S. Environmental Protection Agency
SRA International, Inc.
Prepared by
The Mountain-Whisper-Light Statistics and RIDOLFI Inc.
February 2014
Nez Perce Tribe
Volume III - Appendix H
December 2016 Page H-3
-------
EXECUTIVE SUMMARY
Survey Purpose and Approach
The Tribal Governments in the State of Idaho are collaborating with the U.S. Environmental
Protection Agency (EPA) Region 10, the State of Idaho, and other stakeholders to develop
methods for gathering data on fish consumption rates (FCRs), which includes all freshwater and
marine finfish and shellfish. A survey is being designed to obtain data necessary for determining
fish consumption rates for the Tribes in Idaho, exploring both current and heritage rates. An
additional objective of the survey is to determine how current fish consumption rates might
increase if fisheries resources are improved. This information will be useful for developing water
quality standards that are protective of the current and future health of the Tribes and of other
Idaho residents. Water quality is of great importance to the Native American Tribes in Idaho,
since a substantial portion of their diet is derived from aquatic sources, and water and aquatic
resources play an important cultural and spiritual role for them. It has been documented
elsewhere in the Pacific Northwest (e.g., Puget Sound and the Columbia River) that Tribes
consume far more fish and shellfish than the general U.S. population. In addition, reported
historic fish consumption rates are very high. EPA is therefore interested in investigating FCRs
for Idaho Tribes to support development of Tribal ambient water quality criteria (AWQC) to
protect high fish consuming populations.
Development of the survey design involved informational visits to the Idaho Tribes, including an
open exchange of interests, concerns, and ideas; collection of relevant information on culture,
history, fisheries, environment, and Tribal objectives; investigation of statistical methods and
issues; development of an appropriate statistical methodology for the current fish consumption
survey and an approach for documentation of heritage rates; preparation of a multi-part survey
questionnaire, including screening, two 24-hour dietary recalls, and food frequency
questionnaire; calculations to support a statistically valid design; and coordination with involved
agencies, tribes, consortia, and consultants. This report describes the proposed survey design for
the Nez Perce Tribe (NPT).
Current Survey and Historic Assessment
There are three eras of importance for a fish consumption study: the past, present, and future.
Over an extended period of time, the Tribes have experienced environmental and social changes
that have reduced fish abundance, access to fish, safety of fish consumption, and fish
consumption itself. The Tribes are seeking to increase fish availability, fish safety (i.e., free from
contamination), and fish consumption in the future. Thus, current consumption rates do not
reflect the Tribe's past nor its future goals. Assessing consumption through a current, cross-
sectional survey will provide relatively precise information about current consumption only. For
the overall goals of this survey project, the current consumption rates should not be considered in
isolation. Assessing past consumption through an assessment of historical materials and,
potentially, interviews with some older individuals whose history reaches back a long lifetime
may be highly informative, but rates so derived are likely not as precise because they involve
longer-term recall and unknown quality and completeness of past documentation.
Since the results of the survey may be used for water quality regulation, it is intended that rates
and ancillary materials will support that use. The strength of the current rates is the methodology
and the ability to compare them to contemporary rates for other populations. The strength of the
historical rates is their relevance to the goals of the Tribe, which is to restore fish consumption to
Nez Perce Tribe
Volume III - Appendix H
December 2016 Page H-4
-------
past, higher levels. Future rates may be projected based on anticipated increases in fish
populations resulting from planned or ongoing habitat restoration and supplementation efforts,
and associated increases in fish consumption.
The draft survey design includes a description of the Nez Perce Tribe's story about suppression,
based primarily on existing literature and supplemented with input directly from the Tribes.
Historical fish harvest and fish consumption by Tribal members is presented, as well as causes of
decline in the fish populations, and goals for the future. Additional research and discussion with
Tribal representatives and experts will take place to implement the survey design. During the
survey implementation phase, a more in-depth study of suppression will take place and its
implications for future fish consumption will be considered.
Suppression Effects and Their Implications
According to the National Environmental Justice Advisory Council (NEJAC), a "suppression
effect" occurs when a fish consumption rate for a given population, group, or tribe reflects a
current level of consumption that is artificially diminished from an appropriate baseline level of
consumption for that population, group, or tribe. The baseline level of consumption is
suppressed, and cannot be characterized via a survey of current consumption.
There are circumstances in which suppression effects have implications for an environmental
justice policy that seeks to sustain healthy aquatic ecosystems and to protect the health and safety
of people consuming fish, shellfish, aquatic plants, and wildlife for subsistence, traditional,
cultural, or spiritual purposes. First, a suppression effect may arise when an aquatic environment
and the fish it supports have become contaminated to the point that humans refrain from
consuming fish caught from particular waters. Were the fish not contaminated, these people
would consume fish at more robust baseline levels. Second, a suppression effect may arise when
fish upon which humans rely are no longer available in historical quantities (and kinds), such that
humans are unable to catch and consume as much fish as they had or would. Such depleted
fisheries may result from a variety of affronts, including an aquatic environment that is
contaminated, altered (due, among other things, to the presence of dams), overdrawn, and/or
overfished. Were the fish not depleted, these people would consume fish at more robust baseline
levels. Third, a suppression effect may occur from loss of access to fisheries resources and
changes in social structure such that individuals no longer harvest fish to the same extent as
before, or do not harvest at all.
When environmental agencies employ a FCR that does not capture fully the consumption that is
suppressed - under any scenario in which suppression effects occur - they may set in motion a
sort of downward spiral whereby the resulting environmental standards permit further and further
contamination or depletion of the fish and so diminished health and safety of people consuming
fish, shellfish, aquatic plants, and wildlife for subsistence, traditional, cultural, or spiritual
purposes. This survey is intended to develop the most precise FCRs possible while taking into
consideration historical rates as they relate to restored future rates. An approach is presented for
determining the Tribe's heritage rates based on a critical evaluation of existing historical
literature.
Nez Perce Tribe
Volume III - Appendix H
December 2016 Page H-5
-------
Survey Design and Questionnaire
The target population for the current survey is adult (18+) enrolled Nez Perce Tribal members, a
population which will be geographically defined (e.g., by zip codes within the reservation and
within a reasonable travel distance of the reservation). Sampling will occur with the use of
stratification; strata will be defined by the combination of age, gender, and frequency of
consumption (determined through an initial phone screening process). Potential respondents will
be selected randomly from each stratum and this screening list will include 3 to 5 times as many
individuals as the ultimate effective sample size,7 which was statistically derived to achieve
acceptably precise rates and support the use of modern survey methodology based on 24-hour
dietary recall interviews. The proposed sample size is expected, conservatively, to provide an
estimated mean consumption rate (all species combined, calculated from responses to the food
frequency questionnaire) that has 95% probability of falling within 25% of the population mean,
and to provide an estimated 95th percentile of consumption that has 95% probability of falling
within 40%) of the population 95* percentile of consumption. The sample size is also likely to
provide an acceptable number of respondents with fish consumption on both days of the 24-hour
o
dietary recall interview, enabling use of the methodology for analyzing the recall data.
Trained Tribal representatives will conduct in-person interviews. Each individual surveyed will
complete a food frequency questionnaire and a 24-hour dietary recall interview focused on fish
consumption behavior. A subsample of individuals will subsequently be contacted by phone for a
second 24-hour recall interview after several days. The food frequency questionnaire will
ascertain species-specific frequency of consumption, typical quantities consumed by fish-eating
period, sources of fish consumed, and preparation methods. Portion size characterization will be
facilitated through use of models. Species identification will be facilitated by use of photographs.
Hard copy and electronic data will be handled under strict confidentiality and quality
assurance/quality control protocols.
In addition to the approach presented for critically reviewing existing literature to determine the
Tribe's heritage rates and future aspirations for consumption, the survey questionnaire will
include qualitative questions related to changes in fish consumption over time. The survey
questionnaire presented to respondents during the in-person interviews will include questions
related to changes in fish consumption and fishing activities compared to the past, reasons for
changed fish consumption, and future consumption goals. These inquires will provide additional
lines of evidence regarding heritage rates.
Survey Data Analysis and Reporting
In addition to data collection activities, the draft survey design includes a description of methods
for data management, confidentiality, analysis, and reporting. The results of the suppression
study for each Tribe (including fish consumption rates and supporting materials) will be
presented in a final report along with the results of the current consumption survey. Reported fish
consumption rates from the implementation of the current consumption survey will include the
mean (average) and various percentiles of consumption up to the 95th percentile—and beyond, if
warranted. The precision (margin-of-error) for certain rates (e.g., mean, median, 90th and 95th
7 See subsections "FFQ Sample Size" and "24-Hour Dietary Recall Sample size" for details on sample size methodology.
8 The "NCI method", described later, will be used to analyze the 24-hour dietary recall data. The NCI method may be used only if
there is a sufficient number of respondents with fish consumption on both days of the 24-hour recall interviews.
Nez Perce Tribe
Volume III - Appendix H
December 2016 Page H-6
-------
percentiles) will also be presented. Rates based on the food frequency questionnaire will be
presented for population sub-groups defined by age, gender, and other characteristics in grams
per day (and for some analyses, in grams per kilogram of body weight per day). Rates for fish
species groups (e.g., anadromous, resident freshwater, and marine species) will also be
presented. Data from the 24-hour recalls will be used (and assessed by the 'NCI method' where
possible) to provide rates for all species combined and, if supported by the data, for population
sub-groups and for some species groups. The report of findings will include a description of the
survey operations performed and statistical analyses, results of both the current survey and
heritage rate study, a discussion of the data, including a comparison of the fish consumption rates
derived from both the FFQ and the 24-hr recall surveys, and supporting materials.
Nez Perce Tribe
Volume III - Appendix H
December 2016 Page H-7
-------
TABLE OF CONTENTS
SURVEY DESIGN REPORT
EXECUTIVE SUMMARY 4
1.0 INTRODUCTION AND BACKGROUND 13
1.1 Survey Background and Purpose 13
1.2 Procedures Used to Develop Design Document 14
1.3 Survey Objectives for the Nez Perce Tribe 14
1.4 Role of Current Survey and Historic Assessment 15
2.0 TRIBAL PERSPECTIVE ON SUPPRESSION 16
2.1 Suppression Effects and Their Implications 16
2.2 Historical Fish Harvest and Consumption 17
2.3 Causes of Decline in Fish Populations 19
2.4 Vision for the Future 21
2.5 Estimating Heritage Fish Consumption Rates 25
3.0 SURVEY DESIGN: TARGET POPULATION 27
3.1 Target Population to be Sampled 27
3.2 Phasing-in of the Survey 27
3.3 Sampling the Population 28
3.3.1 Sample Stratification 28
3.3.2 Sample Selection 29
4.0 SURVEY DESIGN: DATA COLLECTION 32
4.1 Survey Methods 32
4.1.1 Selection of In-Person Interviews vs. Other Methods 32
4.1.2 Use of Photographs and Portion Size Models 33
4.1.3 Use of Tribal Interviewers 33
4.2 Measurement Method 33
4.3 Sample Size 35
4.3.1 Screening of Participants 35
4.3.2 FFQ Sample Size 35
4.3.3 24-Hour Dietary Recall Sample Size 37
4.4 Questionnaire Development 38
4.4.1 Telephone Screening 39
Nez Perce Tribe
Volume III - Appendix H
December 2016 Page H-8
-------
4.4.2 Interview Introduction 39
4.4.3 24-Hour Dietary Recall 39
4.4.4 Food Frequency Questionnaire 40
4.4.5 General Information 40
4.4.6 Photographs and Portion Models 41
4.4.7 In-House Testing and Revisions of Questionnaire 41
4.4.8 Pilot Testing of Questionnaire and Field Operations 42
4.5 IRB Approval 42
4.6 EPA Human Subjects Review 43
5.0 SURVEY OPERATIONS 44
5.1 Interviewing 44
5.1.1 Intervi ewer S el ecti on 44
5.1.2 Interviewer Training 44
5.1.3 Procedure Manual and Training for Interviewers and Supervisors 45
5.1.4 Scheduling and Monitoring Interviewers and Activities 46
5.1.5 Recording Interviewer Responses 46
5.1.6 Integrity and Handling of Questionnaire Hardcopy 47
5.2 Contact with Respondents 47
5.2.1 Initial Contact by Mail and Telephone 47
5.2.2 In-Person Interviews 47
5.2.3 Follow-up Call and Re-Interview 48
5.3 Tribal Collaboration in Field Operations 48
5.4 Key Entry of Questionnaire, Validity Checks, and Storage 48
5.4.1 Field Validity Checks and Re-interview 49
5.4.2 Handling Missing Values 49
5.4.3 Naming and storage of electronic files 49
5.4.4 Back-up and Transfer Protocols 50
5.5 Sensitive Information50
5.6 Confidentiality and Data Management 50
5.6.1 Confidentiality of Hardcopy and Electronic Files 50
5.6.2 Communicating Confidentiality to Participants 50
Nez Perce Tribe
Volume III - Appendix H
December 2016 Page H-9
-------
6.0 ANALYSIS, REPORTING, CLOSE-OUT OF STUDY 52
6.1 Analysis of FFQ results 52
6.2 Analysis of 24-hour Recalls 53
6.3 Reporting of Results 53
6.4 Peer Review 55
6.5 Archiving, Ownership, Sharing of Data 55
7.0 DESIGN TEAM, ACKNOWLEDGEMENTS, AND RESOURCES 56
7.1 Names and affiliation56
7.2 Acknowledgements 56
7.3 Resources 56
7.3.1 Guidance, Regulations, and Other Agency Reports 57
7.3.2 Fish Consumption Surveys and Survey Methodology 58
7.3.3 Traditional Lifeways and Suppression Studies 61
8.0 REFERENCES CITED 64
LIST OF FIGURES
Figure 2-1 Abundance-Based Tribal Harvest Goals
Figure 4-1. Precision of mean and selected percentile estimates vs. sample size
LIST OF TABLES
Table 2-1 Abundance Thresholds for Certain Snake River Anadromous Fish
Table 3-1. Number of adult Tribal members by distance from Tribal Reference point
defined by zip code of residence
Table 3-2. Hypothetical strata based on three stratifying factors: age, gender and
frequency of fish consumption
Nez Perce Tribe
Volume III - Appendix H
December 2016 Page H-10
-------
LIST OF APPENDICES
Appendix A. Survey Questionnaire
Note: In the Nez Perce design document as originally issued, Appendix A referred to the survey
questionnaire, contained in a separate file. References to Appendix A in this design report now
refer to the latest version of the questionnaire, found in Appendix A of Volume III.
Nez Perce Tribe
Volume III - Appendix H
December 2016 Page H-l 1
-------
LIST OF ABBREVIATIONS AND ACRONYMS
AWQC
CDC
CV
CWA
EPA
ESA
FCR
FFQ
FOIA
HSRRO
ICC
IRB
NCI
NEJAC
NHANES
NPDES
NPT
NPTEC
PHI
PI
Pll
SFTP
USDA
g/day
g/kg-day
ambient water quality criteria
Center for Disease Control and Prevention
Curriculum vitae
Clean Water Act
U.S. Environmental Protection Agency
Endangered Species Act
fish consumption rate
food frequency questionnaire
Freedom of Information Act
Human Subjects Research Review Official
Indian Claims Commission
Institutional Review Board
National Cancer Institute
National Environmental Justice Advisory Council
National Health and Nutrition Examination Survey
National Pollutant Discharge Elimination System
Nez Perce Tribe
Nez Perce Tribal Executive Committee
Protected Health Information
Principal Investigator
Personally Identifiable Information
Secure File Transfer Protocol
U.S. Department of Agriculture
LIST OF UNITS
grams per day
grams per kilogram of body weight per day
Nez Perce Tribe
Volume III - Appendix H
December 2016 Page H-12
-------
1.0 INTRODUCTION AND BACKGROUND
The Tribal Governments in the State of Idaho are collaborating with the U.S. Environmental
Protection Agency (EPA) Region 10, the State of Idaho, and other stakeholders to develop
methods for gathering data on fish consumption rates (FCRs) in Idaho. This effort is underway to
support development of water quality standards. This survey has been designed to obtain data
necessary for determining fish consumption rates for the Nez Perce Tribe (NPT). The survey is
focused on both current and heritage rates. This information will be useful in developing water
quality standards that are protective of the health of Tribal members as well as of other residents
of Idaho.
1.1 Survey Background and Purpose
Water quality is of great importance to the Native American Tribes in Idaho, since a substantial
portion of their diet is derived from aquatic sources, and water and aquatic resources play an
important cultural and spiritual role for them. EPA Region 10 is conducting fact finding to assist
Tribal governments in Idaho to identify fish consumption rates9 that are appropriate for use in
setting Tribal ambient water quality criteria (AWQC) to protect human health. Idaho Tribal
FCRs may also be of use to the State of Idaho as Idaho AWQC undergo revision.
The numeric value for a particular AWQC is inversely dependent on the FCR used to derive it.
As the FCR increases, the AWQC becomes lower, or more stringent (and, therefore, more
protective of human health). This is particularly true for bioaccumulative chemicals (i.e.,
chemicals that dissolve in fat and increase in concentration at higher levels of the food chain).
It has been documented elsewhere in the Pacific Northwest (e.g., Puget Sound and the Columbia
River) that tribes consume far more fish and shellfish than the general U.S. population. EPA is
thus interested in investigating FCRs for Idaho Tribes to support development of AWQC to
protect high Tribal fish consuming populations.10
EPA has a national goal, established by the Clean Water Act (CWA), to protect water quality so
that fish and shellfish thrive and can be safely eaten by humans. AWQC serve as an important
tool in these efforts. AWQC are used by the CWA National Pollutant Discharge Elimination
System (NPDES) to establish permits for allowable levels of contaminant discharge to the
Nation's waters as well as other water quality management tools to reduce toxics and protect
human health. Protection of tribal health is an important consideration for these regulatory
efforts.
The purpose of this report is to provide a survey design for collecting Tribal fish consumption
information for the Nez Perce Tribe. The information resulting from implementation of the
survey can be used to set AWQC for Tribal waters. This survey effort will help Tribes build
capacity for measuring FCRs, inform tribal fisheries management, and document the importance
of fish in tribal culture and lifeways. The survey results may also be useful for the State of Idaho
in its decision-making process for development of water quality standards.
9 A fish consumption rate (FCR) is the amount of fish and shellfish (by weight) that is consumed by a person on a daily or annual
basis.
10 EPA is also interested in protecting the health of other high fish consuming populations (e.g., recreational anglers or ethnic
minorities). The State of Idaho is currently preparing a survey to determine FCRs for recreational anglers.
Nez Perce Tribe
Volume III - Appendix H
December 2016 Page H-13
-------
1.2 Procedures Used to Develop Design Document
The development of this survey design included informative visits with the five Idaho Tribes on
their reservations, including an open exchange of interests, concerns, and ideas; collection and
review of relevant information on culture, history, fisheries, environment, and Tribal objectives;
investigation of statistical methods and development of an appropriate approach for the fish
consumption survey; drafting a multi-part survey questionnaire, including questions on past,
current, and future consumption patterns; calculations to support a statistically valid design for
each of the Tribal surveys; and coordination with involved agencies, tribes, consortia, and
consultants.
The Tribal visits helped the survey team develop a working relationship with each of the Tribes
and provided critical information for the survey design. The type of information gathered
included the Tribes' objectives for the survey; the type of data compiled in their tribal registers
(to be used for sample selection); existence of and content of historical records on fisheries
resources; issues on language, travel and communication; planning for tribal hosting of and
publicity around the surveys; issues of confidentiality of Tribal data and future survey records;
and discussion of tribal capabilities for carrying out duties during the implementation phase.
Historical reports, past questionnaires, guidance documents, literature articles, and study
methodologies were reviewed. Specific topics of interest relevant to this work included fish
species, preparation methods, ceremonial uses, and suppressed consumption. As available,
ethnographic information for each Tribe was reviewed. A list of additional resources related to
this effort are provided in Section 7 of this report.
Design development included the evaluation of appropriate methodologies for a fish
consumption survey; defining the population of interest; drafting a questionnaire based on survey
objectives; performing calculations to support a statistically valid design for each of the Tribal
surveys; incorporating methods to account for the effect of suppressed consumption; and
specifying key elements of the survey operation.
The State of Idaho is also planning to implement fish consumption surveys. Coordination with
the State of Idaho survey involved periodic conference calls with the survey design teams,
agencies, Tribes and consultants to discuss technical topics related to the survey design. For
example, methods of accessing survey participants, sampling frames, sharing of questionnaires
and documentation from past surveys, defining consumers and non-consumers, species lists, and
identification of survey components were discussed and may promote comparison of the final
results from multiple surveys.
1.3 Survey Objectives for the Nez Perce Tribe
The Nez Perce Tribe has treaty reserved fishing rights within the Columbia Basin and Snake
River basins. In the Snake Basin, the Nez Perce Tribe has quite possibly the largest number of
tributary salmon and steelhead fisheries which can often occur year- round across the states of
Washington, Oregon and Idaho. The Nez Perce Tribe has usual and accustomed fishing places
throughout 13 million+ acres that have been found to been exclusively used and occupied by the
Tribe (including the major portions of the Snake, Tucannon, Imnaha, Grande Ronde, Salmon and
Clearwater Rivers and their drainages); the mainstem Columbia River; and other locations in the
Columbia/Snake Basin.
Nez Perce Tribe
Volume III - Appendix H
December 2016 Page H-14
-------
The Nez Perce Tribe's primary objective for the fish consumption survey is to support
development of more stringent water quality standards that are protective of tribal members'
consumption of fish. The Tribe's culture is and always has been intimately tied to fish, which is a
staple of their diet and an integral part of their society; poor water quality impedes fish survival
and can affect both the quantity and availability of fish that can be harvested and safely
consumed by tribal members. The NPT has a vision of restoring fish species native to the Nez
Perce Treaty Territory. To accomplish this vision, the Tribe has engaged in managing the
resident and anadromous fish species in the streams, lakes, and watersheds within their
management authority in an effort to rebuild habitat and restore opportunities for fish harvest.
Their goal is that fish will be found in all available habitats and will provide fishing opportunities
for present and future generations. An objective of the Tribe is that results of this survey and the
resulting water quality standards should support the Tribe's expectation of an enhanced fishery
and should be adequately protective of fish consumption by the Tribe in the future.
1.4 Role of Current Survey and Historic Assessment
There are three eras of importance for a fish consumption study: the past, present, and the future.
Considering the past, over an extended period of time the NPT has experienced environmental
and social changes that have reduced fish abundance, access to fish, safety of fish consumption,
and fish consumption itself. The Tribe is seeking to increase fish availability, fish safety (i.e.,
free from contamination), and fish consumption in the future. Thus, current consumption does
not reflect the Tribe's past nor its goals. Assessing consumption through a current, cross-
sectional survey will provide relatively precise information about current consumption only. For
the overall goals of this survey, the current consumption rates should not be considered in
isolation. Assessing past consumption through an assessment of historical materials and,
potentially, interviews with some older individuals whose history reaches back a long lifetime
may be highly informative, but rates so derived are likely not as precise because they involve
longer-term recall and unknown quality and completeness of past documentation.
The rates and supporting materials generated by this study will be used in water quality
regulation. The strength of the current rates is that they are derived via a technically defensible
methodology and that these rates can be compared to those of other populations. The strength of
the heritage rates is their relevance to the goals of the Tribe. Future rates may be projected based
on anticipated increases in fish populations resulting from planned or ongoing habitat restoration
and supplementation efforts, and associated increases in fish consumption.
Nez Perce Tribe
Volume III - Appendix H
December 2016 Page H-15
-------
2.0 TRIBAL PERSPECTIVE ON SUPPRESSION
This section describes the Nez Perce Tribe's perspective on suppression, based primarily on
existing literature and supplemented with input directly from the Tribe. Historical fish harvest
and fish consumption by Tribal members is presented, followed by causes of decline in the fish
populations, and vision for the future. Additional research and Tribal input will be required
during the survey implementation phase to account for suppression and the implications for
future fish consumption.
2.1 Suppression Effects and Their Implications
According to the National Environmental Justice Advisory Council (NEJAC, 2002), a
"suppression effect" occurs when a fish consumption rate for a given population, group, or tribe
reflects a current level of consumption that is artificially diminished from an appropriate baseline
level of consumption for that population, group, or tribe. The more robust baseline level of
consumption is suppressed, inasmuch as it does not get captured by the current FCR.
There are circumstances in which suppression effects have implications for an environmental
justice policy that seeks to sustain healthy aquatic ecosystems and to protect the health and safety
of people consuming fish, shellfish, aquatic plants, and wildlife for subsistence, traditional,
cultural, or spiritual purposes. First, a suppression effect may arise when an aquatic environment
and the fish it supports have become contaminated to the point that humans refrain from
consuming fish caught from particular waters. Were the fish not contaminated, these people
would consume fish at more robust baseline levels. Second, a suppression effect may arise when
fish upon which humans rely are no longer available in historical quantities (and kinds), such that
humans are unable to catch and consume as much fish as they had or would. Such depleted
fisheries may result from a variety of affronts, including an aquatic environment that is
contaminated, altered (due, among other things, to the presence of dams), overdrawn, and/or
overfished. Were the fish not depleted, these people would consume fish at more robust baseline
levels. Third, a suppression effect may occur from loss of access to fisheries resources and
changes in social structure such that individuals no longer harvest fish to the same extent as
before, or do not harvest at all.
The implications for environmental justice policy will depend in part upon which of these
scenarios accounts for the suppression effect observed. They will also depend upon how the
more robust "baseline" level is defined - an exercise that itself raises important environmental
justice issues. This question of an appropriate "baseline" will in turn be related to the particular
group affected. In some cases, for example, a tribe will be able to cite a historical "point of
reference" that would describe an appropriate baseline in terms of environmental quality,
geographic delineation, and treaty rights. In each case, there may be important questions of
history, culture, and aspiration to be considered in determining an appropriate baseline; that is to
say, an appropriate baseline might mean examination of what people had consumed as well as
aspiration for what people would consume were there "fair access for all to a full range of
resources," (NEJAC, 2002) or were the conditions fulfilled for full exercise of treaty- and trust-
protected rights and purposes.
When environmental agencies employ a suppressed FCR - under any scenario in which
suppression effects occur - they may set in motion a downward spiral where inappropriately lax
environmental standards permit further and further contamination or depletion of the fish and so
diminish health and safety of people consuming fish, shellfish, aquatic plants, and wildlife for
Nez Perce Tribe
Volume III - Appendix H
December 2016 Page H-16
-------
subsistence, traditional, cultural, or spiritual purposes. This survey is intended to develop the
most precise FCRs as possible while taking into consideration heritage rates as they relate to
aspired future rates. An approach is presented for determining the Tribe's heritage rates based on
a critical evaluation of existing historical literature. Results of the heritage rate study will be
presented with supporting materials in the final survey results report.
2.2 Historical Fish Harvest and Consumption
The Nez Perce are a large Northwest tribe with a culture tied closely to fish. Since time
immemorial, the Tribe occupied a territory covering more than 13 million acres that included
what is today north central Idaho, southeastern Washington, and northeastern Oregon. The Nez
Perce subsistence cycle involved traveling year to year on the same well-traveled routes through
the canyons of the Snake, Tucannon, Clearwater, Grande Ronde, Imnaha and Salmon Rivers,
primarily to follow the salmon runs. In addition to those rivers and their tributaries, the Nez
Perce historically took part in the fishing and trading that occurred between several of the
region's tribes at Celilo Falls on the Columbia River, among other locations of the Columbia
Basin.
The Tribe has always fished. Their economy and culture evolved around Northwest fish runs.
Their persistence can be attributed in large part to the abundance of fish, which has served as a
primary food source, trade item and cultural resource for thousands of years. Settlement by
others in the last 150 years has disrupted people of the Tribe and the natural resources (NPT,
2005).The degree to which the Tribe is culturally coupled to fish was recognized in treaties
signed between the Tribe and the United States Government. The same treaties that confined the
Tribe to a fraction of their former territory also guaranteed their access to fishery resources.
Article III of the Treaty of 1855 guarantees to the Tribe:
"The exclusive right of taking fish in all the streams running through or bordering said
reservation ... as also the right of taking fish at all usual and accustomed places in common
with citizens of the Territory." Treaty with the Nez Perces, 12 Stat. 957 (1859).
The 1855 Treaty Council at Walla Walla and the Treaty negotiations reflect the Tribe's inherent
tribal sovereignty and its "aboriginal title" to land. At the Treaty Council, the United States
sought to clear title to lands; the Nez Perce sought to reserve and maintain a homeland
("Reservation") and reserve its aboriginal rights and way of life. The Nez Perce would not have
signed this treaty without first receiving assurances that these rights, including the right to fish,
would be protected into the future. Additional treaties between the two sovereigns have been
made, but the reserved fishing right has remained unchanged since 1855.
In its 1855 Treaty, the Nez Perce reserved a significant portion of their aboriginal land (about 8
million acres). And, this Nez Perce homeland contained, as the United States recognized, many
of the best fisheries:
Gov. Stevens said: "Here (showing a draft on a large scale) is a map of the Reservation.
There is the Snake River. There is the Clear Water river. Here is the Salmon river. Here is
the Grande Ronde river. There is the Palouse river. There is the El-pow-wow-wee. This is a
large Reservation. The best fisheries on the Snake River are on it...
Moreover, in addition to this homeland, Nez Perce leaders insisted on reserving off-reservation
hunting, fishing, gathering, and pasturing rights. The minutes of the treaty negotiations reflect
Nez Perce Tribe
Volume III - Appendix H
December 2016 Page H-17
-------
Governor Stevens' repeated assurances, on behalf of the United States, that the treaty would
reserve these off-reservation rights to the Nez Perce Tribe:
You will be allowed to pasture your animals on land not claimed or occupied by settlers,
white men. You will be allowed to go on the roads, to take your things to market, your horses
and cattle. You will be allowed to go to the usual and accustomedfishing places andfish in
common with the whites, and to get roots and berries and to kill game on land not occupied
by the whites; all this outside the Reservation:"
Gov. Stevens said: "I will ask of Looking Glass whether he has been told of our council.
Looking Glass knows that in this reservation settlers cannot go, that he can graze his cattle
outside of the reservation on lands not claimed by settlers, that he can catch fish at any of the
fishing stations, that he can kill game and can go to Buffalo when he pleases, that he can get
roots and berries on any of the lands not occupied by settlers...
Fish, as a staple of the Nez Perce diet, have always been an integral part of the Nez Perce
society. Principal to the Nez Perce diet were the anadromous fish species that inhabit the rivers
of the inland northwest. This is corroborated by other existing information such as those from
federal court proceedings.
For example, in its 1967 decision concerning the Nez Perce Tribe, the Indian Claims
Commission (ICC) made comprehensive findings based on detailed anthropological evidence
from both the United States and the Nez Perce Tribe, of the Tribe's area of "exclusive use and
occupancy" and "aboriginal ownership". The ICC determined that the Nez Perce had "exclusive
use" and occupancy of 13,204,000 acres of land and "that salmon fishing was one of the major
sources of subsistence since the main rivers through the area, which include the Snake, the
Clearwater, the Salmon, and their branches, were well supplied with this fish in aboriginal
times." It also concluded that their seasonal "cycle consists of specific times of the year for
fishing for salmon, digging camas and other roots, hunting the game"; this "economic cycle can
generally be summarized as ten months salmon fishing and two months berry picking, with
hunting most of the year."11
During the time that the treaty was negotiated, the salmon resource reserved by the Nez Perce
came from ".. .river systems that were biologically functional and fully productive..." (Meyer
Resources, Inc., 1999). The decline of salmon productivity since the mid-1800's to present, does
not alter, change, or abrogate the Nez Perce treaty right to take fish. This right to take fish
represents an inherent right that the Nez Perce have held since time immemorial. The fishing
right is as important to the Nez Perce today as it was before contact with non-Indians.
The Nez Perce governed where fishing occurred, how many fish were to be harvested, who could
participate, how to use the resource, and ways to honor and perpetuate the resource. They
developed ways to harvest large amounts of fish. These were documented as proven methods to
catch the substantial numbers of salmon and steelhead (as well as other species of fish). The
complex, elaborate, and efficient Nez Perce fishing techniques described below document the
11 The ICC was created by Congress in 1946 to hear claims by Indian tribes for, among other things, compensation for the taking of
aboriginal lands by the United States without fair payment. Compensable aboriginal title was required to be based on "actual and
exclusive use and occupancy 'for a long time' prior to the cession, transfer, or loss of the property." It provided historical information
regarding Nez Perce village sites, uses of natural resources, and range and extent of natural resource use.
Nez Perce Tribe
Volume III - Appendix H
December 2016 Page H-18
-------
extent of their reliance on this valuable resource and the importance of fish to its society and
cultural identity.
Whenever possible, the Nez Perce historically and contemporarily have regularly fished for the
following species: Chinook, Silver, Coho, and Sockeye varieties of salmon; Dolly Varden, Cut
Throat, Brook, Lake, and Rainbow varieties of trout; several species of suckers, white fish,
sturgeon, squaw fish, lampreys, and some shellfish (freshwater clams). In order to harvest these
fish species, the Nez Perce developed a number of fishing techniques and methods: weirs and
traps; dipping platforms (either natural or man-made); fish walls and dams; canoes; spears; hook
and line; gaffs; and variety of nets (dipnets, set nets and throw nets).
The expansive territory of the Nez Perce people was rich in rivers and streams abundant in fish
life. Bands fished from the Snake, Salmon, Clearwater, Imnaha, Grand Ronde, Selway,
Tucannon, Rapid River and many other rivers within and outside its homeland and territory. As
with other tribes, the Nez Perce did not limit their fishing to salmon. Research has been
conducted by a number of people in an effort to determine how many fish were historically
harvested by the Nez Perce. There are a number of methods to estimate amount of fish harvested
and consumed by the Nez Perce (commonly expressed in numbers of fish harvested and annual
per capita consumption). Anthropologist Deward Walker, Jr. estimated that each Nez Perce
consumed over 500 pounds of fish each year (CCRH, 2013).
Others (as cited in Scholz et al., 1985) have estimated an annual per capita fish consumption for
the Nez Perce Tribe of 1,000 pounds per year. This range of rates is equivalent to fish
consumption rates of about 620 to about 1,240 grams per day. These values are represented as
"pounds per capita". While estimates, this illustrates the general magnitude of harvest that
occurred.
In addition to salmon and steelhead, the Tribe has traditionally harvested Snake River white
sturgeon for subsistence purposes. Tribal elders confirm the historical presence of white sturgeon
throughout the Snake River, mainstem Salmon River, the Clearwater River from its mouth to
above Orofino, Idaho, as well as seasonal migrations into the Grande Ronde River (Elmer Crow,
Nez Perce Tribe Department of Fisheries Resources Management, Personal Communication,
2014). In addition to being an important food source, white sturgeon served many purposes in
the culture of the Tribe. White sturgeon blood was used to make glue; the hides were used for
bow cases and quivers, and for water proofing footwear. However, subsistence fishing has been
severely limited as a result of low white sturgeon numbers between Hells Canyon and Lower
Granite dams (all as cited in NPT, 2005).
The traditional way of life for the Nez Perce (e.g. gathering, harvesting, ceremonies, and
traditions) depends on continuance of the circle of life for all native species (plants and animals).
To the Nez Perce the rights reserved under the Treaty of 1855 must be protected such that the
enjoyment of these rights resembles that envisioned by the treaty signers and Nez Perce leaders.
2.3 Causes of Decline in Fish Populations
Nez Perce tribal elders believe that one of the greatest tragedies of this century is the loss of
traditional fishing sites and Chinook salmon runs on the Columbia River and its tributaries. They
believe the circle of life has been broken and ask us to consider what the consequences of
breaking that circle may mean for future generations. In many ways the loss of the salmon
mirrors the plight of the Nez Perce people. The elders remind us that the fates of humans and
Nez Perce Tribe
Volume III - Appendix H
December 2016 Page H-19
-------
salmon are linked (Landeen and Pinkham, 1999). This dependence on fish to meet dietary,
spiritual, and basic subsistence needs is still a prevailing necessity of Nez Perce life. To this day,
the right to a "fair share" of the salmon harvest by the Nez Perce Tribe does not occur because of
the impacts to these fish by non-Indian activities and development in the Columbia and Snake
basins.
The Nez Perce lived in the heart of salmon country - along the Salmon, Snake, Grande Ronde,
Imnaha, Clearwater and Tucannon rivers; which historically were major salmon and steelhead
producers. The Nez Perce have lived through and experienced the extirpation of entire
populations of fish by blocking and altering of thousands of miles of rivers and streams as result
of dams. The Hells Canyon, Oxbow and Brownlee on the Snake River, Wallowa Lake Dam on
the Wallowa River, Dworshak Dam on the North Fork Clearwater, the eight major dams on the
Columbia and Snake rivers, and the many other smaller projects, have individually and
collectively impacted fish, and thus the Nez Perce ability to fish for them.
The environment and water that support fish has been altered due to human development and
enterprise over the past century and a half. This human progress has come at a cost to the fish
species and "salmon people." Current productivity of salmon- producing streams is much lower
than it was historically. Many of the fish species either face extinction or are in seriously
depressed conditions. As a result, tribal harvest in the present day is only a very small fraction of
what the Nez Perce harvested in the mid- 1800's. Although hard to quantify, it is probable that
until recently harvest has been less than 1% of historic harvest levels prior to 1855.
Causes contributing to salmon and steelhead decline encompass a variety of human activities and
anthropogenic and natural phenomena. These include the following: commercial, recreational
and subsistence fishing; freshwater and estuarine habitat alteration due to urbanizing, farming,
logging, and ranching; dams built and operated for electricity generation and flood control; water
withdrawals for agricultural, municipal, or commercial needs; stream and river channel
alterations; hatchery production; predation by marine mammals, birds, and other fish species;
competition with other fish species; diseases and parasites; and reduction in annual nutrient
distribution from spawned-out salmon to the local ecosystem. These activities continue to affect
fish.
Salmon and steelhead runs in the Snake Basin are not as abundant or productive as they were
historically. Snake River Chinook salmon (spring, summer and fall runs, and sockeye) and
steelhead are listed under the Endangered Species Act (ESA). Coho and Chinook salmon were
extirpated in the Clearwater River subbasin in the 1990s, and steelhead were at very depressed
levels.
Snake River spring/summer Chinook salmon were historically found spawning in the Snake
River tributaries of the Clearwater, Salmon, Weiser, Payette, and Boise Rivers. A review of run
size for Snake River of spring/summer Chinook salmon is provided by Matthews and Waples
(1991). Their summary of research on run size reports historic runs in the Snake River probably
exceeded one million fish annually in the late 1800s. By the mid-1900s, the abundance of adult
spring and summer Chinook salmon had greatly declined to near 100,000 adults per year in the
1950s. Since the 1960s, counts of spring and summer Chinook salmon adults have declined
considerably at the lower Snake River dams (IDFG, 2013).
The construction of hydroelectric dams on the main stem Snake and Columbia Rivers blocked
access to nearly half of the historic spawning habitat and reduced survival of juveniles and adults
Nez Perce Tribe
Volume III - Appendix H
December 2016 Page H-20
-------
migrating to and from the ocean. Additional effects from hydroelectric dams and water storage
projects have resulted in altered hydrographs and water temperature regimes affecting run timing
of juveniles and adults. Diversions in spawning and rearing streams have caused direct mortality,
loss of habitat and migration barriers. Land management activities have resulted in degraded
habitat with the loss of riparian cover, sedimentation and artificial barriers to passage. The
addition of hatchery programs to mitigate for lost habitat and survival of fish have introduced
genetic concerns about effects to wild stocks. Declining water quality from increasing
development in and along river and tributary streams can affect fish populations. Introductions of
non-native fish in some waters can increase predation and competition with juvenile fish (IDFG,
2013).
Salmon runs in the Clearwater River Subbasin were virtually eliminated by the construction of
hydroelectric dams (Mathews and Waples, 1991). In 1910, the Harpster Dam, constructed on the
lower South Fork Clearwater River, prevented all fishes from returning upstream of Harpster, ID,
and eliminated access to over 95% of the watershed and its high quality spawning grounds
(Schoning, 1940). In 1927, the Washington Water Power Diversion Dam constructed just above
the mouth of the Clearwater River eliminated all upriver salmon runs (Parkhurst, 1950; USFWS,
1962). A crude fish ladder was built on the lower Clearwater River dam, which allowed
steelhead passage during higher flow periods, but proved almost impassible during lower flows
when salmon arrived (Parkhurst, 1950). The ladder was not modified for a period of 12 to 14
years; eliminating all late returning fish, like coho and fall Chinook salmon (all as cited in
Everett et al, 2006).
The cumulative loss of anadromous fish to the Nez Perce Tribe as a result of these two dams was
substantial (Cramer et al., 1993). The Harpster Dam was removed in 1963 and the lower
Clearwater River dam was removed in 1972, making available most of the salmon production
areas in the drainage. However in 1971, Dworshak Dam was built just upstream of the mouth of
the North Fork Clearwater River. Dworshak Dam lacks fish passage, resulting in the permanent
loss of productive salmonid spawning aggregates and high quality habitat. The lower Clearwater
River temperature regime continues to be altered by Dworshak Dam, resulting in warmer water
in the winter and cooler water in the summer (Arnsberg et al., 1992, Arnsberg and Statler, 1995;
all as cited in Everett et al., 2006).
Currently, a majority of the fisheries that occur in the Snake River basin are supported by
hatchery programs. All of the anadromous fish hatcheries in the Snake River basin are mitigation
hatcheries for the development of hydroelectric dams. All of the returns from these hatcheries
pass through or return to the Nez Perce Tribe's usual and accustomed fishing places.
2.4 Vision for the Future
The Nez Perce Tribe has a vision of restoring all fish species native to the Nez Perce Treaty
Territory. To that end, the Tribe has engaged in management of all fish species- both resident
and anadromous - for all streams, lakes and watersheds within their management authority. The
Tribe is involved in these efforts to protect implementation of treaty rights, to restore species and
conditions consistent with the treaty, and to protect the long-term productivity of their natural
resources.
Today, maintaining a healthy 13-plus million acre watershed and improving survival of salmon
and steelhead under the auspices of the 1855 Treaty, rests with the Tribe's Department of
Fisheries Resources Management program and policy direction from the Nez Perce Tribal
Nez Perce Tribe
Volume III - Appendix H
December 2016 Page H-21
-------
Executive Committee (NPTEC), the governing body of the Nez Perce Tribe. Native fish within
the Nez Perce Country depend on healthy habitats, healthy watersheds, and healthy ecosystems.
Sound fisheries and habitat management actions will be implemented to improve survival,
production, recovery and restoration of all populations of native anadromous and resident fish
species and their habitats throughout the Nez Perce Tribe's usual and accustomed fishing places.
It is the Tribe's desire that all species and populations of anadromous and resident fish and their
habitats will be healthy and harvestable throughout the Nez Perce Tribe's usual and accustomed
fishing places.
As described in the Department's Strategic Management Plan (NPT, 2013), Tribal member use
of and access to all treaty rights and resources guaranteed under the Treaty of 1855 guide's the
department's restoration program and actions:
• All native anadromous fish and resident fish have had long-standing cultural significance
to the Nimiipuu, including: subsistence value, ceremonial and spiritual value, medicinal
value, economic or commercial value, and intrinsic value.
• Native fish populations thrive best under natural or normative conditions to which they
are best adapted.
• Natural ecosystems have been and will continue to be increasingly stressed and altered by
human activities and population levels.
• When historic natural conditions are not achievable, altered ecosystems should function
adequately enough to maintain harvest opportunities.
• The entire life cycle of a species must be successfully carried out (from egg through
adulthood) for that species or population to persist.
• Failure to serve a species' needs, at any life history stage, can lead to extirpation of
populations.
• Federal governmental agencies have treaty trust responsibilities; their actions must
recognize the treaties as federal commitments and their actions must be taken in support
of a tribe's ability to exercise rights guaranteed in the treaties.
The following goals seek to secure the integrity of populations and habitat features essential to
anadromous and resident fish:
• Achieve and maintain fish abundance in tributary-specific areas at levels sufficient to
support: 1) population persistence, 2) harvest, and 3) ecological processes.
• Achieve and maintain diverse and productive ecosystems with species composition and
productivity consistent with historic conditions.
• Achieve and maintain adult spawner distribution consistent with historically utilized
tributaries (includes within and across tributary spatial scales).
• Achieve and maintain fish population genetic diversity at levels adequate for population
persistence and consistent with historic conditions.
• Ridge top to ridge top watershed protection and restoration for rearing and spawning
habitats and protection of water quality.
• Supplementation approach "putting fish in the rivers" with hatchery tool.
• Protection and providing flows, water quality and passage for upstream and downstream
migrants.
• Participate in Pacific Salmon Treaty and US v Oregon for ocean and in-river harvest
management.
Nez Perce Tribe
Volume III - Appendix H
December 2016 Page H-22
-------
• Allow an abundance of spawners to protect the resource for future generations.
• Monitor our activities and the runs to determine how things are faring.
• Harvest opportunities currently available will be protected and enhanced.
The Nez Perce Tribe continues to protect and enhance abundance of fish through natural
production and artificial production in the form of hatcheries. Hatcheries for salmon and
steelhead in the Columbia Basin were developed as a necessary mitigation tool to compensate for
the fishery losses that resulted from the impacts of increased human settlement that began soon
after ratification of the Treaty of 1855.
Accordingly, hatcheries represent a promise to those who have always depended on the salmon
for culture, sustenance, and livelihood to replace the fish that are and were diminished as a result
of human development of salmon habitats. In the Snake River Basin, all but one of the hatcheries
(Kooskia), were built specifically to mitigate for the impacts of the development and operation of
hydroelectric dams (Dworshak, Brownlee, Hells Canyon, Oxbow, Lower Granite, Little Goose,
Lower Monumental, Ice Harbor, McNary, John Day, The Dalles, and Bonneville dams). These
hatchery programs play a very important role in meeting congressionally mandated mitigation
obligations and treaty trust responsibility to protect and maintain tribal treaty reserved fisheries.
The Department has been a leader in implementing supplementation programs and hatchery
reform. Tribal goals for supplementation programs are: increased abundance (both total and
natural origin) and spatial structure; maintenance of culturally and economically important tribal
salmon fisheries; contribution to non-Indian fisheries; and restored ecosystem processes and
health.
The Fisheries program has over 150 employees and operates on a budget derived from more than
50 contracts. There are 7 divisions within the program: Administration, Conservation
Enforcement, Harvest, Production, Research, Resident Fish and Watershed. The Fisheries
program works throughout the ceded lands and has offices in Powell, Red River, Grangeville,
Orofino, McCall, Sweetwater, Lapwai and Joseph, OR. Tribal staff coordinate and interact with
State, Federal and Tribal agencies and committees and private entities in assessing and
implementing fish recovery and restoration plans and actions.
The Department has engaged in a significant body of work throughout its U&A areas -
implementing more restoration actions within the Snake River basin than perhaps any other
single entity or agency. The aquatic habitat is subject to a diverse array of natural and
anthropogenic influences and impacts and given the synergistic effect of watershed health on
aquatic habitat quality, the Department employs a "ridge-top to ridge-top" approach to
restoration.
The Department adopted abundance-based reference points (thresholds) for certain anadromous
fish to assist in development of long-term management strategies and to guide the
implementation of short-term management actions to achieve both broad and population-specific
salmon rebuilding goals. Adult salmon abundance (or escapement) objectives are our primary
measure for quantifying goals and are generally defined as the number of adults and jacks in
each population that return to their river of origin.
These identified abundance thresholds serve as useful decision criteria that trigger specific
actions (e.g. harvest rates or initiation and other management actions). Populations at very
depressed to critically low levels require "more aggressive actions and demand a more rapid
Nez Perce Tribe
Volume III - Appendix H
December 2016 Page H-23
-------
population response than populations fluctuating at higher, less risky levels of abundance."
Reference abundances or population designations specified in this section include the designated
escapement objective, and the ecological escapement objective for four focal species,
spring/summer Chinook, steelhead, and fall Chinook (see Table below). The following are
descriptions for each threshold type.
• Viable abundance thresholds are considered the size at which a population maintains
essential genetic diversity, and at which there is negligible risk of long-term extinction
given contemporary levels of environmental variability. They are the minimum
abundance for a healthy population.
• Sustainable Escapement Objectives describe the numbers of returning adults that would
annually sustain substantial spawning as well as harvest for tribal and non-tribal fisheries.
It is assumed that escapement sizes reflecting these values would also encompass healthy
tribal and non-tribal fisheries downriver.
• Ecological Escapement Objectives refer to the escapement level at which sustainable
spawning abundance is maximized within a population, the full utilization of available
spawning and rearing habitat is promoted, and the ecosystem-level processes (e.g.,
nutrient redistribution) for multiple species are fostered. Historical salmon and steelhead
escapement to the Columbia and Snake river basins was 8-16 million and 500,000 - 2
million, respectively (NPPC, 1986; CBFWA, 1990; Chapman, 1986; Fulton, 1968).
According to tribal knowledge, escapement at those historic levels to tributary-specific
areas resulted in "fish so thick you could walk across their backs."
The following table depicts these abundance thresholds for certain fish species.
Table 2-1. Abundance Thresholds for Certain Snake River Anadromous Fish
Species
Major
Population
(iron p
Population! s)
X'iahle
Abundance
Sustainable
1 lar\ est (ioal
Ixolouical
l-scapement
(ioal
Spring/Summer
Chinook
7
41
31,500
215,900
669,000
Fall Chinook
1
1
3,000
39,110
86,300
Steelhead
6
25
25,500
330,200
602,000
The Nez Perce Tribe intends to increase and expand the level of harvest or fishing areas for
salmon and steelhead at all Nez Perce usual and accustomed places, including those in the Snake
Basin, in a way that balances conservation needs of the fish with the right to take fish. This can
be achieved through a biologically-sound harvest management philosophy and harvest rate
schedules keyed to the status and trends in abundance and productivity of fish resources.
Generally, abundance-based tribal harvest strategies can be designed to account for annual
variation in total fish run size and run composition. This is illustrated in the Figure below.
Nez Perce Tribe
Volume III - Appendix H
December 2016 Page H-24
-------
CD
£
CD
X
ra
jZi
H
TD
~0
CD
High
Moderate
1 nw
Minimal
Extintion Viable
Sustainable
Ecological
Adult Escapement Threshold
Figure 2-1. Abundance-Based Tribal Harvest Goals
As returns increase, the Nez Perce Tribe expects to increase the relative magnitude of tribal
harvest and fishing effort and fish consumption.
When restoration efforts result in sustainable returns, the Tribe anticipates that Tribal harvest
will increase and fish consumption rates will rise when fish populations attain "sustainable
abundance" and "ecological abundance" levels of adult escapement. Ultimately, the goal is to
achieve a harvest consistent with pre-Treaty harvest levels. Simply put, the Tribe's goal is to
rebuild the Snake River fishery to healthy, self-sustaining levels that will in turn support
sustainable treaty fisheries.
2.5 Estimating Heritage Fish Consumption Rates
Based on discussions with Tribal representatives and other experts on the issues of suppression
and heritage fish consumption rates, the survey design team recommends that, as part of the
survey implementation phase, heritage fish consumption rates be estimated for each of the
individual Tribes. The design team believes that current survey respondents may provide useful
information and context regarding heritage consumption rates, but that the approach to
estimating heritage rates should be primarily based on a comprehensive review and evaluation of
literature that is relevant to heritage rates, including historical accounts and modern studies of
heritage consumption rates.
For Tribes that harvest fish from the Columbia River basin, there is a significant volume of
literature to form the basis for quantitative estimates of fish consumption rates, or ranges of rates.
Information includes ethnographic studies, personal interviews, historical harvest records,
archaeological and ecological information, and nutritional and dietary information.
Nez Perce Tribe
Volume III - Appendix H
December 2016 Page H-25
-------
During project implementation, the survey team will compile and evaluate relevant available
information regarding heritage consumption rates specific to the NPT. The development of
estimates of heritage rates should include a thorough discussion of the types of information
available regarding consumption, a discussion of the methodologies used to develop the
estimates, and a discussion of factors affecting the uncertainty associated with the estimates.
Finally, the implementation team should develop a quantitative estimate of a heritage fish
consumption rate or range of rates for the Tribe.
One aspect of the quantitative assessment will be a compilation and analysis of historic and
heritage information across the region (primarily for the Idaho Tribes). The purpose of this
compilation and analysis will be, to the extent possible, to reduce the uncertainty associated with
individual heritage rates or update the rate calculations by a statistical methodology that uses
data for multiple Tribes, locations, and times. An analysis which shows consistency in
relationships among these variables will support the individual heritage rates. Further, it may be
possible to estimate a range of rates for the Tribe based on a joint (multivariate) analysis of
heritage, including tabular and graphical displays and numeric estimates of a plausible range.
Nez Perce Tribe
Volume III - Appendix H
December 2016 Page H-26
-------
3.0SURVEY DESIGN: TARGET POPULATION
This section describes the survey design approach as it relates to the target sample population
and sampling frame, including phasing in of multiple surveys.
3.1 Target Population to be Sampled
The target population for the survey is enrolled adult members of the NPT, age 18 and over. The
population to be sampled in this survey can be tentatively defined as enrolled adults (age 18+)
who live within a specified geographic area around the NPT Reservation, e.g., a distance
reflecting up to a reasonable drive time, such as 1-2 hours. While a distance cut-off may appear
arbitrary for a population definition, some kind of practical cut-off is needed, since some tribal
members may reside at great distances from the reservation. Distance will be defined by zip code
or location of residence in relation to a central site for interviewing. The site or sites will be
identified in cooperation with the Tribes. Due to the expected high correlation of diets and the
substantial time per interview, the survey will be limited to enrolled tribal members and will not
include non-tribal spouses or other non-tribal adults. The residential location of all members will
be checked with the Tribes just prior to the sample selection. The specific tribal members in the
population to be sampled will be identified from the Tribal enrollment roster in cooperation with
the Tribal authorities.
Among the adult population, there will be a sub-population of non-consumers of fish, and these
people would be detected in an initial telephone screening (described in Section 4.4.1). For the
non-consumers, defined as those who have not eaten fish in the last year, the screening will
determine the reasons for non-consumption, such as taste (dislike of fish), concern about
advisories or pollution, or other reasons. No further information will be collected for non-
consumers (some demographic information will be available from the tribal enrolment roster),
and the main focus of the effort on the fish consumption survey will be confined to fish
consumers only.
An exclusion from the sample, if they should be selected in the sampling process, is persons
living in an institutional setting (e.g., nursing homes). The reason for the exclusion is that in this
special population, expected to be small, a totally different questionnaire and data collection
method would be needed. Secondly, an institutionalized person is usually not free to make
decisions about their fish consumption, and it is not clear to what extent that consumption
represents the tribal way of life.
Another exclusion is the tribal sub-population of children and young adults (age <18 years). This
demographic group has been excluded at this time to avoid a potential insufficient sample size in
an effort to shorten an already detailed and lengthy interview process for each adult interviewed
and collectively ensure an adequate number of adult interviews within the resources available.
3.2 Phasing-in of the Survey
The design team recommends that the survey implementation be carried out in phases, with one
or two Tribes selected initially to start. It is likely that a great deal will be learned about what
works well and what does not work during the early part of the survey. While the survey design
is certainly intended to provide an excellent framework for all of the Tribes, it is inevitable that
important working points will be learned as the implementation team proceeds, collaborating
with these unique populations. Thus, the survey might start with one or two Tribes and then
proceed to a second and a third, etc., at short intervals. Alternatively, the survey may start with
Nez Perce Tribe
Volume III - Appendix H
December 2016 Page H-27
-------
one Tribe but then proceed with the other Idaho Tribes with a modest delay after that. This is a
decision that is best made closer to initiation of the survey. The survey team will communicate
with tribal fishery staff to determine the several seasons of fishing and fish consumption. The
survey will be scheduled to overlap significant seasonal periods.
3.3 Sampling the Population
The enrollment roster of the NPT will be the sampling frame and basis for sample selection. The
roster is expected to be reasonably complete and up to date, since tribal membership includes
benefits that motivate enrollment. The enrollment roster is expected to include age, birth date,
gender, address (including zip code) and other fields.
The population to be sampled will be limited to specified zip codes or other location indicators.
As noted earlier, the locations will be selected in order to accommodate a reasonable amount of
travel time for members to attend a central site for interviewing. It may be possible for some
interviews to be conducted closer to or at a respondent's home when there are issues of health
and ability to travel.
The Nez Perce Tribe has supplied the data on their adult population counts by zip code of
residence. The design team will use the data to fill in Table 3-1 for the NPT. The table will help
the design team and the Tribes to decide on the geographic area from which survey participants
will be selected.
Table 3-1. Number of adult Tribal members by distance from Tribal reference point
defined by zip code of residence
Distance (miles)
No. of members
Zip codes included
<5
N
11111, 22222, 33333, etc.
5 to <10
N
44444, 55555, etc.
40 to <50
N
88888, 99999, etc.
Etc.
1. Sample Stratification
The eligible adult population (defined by age 18+ and an eligible zip code of residence) will be
sampled using stratification. "Strata" are simply population groups defined by some
characteristic. For example, six strata might be defined by age and gender to include young
adults, the middle aged, and Tribal elders, classified separately by each of the two genders. One
use of stratification is to insure that the sample will represent the population faithfully. For
example, if six strata (not necessarily age-related) cover the whole population and have about
one-sixth of the population each, then one-sixth of the sample can be drawn from each stratum.
An ideal stratifying factor for this survey would be defined by an a priori indication of level and
frequency of fish consumption. High-level consumers are needed since there is particular interest
in the higher percentiles of fish consumption, which the high consumers would tend to define.
Nez Perce Tribe
Volume III - Appendix H
December 2016 Page H-28
-------
Secondly, as explained later, frequent consumers (who also tend to be high-level consumers) are
needed for the survey's planned use of a particular method (National Cancer Institute or NCI
method) to estimate the fish consumption distribution from two or more 24-hour dietary recall
interviews. In the use of the methodology to analyze the 24-hour recall interview data, it is
important to have enough respondents with two days of fish consumption. Currently, age,
gender, and location (defined by zip code) are the only candidates in the roster for the NPT that
might define higher vs. lower level consumers. Fish consumption rates in relation to age show
mixed results for the Native American surveys in the Pacific Northwest. Thus, the phone
screening process (Section 4.4.1) is needed to identify frequent consumers who may, then, have a
higher probability of consuming fish on the second of the two days of 24-hour dietary recall. The
second interview will occur within a time window (yet to be specified) probably of one to four
weeks after the initial interview. The time window will be selected to yield an independent eating
occasion but not so long that seasonal effects (e.g., associated with fish availability) will
influence fish consumption.
Strata will be defined by the combination of age, gender, and frequency of consumption, with
frequency determined from the phone screening process. The age-by-gender composition of the
NPT has already been provided by the Tribe. The age group breakdown will be helpful in
forming initial strata, which will then be sub-divided by at least two frequency categories, such
as consumption of fish 'two or more times/week' vs. 'less than twice per week.' Again, these
strata will both insure that the population can be well represented by the sample selected, and in
addition, allow over-sampling of the high-frequency strata. An oversimplified stratification is
shown in the table below.
Table 3-2. Hypothetical strata based on three stratifying factors: age, gender and frequency
of fish consumption
Siralum
(iender
Age group
Consumption
I'requencv
A
Male
18-44
< 2x per week
B
Male
18-44
> 2x per week
C
Male
45+
< 2x per week
D
Male
45+
> 2x per week
E
Female
18-44
< 2x per week
F
Female
18-44
> 2x per week
G
Female
45+
< 2x per week
H
Female
45+
> 2x per week
2. Sample Selection
Once the strata are defined in terms of age, gender and frequency of consumption, potential
respondents for screening will be selected randomly from each age-gender stratum (combining
the frequency strata). If there are appropriate non-disclosure agreements and adequate security
and confidentiality procedures in place, and if the NPT agrees, a copy of the enrollment file with
fields needed for sample selection can be transferred to the implementation team and then
Nez Perce Tribe
Volume III - Appendix H
December 2016 Page H-29
-------
deleted (including derived files) after there is no further need for the file or after a mutually
agreed period has expired.
If the NPT does not wish to "loan" the enrollment file for sample selection purposes, an alternate
procedure of sample selection can be used. In order to preserve the confidentiality of Tribal
members listed in the electronic enrollment file, the enrollment office will be asked to take the
following steps.
1. Apply any member exclusions (such as non-eligible zip codes and persons less than 18
years of age) and save a copy of the resulting file.
2. Add a field defining age and gender for each person. These strata labels will appear for
each person in the file.
3. Sort the file in random order. Almost any random sort software can be used here.
4. Starting with the randomly sorted file from the previous step, add a field with a new
sequential survey identification number ("surveylD"), which should be a sequential
number, e.g., 1, 2, 3, .... The correspondence between this unique survey ID number and
the Tribes' unique ID number will allow communication between the survey
implementation team and the enrollment office, as needed. Due to the random sort prior
to this step, the assigned survey ID number will be non-informative about any member
characteristics—a helpful step in preserving confidentiality.
5. Save a file which contains only the new survey ID number, and selected demographic
data (e.g., gender, age in grouped categories). Transfer this file to the implementation
team.
6. The implementation team will select the sample from the file provided by the Tribe and
return the file of the selected sample to the Tribe. The implementation team will work
with the Tribe to generate a list of the sample suitable for phone screening (including
names and contact information).
The implementation team will select the specified respondent count for screening from each
stratum by random selection. This process should be carried out under the supervision of the
statistician working with the implementation team. See the section on sample size for the
specified sample count for the NPT.
The random selection process will generate a list of potential respondents for the screening step.
This screening list will include 3 to 5 times as many individuals as the ultimate effective sample
size, since a number of individuals may need to be screened to identify each frequent consumer.
The screening list will be divided into 4 to 5 sections corresponding to waves of screening.
Within each section, the age-by-gender composition of the list will be similar to the composition
of the Tribe.
By screening in several waves, the implementation team can examine initial results to better
understand the population as well as determine what screening methods will yield a higher
percentage of frequent consumers from the first or early waves. This allows the team to refine a
sampling plan so that resources are allocated most effectively. For the 24-hour recall component
of the interviews, it is especially important to obtain a large enough number of people who
consume fish on both recalls. The implementation team will need to focus the selection effort on
identifying people who are likely to meet this condition in order to provide the best chance of
Nez Perce Tribe
Volume III - Appendix H
December 2016 Page H-30
-------
obtaining data suitable for use with the NCI method. After the initial full interview, frequent
consumers can be given a higher probability of selection for the additional second 24-hour recall
interview. All initially interviewed respondents (supplying food frequency interview and an
initial 24-hour dietary recall report) will have a positive probability of selection for the second
24-hour recall. However, frequent consumers will be assigned a higher probability of selection.
While all respondents supplying an initial 24-hour dietary recall will have some probability of
selection for the second 24-hour recall, not all of them will be selected. Nevertheless, all of those
selected for the second 24-hour-recall will be selected on a probability basis from the first recall
and not by a categorical selection that absolutely excludes some first-recall respondents.
A list of respondents to be interviewed in person (in waves, corresponding to the sections of the
screening list) will be generated by the screening process. The initial screening list will be turned
over to Tribal members hired to help with the survey, and they will carry out the screening
process under the direction of the implementation team.
Nez Perce Tribe
Volume III - Appendix H
December 2016 Page H-31
-------
4.0SURVEY DESIGN: DATA COLLECTION
This section describes the survey design approach as it relates to the survey method,
measurement method, sample size, and questionnaire development.
4.1 Survey Methods
Based on our experience, in-person interviews are superior to many other survey research
modes for many reasons; however, for most studies, in-person interviews are cost prohibitive
and a compromise must be achieved between "best practices" and budget constraints. In-
person interviews allow the respondent to see survey aids (in the case of this study,
photographs and models) and to establish a face-to-face connection with the interviewer. In
addition, respondents generally tolerate longer in-person interviews than telephone or other
interview modes (Doyle, 2005).
1. Selection of In-Person Interviews vs. Other Methods
Based on a review of the literature and decades of experience, we have identified several
possible modes for this study. Below is an examination of various modes but, in a summary, we
recommend in-person interviews for this survey. They are a superior solution for this project due
to their inherent cultural advantages and the expected length of the interview for this survey.
Although mail surveys are generally less expensive than other modes (in-person, telephone,
online), they suffer from poor response rates. Without a staff member prompting the potential
respondent to complete the interview, it is very easy for recipients to discard the questionnaire
without opening it. Further, self-administered mail questionnaires are rife with opportunities for
respondents to provide incorrect, improper, or no answers to questions that they do not
understand or do not care to answer. A telephone interview, an in-person interview, and online
interview can all be structured in a way to alert the respondent when they've failed to answer a
question or gone outside the choice parameters—a mail questionnaire cannot do that. Based on
our research, mail questionnaires are insufficient for high-quality data collection, especially for
long interviews. (The anticipated length of this interview is approximately one hour.) Finally,
mail surveys exclude members of the target population who are not literate.
Telephone studies are a popular mode of survey research, allowing for centralized management
of the sample frame, the interviewers, and project administration. Telephone surveys, when
programmed with computer-assisted telephone interviewing software, can include complex skip
patterns and other calculations which are less feasible with mail surveys and in-person
interviews. Telephone studies allow convenient monitoring and supervision of the interviewing
staff, ensuring consistent administration of the questionnaire. However, telephone studies lend
themselves to social desirability bias, the notion that a respondent seeks to provide answers
which will increase the likelihood that the interviewer "likes" the respondent (Maguire, 2009).
Further, telephone studies are limited to respondents with telephones, obviously; it is difficult to
ensure 100% coverage within the sampling frame if it is based on the telephone alone.
The telephone approach also has another disadvantage for dietary surveys. With a telephone
interview it is more challenging to use visual aids for identifying species and quantifying
portions. While materials might be mailed or emailed in advance of the interview, that is another
level of complexity for the survey and the respondent, and it may be difficult to have the proper
conjunction of pre-sent materials and the specified interview appointment. Further, the planned
Nez Perce Tribe
Volume III - Appendix H
December 2016 Page H-32
-------
interview goes into some detail on a number of topics and the hour or hour-plus duration of a
phone interview may lose cooperation and accuracy of reporting.
2. Use of Photographs and Portion Size Models
There are different ways to measure respondent food consumption, including administering
questions verbally, with or without visual aids. The use of aids such as photographs and portion
size models is a well-accepted measurement device when collecting respondent-reported data.
This is consistent with other, large-scale, ongoing survey research projects, such as the National
Health and Nutrition Examination Survey (NHANES), which uses portion size models for its
initial in-person 24-hour dietary recall. The portion model representation will include composite
dishes, such as stews, chowders and other mixtures.
In order to ensure the most accurate self-reported data about past food consumption, we strongly
recommend the use of either photographs, portion size models, or a combination of both for this
survey. Although photographs lack the tactile and 3-dimensional visual appeal of portion size
models, they have been shown to be equally as effective (providing accurate measurement) as
portion size models (Thompson and Subar, 2013). During the pilot test, portion models should be
used to verify their efficacy.
The design team is collecting displays to use as species and portion-size choices for use in the
interviews. See Section 4.4.6 for more information about development of these portion size
models and other visual displays that will be useful tools for respondents to indicate fish
consumption types and quantities during survey implementation.
3. Use of Tribal Interviewers
This project represents an important step in the evaluation of fish consumption among native
populations in Idaho. To encourage participation from respondents, professional interviewers
will administer the questionnaire to each respondent. The interviewing staff will be selected,
hired, and trained from among NPT members. Tribal representatives reported that Tribal
interviewers are necessary to gain and maintain respondent trust. Further, Tribal interviewers are
familiar with the local area.
Complementary goals during the survey include decreasing respondent burden and increasing
respondent comfort. We expect that an interviewer who shares heritage with the respondent can
more easily identify and adhere to cultural norms and sensitivities. The interviewer may be more
attuned to the respondent's background, living situation, and local conventions and events. In
short, we expect greater affinity between respondents and interviewers who are from the same
Tribe than between respondents and interviewers who are not Tribal members. Additionally, this
study covers a broad geography in rural Idaho. In addition to our efforts to match interviewers to
anticipated socio-demographic characteristics of respondents, by using local Tribal interviewers,
study and travel costs may be reduced.
4.2 Measurement Method
The survey will use two methods to measure current fish consumption. The first method will be
based a food frequency questionnaire (FFQ) which ascertains species-specific frequency of
consumption and typical quantities eaten per eating occasion. The questionnaire will also allow
these quantities to vary by 'season' with up to two periods per species. A 'season', as the term is
used here, is one or more periods when the respondent reports consuming fish at a rate different
Nez Perce Tribe
Volume III - Appendix H
December 2016 Page H-33
-------
than that of other periods during the year. Some species may be consumed by a particular
respondent year-round at about the same rate, and that respondent would have one season (over
one year) for that species. Consumption on ceremonial occasions and other special events will be
covered by separate questions. See the questionnaire section of this document for the questions
and wording of the FFQ (Appendix A).
The principle behind the FFQ is as follows. Briefly, a respondent's frequency of occasions of
consumption of fish (per day, week, or month) multiplied by the typical quantity eaten per
occasion will give the total quantity eaten per day, week, or month. This quantity is easily
converted to total annual consumption, which, divided by 365 days, will yield an average
quantity of the given fish species eaten per day. A straightforward extension of this basic
method, described later, can include seasonal variation and consumption at special events.
The strength of the FFQ is that average frequency and quantities of fish consumption are
reported directly by the respondent. The weakness of the FFQ is that the respondent is relying on
memory and must internally average their varying frequencies and varying quantities of
consumption to come up with 'typical' values.
The second method is based on the respondent's recall of fish consumption during two or more
specified 24-hour periods. Each period is the day before an in-person or telephone contact. The
second (and later) interviews will be matched on the weekday vs. weekend occurrence of the
initial 24-hour recall interview for a given respondent. The reason for this day-matching is to
hold other variables relatively constant so that the variation between days of consumption is
random variation in consumption per se and is not influenced by other weekly cycles of eating.
For example, the difference between weekday and weekend fish consumption may be a fixed
average difference and not simply random variation. (With a substantially larger sample size than
will be used in this survey, the NCI method, by using certain information collected about each
eating occasion, could accommodate a mixture of weekday and weekend fish consumption per
respondent.)
The second step in working with the 24-hour recall surveys is use of the 'NCI method' to
analyze the data collected (Tooze, et al., 2006). The NCI method uses some assumptions and
12
statistical models to generate a fish consumption distribution that is consistent with the
observed data in the two 24-hour dietary recalls.
A strength of the NCI method is that the respondent is having to remember only items and
quantities consumed on the previous day. A weakness of the NCI method is that some strong
(but reasonable) assumptions are needed to generate the distribution of average daily intake for a
population. An additional weakness of the NCI method in the context of a fish consumption
study is that it may be able to supply consumption estimates only for all fish species combined
and for one or two frequently consumed species. For the less frequently consumed species there
may be too few consumption 'hits' on the sampled recall days to support a meaningful analysis.
The design team recommends that the questions on the 24-hour recall be constructed to support
estimates of frequency of consumption for a) all species combined, b) anadromous species, c)
freshwater resident species, and d) marine species. The ability to make the consumption
estimates for each of the individual species groups a, b, c, and d using the NCI method depends
12 By 'distribution' in this report we are referring to values of the mean, median, and higher percentiles of the population's fish
consumption rates. 'Distribution' has a more technical definition in the statistical literature.
Nez Perce Tribe
Volume III - Appendix H
December 2016 Page H-34
-------
on having an adequate number of respondents who report eating from the species group on both
of the two 24-hour recall interviews. However, even if the NCI method cannot be used, the FFQ
will be designed to allow calculation of the consumption rate distribution for each of the major
species, for all species combined, and for various groups of species.
The FFQ and the 24-hour questionnaires that will be used to support the fish consumption
estimates can be viewed in Appendix A of this document.
4.3 Sample Size
Multiple sample sizes are considered here, corresponding to the following survey components:
• Initial telephone screening operation to identify non-consumers and high consumers
• Food frequency questionnaire (FFQ)
• 24-hour recalls (1st and 2nd recall days)
Some strata (or groups) of respondents will be sampled at a higher rate than others. For example,
when characteristics of more frequent consumers or high consumers of fish are identified, a
stratum of these tribal members will be sampled at a higher rate than members not in this
stratum. Currently, the design team recommends that the high or frequent consumers be
identified by the initial telephone screen. If one-quarter of the consumer population consists of
high consumers, they may be sampled at four times the rate as the lower-level consumers,
resulting in more than 50% of the sample consisting of high consumers. In the statistical analysis
following data collection, each sampled high consumer would carry one-quarter of the weight
compared to a low-end consumer in order to represent the entire population in an unbiased way.
However, despite their quarter-weight, the extra sampling of high-end consumers will provide
greater precision in estimation of the higher percentiles of fish consumption—percentiles of great
importance in water quality regulation. Also, the over-sampling of high consumers will provide a
better basis for carrying out the NCI method of analyzing the 24-hour recall data.13
For each sampling operation considered, the driving factor in selection of a sample size is the
trade-off between precision of an estimate—which improves with increasing sample size—and
the mounting cost of a survey as sample size increases
1. Screening of Participants
An initial telephone screening call will be carried out to identify any non-consumers of fish and
note reasons for non-consumption (described in more detail in Section 4.4.1). Non-consumers
will not receive a personal interview.
2. FFQ Sample Size
Prior to presenting notes on sample size for this survey, a caveat is that the final sample size will
depend on results from the survey pilot testing and telephone screening as well a critical
dependence on resources available to this project to carry out the surveys for the Idaho Tribes.
The desired effective sample size for the FFQ will be approximately 140 fish-consuming
individuals. The "effective" sample size is smaller than the number of individuals sampled,
13 One of the assumptions of the NCI method is that the within-person variance of the logarithm of the quantity consumed on a day
with fish consumption is constant across all levels of consumption. If the assumption is true, there is no disadvantage to over-
sampling high consumers. It may be possible to check this assumption if there is a sufficient number of respondents with two days
of consumption.
Nez Perce Tribe
Volume III - Appendix H
December 2016 Page H-35
-------
because high consumers will be over-sampled in proportion to their numbers in the population.
The effective sample size here takes into account the statistical weight given to each individual.
A speculative guess is that 25% of consumers14 in the Tribe will be high consumers and if the
high consumers are sampled at a fourfold rate compared to the low-consuming balance of the
consuming population, then approximately 245 individual respondents will be included in the
sample. The 245 individuals would include approximately 105 low consumers and 140 high
consumers. The 140 high consumers would each have one-quarter statistical weight, yielding an
effective sample size of 35 high consumers. (The full 140 high consumer respondents would be
included in the analysis, but four high consumers carry the same statistical weight as one low
consumer, thus the effective sample size of 140/4 = 35 for high consumers.) The 105 low
consumers plus the effective sample size of 35 high consumers yields a total effective sample
size of 140.
Based on some preliminary simulation analyses, 140 completed FFQ questionnaires from
randomly selected Tribal members would yield a mean consumption rate with a 95% probability
of falling within +/- 25% of the true population value.15 This is a conservative estimate of
precision (i.e., precision would likely be better), because the effective sample size of n = 140
stems from a much larger sample size of individuals, due to over-sampling of high-consumers.
Under the same conservative assumptions, the 90th and 95th percentiles will have 95% probability
of falling within about 40% of the true population value. Figure 4-1 shows the relationship
between sample size and precision. It is apparent from the diagram that achieving high precision
for the higher percentiles requires quite large sample sizes.
In order to yield approximately 140 high consumers and based on 25% high consumers and 30%
refusals or no contact, the screening list will need to include approximately 800 individuals.16
The proportion of the population who are high consumers and the survey non-participation rate
are speculative. For that reason, a phased start to the survey, as described in Section 3.2, is
important with the implementation team learning from each wave of screening and then
adjusting methods for the next wave.
14 The Columbia River Inter-Tribal Fish Commission survey results (CRITFC, 1994) reported that 38% of adult fish consumers had
two or more fish meals per week (Table 8). Given that some respondents may have consumed two or more of their weekly fish
meals on a single day, the value of 25% of respondents consuming fish on two or more days per week (i.e., high consumers) may
be a reasonable value to assume for this work.
15 The simulations were samples of size n = 100, 200 and higher from hypothetical surveys of populations with a lognormal
distribution offish consumption rates for consumers only. Different populations were considered to have mean consumption rates
varying from low to medium to high (mean ± SD of 19 ± 21 g/day, 82 ± 128 g/day and 214 ± 273 g/day, respectively). For each
population and sample size 10,000 simulated 'surveys' of the given sample size were drawn and the sample mean, median and 90th
and 95th percentiles were calculated. From the simulation distribution of a descriptive statistic, such as the mean, the 2.5th and 97.5th
percentiles of the descriptive statistic were calculated. This range, though not a confidence interval, shows estimated limits within
which 95% of survey results for the specific statistic would be expected to fall for the given population and sample size. Across the
low, medium and high fish consumption populations the maximum percentage difference of the limits from the true mean was 25%
for a sample size of 140 (using linear interpolation between sample sizes of n = 100 and 200). For the 95th percentile of consumption
the corresponding maximum percentage deviation from the true 95th percentile was 39%.
16 Approximately 200 high consumers would need to have contact attempts in order to yield 140 net high consumers after a 30%
loss rate. If 25% of Tribal members are high consumers, 800 Tribal members (of any consumption rate) would need to be contacted
to find the 200 net high consumers. The low consumers can be selected from the remaining 600 Tribal members—the balance of
the 800 who are not high consumers.
Nez Perce Tribe
Volume III - Appendix H
December 2016 Page H-36
-------
Impact of Sample Size (High Rate)
Mean
Median
90th PcL
95th Pet
True Rate
~~r~
50
100
200
400
Sample Size
Figure 4-1. Precision of mean and selected percentile estimates vs. sample size The upper
and lower bounds for each estimate are expressed as a ratio to the true value. In 95% of surveys
drawn from a population with a lognormal distribution of consumption rates, the estimated value
of the statistic is expected to fall between the bounds corresponding to the survey's sample size
(bounds are approximate from simulation).
3. 24-Hour Dietary Recall Sample Size
All of the expected 245 individual respondents will complete the first 24-hour dietary recall
assessment. All of these 245 respondents will have the possibility of selection for the 2nd 24-hour
recall interview, but the probability of selection will increase with increasing (grouped) quantity
and frequency of consumption as determined from the FFQ. The goal is to adjust the net number
sampled on day 2 of the recall to yield at least 50 respondents with fish consumption on both
days of the 24-hour recall.17
The implementation team will need to: a) choose a cut-off that defines frequent consumers in
terms of the frequency of consumption (and possibly the quantity eaten on day 1 of the recall),
and b) determine selection probabilities for day 2 of the 24-hour recall in order to have at least an
expected 50 individuals with fish consumption on both days 1 and 2 of the 24-hour dietary recall.
The key parameters in this calculation will be an estimated survey non-participation rate (refusal,
17 The minimum number of respondents—50—who consume fish on both days of the 24-hour recall has been suggested by Dr.
Kevin Dodd, one of the developers of the NCI method. This minimum sample size is based on the precision of a variance estimate.
To put the n = 50 in perspective, standard deviations (SD) based on 25, 50, or 75 degrees of freedom for samples drawn from a
normal distribution would have 95% confidence limits that differ from the estimated SD by no more than 39%, 25% or 19%,
respectively. Thus, n = 50 has an associated 25% level of precision, which is fair (not excellent) precision.
Nez Perce Tribe
Volume III - Appendix H
December 2016 Page H-37
-------
no contact, etc.) projected to the day 2 attempted contact, the percentages of day 1 recall
respondents who consume at various frequencies, and the day 1 quantity of fish consumed.
As a side note, it is possible that the number of sampled individuals with two recall days of fish
consumption will not be sufficient to yield a meaningful estimate of the fish consumption
distribution using the NCI method. In that case, the data from multiple Tribes may be pooled and
used with the NCI method, introducing the Tribe as a categorical covariate or as the person-
specific fish consumption rate for the species group being evaluated. That procedure will yield a
distribution for each Tribe. However, some assumptions about commonality among the Tribes of
certain statistics of the distributions will need to be tested and noted.
4.4 Questionnaire Development
A survey questionnaire, provided in Appendix A, was developed to help determine the fish
consumption rate of the NPT. The purpose of the questionnaire is to ask Tribal members about
their dietary patterns and activities related to fish consumption in the past 24 hours as well as in
the preceding 12 months to determine current fish consumption rates. This will be accomplished
by conducting two 24-hour dietary recall interviews (the second of which will be administered
after a week, but within four weeks after the first recall interview) and a food frequency
questionnaire, as discussed above. The second 24-hour recall will be administered to a randomly
selected sample of the first-interview respondents, weighted toward those determined to consume
fish more frequently, based on the first interview. Data will be collected regarding fish species
consumed, frequency of consumption, and portion size, with additional information gathered
about parts eaten, preparation methods, and special events. Data will also be collected regarding
changes in fish consumption patterns from the past and expectations for future consumption to
develop a more accurate FCR that is not restricted by current-day suppression factors.
The questionnaire is written such that the trained interviewer can clearly follow the line of
questioning, read each question verbatim, and record (in written form, by check box or circling)
the information given by each respondent in the space provided in a consistent manner. Words to
be spoken by the interviewer are identified in bold text on the questionnaire, and each question
will be asked in numeric order. Written information will only be recorded on the questionnaire
form by the interviewer. Entry codes, species displays, and portion displays will be used during
the interviews.
Past fish consumption surveys were reviewed, in addition to recent survey questionnaires
developed by the Center for Disease Control and Prevention (CDC), for guidance in selecting
wording for the current questionnaire. These resources are listed in Section 7. The questionnaire
will be pre-tested (during a pilot survey) and revised as necessary prior to implementation. The
questionnaire is organized according to the following sections, which are discussed in more
detail below:
1. Telephone Screening
2. Interview Introduction
3. 24-Hour Dietary Recall
4. Food Frequency Questionnaire
5. General Information
6. Second 24-Hour Dietary Recall
Nez Perce Tribe
Volume III - Appendix H
December 2016 Page H-38
-------
1. Telephone Screening
Potential respondents will first be contacted by telephone. The initial phone contact will provide
an opportunity to screen for fish consumers versus non-fish consumers and to discern why fish is
not being eaten by the non-consumers. For those who do eat fish, an in-person interview will be
scheduled with the respondent for a later date, if they are willing. The selection (or non-
selection) of a tribal member reached through a screening call will be based on the survey's
progress in filling in the required sample counts for each population stratum.
Each respondent will have his or her own Telephone Screening Contact Log. The Telephone
Screening Contact Log will be maintained separate from the interview forms, as the contact log
will provide the only documentation linking the respondent's name with the respondent's
randomly assigned identification number. Subsequent interview forms will only include the
respondent identification number to maintain confidentiality of the respondent.
This section of the questionnaire provides statements for the caller (interviewer) to make over the
telephone and a log to record every contact attempt. If multiple attempts are made, the
interviewer placing the call may vary (and may be different from the person who ultimately
conducts the interview). The Telephone Screening Contact Log will include the date, day of the
week, and time of the call, name and identification number of the interviewer making the call,
results of the call according to the entry codes provided, and whether or not the respondent
consumes fish. If an in-person interview is scheduled over the telephone, the date and location of
the interview will be recorded on the contact log.
2. Interview Introduction
The primary in-person interview will begin by documenting basic identifying information about
the interview (who, when, where) and introducing the respondent to the project and the purpose
of the interview. Administrative information will be recorded before (or as) the interview begins
and will include the interviewer's name and assigned identification number, the respondent's
assigned identification number (no name), and the date, day, start time, and location (city, state,
and venue) of the interview. After the administrative information is recorded, the interviewer
will read the introductory narrative to the respondent to formally begin the interview. The
respondent will be reminded that that their information will remain confidential. The primary in-
person interview includes three parts, the 24-hour dietary recall, the FFQ, and general
information. A second 24-hour dietary recall survey will be conducted for a subset of
respondents by telephone.
3. 24-Hour Dietary Recall
Following the introduction, a 24-hour dietary recall questionnaire will be administered to collect
information on fish dietary patterns during the previous day. The objective of this component of
the survey is to estimate total intake of fish that was consumed during the 24-hour period prior to
the interview from midnight to midnight. The interviewer will read the questions in numeric
order and complete the table, entering and circling answers as provided by the respondent.
The primary series of questions relate to the types of fish eaten over the past 24 hours, the
quantity, preparation method, and source of the fish eaten. Once the interviewer has verified
whether the respondent ate fish during the previous 24 hours, the interviewer will inquire about
fish eaten during each occasion over those 24 hours, including species type (to be coded later),
Nez Perce Tribe
Volume III - Appendix H
December 2016 Page H-39
-------
portion size (quantity), preparation method, and source of each fish meal or snack consumed by
the respondent. Species and portion displays will be used.
A representative selection of respondents, weighted toward those identified as being high fish
consumers, will be contacted for a second (separate) 24-hour dietary recall survey by telephone
after a week, but within four weeks after the first interview. The second 24-hour dietary recall
questionnaire will mimic the first, repeating the same inquiries as administered during the
primary 24-hour dietary recall interview. The method of identifying species and sizing portions
on the second 24-hour interview (by phone) is still being determined, but it is likely that it will
use either displays left with the respondent at the initial interview or else delivered to the
respondent.
4. Food Frequency Questionnaire
Following the 24-hour dietary recall, an FFQ will be administered to collect information on fish
dietary patterns and associated activities over the past year. The objective of this component of
the survey is to estimate total intake of fish that was consumed over the previous 12 months as
well as to gather information about fishing activities and other factors that may affect
consumption. The interviewer will read the questions in numeric order and complete the table in
the questionnaire.
The first series of questions relate to the species, frequency, and quantities of fish eaten. If
consumption varies with high and low-eating periods, questions will be asked for each period.
Once the interviewer has verified whether the respondent ate fish during the previous 12 months,
the interviewer will inquire about which type of species were eaten, the number of portions or
frequency that each type was eaten, and typical portion sizes. Species and portion displays will
be used.
Information will be gathered regarding parts of fish consumed, methods of preparation, and
sources of fish consumed over the past 12 months. Information will also be gathered about
activities associated with fish consumption, including special events, such as feasts and
ceremonies, as well as fishing activities. Finally, several questions will attempt to gather more
qualitative data on changes in fish consumption compared to the past and about intentions for
fish-consumption in the future.
5. General Information
General information will be collected at the end of the primary in-person interview.
Demographic information will be recorded, including the respondent's gender, date of birth, age,
height, weight, residence on or off reservation, education level, and household income. These
items are being collected to provide sub-groups for rate-reporting, to support calculations of rates
in other formats (e.g., g/kg-day), or to attempt to identify characteristics of high vs. low
consumers of fish. After the demographic information is recorded, the interviewer will ask
female respondents about their breastfeeding history (linkage to child health).
The interviewer will conclude the interview by reading the statements of appreciation, inquiring
about future contact. At that point, the interviewer will record the end time (and calculated
length) of the interview. Following the interview, the interviewer will record their opinion of the
respondent's level of participation (cooperation and reliability) and acknowledge that they
recorded the information truthfully and to the best of their ability by signing an attestation of
authenticity.
Nez Perce Tribe
Volume III - Appendix H
December 2016 Page H-40
-------
6.
Photographs and Portion Models
Portion models and graphics (photographs or other representations) will be used during the 24-
hour recall and food frequency questionnaires and will be comparable to the U.S. Department of
Agriculture (USDA) portion size booklet (and accompanying measuring implements) that is used
by NHANES for national dietary surveys. These models will provide a visual display of
quantities of fish consumed during each meal. These models will be reviewed and tested by the
implementation team prior to survey interviews, and they will be evaluated for usefulness and
appropriateness by the Tribes (and modified, if needed) during pilot testing of the questionnaire.
The portion displays have not been fully evaluated by the survey team yet, but following are
some general considerations in the selection and use of the final portion displays.
There may be a need to calibrate the portion displays to physical weights of the species
represented and for each specific portion size shown in the display. Any portion displays should
show the portions as actual (100%) size. If possible, the display should be shown to the
respondent at a distance similar to the distance between a person and their meal, without being
intrusive of personal space. This could usually be accomplished by handing the display to the
respondent and asking them to indicate the particular portion mark within the display that
corresponds to their consumption in response to a question.
All portion displays will have a specific code attached to them, and a separate table (to be used
during data analysis) will show the volume and/or weight-per-species corresponding to each
portion mark in the display. To maintain efficiency of the interview, the respondent will answer
questions in terms of simple portion marks or codes on each display, saving the interviewer a
table look-up for the species-specific weight of the noted portion.
Dishes such as stews, chowders, casseroles, and special composite dishes unique to the NPT will
have their own portion models to indicate serving sizes. For example, measuring bowls will be
used for respondents to identify portions of liquid dishes (with a fish ingredient list pre-
determined). The survey team will identify the tribal-specific dishes (only those which include
fish as a component) and obtain approximate recipes for conversion of visual portion sizes to
weight of fish by species. Other composite dishes that are reported will be handled using
standard recipes (such as that complied by the EPA) to convert respondent-reported quantities
consumed to weight of fish consumed.
7. In-House Testing and Revisions of Questionnaire
In order to create the most effective questionnaire, the research design team identified the
information of interest and crafted an initial design that was modeled after other questionnaires
from recent, similar studies. Survey research experts from Pacific Market Research reviewed the
questionnaire, along with statistical and subject matter experts.
Prior to widespread implementation, the questionnaire will be administered and tested among
team members for content and length. After passage of that test, the questionnaire will be
administered to a small subset of the target population. Following this "pilot test," sample
respondents will be interviewed about their experience with the questionnaire, including:
• Was your overall impression of your interview experience positive or negative, and why?
• Which questions were challenging? If any were challenging, what might make them
easier?
Nez Perce Tribe
Volume III - Appendix H
December 2016 Page H-41
-------
• Keeping in mind that the study topic is fish consumption, are there any questions that
ought to have been asked but weren't?
• Are there any questions which seemed unnecessary?
Each step of the process allows for questionnaire revisions as appropriate. Significant revisions
and/or additions to the questionnaire deserve further testing.
8. Pilot Testing of Questionnaire and Field Operations
The pilot test will cover most of the survey procedures, including screening, invitation and first
contact, interview using the questionnaires (FFQ and 24-hour recall), field review and key entry.
The persons selected for the pilot test will exercise all dimensions of the questionnaire. It is
likely that 15-20 persons, at least, will be needed for an adequate pilot test.
Questionnaires may be revised continuously while the pilot test is underway, but substantial
revisions may require additional pilot interviews to test new questions or new wording and
formats. The following characteristics of pilot test respondents (who will not be eligible for
inclusion in subsequent sample selection) will be covered.
• Age: elders and younger members
• Gender: males and females
• Lifestyle: modern and traditional
• Fishing: fishers and non-fishers
• Source of fish: primarily eat at home vs. eat out frequently
• Income: low-income and high-income
• Food preparation: respondents who do and do not usually prepare food for the household
A pilot test respondent may cover more than one dimension. For example, elder fishers may
contribute to understanding the questionnaire performance on both elders and fishers. However,
other combinations of characteristics with an elder and with a fisher should also be tested.
Additional pilot test participants may be added until the various dimensions have been fully
covered. During the pilot test it is important to interview different types of respondents so that all
iterations of the questionnaire can be addressed. The pilot test should include the anticipated
final questionnaire as well as other tools related to it, such as portion size models and
photographs.
4.5 IRB Approval
In order to meet accepted standards of protection for survey respondents, we will seek
Institutional Review Board (IRB) approval of the survey design. We have identified Quorum
Review IRB, a commercial IRB service, as a vendor for this purpose. The process consists of
preparing a set of documents (see list below), working with the IRB for pre-review of the
application, revising the application based on the pre-review, and then submitting the revised
application for full review.
The following list provides an example of the documents needed for the IRB application; many
forms and examples are available on the Quorum Review IRB website, at
http://www.quorumreview.com/forms/.
• Submission forms, which include administrative details about the study, study locations,
and study team.
Nez Perce Tribe
Volume III - Appendix H
December 2016 Page H-42
-------
• Study Protocol, including discussion of the purpose and benefits of the study, potential
risks to the respondents, description of the study methods, selection criteria for
respondents, and procedures to protect confidentiality.
• Curriculum vitae (CV) and other credentials of the Principal Investigator (PI). Only one
PI is needed for the IRB application if that PI will be responsible for the protection of
human subjects.
• Survey documents, including survey forms, consent forms, and any other written material
which will be provided to respondents.
The goal of pre-review with the IRB prior to full submission is to improve the quality and
completeness of the submission. Quorum Review provides a pre-review service for this purpose.
The expected timeline for IRB approval is about 1 week from submission of all documents,
depending on whether the pre-review identifies any issues. Since this survey is purely behavioral
and risk to the study participants is minimal, we expect that it will qualify for expedited review.
4.6 EPA Human Subjects Review
In addition to IRB review and approval, the survey will need review and approval from the EPA
Human Subjects Research Review Official (HSRRO). The process consists of submitting an
application and supporting documents to the HSRRO. The IRB review and approval is one input
to the HSRRO review process. The HSRRO has final authority for review of human subjects
research supported by the EPA. The following documents are needed for submission to the
HSRRO; additional documents may also be requested:
• Application memorandum using a template provided by the HSRRO, which includes a
brief discussion of the value of the research, any risk to the subjects from the research,
and the approach for subject selection and informed consent.
• Documents submitted to the IRB, including the study design and survey documents such
as consent forms, survey forms, and recruitment material.
• Documents received from the IRB, such as review comments and letters of approval or
exemption.
The HSRRO review process takes place after IRB approval and prior to commencement of the
survey.
The EPA provides educational resources for investigators to clarify human subjects research
policies, such as the online tutorial "Human Subjects Research at the Environmental Protection
Agency: Ethical Standards and Regulatory Requirements" at
http://www.epa.gov/osa/phre/phre course/index.htm. The survey team will pursue and manage
the human subjects approval process with EPA.
Nez Perce Tribe
Volume III - Appendix H
December 2016 Page H-43
-------
5.0 SURVEY OPERATIONS
This section describes the field operations, including interviewing and contacting participants, as
well as pilot testing and key entry of the questionnaire.
5.1 Interviewing
This section describes the selection and training of individuals who will administer the survey
interviews; procedures for conducting the interviews; scheduling, monitoring, and recording
interviews; and proper handling of the questionnaires.
1. Interviewer Selection
Interviewing positions will be filled in collaboration with the Tribal authorities with agreement
on selection by both parties. Once hired, the interviewers will report to the survey team. Ideally,
the Tribes will recruit or propose two to three individuals for each interviewer position.
Additionally, the survey team hopes that the NPT will promote participation in this study, both
for respondents and interviewers. For those who apply for the interviewing position, a survey
team staff member will explain the job duties; those whose qualifications appear promising will
be invited to complete various skills and aptitude tests that cover:
• Education
o High school diploma or GED
o 9th grade reading level
¦ Reading sample survey script: silently and aloud
¦ Comprehension and clarity
• Clerical skills
o Legible hand-writing
o Spelling
o Grammar
• Employment availability: part-time work for 9-12 months
• Transportation
• 18+years old
• Courtesy and professionalism
• Ability to think "on one's feet" and to adapt to changing conditions
• Good communication skills
• Reliability
• Ability to follow directions, as it is important that surveys be administered using a
common, scripted approach to maximize objectivity and to enhance comparability of
answers.
2. Interviewer Training
Interviewers will be trained to follow "best practices" when it comes to in-person interviews.
This classroom component of the training is expected to last approximately 4-8 hours. It will
begin with an overview of survey research, including a brief history of its utility and the
importance of its role. The training will include general and specific interviewing techniques and
skills. In addition to an explanation of the origin of this survey, interviewers will receive survey-
related materials and information about the critical nature of the project. As part of the training,
Nez Perce Tribe
Volume III - Appendix H
December 2016 Page H-44
-------
the survey staff will themselves need some instruction in practices that are acceptable to or
unacceptable to Tribal respondents. These important cultural points will be included in the
training.
Interviewers will be exposed to general survey research principles related to interviewing.
Objective data collection will be emphasized, as will the need to listen closely to what the
respondent says and record it accurately. Interviewers will learn how to probe, clarify and check
open-ended answers to ensure that they've elicited and captured all relevant information from the
respondent. Most importantly, interviewers will participate in a lengthy and in-depth mock
interview session during which the interviewer works directly with a supervisor or another co-
worker to try out the questionnaire and what they've learned. The supervisor will provide the
interviewer with challenging but realistic answers to the questions.
Special attention will be devoted to cultural aspects which might prove challenging during
verbatim administration of the questionnaire. For example, if a respondent does not understand a
question, a typical interviewing technique is to repeat the question and to answer the
respondent's inquiries with, "I can't interpret the question for you. It is whatever the question
means to you." If the pilot test uncovers survey items which are unclear, additional probes and
prompts will be developed in order to minimize interviewer interpretations while in the field.
3. Procedure Manual and Training for Interviewers and Supervisors
All interviewers and supervisors will undergo a comprehensive training prior to beginning work
on this project. The training will include basic and advanced topics necessary to successfully
conduct in-person survey research. Below is an example agenda for the training sessions which
would be required for all interviewing staff.
• Introduction of survey staff and implementation team
• Proj ect b ackground
• Overview of survey research
• Confidentiality requirements
o Dealing with Personally Identifiable Information (PII)
o What to do if you know the respondent
• Exploration of question types
o Close-ended items
¦ Numeric items
¦ Scale items
o Open-ended items
• Importance of precision and accuracy when recording answers
• Objective research: non-bias by interviewer
• Techniques to probe and clarify
• Building rapport with respondents
o Being courteous and respectful
o Addressing challenging respondents
¦ Older
¦ Hard-of-hearing
¦ Angry
• Review of questionnaire
Nez Perce Tribe
Volume III - Appendix H
December 2016 Page H-45
-------
• Quality control measures
o Self-monitoring
o Supervisor/data entry controls
o Call-backs and verification
o Statistical tests
• Productivity targets
• Logistics related to appointments, survey administration, etc.
o Reimbursement for expenses
o Contact information for all staff
4. Scheduling and Monitoring Interviewers and Activities
The process for assigning in-person interviews will be administered by the survey team's
scheduler, who, initially, will be an employee of Pacific Market Research—one of the three
firms which will be carrying out the survey implementation work. The scheduler will work
closely with the interviewers to ensure that the in-person interviews are scheduled only during
hours when the interviewers are available. Over time, some or all of the scheduling responsibility
might be transferred to the interviewers with continued monitoring by the survey staff. Based on
the estimated interview length, we anticipate that it will be possible for an interviewer to
complete two interviews per day. This is expected to be the target quota for the interviewers,
given the length of the interviews and activities associated with each interview. This depends on
many factors, including the distance that the interviewer must travel, road conditions, and
whether the respondents show up when agreed. We recommend setting a target of at least one
half of all interviews being conducted at a central location on each reservation.
Consideration will be extended for respondents with mobility problems, ensuring that their
responses are gathered even if they are homebound. Accounting for respondent availability and
interviewer workload, interviews will be scheduled seven days a week starting as early as 8:00
a.m. with no interview beginning later than 8:00 p.m. To the extent possible, a primary goal is to
minimize respondent burden; one way to do this is to offer an assortment of times and
convenient locations for the interviews.
Any issues of calendar sensitivity (such as avoiding or minimizing interviews on Sundays or
special occasions) will be addressed in conjunction with the Tribes prior to the commencement
of interviewing. The survey implementation team will work with the Tribes to jointly design an
initial approach to respondents that is consistent with the Tribes' way of carrying out activities
and is also consistent with accepted scientific survey practice.
5. Recording Interviewer Responses
Interviewers will record interview answers on the hardcopy questionnaire. They will also record
start date, the start time, the completion date, and the end time. Writing will be tidy and easily
readable. Stray marks or mistakes will be corrected as necessary prior to handing off the
completed questionnaire for data entry.
During data entry, the entry staff will review the questionnaires as they enter them. If the
supervisor or the data entry personnel observe missing data or other problematic aspects with the
questionnaire, it will be referred to the original interviewer for review and correction as
appropriate.
Nez Perce Tribe
Volume III - Appendix H
December 2016 Page H-46
-------
6.
Integrity and Handling of Questionnaire Hardcopy
The completed questionnaires will be protected by interviewers until the questionnaires have
been delivered to the data entry staff or a secure holding area. Questionnaires must not be left out
where non-survey staff might gain access to them. Instead the questionnaires should be kept with
the interviewer, within his/her physical control, or in a locked area prior to handing off to data
entry.
5.2 Contact with Respondents
Respondents will first be contacted by mail and/or Tribal newsletter to introduce the project in
general. Respondents will then be contacted by telephone, followed by a selection of those
respondents who are willing to participate in the in-person interview(s).
1. Initial Contact by Mail and Telephone
Initial contact with respondents will be by letter or postcard, alerting respondents that the survey
is forthcoming and that their opinions are important. Follow-up contact will occur via telephone
(up to 15 call attempts before assigning a record as deceased or otherwise ineligible). During the
telephone call, respondents will be screened for fish consumers versus non-fish consumers, and
an attempt will be made to schedule an appointment for an in-person interview with fish
consumers.
The implementer will coordinate with individual Tribes to identify motivating factors such as
incentives or other valuable rewards for prospective respondents. EPA funds cannot be used for
remuneration but we strongly recommend providing a token of gratitude in order to establish
good will and boost the response rate. Without incentives there is danger of survey failure due to
a low response rate. If the main motivation for the respondents in this project is a sense of
altruism, it is all the more important that the interviewers are extremely assertive and persuasive
in convincing prospective respondents to participate. In order for the survey to be successful, the
Tribal leadership will need to play a central role in informing the Tribe about the survey and
promoting cooperation with the survey.
When contacting respondents by telephone, some individuals are expected to refuse to
participate. The initial counterpoint to a respondent refusal is to explain the importance of the
respondent's opinions and experiences in the study, sharing with him/her how the results will
benefit the Tribes and community. If he/she still refuses, the interviewer will put the number
back in the system, allowing several days to pass before attempting the number again. Call-back
conversion attempts are often handled by "conversion experts," different from the original
interviewer, which may be applied as necessary. Interviewers will use standard survey research
practices to try to convert initial refusals to cooperative participants.
2. In-Person Interviews
Data collection will take place either in the respondent's home or in a central, public location.
Part of the goal of the research is to promote a feeling of confidence and good will among the
prospective respondents in order to conduct as many completed interviews as possible. To this
end, we recommend conducting the interview in a location where the respondent feels
comfortable and safe. The interviewer will either meet the respondent in a mutually agreed
location or go to the respondent's home. Background materials relevant to the survey will also be
provided to the respondent in advance.
Nez Perce Tribe
Volume III - Appendix H
December 2016 Page H-47
-------
At each interview's conclusion, the interviewer will graciously thank the respondent for his/her
time, reiterate the importance of the study results, and quickly review the questionnaire so that
the interviewer may administer follow-up questions for any items which have missing
information. To the extent possible, interviewers will record interview feedback from
respondents. This includes praise and complaints from respondents. Feedback will be provided to
the scheduler or the supervisor at the end of each day. Interviewers are required to provide the
outcome or disposition of each interview attempt as soon as possible after the attempt or at the
conclusion of each day, whichever comes first. The disposition will be recorded in a master
database so that the result is available for immediate and later analysis.
3. Follow-up Call and Re-Interview
For quality control purposes, we recommend a follow-up call to every respondent. The follow-up
call or verification call is intended to provide a double-check of the interview. Some respondents
who receive a follow-up call will merely be asked whether they participated in the survey. But a
sub-sample of the entire group will be asked to validate their data. By asking some of the same
questions again, the researchers can test the reproducibility of the data. The questions will be
selected to represent major sections of the questionnaire and will avoid questions with complex
or long lead-in development.
5.3 Tribal Collaboration in Field Operations
It is recommended that a primary technical contact for survey operations be identified by the
Tribes. This contact person will be responsible for collaborating with the survey implementation
team, providing access to the Tribal facilities for conducting interviews, assisting with the
logistics of contacting and following-up with survey participants, and keeping the Tribal
leadership and membership informed of the status of the survey.
To create and roll out a successful survey, it is critical to obtain Tribal support initially,
particularly Tribal leadership, and to develop and maintain the relationship and support
throughout the project. From the implementation team this requires familiarity with quantitative
survey research as well as cultural sensitivity. The implementation team must be available to the
Tribal representatives to address any outstanding survey issues. Two-way communication is
crucial.
5.4 Key Entry of Questionnaire, Validity Checks, and Storage
Data collection will be conducted with hard copies of the questionnaire. After the data have been
recorded on the questionnaire, information will be keypunched or entered onto digital media.
This provides an extra level of redundancy as well as, and more importantly, an automated
method of organizing and eventually analyzing the data.
Many data entry software packages are available and they allow quick, efficient, reliable and
secure data entry. Some of these include: SPSS Data Collection Data Entry, Voxco Interviewer
Suite/Command Center, EpiData Software, Survey Analytics iPad Survey Tool, snap Surveys,
Confirmit and even Excel. Pricing varies depending on the vendor and the type of solution, from
many thousands of dollars to a nominal (or even no) fee for open source applications. Each
software package has its benefits and drawbacks, but for this project we recommend SPSS Data
Collection. For security purposes, sample files and data files shall be encrypted.
Nez Perce Tribe
Volume III - Appendix H
December 2016 Page H-48
-------
Best practices demand that data entry is verified. This can be accomplished by spot-checking
randomly selected data points in every nth interview or entering all responses for every nth
interview twice. The most reliable way to check the accuracy of the data entry is to perform
100% verification. This means that all data points for every interview are entered twice. We
strongly recommend 100% verification.
To effect reliable data verification, two or more parties will be involved in the process. An initial
keypunch operator enters the data for one interview; it is verified (re-entered) by a different
keypunch operator. Each record or line of data related to the questionnaire is checked against its
respective original record. If discrepancies are found, a supervisor or other staff member will
review both of the electronic records and the hard copy of the questionnaire to determine which
data entry point is correct.
Error rates will be tracked among survey responses in general and also by cross-tabulating
responses by various demographic or other information, and looking for anomalies or
statistically significant differences.
1. Field Validity Checks and Re-interview
Of the many places where an error can be introduced into the data, the collection point is among
the first. A typical way to test for interviewer errors is to re-contact some respondents and re-ask
several questions. Due to the additional burden on respondents during this follow-up process, it's
unrealistic to administer the entire interview again; instead a subset of questions may be asked to
validate the data recorded by the interviewer. Not all respondents will be re-contacted. In the
event that significant differences are found (between the originally recorded answers and the
validation answers), the interview for that respondent will either be discarded or a new
interviewer will be sent to administer the full questionnaire again. Each interviewer's work will
be evaluated for consistency and accuracy. Selected questions will be re-asked of a selected sub
sample.
2. Handling Missing Values
Missing survey data, whether because of survey design problems, interviewer error, respondent
misunderstanding or simply refusal to answer questions, can be problematic for any project.
Ideally there will be no missing data. In the event that a record is missing some of its data—and
it is due to respondent-caused factors—there are several acceptable steps for adjusting the data to
accommodate missing values. By using data analysis software we can impute new values where
once the data were missing. That is, based on the values in other, similar cases, data can be
pushed into the records which had missing data. The replacement data might be based on
copying a value from a random case, mean substitution, regression, or multiple imputation.
Generally, the most robust method is with multiple imputation; we recommend using multiple
imputation for this project. This will be implemented during analysis.
3. Naming and storage of electronic files
Data files will be stored and named according to the specifications of the selected data entry
software. Generally the file name suffix should be a concise but descriptive annotation of the
file's contents and the date of last revision. For example, a data file created in Excel which holds
information about the NPT should be named fishconsumption NPT'2014 04 23.xls, where
"fish_consumption" describes the study, "NPT" identifies the Tribe and "2014_04_23" is the
date that the file was last modified. In most cases the file extension will depend on the data entry
Nez Perce Tribe
Volume III - Appendix H
December 2016 Page H-49
-------
software. Some systems do not allow long file names. In this case, the file name will be
shortened to convey as much information as possible without exceeding file-naming rules for the
respective operating system.
4. Back-up and Transfer Protocols
Data back-ups shall be completed on a basic grandfather-father-son rotation schedule. Backups
will be completed daily, weekly, and monthly. Media for daily back-ups are rotated daily,
weekly back-ups are rotated weekly, and monthly back-ups are rotated monthly. For example, a
back-up is completed each day. After the initial back-up, additional back-ups will be incremental
(i.e., backing up only the files which have changed since the previous back-up).
The transfer of files which contain Personally Identifiable Information (PII) or Protected Health
Information (PHI) shall be conducted via secure messaging or via a Secure File Transfer
Protocol (SFTP) site. Sensitive data must not be transmitted via "regular" e-mail or other
unsecured means.
5.5 Sensitive Information
During the administration of this survey, the Tribe will provide information about their
membership. Some of this information is considered "sensitive information" and must be
protected from disclosure. Sensitive information includes PII and PHI. Various laws and
regulations affect the handling of PII and PHI.
5.6 Confidentiality and Data Management
Tribal Committees and the Tribal Office of Legal Council will be included in discussions and
plans to maintain the confidentiality of the data during the survey operation. All survey staff will
be required to sign a Proprietary Information Agreement and a Non-Disclosure form prior to
gaining access to private or sensitive information and certainly before beginning work on the
data collection. The agreement will include confidentiality during the interviews and
confidentiality of the survey results.
1. Confidentiality of Hardcopy and Electronic Files
Hardcopy questionnaires, with data on them, whether completed or not, must be stored in a
secure location if they include PHI or PII. A secure location is an area that cannot be easily
breached by the public or by non-authorized personnel. An example of a secure location is within
a safe, a locked filing drawer or sometimes a locked office. However, a locked office is often
insufficient as custodial staff or other workers might have access to the area.
Data files which contain PII or PHI shall be stored on secure password-protected devices. In this
case a password-protected device is an electronic medium which requires a unique username (not
shared among users) and a strong password in order to access the file. The strong password
should include a combination of alphanumeric characters, with uppercase and lowercase letters
and numbers. The file should be encrypted using at least AES 256-bit security.
2. Communicating Confidentiality to Participants
Respondents will be informed in advance and again at the beginning of the interview that their
survey responses will remain confidential and that all research results will be reported in an
aggregate manner. No individually-identifiable data or answers will be shared with anybody
outside of the survey staff. The respondents will be assured that they can safely and honestly
Nez Perce Tribe
Volume III - Appendix H
December 2016 Page H-50
-------
answer the questions, since they will remain anonymous after completion of the interview.
Respondents will be advised that a Freedom of Information Act (FOIA) request might nullify the
study sponsor's promise of confidentiality. However, the usefulness of the data, on an individual
level is dubious: a FOIA request is unlikely to affect divulgence of individual information.
The EPA and the NPT have yet to agree on and sign confidentiality agreements; communication
to the respondents will be specified (and reviewed by the Tribes) after such agreements are in
place. The survey will not proceed on administering any interviews with tribal members until
confidentiality agreements are in place between the NPT and EPA and the survey has received
both IRB and EPA Human Subjects approval.
Nez Perce Tribe
Volume III - Appendix H
December 2016 Page H-51
-------
6.0 ANALYSIS, REPORTING, CLOSE-OUT OF STUDY
This section discusses the methods for analyzing data collected from the FFQ and 24-hour
dietary recall surveys, as well as final reporting and completion of the study.
6.1 Analysis of FFQ results
The data collected from the FFQ will enable a fish consumption rate (g/day) to be determined for
each sampled individual. For an individual, the rate can be determined for each species or
species group (anadromous, resident freshwater, and marine). Briefly, an annual amount
consumed arising from consumption in a particular season can be calculated per species from the
typical portion size (grams) consumed for that species multiplied by the frequency of
consumption, then multiplied by the duration of the season (or period). The sum of this total
seasonal quantity for the two seasons yields an annual quantity. Secondly, the amount consumed
(grams) in ceremonial or special events can be calculated from the typical consumption amount
at those events multiplied by the number of such events attended per year by the individual. This
can be added to the total amount for two seasons to yield a total consumption for a year. Division
by 365.24 days (taking into account leap years) will yield a daily amount in grams per day for
the given species. The daily consumption rate for a species group can be calculated for an
individual by summing the daily rates for the individual species included in the group. Some
selected analyses can be carried out to express consumption rates in grams per kilogram of body
18
weight per day (g/kg-day), since some consumption studies report rates in these units.
The computation of means, medians and other percentiles will need to take into account the
stratification and weighting used in the sampling, as well as any correlation among respondents'
data introduced by the occurrence of two sampled adults in the same household.
Quantities reported for the NPT should be accompanied by appropriate indications of uncertainty
and, where applicable, an estimate of variation across individuals. All means reported for fish
consumption rates or for other variables should be accompanied by standard deviations along
with a notation of the weighted and unweighted sample size underlying the calculation. Other
estimated quantities (aside from means), such as percentiles of the fish consumption distribution,
should be reported with standard errors and, for rates that are likely to be considered for setting
water quality standards or other regulatory actions, the estimate should be accompanied by 95%
confidence intervals. Again, for percentiles and other quantitative estimates, the underlying
weighted and unweighted sample size should be noted.
There are several methods available for computing percentiles of an empirical distribution. See
Hyndman and Fan (1996), for a discussion of the different methods. The design team
recommends the calculation of type 7 percentiles, as noted in the Hyndman article.
A number of other quantities and responses are collected in relation to the FFQ. These quantities
will consist of continuous variables (such as age) and categorical variables (such as gender or
education). The continuous variables can be summarized by means (and medians if there are
highly skewed distributions), standard deviations, minimum and maximum values and, if
appropriate, standard errors. Categorical variables can be summarized by percentages per
category. The total sample size underlying each set of summary statistics for variables should
also be shown.
18 Body weight data will be collected with general demographic information during the in-person interviews
Nez Perce Tribe
Volume III - Appendix H
December 2016 Page H-52
-------
Confidence intervals (95% level) for the various statistics can be calculated by several methods.
The choice of method depends heavily on the distribution of the values used to calculate the
statistics and on the sample size. For the larger sample sizes (e.g., over 100), the nonparametric
Bootstrap will usually work well for the mean, median and percentiles near the median, but other
methods may be needed for the higher percentiles. (The Bootstrap method will need to be
adapted to the particular weighting and stratification scheme used for the NPT.) Experiments
with the Bootstrap for 95% confidence intervals for various percentiles or the mean from random
samples from a lognormal distribution show less than 95% coverage for samples sizes of the
magnitude discussed in this report. For the 90th and 95th percentiles (and possibly other nearby
percentiles), non-parametric confidence intervals can be based on the ranking method described
by Hollander and Wolfe (1999).
Alternatively, if the distribution appears close to the lognormal or another distribution that can be
specified in closed form, the parametric bootstrap can be used. For example, a lognormal
distribution can be fitted to the data (taking account of weighting) and the bootstrap algorithm
can be applied to calculate percentiles for samples drawn from the fitted distribution, again
taking account of weighting and stratification. In fitting a distribution to the data, another method
that may be useful is to fit a broken-stick spline to the Q-Q plot (using normal distribution
quantiles). The parametric bootstrap can then be carried out with the fitted distribution.
6.2 Analysis of 24-hour Recalls
The 24-hour recall data will be analyzed using the "NCI method." An example of analysis of fish
consumption data using the NCI method, along with a heuristic description of the method can be
found in Polissar, et al., 2012. Dr. Kevin Dodd of the NCI, one of the developers of the method,
has offered to assist in implementation of the method for the Idaho Tribes. The implementation
team statistician will be in touch with Dr. Dodd to carry out this work. Helpful references for this
method can be found in Tooze, et al., 2006; Dodd, et al., 2006; and Kipnis, et al., 2009. An
excellent series of webinars, including a talk and materials by Dr. Janet Tooze on the NCI
method, are available at http://riskfactor.cancer.gov/measurementerror/. The SAS software for
the method is available from Dr. Dodd at NCI and it will need to be adapted to this specific
survey methodology. Confidence intervals are not provided by the methodology, but they may be
computed by some potentially computationally extensive methods.
As noted previously, there may not be a sufficient sample size of respondents with two fish
consumption days from the two 24-hour recall interviews to support the NCI method for the NPT
considered alone. In that case it may be possible to estimate fish consumption rates for the NPT
by pooling data with other Tribes (for this purpose alone) and then using a covariate or
covariates to generate a unique NPT distribution of consumption rates. The covariate might be
either a tribal indicator variable or else the respondent-specific consumption rate from the food
frequency questionnaire.
6.3 Reporting of Results
The results of this survey are likely to be used for years ahead, if not decades, therefore a very
complete report should be prepared. Some of the tribal fish consumption surveys in the Pacific
Northwest continue to be used for environmental regulation more than 20 years after their
completion. This survey will likely also have that long-term utility.
Nez Perce Tribe
Volume III - Appendix H
December 2016 Page H-53
-------
In addition to the report describing the methods and results of the survey, the implementation
team may also prepare a short procedural history of the survey, including lessons learned and
changes in design made during the survey operation. Such a report will help users of the results
to understand the context of data collection more thoroughly.
The suggested format for the report on suppression and quantitative rates is the commonly used
sequence of:
• Executive summary
• Introduction (including background and motivation)
• Methods (including methods for survey design, survey operations and statistical methods
for data analysis—for both the suppression study and the current consumption survey)
• Results (extensive tables and displays along with textual commentary) on the suppression
study and the current consumption survey
• Discussion (including main findings, comparison of the rates from the FFQ and the NCI
method, strengths, weaknesses, remaining uncertainty, potential applications of the
results in water quality regulation and conclusions)
• References
• Appendices (including more detailed tables than presented in the body of the report,
technical notes, and other supporting material)
• Acknowledgments (thanking, in particular, tribal council, tribal respondents and tribal
staff)
The suppression study will fit into this framework as well, as part and parcel of the report. There
have been many studies of historic rates and suppression in the past, but their isolation from a
report on current rates may have denied them the attention they deserve. The primary
quantitative results from the suppression study are likely to be mean (average) consumption per
day with a plausible range bracketing the mean. To the extent possible, the rates will be
categorized by broad species groups.
The methods section of the report can include plain-language description of methods, but highly
technical material should be placed in the appendices. This should be a report whose main body
is very readable by Tribal leaders and managers, environmental scientists, political leaders,
regulatory staff, and by anyone with previous exposure to the topic.
The main results such as the mean, median, and percentiles of fish consumption for all species
combined and for various species groups can be presented in tabular and graphical format in the
main body of the report. The various rates can be presented for age, gender, income and
educational attainment groups, but more detailed tables (e.g., with more percentiles, more sub-
divided groups, and with confidence intervals) can be presented in the appendices. The
implementation team should keep in touch with the team conducting the surveys for the State of
Idaho and attempt to include tables in the report that have comparable species and demographic
groups as the main tables of the State surveys.
The State of Idaho will be surveying anglers (in addition to their survey of the general
population) and the NPT's report can also report on Tribal anglers who are sampled within the
survey process. The anglers may be defined by, for example, having fished at least a certain
number of times during a defined period (using questions included in the in-person interview).
The extent of results reported for anglers will depend on the number of anglers encountered.
Nez Perce Tribe
Volume III - Appendix H
December 2016 Page H-54
-------
6.4 Peer Review
The design team recommends that a technical peer review panel be convened. The topic of fish
consumption rates is controversial, and there are always opportunities for mistakes in a survey as
large and complex as this one. The panel may consist of an environmental scientist familiar with
issues in fisheries and fish consumption, a PhD-level statistician familiar with surveys, a scientist
familiar with reconstruction of heritage consumption rates, and a support or reference person
who is familiar with the use of FCRs for environmental regulation.
6.5 Archiving, Ownership, Sharing of Data
The EPA management staff for this project will be communicating with the Idaho Tribes, with
this design team, and with other EPA staff to develop a globally satisfactory policy for
confidentiality and ownership of, access to, and potential sharing of the data developed from this
survey. The design team has provided input on this process and various issues related to this
topic. The formal agreement on ownership of current and future access to the survey electronic
and hardcopy data will be an agreement between EPA and the Tribes, it is anticipated. A survey
team representative(s) may also be a signer - in the role of one implementing parts of the
agreement. The survey team will request to review and comment on any proposed agreements to
ensure that there is compatibility between the agreements and survey operations, planned data
analysis, and final reporting.
Undoubtedly the results of this survey will be a precious resource for the Tribe and others,
documenting the status of fish consumption and factors affecting it both historically and at this
time. Future aspirations for fish consumption are also covered.
Given the present and future importance of the survey results, it will be important to archive the
material carefully. The quantitative data should be saved in electronic system and text files,
accompanied by data dictionaries, including the name of each variable (field), its definition and
meaning, file position and width, and codes used with a definition of each code. At least two
copies of the files should be kept on external media and the two or more sets of files should be
maintained in widely separate locations to avoid common loss in case of a disaster. At least
annually (signaled by a tickler file) a copy should be made of each set of files (and verified) to
avoid loss through physical deterioration of media. As storage modes change over time (e.g., the
past transition from tape to disc), the storage mode of the survey files should be kept up to date.
Nez Perce Tribe
Volume III - Appendix H
December 2016 Page H-55
-------
7.0 DESIGN TEAM, ACKNOWLEDGEMENTS, AND RESOURCES
The survey design team coordinated with the Idaho Tribes, EPA, and the State of Idaho to
develop this survey design. Various resources were compiled and reviewed as much as possible
to support design development.
7.1 Names and affiliation
The survey design was conducted as a collaboration between The Mountain-Whisper-Light
Statistics (TMWL) and RIDOLFI Inc., with support from Pacific Market Research (PMR), and
consisted of the following key team members:
• Dr. Nayak Polissar of TMWL
• Dr. Derek Stanford of TMWL
• Callie Ridolfi of RIDOLFI Inc.
• William Beckley of RIDOLFI Inc.
• Kristin Callahan of RIDOLFI Inc.
• Anthony Salisbury of PMR
7.2 Acknowledgements
The survey design team would like to thank the following Tribal representatives for their support
and input during the design phase:
• Silas Whitman, Nez Perce Tribal Executive Committee (NPTEC) Chairman
• Joel Moffett, NPTEC Vice-Chairman
• McCoy Oatman, NPTEC Treasurer
• Anthony Johnson, NPTEC Secretary
• Daniel Kane, NPTEC Asst. Sec./Treasurer
• Leotis McCormack, NPTEC Chaplain
• Samuel Penney, NPTEC Member
• Albert Barros, NPTEC Member
• Brooklyn Baptiste, NPTEC Member
• Julie Kane, Managing Attorney, Office of Legal Counsel
• Michael Lopez, Staff Attorney, Office of Legal Counsel
• David Cummings, Staff Attorney, Office of Legal Counsel
• Carla Timentwa, Enrollment and Chair of General Council
• James Holt, Director of Water Resources Division
• Ken Clark, Water Quality Program Coordinator
• Joseph Oatman, Deputy Program Manager, Department of Fisheries Resource
Management
• Patrick Baird, Tribal Historic Preservation Officer, Cultural Resources
• Nakia Williamson, Tribal Ethnographer, Cultural Resources
7.3 Resources
A list of resources pertinent to developing and implementing a FCR survey is presented below,
including agency guidance documents, existing surveys and methodology reports, and traditional
lifeways and suppression studies. These resources, in addition to the references cited within this
Nez Perce Tribe
Volume III - Appendix H
December 2016 Page H-56
-------
design report (Section 8), will provide additional guidance, background information, and
research to support implementation of the survey.
1. Guidance, Regulations, and Other Agency Reports
Idaho Department of Health and Welfare (IDHW). 2013. Eat Fish, Be Smart, Choose Wisely, A
guide to Safe Fish Consumption for Fish Caught in Idaho Waters. Bureau of Community
and Environmental Health.
U.S. Environmental Protection Agency (EPA). 1998. Guidance for Conducting Fish and Wildlife
Consumption Surveys. Office of Water. EPA-823-B-98-007. November.
U.S. Environmental Protection Agency (EPA). 2000a. Methodology for Deriving Ambient Water
Quality Criteria for the Protection of Human Health. Office of Water, Office of Science
and Technology. EPA-822-B-00-004. October.
U.S. Environmental Protection Agency (EPA). 2000b. Guidance for Assessing Chemical
Contaminant Data for Use in Fish Advisories. Volumes 1-4. Office of Water. EPA-823-
B-00-007. November.
U.S. Environmental Protection Agency (EPA). 2002. Columbia River Basin Fish Contaminant
Survey, 1996-1998. EPA Region 10. EPA 910-R-02-006.
U.S. Environmental Protection Agency (EPA). 2006. Paper on Tribal Issues Related to Tribal
Traditional Lifeways, Risk Assessment, and Health & Well Being: Documenting What
We've Heard. The National EPA-Tribal Science Council. April.
U.S. Environmental Protection Agency (EPA). 2007. Framework for Selecting and Using Tribal
Fish and Shellfish Consumption Rates for Risk-Based Decision Making at CERCLA and
RCRA Cleanup Sites in Puget Sound and the Strait of Georgia. EPA Region 10. August.
U.S. Environmental Protection Agency (EPA). 2013a. Human Health Ambient water Quality
Criteria and Fish Consumption Rates Frequently Asked Questions. January. Available
online:
http://water.epa.gov/scitech/swguidance/standards/criteria/health/methodology/index.cfm.
U.S. Environmental Protection Agency (EPA). 2013b. National Rivers and Streams Assessment
2008-2009, A collaborative Survey, DRAFT. Office of Wetlands, Oceans and
Watersheds, Office of Research and Development. EPA/841/D-13/001. February 28.
U.S. Environmental Protection Agency (EPA). 2013c. Fish Consumption in Connecticut,
Florida, Minnesota, and North Dakota. Office of Research and Development.
EPA/600/R-13 -098F. August.
U.S. Environmental Protection Agency (EPA) and Toxicology Excellence for Risk Assessment
(TERA). 1999. Comparative Dietary Risks: Balancing the Risks and Benefits of Fish
Consumption. Results of a Cooperative Agreement between EPA and TERA. August 6.
Nez Perce Tribe
Volume III - Appendix H
December 2016 Page H-57
-------
2.
Fish Consumption Surveys and Survey Methodology
Columbia River Inter-Tribal Fish Commission (CRITFC). 1994. A Fish Consumption Survey of
the Umatilla, Nez Perce, Yakama, and Warm Springs Tribes of the Columbia River
Basin. Technical Report 94-3. October.
Freimund, J., M. Lange, and C. Dolphin. 2012. Lummi Nation Seafood Consumption Survey.
Water Resource Division, Lummi Natural Resources Department. Prepared for Lummi
Indian Business Council. August, 31.
Groves, R.M., F.J. Fowler, Jr., M.P. Couper, J.M. Lepkowski, E. Singer, and R. Tourangeau.
2013. Survey Methodology, 2nd Edition.
Harper, B.L, B. Flett, S. Harris, C. Abeyta, and F. Kirschner. 2002. "The Spokane Tribe's
multipathway subsistence exposure scenario and screening level RME. " In: Risk
Analysis. 22: 3, 513 - 526.
IDM Consulting. 1997. Establishing Alaska Subsistence Exposure Scenarios. ASPS #97-0165.
Prepared for Alaska Department of Environmental Conservation. September 1.
Kissinger, L., 2010. "Development of a computer-assisted personal interview software system
for collection of tribal fish consumption data. " In: Risk Analysis. 30: 12, 1833-1841.
Landolt, M.L, F.R. Hafer, A. Nevissi, G. van Belle, K. Van Ness, and C. Rockwell. 1985.
Potential Toxicant Exposure Among Consumers of Recreationally Caught Fish from
Urban Embayments ofPuget Sound. Memorandum to Pacific Office, Coastal and
Estuarine Assessment Branch, Ocean Assessments Division, Office of Oceanography and
Marine Assessment, National Ocean Service, National Oceanic and Atmospheric
Administration (NOAA). November.
Landolt, M.L, D. Kalman , A. Nevissi, G. van Belle, K. Van Ness, and F. Hafer. 1987. Potential
Toxicant Exposure Among Consumers of Recreationally Caught Fish from Urban
Embayments of Puget Sound: Final Report. Memorandum to Pacific Office, Coastal and
Estuarine Assessment Branch, Ocean Assessments Division, Office of Oceanography and
Marine Assessment, National Ocean Service, National Oceanic and Atmospheric
Administration (NOAA). April.
Mayfield, D.B., S. Robinson, and J. Simmonds. 2007. "Survey of Fish Consumption Patterns of
King county (Washington) Recreational Anglers. " In: Journal of Exposure Analysis and
Environmental Epidemiology. Natural Publishing Group. 17: 7, 604-612. February.
McCallum, M. 1985. Seafood Catch and Consumption in Urban Bays ofPuget Sound.
Washington State Department of Social and Health Services. January.
Merrill, T. and M. Opheim. 2013. Assessment of Cook Inlet Tribes Subsistence Consumption.
Revised. Seldovia Village Tribe, Environmental Department. Prepared for the SVT
Council. May 20.
Murray, D.M. and D.E. Burmaster. 1994. "Estimated Distributions for Average Daily
Consumption of Total and Self-Caught Fish for Adults in Michigan Angler Households. "
Revised. In: Risk Analysis. 14: 4, 513-519. February.
Nez Perce Tribe
Volume III - Appendix H
December 2016 Page H-58
-------
Ochsner, Jean. 1996. Technical Memorandum on the results of the 1995 Fish Consumption and
Recreational Use Surveys - Amendment No. 1. Memorandum to Chee Choy. Adolfson
Associates, Inc. April 19.
Patterson, R.E., A.R. Kristal, L.F. Tinker, R.A. Carter, M.P. Bolton, and T. Agurs-Collins. 1999.
Measurement Characteristics of the Women's Health Initiative Food Frequency
Questionnaire. Ann Epidemiol. 1999;9(3): 178-187.
Pierce, D. 1981. Commencement Bay Seafood Consumption Study. Preliminary Report. Tacoma-
Pierce County Health Department. December.
Polissar, N., M. Neradilek, A.Y. Aravkin, P. Danaher, and J. Kalat. 2012. Statistical Analysis of
National and Washington State Fish Consumption Data, DRAFT. The Mountain-
Whisper-Light Statistics. September 18.
Prentice, R.L., Y. Mossavar-Rahmani, Y. Huang, L. Van Horn, S.A.A. Beresford, B. Caan, L.
Tinker, D. Schoeller, S. Binghamy, C.B. Eaton, C. Thomson, K.C. Johnson, J. Ockene,
G. Sarto, G. Heiss, and M.L. Neuhouser. 2011. Evaluation and Comparison of Food
Records, Recalls, and Frequencies for Energy and Protein Assessment by Using
Recovery Biomarkers. Am J Epidemiol. 2011; 174(5):591—603.
RIDOLFI Inc. (Ridolfi). 2007. Yakama Nation Exposure Scenario for Hanford Site Risk
Assessment. Prepared for the Yakama Nation ERWM Program. September.
Ruffle, B., D.E. Burmaster, P.D. Anderson, and H.D. Gordon. 1994. "Lognormal distributions
for fish consumption by the general U.S. population. " In: Risk Analysis. 14: 4, 395-404.
January 3.
Sechena, R., C. Nakano, S. Liao, N. Polissar, R. Lorenzana, S. Truong, and R. Fenske. 1999.
Asian and Pacific Islander Seafood Consumption Study. EPA Environmental Justice
Community/University. EPA 910/R-99-003. May 27.
Singer, E. and C. Ye. 2013. "The use and effects of incentives in surveys. " In: The ANNALS of
the American Academy of Political and Social Science. 645: 112.
Spokane Regional Health District. 1998. 1998 Fish Consumption Survey.
Assessment/Epidemiology Center. Prepared for the Washington State Attorney General's
office and the Department of Ecology.
Sun Rhodes, N.A. 2006. Fish Consumption, Nutrition, and Potential Exposure to Contaminants
Among Columbia River Basin Tribes, A Thesis. Prepared for Department of Public Health
and Preventive Medicine, Oregon Health and Science University. April.
Suquamish Tribe. 2000. Fish Consumption Survey of the Suquamish Indian Tribe of the Port
Madison Indian Reservation, Puget Sound Region. Port Madison Indian Reservation,
Fisheries Department. August.
Towksjhea, A., S. Iwenofu, L. Kissinger, and A.H. Williams. RARE Project Tribal Seafood
Consumption Survey Software. Quinault Indian Nation.
Toy, K.A., N.L. Polissar, S. Liao, and G.D. Mittelstaedt. 1996. A Fish Consumption Survey of
the Tulalip and Squaxin Island Tribes of the Puget Sound Region. Department of
Environment, Tulalip Tribes and Mountain-Whisper-Light Statistical Consulting.
October.
Nez Perce Tribe
Volume III - Appendix H
December 2016 Page H-59
-------
Tran, N.L., L.M. Barraj, X. Bi, L.C. Schuda, and J. Moya. 2012. "Estimated long-term fish and
shellfish intake—national health and nutrition examination survey. " In: Journal of
Exposure Science and Environmental Epidemiology. Nature America. 23, 128-136.
October 10.
U.S. Department of Health and Human Services. 1989. The Relationship of Human Levels of
Lead and Cadmium to the Consumption of Fish Caught in and Around Lake Coeur
D 'Alene, Idaho. Final Report. Agency for Toxic Substances and Disease Registry,
Division of Health Studies. Technical Assistance to the Idaho State Health Department
and the Indian Health Service. September.
U.S. Environmental Protection Agency (EPA). 2002. Estimated Per Capita Fish Consumption in
the United States. August.
Washington State Department of Health (WADOH). 1997. Anglers Who Frequently Fish Lake
Roosevelt. September.
Washington State Department of Health (WADOH). 2001. Lake Whatcom Residential and
Angler Fish Consumption Survey. April.
Westat. 2012. Upper Columbia River Site Remedial Investigation and Feasibility Study Tribal
Consumption and Resource Use Survey. Final Report. Prepared for the U.S.
Environmental Protection Agency. June 22.
Western Institutional Review Board (WIRB). 2013. A Guide for Researchers.
Nez Perce Tribe
Volume III - Appendix H
December 2016 Page H-60
-------
3.
Traditional Lifeways and Suppression Studies
Baumhoff, M.A. 1963. "Ecological determinants of aboriginal California populations." In:
University of California Publications in American Archaeology and Ethnology.
Cambridge University Press. Vol. 49, No. 2, pp.155 - 236. May 28.
Burger, J. 1999. American Indians, Hunting and Fishing Rates, Risk, and the Idaho National
Engineering and Environmental Laboratory. Environmental Research, Section A, 80,
317-329.
Craig, J. A. and R.L. Hacker. 1938. The History and Development of the Fisheries of the
Columbia River. United States Bureau of Fisheries. Bulletin No. 32.
Dall, W.H. 1897. "Alaska and Its Resources." Boston Lee and Shepard Publishers. Digitized by
Google.
Donatuto, J. and B.L. Harper. 2008. "Issues in evaluating fish consumption rate for native
American tribes. " In: Risk Analysis. 28: 6, 1497-1506.
Dunn, L. and W.S. White. 2013. The Incidence of Disease in the Lower Elw ha Klallam Tribe as
Related to the Decrease in Fish Consumption, Specifically Salmon and Shellfish. Lower
Elwha Klallam Tribe.
Environmental Health Perspectives (EHP). 2013. Meeting the Needs of the People, Fish
Consumption Rates in the Pacific Northwest. Vol. 121, No. 11-12, p. A335-339.
November-December.
Harper, B.L. 2007. Traditional Tribal Subsistence Exposure Scenario and Risk Assessment
Guidance Manual. EPA-STAR-J1-R831046. August.
Harper, B. and D. Ranco. 2009. Wabanaki Traditional Cultural Lifeways Exposure Scenario.
Prepared for EPA. July, 9.
Harper, B., A. Harding, S. Harris, and P. Berger. 2012. Exposure Assessment Articles:
Subsistence Exposure Scenarios for Tribal Applications. Human and Ecological Risk
Assessment, 18:810-831.
Harris, S.G., and B.L. Harper. 1997. "A native American exposure scenario. " Revised. In: Risk
Analysis. 17: 6, 789-795. April 29.
Harris, S.G., and B.L. Harper. 2000. "Using eco-cultural dependency webs in risk assessment
and characterization of risks to tribal health and cultures. " In: Environmental Science
and Pollution Research (ESPR). Ecomed Publishers. Special Issue 2, 91-100.
Harris, S.G. and B.L. Harper. 2004. Exposure Scenario for CTUIR Traditional Subsistence
Lifeways. Department of Science & Engineering, Confederated Tribes of the Umatilla
Indian Reservation, Pendleton, Oregon.
Hewes, G.W. 1947. Aboriginal use of fishery resources in northwestern North America. Doctoral
dissertation, University of California, Berkeley.
Hewes, G.W. 1973. Indian fisheries productivity in precontact times in the Pacific salmon area.
Northwest Anthropological Research Notes 7:133-155.
Nez Perce Tribe
Volume III - Appendix H
December 2016 Page H-61
-------
HistoryLink.org. 2013. Northwest Indian Fisheries Commission. HistoryLink File #9786.
Accessed May 2013. Available:
http://www.historylink.org/index.cfm?DisplavPage=output.cfm&file id=9786.
Hunn E.S. 1990. Nch "i-Wana, The Big River: Mid-Columbians and Their Land. University of
Washington Press, Seattle, Washington.
Hunn, E.S. and C.L. Bruneau. 1989. Estimates of Traditional Native American Diets in the
Columbia Plateau. Work supported by the U.S. Department of Energy and Pacific
Northwest Laboratory. August.
O'Neill, C.A. 2000. Variable Justice: Environmental Standards, Contaminated Fish, and
"Acceptable " risk to Native Peoples. Sanford Environmental Law Journal. 19 stan. Envtl.
L.J.3. January.
O'Neill, C.A. 2007. Protecting the Tribal Harvest: The Right to Catch and Consume Fish. J.
Envtl. Law and Litigation. Vol. 22, 131.
O'Neill, C.A. 2008. Environmental Justice in the Tribal Context: A Madness to EPA's Method.
Environmental Law, Lewis & Clark Law School. Issue 38:2.
O'Neill, C.A. 2013. "Fishable Waters." In: American Indian Law Journal. Vol. 1, Issue 2, pp.
181-284.
Ray, V.F. 1977. Ethnic Impact of the Events Incident to Federal Power Development on the
Colville and Spokane Indian Reservations.
Riley, D.M., C.A. Newby, and T.O. Leal-Almeraz. 2006. "Incorporating ethnographic methods
in multidisciplinary approaches to risk assessment and communication: cultural and
religious uses of mercury in Latino and Caribbean communities. " In: Risk Analysis. 26:
5, 1205-1221.
Sanger, D. 1988. "Maritime adaptations in the gulf of Maine. " In: Archaeology of Eastern North
America. Eastern States Archeological Federation. 16, 81-99.
Schalk, R.F. 1986. Estimating salmon and steelhead usage in the Columbia basin before 1850:
the anthropological perspective. Northwest Environmental Journal. Vol. 2 (2), p. 1-26.
Sullivan, R.J. 1942. "The Ten'a food quest." In: The Catholic University of America
Anthropological Series. Dissertation. The Catholic University of America Press. No. 11.
Swindell, E.G. 1942. Report on source, nature and extent of the fishing, hunting and
miscellaneous related rights of certain Indian tribes in Washington and Oregon together
with affidavits showing location of a number of usual and accustomedfishing grounds
and stations. United States Department of the Interior, Office of Indian Affairs, Division
of Forestry and Grazing. Los Angeles, California. July.
Tiller, V.E.V. 2005. Tiller's Guide to Indian Country, Economic Profiles of American Indian
Reservations.
Walker, D.E. 1967. "Mutual cross-utilization of economic resources in the plateau: an example
from aboriginal Nez Perce fishing practices." In: Washington State University Laboratory
of Anthropology Report of Investigations. No. 41, p. 1-70.
Nez Perce Tribe
Volume III - Appendix H
December 2016 Page H-62
-------
Walker, D.E. 1992. "Productivity of tribal dipnet fishermen at celilo falls: analysis of the Joe
Pinkham fish buying records." In: Northwest Anthropological Research Notes. Vol. 26:
No. 2, pp. 123 - 135.
Walker, D.E. (ed). 1998. Handbook of North American Indians, Volume 12: Plateau.
Washington, DC: Smithsonian Institution.
Wolfe, R.J. and R.J. Walker. 1987. "Subsistence economics in Alaska: productivity, geography,
and development impacts." In: Arctic Anthropology. Vol. 24, No. 2, pp. 56-81.
Nez Perce Tribe
Volume III - Appendix H
December 2016 Page H-63
-------
8.0 REFERENCES CITED
Center for Columbia River History (CCRH). 2013. Nez Perce, Nee-Me-Poo. Available:
http://www.ccrh.org/comm/river/profile/nezperce.php. Accessed September 19.
Coulston, A.M., C.J. Boushey, and M.G. Ferruzzi. 2001. Nutrition in the Prevention and
Treatment of Disease. Third Edition, p. 12. Available:
http://appliedresearch.cancer.gov/diet/adi/thompson_subar_dietary_assessment_methodol
ogy.pdf.
Dodd, K.W., P.M. Guenther, L.S. Freedman, A.F. Subar, V. Kipnis, D. Midthune, J.A. Tooze,
and S.M. Krebs-Smith. 2006. "Statistical methods for estimating usual intake of nutrients
and foods: a review of the theory." J Am Diet Assoc 2006; 106(10): 1640-50.
Doyle, J.K. 2005. Face-to-Face Surveys. Available:
http://www.wpi.edu/Images/CMS/SSPS/Doyle_-_Face-to-Face_Surveys.pdf.
Everett, S., C. Beasley, R. Johnson, C. Davenport, and R. Larson. 2006. "Nez Perce Tribal
Hatchery Project; Coho Salmon Master Plan, Clearwater River Basin", 2003-2004
Annual Report, Project No. 198335000, 134 electronic pages, (BPA Report DOE/BP-
00004035-1). FishPro Division ofHDR Engineering.
Hollander, M. and D.A. Wolfe. 1999. Non-parametric Statistical Methods, Second Edition,
Wiley.
Hyndman, R.J. and Y. Fan. 1996. "Sample Quantiles in Statistical Packages". American
Statistician (American Statistical Association); 50 (4): 361-365. November.
Idaho Department of Fish and Game (JDFGChinook Salmon (Snake River Spring Summer Ran)
Conservation Status. Available: http://fishandgame.idaho.gov/ifwis/cwcs/
pdf/Chinook%20Salmon%20%28Snake%20River%20spring_summer%20run%29.pdf.
Accessed September 13,
Kipnis V., D. Midthune, D.W. Buckman, K.W. Dodd, P.M. Guenther, S.M. Krebs-Smith, A.F.
Subar, J.A. Tooze, R.J. Carroll, and L.S. Freedman. 2009. "Modeling data with excess
zeros and measurement error: application to evaluating relationships between
episodically consumed foods and health outcomes." Biometrics 65, 2009, pp. 1003-1010.
Landeen, D. and A. Pinkham. 1999. Salmon and His People: Fish and Fishing in Nez Perce
Culture.
Maguire, K.B.. 2009. Does mode matter? A comparison of telephone, mail, and in-person
treatments in contingent valuation surveys. J Environ Manage. Aug;90(l 1 ):3528-33.
Available: http://www.ncbi.nlm.nih.gov/pubmed/19647362.
Meyer Resources, Inc. 1999. Tribal Circumstances and Impacts fi'om the Lower Snake River
Project on the Nez Perce, Yakama, Umatilla, Warm Springs, and Shoshone Bannock
Tribes. Prepared for CRITFC, funded by USACE.
National Environmental Justice Advisory Council (NEJAC). 2002. Fish Consumption and
Environmental Justice. Report developed from NEJAC Meeting of December 3-6, 2001
(Federal Advisory Committee to the EPA). November.
Nez Perce Tribe
Volume III - Appendix H
December 2016 Page H-64
-------
Nez Perce Tribe (NPT). 2005. White Sturgeon Management Plan in the Snake River Between
Lower Granite and Hell's Canyon Dams. Department of Fisheries Resource
Management. Lapwai, ID. September.
Polissar, N., M. Neradilek, A.Y. Aravkin, P. Danaher, and J. Kalat. 2012. Statistical Analysis of
National and Washington State Fish Consumption Data, DRAFT. The Mountain-
Whisper-Light Statistics. September 18.
Scholz, A., K. O'Laughlin, D. Geist, D. Peone, J. Uehara, L. Fields, T. Kleist, 1. Zozaya, T.
Peone, and K. Teesatuskie. 1985. Compilation of Information on Salmon and Steelhead
Total Run Size, Catch and Hydropower Related Losses in the Upper Columbia River
Basin, above Grand Coulee Dam. Fisheries Technical Report No. 2. Upper Columbia
United Tribes Fisheries Center, Eastern Washington University, Department of Biology.
Cheney, Washington 99004. December.
Smithsonian Institution. 1907. Handbook of American Indians North of Mexico. Bureau of
American Ethnology, Bulletin 30. Ed. F.W. Hodge. Washington Government Printing
Office. Tooze, J., et. al. 2006. A new statistical method for estimating the usual intake of
episodically consumed foods with application to their distribution. Journal of the
American Dietetic Association 106:10, 2006, pp. 1575-1587.
Nez Perce Tribe
Volume III - Appendix H
December 2016 Page H-65
------- |