United States Office of Water EPA 570/9-90-015
Environmental Protection Office of Pesticides and November 1990
Agency Toxic Substances
National Survey of Pesticides
in Drinking Water Wells
Phase I Report
-------
National Pesticide Survey
Phase I Report
US. Environmental Protection Afency
Region 5, Library (PI-12JJ
77 West Jackson Boulevard, 12th
ll 60604-3590
November 1990
-------
-------
Acknowledgements
The following organizations participated in the preparation of this report:
U.S. EPA Office of Water
U.S. EPA Office of Drinking Water
U.S. EPA Office of Pesticides & Toxic Substances
U.S. EPA Office of Pesticide Programs
U.S. EPA Technical Support Division
U.S. EPA Environmental Chemistry Section
U.S. EPA Regions 1 - 10
EPA National Pesticide Survey Regional/State/County Work Group
ICF Incorporated
Stretton Associates, Incorporated
Technology Applications Incorporated
Westat Incorporated
In addition to the organizations listed above, the following also assisted in the Survey:
Each of the 50 States
County Agricultural Extension Agents throughout the country
National Association of County Health Officials (NACHO)
National Association of Counties (NACO)
State FIFRA Issues Research and Evaluation Group (SFIREG)
National Pesticide Survey: Phase I Report
-------
Table of Contents
Page
Executive Summary vii
Chapter One: Introduction i
Introduction 1
1.1 Goals of the National Pesticide Survey 1
1.2 Overview 2
1.3 How to Use This Report 4
1.4 Additional NFS Information 5
Chapter Two: Survey Design 7
Introduction 7
2.1 Key Elements of Survey Design 7
2.2 Community Water Systems 15
2.3 Rural Domestic Wells 18
Chapter Three: Implementation of the Survey 23
Introduction 23
3.1 Key Participants 23
3.2 Water Sampling and Data Collection 24
3.3 Questionnaires 28
3.4 Survey Operations Management 28
3.5 Data Management 30
3.6 Communications 31
Chapter Four: Analytical Chemistry 35
Introduction 35
4.1 Analyte Selection and Methods Development 35
4.2 Analytical Laboratories 36
4.3 Minimum Quantification and Reporting Limits 36
4.4 Confirmation of Positive Detections 40
4.5 Quality Control 40
4.6 Precision and Accuracy 41
4.7 Analyte Stability Studies 41
4.8 False Positives and False Negatives 42
Chapter Five: Quality Assurance/Quality Control 43
Introduction 43
5.1 Overview of the Quality Assurance Program 43
5.2 Quality Control 47
5.3 Audits 47
5.4 Corrective Action Summary 48
National Pesticide Survey: Phase I Report
-------
vi Table of Contents
Chapter Six: Findings and Results
Introduction 51
6.1 National Estimates 53
6.2 National Estimates for Pesticides 58
6.3 Estimated Concentrations of Frequently Detected
Analytes and Comparison to Regulatory and Health Limits 63
6.4 Characteristics of CWS Wells and Rural Domestic Wells 69
6.5 Special Interest Population Domains 76
6.6 Additional Analyses to Appear in Phase II Report 84
Chapter Seven: Survey Products 89
Introduction 89
7.1 Survey Products 89
Glossary and List of Acronyms 91
Appendices
Appendix A: Background A-l
Appendix B: Statistical Design and Analysis B-l
Appendix C: Implementation of the Survey C-l
Appendix D: Questionnaire Data and Questionnaires D-l
Appendix E: Survey Analytes and Analytical Chemistry Tables E-l
National Pesticide Survey: Phase I Report
-------
Executive Summary
The U.S. Environmental Protection Agency (EPA) has completed its five-year National Survey of
Pesticides in Drinking Water Wells (the Survey or NPS). The Survey is the first and perhaps the most
extensive monitoring survey ever undertaken to evaluate the presence of pesticides, pesticide degradates, and
nitrate in drinking water wells in the United States. EPA's Phase I Report provides national estimates of the
occurrence and frequency of detections of nitrate and pesticides in drinking water wells. The Phase I results
of the Survey indicate that at least half of the nation's drinking water wells contain detectable amounts of
nitrate, with a small percentage at concentrations higher than EPA's regulatory and health-based limits for
drinking water (about 1.2% of community water system (CWS) wells and about 2.4% of rural domestic wells).
Based on the results of the Survey, EPA estimates that about 52.1% of the 94,600 community water
system wells in the United States contain nitrate; about 10.4% contain one or more pesticides; and about 7.1%
may contain both. Of the approximately 10.5 million rural domestic wells, EPA estimates that about 57.0%
contain nitrate, 4.2% contain one or more pesticides, and about 3.2% contain both. Exhibit 1 illustrates these
proportions. EPA estimates that less than one percent (0.6%) of rural domestic wells containing pesticides,
or approximately 60,900 wells, contain at least one pesticide over a Maximum Contaminant Level (MCL) or
Lifetime Health Advisory Level (HAL). None of the detections of pesticides or pesticide degradates for CWS
wells were above the MCL/HAL. Nevertheless, based on its statistical analysis of the Survey results, EPA
estimates that less than 1% (0.8%) of community water system wells contain pesticides at concentrations
higher than the MCL or HAL for those chemicals for which an MCL or HAL has been established. These
results are illustrated by Exhibit 2.
The pesticides detected most frequently in the Survey were DCPA acid metabolites and atrazine.
DCPA acid metabolites are degradates of DCPA, which is used primarily as a weed killer on lawns, turf, and
golf courses, and also on a variety of fruits and vegetables. In addition, ten other pesticides were detected at
concentrations exceeding the Survey's minimum reporting limits. The concentrations of pesticides and
pesticide degradates detected were usually well below levels of health concern.
These Survey results indicate that the. proportion of wells nationwide found to contain any particular
pesticide or pesticide degradate is low. Considering only the proportion of wells containing pesticides over
the EPA levels of health concern, Survey results do not demonstrate any immediate widespread health
problem. Survey results also show, however, that substantial numbers of wells, particularly rural domestic
wells, could be affected by the presence of one or more pesticides. In addition, substantial numbers of wells
are affected both by the presence of nitrate and by nitrate over EPA levels of health concern. These results
indicate that there is need for continued attention to ground-water protection and additional analysis of the
issue. A Phase II report, to be released in 1991, will discuss EPA's assessment of the causes and factors
associated with the occurrence of pesticides and nitrate in drinking water wells.
Five pesticides - alachlor, atrazine, dibromochloropropane (DBCP), ethylene dibromide (EDB), and
gamma-HCH (Lindane) - were detected in rural domestic wells at levels above their respective MCLs/HALs.
MCLs are enforceable standards established by EPA pursuant to the Safe Drinking Water Act as the maximum
permissible level of a contaminant that is allowed in a public water supply. Although they do not apply as
a matter of law to rural domestic wells, MCLs were used as a standard of quality for source drinking water.
HALs are the maximum concentration of a contaminant in water that may be consumed safely over an average
human lifetime.
National Pesticide Survey: Phase I Reoort
-------
viii Executive Summary
Exhibit 1
Estimated Percent of Community Water System Wells and
Rural Domestic Wells in the United States with Nitrate and Pesticides
Community Water System Wells
Rural Domestic Wells
Without Nitrate or
Pesticides (44.6%)
With Nitrate Only
Without Nitrate or
PMtlcidM (42.0%)
With Nitrate Only
(53.8%)
With PestlcWes With Nitrate and
Only (3.3%) Pesticides (7.1%)
With PMtlcidM
Only (1.0%) with NltraM and
PMtlcidM (3.2%)
Exhibit 2
Estimated Percent of Community Water System Wells and
Rural Domestic Wells in the United States Containing Pesticides
Exceeding Health-Based Limits
Community Water System Wells
Above MCL/HAL
Rural Domestic Wells
Above MCL/HAL
(0.8%)
Below MCL/HAL
If 995%)
Bated on upper bound of
the 95% confidence Interval
Below MCL/HAL
(99.4%)
Relatively small proportions of wells in the United States are estimated to contain nitrate over health-
based limits. An estimated 1.2% of community water system wells and an estimated 2.4% of rural domestic
wells contain nitrate above the Maximum Contaminant Level of 10 mg/L, as illustrated by Exhibit 3.
-------
Executive Summary ix
Exhibit 3
Estimated Percent of Community Water System Wells and
Rural Domestic Wells in the United States Containing Nitrate
Exceeding Health-Based Limits
Community Water System Wells
Rural Domestic Wells
Without Mitral*
(47.9%)
Without Nitrate
(43.0%)
With Mitral* Below
10 mgrt. (MCL)
(50.9%)
With Nltrat* Btlow
10 mg/L (MCL)
(54.6%)
With Nltrat* At Or Above
10 mg/L (MCL)
(1.2*)
With Nltrat* At Or Above
10 mg/L (MCL)
Exhibits 4 and 5 show the estimated number and percent of wells in the United States containing each
of the 13 chemicals detected at concentrations above the Survey's minimum reporting limits.
Exhibit 4
Estimated Number and Percent of Community Water System Wells
Containing NPS Anafytes
Analyte
Nitrate
DCPA acid metabolites
Atrazine
Simazine
Prometon
Hexachlorobenzene*
Dibromochloropropane (DBCP)*
Dinc«eb*
Estimated
Number
49,300
-6,010
1,570
1,080
520
470
370
25
95% Confidence
Interval
(lower - upper)
(45,300 - 53,300)
(3,170 - 8,840)
(420 - 2,710)
(350 - 2,540)
(78 - 1,710)
(61 -1,630)
(33-1,480)
(1 - 870)
Estimated
Percent
52.1
6.4
1.7
1.1
0.5
0.5
0.4
95% Confidence
Interval
(lower - upper)
(50.0 - 56.3)
(3.4 - 9.3)
(0.5 - 2.9)
(0.5 - 2.7)
(0.1 - 1.8)
(0.1 - 1.7)
(0.1 - 1.6)
<0.1 (<0.1 - 0.9)
Registration cancelled by EPA.
National Pesticide Survey: Phase I Report
-------
Executive Summary
Exhibit 5
Estimated Number and Percent of Rural Domestic Wells
Containing NPS Analytes
Analyte
Nitrate
DCPA acid metabolites
Atrazine
Dibromochloropropane
(DBCP)*
Prometon
Simazine
Ethylene dibromide*
gamma-HCH
Ethylene thiourea
Bentazon
Alachlor
Estimated
Number
5,990,000
264,000
70,800
38,400
25,600
25,100
19,200
13,100
8,470
7,160
3,140
95% Confidence Interval
(lower - upper)
(5,280,000 - 6,700,000)
(129,000 - 477,000)
(13,300-214,000)
(2,740 - 164,000)
(640 - 142,000)
(590- 141,000)
(160-131,000)
(14 - 120,000)
(1 - 111,000)
(1 - 109,000)
(1 -101,000)
Estimated
Percent
57.0
2.5
0.7
0.4
0.2
0.2
0.2
0.1
0.1
0.1
95% Confidence
Interval
(lower - upper)
(50.3 - 63.8)
(1.2-4.5)
(0.1 - 2.0)
(<0.1 - 1.6)
(<0.1 - 1.4)
(<0.1 - 1.3)
(<0.1 - 1.2)
(<0.1 - 1.1)
(<0.1 - 1.1)
(<0.1 - 1.0)
<0.1 (<0. 1-1.0)
* Registration cancelled by EPA.
There are significant differences in the uses and possible sources of the most commonly detected
Survey analytes. Nitrate and nitrite can occur naturally as well as a result of human activity. (Nitrate and
nitrite were not separately distinguished in the Survey laboratory methods and testing, and the results
presented reflect both nitrate and nitrite, measured as nitrogen). Nitrate is found in some fertilizers, as well
as a result of nitrification of ammonium fertilizers. Nitrate is present in septic systems, animal feed lot wastes,
industrial wastewaters, and sanitary landfills. DCPA, a selective herbicide, is the parent compound of DCPA
acid metabolites. DCPA is also known by the common names Dacthal and chlorthal dimethyl. It is used to
combat certain annual grasses and broadleaf weeds, and is approved for use on turf, ornamentals, and a
number of vegetables. Atrazine is used to control many annual broadleaf weeds and certain grasses, and is
used primarily on corn and soybeans.
A different number of wells might contain pesticides or nitrate than are indicated by these results.
The Survey randomly sampled geographic areas of the country. Local or regional areas with particularly high
or low levels of chemicals may not have been chosen for sampling. EPA scheduled the Survey's well sampling
activities over a 22 month period to help ensure that water samples were obtained during all seasons and
pesticide application cycles. EPA plans to test whether Survey results might be affected by the timing of
sampling. The 95% confidence intervals reflect this variability. In addition, the Survey used a stringent series
of detection limits and confirmation requirements to ensure a high degree of certainty about the detections
that were reported. Additional chemicals might have been present in the wells sampled, at levels too low to
be reported as detections by the Survey.
The Survey's findings and results are the product of a lengthy and complicated study. A joint project
of EPA's Office of Drinking Water (ODW) and Office of Pesticide Programs (OPP), the Survey's design began
National Pesticide Survey: Phase I Report
-------
Executive Summary xi
in 1984, and involved the development of the statistical procedures, analytical chemistry, implementation
methods, health advisories, and quality control procedures necessary to ensure the quality, accuracy, and
reliable communication of Survey results. EPA conducted a pilot study in California, Mississippi, and
Minnesota in 1987 to test Survey implementation and analytical procedures. The design of the Survey was
reviewed and approved by a special subpanel of EPA's independent Federal Insecticide, Fungicide, and
Rodenticide Act (FIFRA) Scientific Advisory Panel in 1985 and again in 1987 after the pilot study. After
extensive planning and preparation, full scale sampling for the National Pesticide Survey began in April 1988,
and involved the participation of hundreds of persons from States, counties, EPA's Regional Offices, and
private citizens. The final sample was collected in February 1990. More than 1300 wells were sampled, some
in every State, and extensive data were collected about the wells' characteristics and condition, in addition to
surrounding circumstances such as nearby pesticide use and agricultural activities, at a total cost of
approximately $12 million.
The Survey's findings will help EPA characterize the nature and extent of pesticide residues in
drinking water wells, and will help the Agency set priorities, prepare guidance, and implement regulatory
programs. The following timeline shows the key events in the life of the Survey.
Exhibit 6
Timeline of Survey Activities
Mgki
*««y
DMlgn
togta
Ptot Study
Sctontfle
Advfcory
PcnriRwlM
MMBo
ji-fcifailp..
Miwwy
PtnriAwlM
End
Plot Study
tegki
Commonly
Wil temping
teg* Aural
OomwtfcWil
temping
Endftnl
OomMtfcWil
wiping
End
Convnunly
WM BftfflpInQ
CompMtonof
Mi Am**
CompMtonor
Chtmttl
Antlyik
AMMMOf
PtBMl
top*
1M4
1QM
a/i?
M7
M7
4/M
7/M
MO
MO
11/90
Discussion of Findings
The National Pesticide Survey has two principal objectives: (1) to determine the frequency and
concentration of pesticides and nitrate in drinking water wells nationwide, and (2) to improve EPA's
understanding of how the presence of pesticides and nitrate in drinking water wells is associated with patterns
of pesticide use and the vulnerability of ground water to contamination. This Phase I Report answers the first
objective. A Phase n report, planned for Spring 1991, will respond to the second objective.
EPA tested well water from wells at community water systems and from rural domestic wells for 127
analytes, including 101 pesticides, 25 pesticide degradates, and nitrate. Detections of the analytes were
reported when the concentration found was greater than the Survey's specified minimum reporting limits. Of
the 127 analytes, the presence of 13 was detected above their minimum reporting limits in wells sampled in
National Pesticide Survey: Phase I Report
-------
xii Executive Summary
the Survey. Four additional analytes were detected for which national estimates could not be made. The
results indicated that nitrate is common in wells, sometimes above health based limits; pesticides are much
less common. The concentrations of pesticides and pesticide degradates were usually lower than health-based
limits.
The most frequently detected chemical was nitrate. EPA estimates, based on Survey results, that about
52.1% of the approximately 94,600 community water system wells and about 57.0% of the 10.5 million rural
domestic wells in the country (about 5.9 million wells) contain levels of nitrate above the NPS minimum
reporting limit of 0.15 mg/L.
Out of all community water system wells nationwide, however, about 1.2% (or approximately 1,130
wells) are estimated to contain nitrate over EPA's Maximum Contaminant Level (MCL) of 10 mg/L. About
2.4% of rural domestic wells, or approximately 254,000 wells, are estimated to contain nitrate over the MCL.
In all, EPA detected nitrate and 16 different pesticides or pesticide degradates in drinking water wells
nationwide. The most commonly found chemicals, other than nitrate, are the pesticide degradate DCPA acid
metabolites and the pesticide atrazine. Other pesticides that were detected at concentrations at or above the
Survey's minimum reporting limit include alachlor, bentazon, dibromochloropropane (DBCP), dinoseb,
ethylene dibromide (EDB), etheylene thiourea (ETU), hexachlorobenzene, gamma-HCH (Lindane), prometon,
and simazine. In addition, three analytes (alpha-chlordane, gamma-chlordane, and beta HCH) were detected
by EPA laboratories at concentrations lower than the minimum reporting limits used by the contract
laboratories. Because EPA laboratories only tested 10% of the sampled wells, national estimates of the
occurrence of these three chemicals are not possible. One analyte, 4-nitrophenol, was detected but a national
estimate could not be produced due to limitations of the laboratory analysis method for this particular
chemical.
EPA estimates that about 10.4% of the community water system wells nationwide contain detectable
concentrations of at least one pesticide. About 4.2% of rural domestic wells contain a detectable pesticide
concentration above the Survey's minimum reporting limit. Because the Survey's findings are based on a
complex statistical design, in which national estimates are developed on the basis of a sample of wells, the best
national estimates fall between a high and a low estimate. For CWS wells, the number of wells containing
pesticides is approximately 9,850 wells, but the number of wells could be as high as 13,400 wells or as low as
6,330 wells. For rural domestic wells, the number of wells containing pesticides is approximately 446,000 wells,
but the number of wells could be as high as 647,000 wells or as low as 246,000 wells.
Of all the pesticides and pesticide degradates detected in the Survey, DCPA acid metabolites and
atrazine were the most commonly found. For DCPA acid metabolites, degradates of DCPA, the maximum
concentrations detected were 7.2 /ig/L (parts per billion) for CWS wells and 2.4 ng/L for rural domestic wells.
The median concentrations for DCPA acid metabolites were 0.34 yxg/L for CWS wells and 0.38 /zg/L for rural
domestic wells. The maximum concentrations of atrazine detected in Survey drinking water samples were
about 0.92 ptg/L for CWS wells and 7.0 ptg/L for rural domestic wells. The median concentrations of atrazine
were 0.26 /ig/L for CWS wells and 0.28 /ig/L for rural domestic wells.
How Did EPA Select Wells?
The Survey was designed to yield results that are statistically representative of the nation's CWS wells
and rural domestic wells. EPA used statistical survey methods to select a nationally representative subset of
CWS wells and rural domestic wells for sampling. The results are not representative of any State or local area.
First, EPA characterized all counties in the U.S. according to pesticide use and relative ground-water
vulnerability - two critical factors affecting the presence of pesticides in drinking water wells. EPA
concentrated on agricultural pesticide use, specified as high, moderate, low, or uncommon pesticide use. EPA
specified areas of greater and lesser relative ground-water vulnerability by using a numerical classification
system called DRASTIC, which considers seven factors that may affect the vulnerability of ground water to
contamination - depth of water, recharge, aquifer media, soil media, topography, impact of vadose
National Pesticide Survey: Phase I Report
-------
Executive Summary xiii
(unsaturated) zone, and hydraulic conductivity of aquifers. The Survey design called for specified numbers of
wells to be sampled from areas of greater and lesser pesticide use and ground-water vulnerability.
To identify CWS wells, EPA randomly selected 7,083 community water systems from a list containing
information on all public water supply systems. To ensure more precise estimation of analyte detection rates,
wells in counties that are more vulnerable to ground-water contamination were slightly over-represented in
the group of selected public water system wells. EPA conducted telephone interviews with representatives of
the 7,083 selected systems and based on the results of the screening process identified systems eligible for
sampling. Water samples from 566 wells in 50 States were collected, 540 of which were used in data analysis
based on well water samples that passed quality assurance requirements.
When selecting rural domestic wells, EPA randomly chose 90 counties in 38 States as areas for
sampling to represent the nation's wide range of agricultural pesticide use and ground-water vulnerability.
EPA assessed ground-water vulnerability within subcounty areas and collected information from county
agricultural extension agents on cropping intensity to further subdivide the counties into areas that are more
or less vulnerable to the presence of pesticides in ground water. Rural domestic wells in locations with high
pesticide use and high vulnerability to ground-water contamination were slightly overrepresented. Of the
eligible wells, 783 wells were sampled, 752 of which were used in data analysis based on well water samples
that passed quality assurance requirements.
Once the wells were selected, EPA scheduled sample collection so that wells were visited randomly
over the 22 month sampling period. This schedule provided well water samples during all seasons and
pesticide application cycles. This approach was used to minimize the effect of seasonal variability.
Sampling Across the Country
From April 1988 to February 1990, EPA collected water samples and well information from over 1,300
community water system wells and rural domestic wells. EPA visited each well once, collecting a minimum
of 17 bottles of well water. State agencies across the country provided the sampling teams to collect samples
from community water systems. EPA's contractors conducted the domestic well sampling. CWS wells were
sampled in every State and rural domestic wells were sampled in 38 States.
At each well sampled, questionnaires were used to collect data necessary for the interpretation of NPS
results. Data included:
observations about the well sampled and the surrounding area;
information from the owner/operator about well construction and agricultural and non-
agricultural pesticide usage on the property where the well was located; and
information from local area experts (such as a county agricultural extension agent) about
crops, pesticide use, and land use within one-half mile of the well.
Chemical Analysis
Each water sample gathered in the National Pesticide Survey was tested for 127 analytes. Through
extensive literature searches, consultation with scientific experts, and assessment of methods in the laboratory,
EPA evaluated existing laboratory methods for testing the analytes. Because of the wide variety of procedures
used in these methods and the large number of analytes to be included in the Survey, EPA needed methods
that could efficiently test for several analytes. Ultimately, EPA used two existing analytical methods and
developed six new methods specifically for the Survey. Five of these six, referred to as multi-residue methods,
are each capable of detecting ten or more pesticides. These new methods have been adopted by EPA for uses
other than the Survey, and are an important product of the NPS.
National Pesticide Survey: Phase I Report
-------
xiv Executive Summary
The chemical analyses of NFS water samples were performed at five contract laboratories and three
EPA laboratories. The contract laboratories were responsible for chemical analyses of water samples collected
in the field by one or more of the established methods. Two EPA laboratories were responsible for managing
contract laboratories, confirming detections of pesticides in samples, and ensuring that quality control
standards were maintained; the third EPA laboratory performed chemical analyses.
Health Advisories
To assist in notifying Survey participants of sampling results, EPA prepared one-page Health Advisory
Summaries to explain in non-technical terms the potential health effects of exposure to pesticides. These
Summaries are based on longer, more technical scientific documents called Health Advisories prepared by EPA
for Survey analytes before the beginning of full-scale sampling.
Communicating Results
In EPA's ten Regions, an NPS contact was identified to enlist the cooperation of State water supply
and pesticide agencies and to answer questions from the media, elected officials, and organizations interested
in the Survey. At the local level, county agricultural extension agents and health officials contributed to Survey
planning and implementation, provided a two-way flow of information between EPA and participating local
communities, and provided expert information on local conditions.
When sampling results showed pesticide and nitrate levels below levels of health concern, EPA sent
appropriate Health Advisory Summaries to well owners and operators along with Survey sampling results,
usually within 2 to 3 months of sample collection.
Whenever the results of the analysis of water samples from either a community water system well or
a rural domestic well indicated that a pesticide was present in the well in a concentration over the MCL or
HAL, and could pose a potential health risk, EPA notified the State in which the well was located by
telephone within 24 hours of confirmed analysis of the sample. The State, in turn, notified the owner or
operator of the well within an additional 48 hours.
Putting Quality First
EPA made an extraordinary commitment to ensuring the validity and accuracy of Survey results. A
quality assurance (QA) program covering every major component of the Survey ensured that the Survey
produced high quality, statistically valid data - useful to both scientists and policy makers. EPA developed
QA Project Plans for well selection, data collection and analysis, well sampling, interviewing, and laboratory
analysis. During the Survey, EPA conducted numerous audits of field and laboratory activities to ensure that
procedures approved as part of the QA program were followed. EPA conducted performance evaluation
studies of the analytical laboratories to monitor laboratory capability.
Summary of Planned Phase II Analyses
EPA plans a number of statistical analyses using the data compiled from chemical analyses of water
samples, NPS questionnaires, and other sources. These analyses will study the association of pesticides and
nitrate in drinking water wells with such factors as fertilizer and pesticide use on the property where the well
is located, use within one-half mile of the well, and use within the county. Studies will also address the
association of pesticides and nitrate in wells with ground-water vulnerability characteristics, including depth
to ground water, recharge, aquifer media, soil media, topography, impact of unsaturated zone, and hydraulic
conductivity. The Phase II Report will analyze the questionnaire databases, the first-stage and second-stage
DRASTIC stratification scoring results, nitrogen fertilizer and pesticide sales, and the Survey analytical results,
to investigate the potential causes and consequences of pesticide residues in drinking water wells.
These analyses will include a study of whether there is an association among questionnaire responses.
The hypotheses that might be tested include the following: Are there more analyte detections of chemicals
National Pesticide Survey: Phase I Report
-------
Executive Summary XV
associated with unconfined aquifers than with confined aquifers? Are nitrate detections associated with septic
units on the property? Are there more analyte detections of chemicals associated with areas where irrigation
is used? EPA is planning scientific investigations to:
Determine the association among detections and well characteristics;
Correlate DRASTIC subscores by individual factors (e.g., depth to ground water) with
detections in rural domestic wells;
Determine associations between pesticide use and pesticide detections in water samples;
Prepare tables showing pesticide detections by county pesticide use estimates for specific
analytes that were detected in the Survey;
Prepare summary statistics such as frequencies of analytes detected in areas where crops are
grown with known associated pesticide use;
Conduct regression modeling for analytes with sufficient detections. These analyses will seek
to identify significant relationships between variables that are not readily discernible from the
analyses of well characteristics presented in the Phase I Report;
Evaluate analytes with few detections by non-statistical review of all available data, such as
marginal comments on questionnaires; and
Compare NPS findings with other studies.
Survey Information
This Phase I Report covers Survey design, implementation, analytical chemistry, quality assurance, and
summary results. In 1991, EPA will release a Phase II Report containing results and findings from analyses
of the relationships between various types of information gathered during the Survey.
National Pesticide Survey: Phase I Report
-------
Chapter One: Introduction
Introduction
The National Survey of Pesticides in Drinking Water Wells (the National Pesticide Survey, the Survey,
or NPS) is perhaps the most extensive monitoring survey ever undertaken to evaluate the presence of
pesticides, pesticide degradates, and nitrate in drinking water wells.1 The U.S. Environmental Protection
Agency (EPA), with the assistance and cooperation of EPA Regional Offices, States, local and county officials,
and private well owners, sampled 1349 drinking water wells throughout the country for 126 pesticides and
pesticide degradates, plus nitrate (a total of 127 analytes).2 The NPS analyzed the water samples using a
variety of sophisticated chemical techniques and collected additional information about the conditions in the
vicinity of the sampled wells. The design and implementation of the Survey required five years ~ at a cost of
over $12 million. EPA has prepared this Phase I report to describe the NPS and to provide the Survey's initial
findings and results. EPA plans to publish a Phase II report that will provide an expanded assessment of the
Survey's findings in spring 1991.
1.1 Goals of the National Pesticide Survey
The National Pesticide Survey was designed to provide, for the first time, a statistically reliable
assessment of the frequency and concentration of selected pesticides, pesticide degradates, and nitrate in the
drinking water wells of the nation. EPA designed the Survey, in addition, to provide data to better understand
how the presence of pesticides or nitrate in drinking water wells is associated with patterns of pesticide usage
and the vulnerability of ground water to contamination.
The focus of the NPS is on the quality of drinking water in wells rather than on the quality of ground
water, surface water, or drinking water at the tap. The Survey's analytic results are statistically representative
of over 10.5 million rural domestic wells and wells at 38,300 community water systems (CWS) that use ground
water across the country. The study has yielded valuable information on the frequency and concentration of
pesticides and nitrate present in rural domestic wells and community drinking water wells on a nationwide
basis. The NPS was not designed to provide a statistically valid assessment of pesticide or nitrate concentra-
tions in wells at the local, county, or State levels.
1 Pesticides are chemical substances used to destroy, control, or repel undesirable plants, insects, fungi, nematodes, or
animals (herbicides, insecticides, fungicides, nematocides, and rodenticides, respectively). Pesticide degradates are breakdown
products of pesticide active ingredients resulting from biological processes (i.e., metabolism) or physical or chemical processes
(i.e., hydrolysis, photolysis, photooridation). The term "pesticides" throughout this report stands for both pesticides and
pesticide degradates, unless otherwise indicated. Nitrate (NO3") is the major nitrogen source assimilated by higher plants.
It is found in some nitrogen fertilizers and/or as a result of nitrification of ammonium fertilizers. Nitrate is produced not only
in soil but also in septic systems, animal feed Jots, industrial waste waters, and sanitary landfills. Nitrite (NO2") can be formed
either as the intermediate product of ammonium (NH/) during the nitrification process or as the intermediate product of
nitrate during the denitrification process. Nitrate and nitrite were not separately distinguished in the Survey laboratory
methods and testing. They were measured as nitrogen (N). In this report, the term "nitrate" stands for both nitrate and
nitrite, unless otherwise indicated.
2 States and private organizations recently have conducted several other surveys of well water, but they have generally
concentrated on either more limited geographic areas or a more limited set of analytes than the NPS. Two of the most
notable were conducted by the State of Iowa and the Monsanto Agricultural Company. The Iowa State-Wide Rural Well-
Water Survey, a one-time sampling survey conducted between April 1988 and June 1989, sampled 686 sites for total coliform
bacteria; nitrate, ammonia, and organic-nitrogen; other major inorganic ions (e.g., chloride, sulfate, calcium); 27 commonly-
used pesticides; and selected pesticide metabolites. The National Alachlor Well Water Survey (NAWWS) conducted by the
Monsanto Agricultural Company between June 1988 and May 1989 collected water samples from 1,430 wells and tested for
alachlor, four other pesticides, and nitrate.
National Pesticide Survey: Phase I Report
-------
Chapter One: Introduction
With information from the Survey, EPA will be able to better design its regulatory programs to target
pesticides of concern and to refine future regulatory initiatives. The Federal Insecticide, Fungicide, and
Rodenticide Act (FIFRA) gives EPA's Office of Pesticide Programs (OPP) authority to regulate the marketing
and use of pesticides. Pesticides that are shown to pose potential hazards by their ability to leach into ground
water could be subject to further registrant monitoring requirements and a range of regulatory actions,
including changes in label directions, use restrictions, placement in the Special Review process, or suspension
or cancellation of the pesticide's registration. EPA's Office of Drinking Water (ODW) also may use
information from the Survey to implement requirements of the Safe Drinking Water Act (SDWA). New
maximum contaminant levels and monitoring requirements may be proposed for pesticides found in drinking
water supplies.
1.2 Overview
Ground water currently supplies drinking water to about 40 percent of the U.S. population,
particularly in rural areas. In 1984, directing growing attention to the question of how to protect our ground
water and citizens who drink it from the impacts of pesticides and nitrate, EPA's Office of Drinking Water
and Office of Pesticide Programs began the NPS to study the extent and causes of the problem. Exhibit 1-1
is a conceptual diagram of the planning, development, implementation, and analytic phases of the NPS. It
charts the development of the Survey, from design and development of analytic methods to Survey
implementation, data analysis, and reporting.
Planning of the NPS was carried out by several internal workgroups convened by ODW and OPP, as
well as by expert consultants hired by EPA. Throughout the design and implementation of the Survey, EPA
also consulted regularly with a workgroup of State and local representatives and with other Federal agencies.
EPA began by organizing an analyte selection workgroup, which chose a preliminary list of the
pesticides and pesticide degradates to be analyzed in the Survey based primarily on their expected leaching
potential, regulatory status under the SDWA, volume of sales nationally, prior occurrence in ground water,
and potential health impacts.
An analytical methods workgroup conducted literature searches and consulted with pesticide analysis
experts to identify appropriate analytical methods for the Survey. An important goal was to expand the
capabilities of existing EPA analytical methods to identify additional analytes. Six new analytic methods
ultimately were developed for the Survey, and an existing EPA method and a modification of an EPA method
were used. In the course of developing analytical techniques, certain potential analytes were dropped and
others added based on the capabilities of the analytic methods. The eight methods eventually developed for
the NPS measured 127 analytes, including 101 pesticides, 25 pesticide degradates, and nitrate.
Following selection of the NPS analytes, an EPA health advisory workgroup developed 99 Health
Advisories for pesticides and nitrate. The Health Advisories include information on pbysicochemical
properties, uses, chemical fate, health effects, treatment, and existing criteria and guidelines. These Health
Advisories are technical, scientific guidance documents developed to assist in evaluating the results of well
sampling and in determining whether the levels found warrant further action. As part of EPA's effort to
improve risk communication, the Agency also prepared and distributed non-technical health advisory
summaries. These summaries explain the health effects of exposure to pesticides and nitrate in non-technical
language. Health advisory summaries were distributed immediately to the well owners and operators of
sampled wells where analytes were found in excess of health-based levels and to residents who might consume
water from such wells.
National Pesticide Survey: Phase I Report
-------
Chapter One: Introduction
Exhibit 1-1
Overview of NFS
\
°
i
1
Analysis and
Confirmation o
Samples
i
Interviews with
Well Owners.
sidenu and Farmen
at
Well Sampling
«
«
M
s
5
2
*M
Analysis and
Confirmation o
Samples
r
Physkal/
Environmental
Dau
"f
u
Q o E u » - o ts) 9 M > « >
Interviews with
Well Owners/
Operalon
Well Sampling
U^CO t/5 3 >-
National Pesticide Survey: Phase I Reoort
-------
Chapter One: Introduction
For the statistical design of the Survey, EPA used contractor support and convened two supporting
workgroups. The data collection program followed EPA's formal Data Quality Objectives (DQOs) process
for identifying data needs, uses, and quality goals, while remaining within resource constraints. EPA's
decisions on the details of survey design were based on the work of a hydrogeologic workgroup, which selected
a method of measuring hydrogeologic vulnerability, and a pesticide-use workgroup, which developed indicators
for determining levels of pesticide use in counties across the country.
In March 1987, EPA launched a pilot study to field test the major components of the Survey and to
provide an opportunity to make any necessary modifications to the Survey design and implementation strategy
prior to initiating the full-scale national Survey. The pilot study was conducted in California, Minnesota, and
Mississippi. From March through August 1987, EPA collected drinking water samples from 48 community
water systems and private wells as part of the pilot study. In addition, EPA-trained State officials and
contractor staff interviewed CWS operators and householders to collect information about well construction
and usage and gather data about well site observations and the area in the vicinity of each well.
In September, 1987, a ten-member subpanel of EPA's FTFRA Scientific Advisory Panel (SAP)
reviewed the overall survey design and the findings of the pilot study. The subpanel provided advice on a
broad range of issues involved in implementing the full Survey. The subpanel review and EPA's own
evaluation of the pilot study enabled EPA to make appropriate revisions and adopt more efficient and cost-
effective alternatives for carrying out the Survey.
The full-scale National Pesticide Survey began in April 1988. Over the next two years, until February
1990, EPA Regional, State, and contractor sampling teams collected water samples and using Survey
questionnaires gathered extensive data about well construction, pesticide use, and other factors. In total, 783
rural domestic wells and 566 community water system wells were sampled. Five contract laboratories and three
EPA laboratories carried out analyses of the 17 or more water samples collected for each of the wells
investigated in the Survey. Chemical analysis of the water samples was completed by May 1990. Throughout
the course of the Survey, EPA followed rigorous quality assurance/quality control procedures in survey design,
data collection, sample collection, chemical analysis, recordkeeping, and reporting to ensure the high quality
and representativeness of the data.
Summary statistics of the NPS data are included in this Phase I Report. EPA is currently conducting
an extensive assessment and further statistical analysis of NPS data, which will be released with the NPS
Phase II report.
1.3 How to Use This Report
This report has been prepared to provide an initial description of the design and implementation of
the EPA National Pesticide Survey, and an explanation of the Survey's initial findings and results. Specifically,
this report is organized as follows:
Chapter Two: 'Survey Design" describes the process used to select representative subsets
of community water system wells and rural domestic wells for sampling. Chapter Two
describes the overall approach used to identify the study population and subpopulations and
to specify the precision of sample estimates, and to respond to the recommendations of the
Scientific Advisory Panel subpanel. Chapter Two also describes key design factors such as
stratification, sample sizes and variances, and temporal variability.
Chapter Three: 'Implementation of the Survey" provides an overview of the key elements
of water sampling and data collection activities during the two-year implementation phase
of the Survey. It describes the training program for sampling teams and other NPS field staff,
scheduling requirements, water sampling methods and procedures, and the administration of
National Pesticide Survey: Phase I Report
-------
Chapter One: Introduction
questionnaires developed for the Survey. Chapter 3 also provides a description of the
extensive network of operations management, data management, and communications systems
used to plan, organize, perform, track, and improve NFS implementation activities.
Chapter Four: 'Analytical Chemistry* describes how EPA chose the pesticides and
pesticide degradates to be tested for in NFS water samples. It identifies and describes the
development of the testing methods used by NFS and the activities of the analytical
laboratories that conducted the testing.
Chapter Five: 'Quality Assurance/Quality Control" discusses the objectives, organization,
and key components of the NFS quality assurance/quality control (QA/QC) program as they
relate to each of the major components of the Survey. In particular, it describes the
importance of the QA/QC program in monitoring and controlling information-gathering
processes so that data would meet Survey design criteria for accuracy, precision,
representativeness, comparability, and completeness.
Chapter Six: "Findings and Results" presents summary statistical results of the chemical
analyses of the water samples taken from the drinking water wells and summaries of
questionnaire data. It discusses key descriptive characteristics and preliminary analysis
patterns of pesticides, pesticide degradates, and nitrate; and discusses additional analyses (e.g.,
relational analyses) to be included in the NFS Phase II report.
Chapter Seven: "Survey Products" includes a brief description of important NFS products
such as Health Advisory Summaries and information on their availability.
There are five appendices to this report:
Appendix A: "Background" provides an in depth review of Survey planning, development,
the pilot study, and Science Advisory Panel reviews.
Appendix B: "Statistical Design" is a detailed discussion of the topics outlined in
Chapter Two.
Appendix C: "Survey Implementation" presents detailed discussions of the topics
summarized in Chapter Three.
Appendix D: "Questionnaire Data and Questionnaires" provides summary statistics of
the data collected using Survey questionnaires and copies of NFS questionnaires administered
during the survey.
Appendix E: "Survey Analytes and Analytical Chemistry Tables" provides summary
information about each analyte, Minimum Quantification Limits (MQLs), and quality control
results for the NFS analytes and analytical laboratories.
1.4 Additional NFS Information
This Phase I Report is part of a series of NFS reports and outreach materials. It will be followed by
a Phase II report that will provide the results of additional analysis of the NFS data. In addition, the Phase
II report will contain detailed information on a number of topics, including analytic methods, quality
assurance, database documentation, and health advisory summaries.
-------
Chapter One: Introduction
EPA has prepared a number of fact sheets on selected topics concerning the National Pesticide Survey.
These fact sheets are available through EPA's Public Information Center (401 M Street SW, Washington DC
20460, 202-382-2080):
Project Summary Summary of Results Survey Design
Analytical Methods Survey Analytes Glossary
Quality Assurance/ How EPA Will Use Fact Sheet for each
Quality Control The NFS Results detected analyte
Additional information on the Survey and on pesticides in general can be obtained from the following
sources:
U.S. EPA Safe Drinking Water Hotline Information on regulation of
1-800-426-4791 (In Washington, DC - 382-5533) pesticides in drinking
Monday-Friday, 8:30 am to 4:30 pm Eastern Time water
National Pesticide Telecommunications Network Information on health
1-800-858-7378 effects and safe
24 hours a day handling of pesticides
U.S. EPA Office of Pesticide Programs (OPP) Docket Background documents
Public Information Branch (H 7506C) for Survey (available
401 M Street, SW for review)
Washington, DC 20460
(703) 557-2805
National Technical Information Service (NTIS) Copies of the
5285 Port Royal Road NPS Phase I Report
Springfield, VA 22161 (available 1991) and
(703) 487-4650 NPS Phase II Report
(when available)
Information on access to data tapes containing data generated by the Survey will be available at the
conclusion of the Survey.
National Pesticide Survey: Phase I Report
-------
Chapter Two: Survey Design
Introduction
This chapter explains the key elements of the NFS survey design and provides an overview of how the
wells were selected for the two categories of wells included in the Survey - community water system (CWS)
wells and rural domestic wells. Section 2.1 describes the key elements of the Survey design. Section 2.2
defines community water systems and describes the methods used to obtain a sample of the population of wells
from those systems. Section 2.3 defines rural domestic wells and describes the methods used to obtain a
sample of the population of those wells. Details of survey design, including the statistical basis of the sample
sizes and stratification, are described in Appendices A and B.
2.1 Key Elements of Survey Design
The National Pesticide Survey was designed to collect samples of drinking water from wells across the
country. The NFS also collected information about well construction, physical surroundings, and land use for
investigation of possible relationships between the analytical results, ground-water vulnerability, and pesticide
use.
EPA and its consultants designed the Survey to yield results that are statistically representative of the
nation's domestic and community water system drinking water wells. Statistical survey methodology was used
to select representative subsets of these wells for sampling. Water samples were then obtained from those
wells. From these sampling results, NFS statisticians can produce national estimates, relating the information
obtained about the wells from which water samples were obtained to the entire population of drinking water
wells.
The key elements of the NPS survey design included the stratification used in the selection of
representative samples of wells, the method used to determine the numbers of wells from which water samples
should be obtained, evaluation of the survey design by an outside panel of scientific experts, and the
development of the NPS questionnaires.
2.1.1 Design Goals
EPA's Office of Pesticide Programs and Office of Drinking Water established the information goals
for the Survey soon after planning began in 1984, and those overall goals in turn formed the basis of the
survey design. The two major design goals of the Survey are:
To determine the frequency and concentration of the presence of pesticides and nitrate in
drinking water wells nationwide; and
To examine the relationships of the presence of pesticides and nitrate in drinking water wells
to patterns of pesticide use and ground-water vulnerability.
Full-scale planning of the NPS began in 1984, and the background and early stages of that process
are described in Appendix A. This chapter provides a conceptual description of the NPS survey design, rather
than an historical account. The basic conceptual steps in the design of the Survey were:
Defining the survey population to be studied;
Defining the criteria for stratification;
National Pesticide Survev: Phase I Renort
-------
8 Chapter Two: Survey Design
Identifying the subgroups of the population (the domains of interest) to form the basis for
the design; and
Specifying the level of precision to be associated with the Survey estimates.
2.1.2 Survey Population
Before determining the size of the sample of the well population that should be selected, the Survey
designers first needed to estimate the size of the well population as a whole.3 Rural domestic wells were
defined as drinking water wells that supply occupied housing units located in rural areas of the United States,
except for wells located on government reservations. The Census Bureau defines rural areas as households
outside of incorporated or unincorporated places with 2,500 or more population and outside of urban fringe
areas. The number of private rural domestic wells was estimated based on 1980 Census Bureau data as 13
million. The estimate of the number of community well systems was based on the Federal Reporting Data
System (FRDS), maintained by the EPA Office of Drinking Water which contains information on public water
supply systems. FRDS listed about 51,000 systems in the United States for the period from July 1984 through
June 1985.
2.1.3 Stratification and How Stratification Criteria were Determined
The NFS used a process known as stratification in selecting the subsets of CWS and rural domestic
wells for sampling. In general, stratified designs are used to improve the accuracy of the estimates by
controlling the distribution of the sample of the population from which data will be collected. The selection
of the samples of the population of CWS and rural domestic wells took place in stages, with additional
selection criteria at each stage. The resulting final samples of the population of CWS and rural domestic wells
were controlled for two specific criteria: pesticide use and ground-water vulnerability in the county or sub-
county area in which the wells to be sampled were located.
The first step in well selection (first-stage stratification) was common to both CWS and rural domestic
wells. Each of the 3,137 counties or county equivalents in the U.S. (which were determined to be
conveniently-sized units for constructing the first-stage sampling frame) was categorized by pesticide use and
ground-water vulnerability. The first-stage sampling frame was stratified in this manner to ensure that samples
were taken from wells located in areas with different levels of pesticide use and ground-water vulnerability.
In 1984 when the survey design was begun, only limited data were available on pesticide use.
Originally, all types of pesticide use -- agricultural, home and garden, industrial, rights-of-way, commercial, and
government - were considered as stratification criteria. Because agricultural use of pesticides was significantly
higher than other uses, agricultural pesticides had been found in ground water, and data on agricultural
pesticide use were better than for other categories of use, agricultural pesticide use was chosen as one of the
stratification criteria.
EPA developed county-level pesticide use estimates for each county, based on 1982 Census of
Agriculture information for crop acreages and on private marketing data (provided by Doane Marketing
Research, Inc.) for pesticides. Total acreage using 63 specific chemicals was calculated for 28 crops. The
acreage to which a specific chemical was applied in a county was estimated; acreages for all chemicals were
summed, and the total expressed as a percentage of the total area of the county. Agricultural pesticide use
was described as high, medium, low, or uncommon, based on the following criteria:
3 These numbers represent the Agency's estimate of the population of wells at the beginning of the Survey. Estimates
of the well population based on the Survey results are smaller: 10.5 million rural domestic wells and about 94,600 wells at
38,300 community well systems.
National Pesticide Survey: Phase I Report
-------
Chapter Two: Survey Design 9
High Pesticide Use. Agricultural pesticides applied to the equivalent of 75 percent or more
of the total county area;
Moderate Pesticide Use. Agricultural pesticides applied to the equivalent of between 30
percent and 74 percent of the total county area;
Low Pesticide Use. Agricultural pesticides applied to the equivalent of between 5 percent
and 29 percent of the total county area; and
Uncommon Pesticide Use. Agricultural pesticides applied to the equivalent of less than
5 percent of the total county area.
In order to examine any relationships between levels of pesticides and nitrate and hydrogeologic
characteristics conducive to ground-water contamination, NFS developed a system for identifying a variety of
hydrogeologic conditions and classifying them into areas of relatively greater or lesser ground-water
vulnerability on a county-wide basis for the purpose of sample stratification. One of the goals of the Survey
was to test this system and determine if it functioned as a useful sorting technique.
The Agricultural DRASTIC classification system, one variation of an indexed scoring system that had
already been developed by the National Water Well Association, was modified for use in the Survey by
Research Triangle Institute. DRASTIC is an acronym for a ranking system for evaluating the vulnerability
of aquifers to pollution beneath a particular land area. It is derived from the seven hydrologic factors
considered in the ranking system: depth of water, recharge (net), aquifer media, soil media, topography
(slope), impact of vadose zone, and conductivity (hydraulic) of the aquifer. Another proposed classification
system, the Great Soil Groups, was not chosen because its component factors apply only to the top 30 inches
of soil. Site-specific classification systems were precluded for a large-scale national survey such as NFS because
of the large amounts of data such systems require. Design of an entirely new system for use by the Survey was
precluded by the cost and time that would have been needed to develop it.
To represent relative ground-water vulnerability, a county-level DRASTIC hazard index was developed
for each of the 3,137 U.S. counties, based on a number of information sources and the experience and
judgment of hydrogeologists familiar with particular regions. The sources of information included:
U.S. Geologic Survey (USGS) 7.5 minute quadrangle maps (or 15 minute quadrangle maps
where 7.5 minute maps were not available);
USGS Geologic Maps and USGS Groundwater Maps (where available);
U.S. Department of Agriculture, Soil Conservation Service Soil Survey Maps; and
Materials from State geologic surveys, Departments of Water and Natural Resources,
Departments of Environmental Protection, regional planning authorities, county and regional
water supply agencies, private water supply firms, hydrogeologic and engineering firms,
professional associations, and local colleges and universities.
These DRASTIC scores were used to divide counties into areas of relatively high, moderate, and low
ground-water vulnerability, as follows:
High. County-level DRASTIC score that identifies the most highly vulnerable 25 percent of
households with wells within each of the pesticide use strata.
Moderate. County-level DRASTIC score that identifies the most highly vulnerable 50
percent of households with wells within each of the pesticide use strata.
National Pesticide Survey: Phase I Report
-------
10 Chapter Two: Survey Design
Low. Remainder of the area frame (i.e., counties with the lowest DRASTIC scores
identifying 25 percent of households with wells within each of the pesticide use strata).
Thus, the DRASTIC scores, together with the pesticide use criteria, result in a matrix of twelve strata. The
strata, and their associated DRASTIC scores, are described in Exhibit 2-1.
Exhibit 2-1
Strata for NPS First-Stage Survey Design
Agricultural
Pesticide Use
High
High
High
Moderate
Moderate
Moderate
Low
Low
Low
Uncommon
Uncommon
Uncommon
Ground-Water
Vulnerability
High
Moderate
Low
High
Moderate
Low
High
Moderate
Low
High
Moderate
Low
Defining
DRASTIC Scores
148 and over
11610147
1 1 5 and under
1 63 and over
113 to 162
112 and under
159 and over
132 to 158
131 and under
1 52 and over
121 to 151
120 and under
2.1.4 Definition of Domains of Interest and Design Assumptions
The NPS survey design identified particular subsets of the entire population of drinking water wells
that were of special interest to the Survey. These subsets, known as domains of interest, constituted categories
of wells about which it was particularly important to obtain information. The seven domains of interest were:
1. Community water system wells nationally;
2. Community water system wells in counties having the highest average ground-water
vulnerability;
3. Rural domestic wells nationally;
4. Rural domestic wells in counties with the highest average pesticide use;
5. Rural domestic wells in counties with the highest average ground-water vulnerability;
6. Rural domestic wells in "cropped and vulnerable" county subregions; and
7. Rural domestic wells in counties with the highest average pesticide use and ground-water
vulnerability.
National Pesticide Survey: Phase I Report
-------
Chapter Two: Survey Design 11
2.1.5 Precision Requirements and Domain Sizes
In order to determine the number of wells to be sampled from each of the domains of interest (the
domain sizes) Survey designers had to specify the level of precision of the Survey that they wanted to achieve.
Different assumptions about the "true" proportion of all wells that contain pesticides were evaluated.
Alternative domain sizes and precision requirements were evaluated from the standpoint of EPA
policy, fiscal, and program considerations. These alternatives, with cost estimates for collecting data, were then
evaluated mathematically to determine the sizes of the sample of the population at each stage within each of
the design strata that met EPA's cost ceilings.
Survey designers ultimately decided to structure the Survey so that it provided a 90 percent probability
of detecting the presence of pesticides in the sample of CWS wells if 0.5 percent of all community water
systems in the country contained pesticides. This is fairly conservative; the probability of detection would of
course be much better if more than 0.5 percent of all CWS in fact contain pesticides.
The domestic well statistical design was intended to yield a range of probabilities of detecting the
presence of pesticides in different domains. The strategy was to employ a sampling design to obtain the
highest power (power being the probability of detecting contamination when contamination actually exists)
in the vulnerable/high use counties and in the cropped/vulnerable subcounty areas of those counties, and to
accept a slightly lower power in the national level estimates. Areas of particular concern were oversampled
in order to yield better, more precise estimates. Thus, the likelihood of the Survey detecting the presence of
pesticides in domestic wells was 63 percent at the national level (assuming a 1 percent rate nationally), but
the Survey had a 97 percent chance of detecting pesticides in cropped and vulnerable areas (assuming a 0.25
percent actual rate in these areas). The precision requirements for the domains of interest eventually chosen
are provided in Exhibit 2-2.
Exhibit 2-2
NPS Domain Sizes and Confidence Levels
Domain
CWS wells nationally
CWS wells in counties with highest average ground-
water vulnerability
Rural domestic wells nationally
Rural domestic wells in counties with highest average
pesticide use
Rural domestic wells in counties with highest average
ground-water vulnerability
Rural domestic wells in cropped and vulnerable parts
of counties
Rural domestic wells in counties with highest average
pesticide use and ground-water vulnerability
Percentage of
the Specified Domain*
0.5%
0.1%
1.0%
0.14%
0.25%
0.25%
0.3%
Probability^
Detection**
90%
60%
63%
75%
75%
97%
47%
* Wells in the domain assumed to contain pesticides as a percentage of all wells in the
population.
** Probability of at least one detection, given the relative domain size.
National Pesticide Survey: Phase I Report
-------
12 Chapter Two: Survey Design
2.1.6 Pilot Study
EPA conducted a pilot study from March to August, 1987, to test the major components of the Survey
and to provide an opportunity for any necessary revisions or modifications before the full Survey. The pilot
study was conducted in three States: California, Minnesota, and Mississippi, to provide geographic diversity
and because of their high level of interest and cooperation.
Water sampling was conducted at 48 wells in the pilot study, including both rural domestic wells and
community water system wells. In addition, interviews were conducted with householders regarding the use
and construction of their wells, and data were collected on conditions within the immediate vicinity of the well
and within a half mDe radius of the well. CWS wells in each pilot State were sampled by State health
department officials. Questionnaires on the construction and characteristics of the wells also were
administered to CWS operators.
The pilot study led to improved procedures and techniques in several areas. Refinements were made
in the interview questionnaires, training manuals, and sampling communications procedures, in anticipation
of the full Survey. In addition, the pilot study raised specific issues related to the Survey design, which were
analyzed and presented to a subpanel of the FTFRA Science Advisory Panel in September 1987.
2.1.7 Design Review
In 1985, at the initial design stage of the Survey, and again in 1987, prior to implementation of the
full Survey, the Survey design was submitted for peer review to a subpanel of EPA's FIFRA Scientific Advisory
Panel. Public meetings of the subpanel were held to discuss the subpanel's conclusions. In October, 1987,
the SAP subpanel commented on four topics:
1. The type and complexity of data needed to meet Survey objectives, particularly the county
DRASTIC assessments and the cropping and pesticide use data;
2. The proposed method for selecting CWS wells for water sampling;
3. The issue of temporal variation (seasonal patterns of pesticide application, precipitation, and
other similar phenomena), as a factor influencing when and in what concentration pesticides
and nitrate impact drinking water wells, and how to address temporal variation in water
sampling; and
4. Issues involving the stability of the analytes in water samples over time.
Recommendations of the SAP subpanel were received and reviewed by EPA. The changes made to the survey
design to incorporate the subpanel's suggestions are described briefly in this chapter. Additional details on
the SAP subpanel review are provided in Appendix A.
2.1.8 Survey Data Objectives and Questionnaire Development
EPA ultimately developed nine questionnaires for the NFS. Following the pilot study and SAP
subpanel review, additional questions collecting well-specific information, pesticide use data, and land use data
were added to the questionnaires. The key information sought and the need for this information is described
below for each questionnaire:
Team Leader Introduction and Well Observation Record. The Team Leader Introduction
was developed to assist in ensuring the proper information was gathered prior to the actual
sampling event. The sampling team leader used the introduction to introduce the sampling
team to the well owner/operator, confirm eligibility, and provide an overview of the Survey
National Pesticide Survey: Phase I Report
-------
Chapter Two: Survey Design 13
objectives. In cases where the CWS was a multiple well system, the Team Leader
Introduction was used to identify the specific well to be sampled through a random selection
process. For rural domestic wells, it was used to obtain written consent from the well
owner/resident to sample the well.
The Well Observation Record was completed by the Team Leader. It recorded the Team
Leader's observations about the well and the surrounding area, including topography, soils,
water treatment and storage systems, well housing, and additional observations deemed
important because of their potential influence on pesticide migration into the well sampled.
Well-specific notes, including a sketch of the well system indicating the sampling port or tap,
and a sketch of local land features around the well also were recorded in the Well
Observation Record.
CWS and Domestic Well Main Questionnaires. These questionnaires collected detailed
information regarding water use, water treatment, well construction, and non-farm and farm
pesticide use on the property where the well is located.
Domestic Well Second-Stage County Agricultural Extension Agent Questionnaire.
This questionnaire gathered cropping and pesticide use information for 84 of the domestic
well survey counties (such data had previously been collected for six counties as part of the
pilot study). Information collected in this questionnaire was used to identify areas ("cropped
and vulnerable") in the counties from which a larger proportion of wells would be selected
(over-sampling areas). Information was gathered on what pesticides were used within the
county to be used for statistical analyses in the Phase II report.
Local Area Expert Questionnaires. These questionnaires were administered to persons
familiar with local conditions and use of land and pesticides surrounding the well being
sampled. For CWS, this information was usually collected from the county agricultural
extension agent. For rural domestic wells, this questionnaire was administered to the county
agricultural extension agent who responded in the Second-Stage County Agent Questionnaire.
The county agent was recontacted after all the wells in a county were sampled to obtain
additional information about local conditions and use of land and pesticides within one-half
mile of the well. Information collected from this questionnaire can be compared to the
general agricultural practice responses in the second-stage questionnaire.
In addition to these questionnaires administered in the field, two screening questionnaires,
administered by telephone, were used to identify the wells to be sampled. The CWS Screening Questionnaire
was developed in part to respond to the SAP subpanel recommendations on selection of CWS wells.
Screening Questionnaires. The screening questionnaires were developed to identify a
national sample of community water systc and rural domestic wells that met specific NPS
eligibility criteria. The CWS Screening Q* ^.tonnaire was designed to confirm the eligibility
of 7,083 CWS, which were randomly selected from more than 51,000 CWS listed in the
Federal Reporting Data System (FRDS), and which used ground water as a source of
drinking water. A random subsample of 599 CWS wells was then identified from the eligible
CWS for inclusion in the community drinking water sampling survey. The DWS Screening
Questionnaire was designed to identify a sample of eligible households possessing a drinking
water well by county subregions and to obtain the well owner's verbal approval for sampling.
National Pesticide Survey Phase I Reoort
-------
14 Chapter Two: Survey Design
2.1.9 Scheduling and Temporal Variability
In response to the recommendation by the SAP subpanel that the Survey design address temporal
variability, EPA selected a method of allocating sampling dates to wells randomly across the data collection
period, separately for each stratum, to minimize the possibility of temporal bias. The capacity of the analytic
laboratories established the ceiling for the number of wells that could be sampled in a particular period.
Within that ceiling, EPA allocated counties selected for domestic well sampling to two-week time periods,
spread over an 18-month period, from September 1988 to February 1990. (Domestic wells selected from
counties in the three States included in the pilot study were sampled from April through August 1988.)
Sampling of community water systems was similarly spread out in two-week periods from August 1988 to
December 1989. Not all CWS wells in a county were sampled in the same time frame. An example of the
sampling pattern eventually adopted for CWS wells in one State is shown in Exhibit 2-3. State sampling teams
scheduled sampling and data collection for a date during one of the two weeks.
Exhibit 2-3
Example of Sampling Schedule for CWS in One State
Sampling Time Frame
Number of Week for Sampling
Community Wells in Confirmed
Well Systems Stratum System With CWS
CWS
CWS
CWS
CWS
CWS
CWS
CWS
CWS
CWS
CWS
CWS
CWS
CWS
CWS
CWS
* Note:
1
2
3*
4
5
6
7
8
9
10
11
12
3*
13
14
Two
4
11
12
8
7
4
11
12
11
7
4
10
12
11
8
wells from
2
3
40
2
2
3
1
2
12
20
1
2
40
2
1
CWS 3 sampled
10/17/88 -
12/12/88 -
01/09/89 -
01/09/89 -
02/20/89 -
03/06/89 -
03/20/89 -
04/17/89 -
05/01/89 -
05/01/89 -
05/15/89 -
06/12/89 -
08/07/89 -
09/04/89 -
12/04/89 -
in different
10/21/88
12/16/88
01/13/89
01/13/89
02/24/89
03/10/89
03/24/89
04/21/89
05/05/89
05/05/89
05/19/89
06/16/89
08/11/89
09/08/89
12/08/89
sampling
Potential
Sampling Week 1
10/17/88 -
12/12/88 -
01/09/89 -
01/09/89 -
02/20/89 -
06/06/89 -
03/20/89 -
04/17/89 -
05/01/89 -
05/01/89 -
05/15/89 -
06/12/89 -
08/07/89 -
09/04/89 -
12/04/89 -
periods.
10/21/88
12/16/88
01/13/89
01/13/89
02/24/89
03/10/89
03/24/89
04/21/89
05/05/89
05/05/89
05/19/89
06/16/89
08/11/89
09/08/89
12/08/89
Potential
Sampling Week 2
10/24/88
01/02/89
01/16/89
01/16/89
02/27/89
03/13/89
03/27/89
04/24/89
05/08/89
05/08/89
05/22/89
06/19/89
08/14/89
09/11/89
12/11/89
- 10/28/88
- 01/06/89
- 01/20/89
- 01/20/89
- 03/03/89
- 03/17/89
- 03/31/89
- 04/28/89
- 05/12/89
- 05/12/89
- 05/26/89
- 06/23/89
- 08/18/89
- 09/15/89
- 12/15/89
The following two sections provide additional details on the unique features of the Survey design for
CWS and rural domestic wells, respectively.
National Pesticide Survey: Phase I Report
-------
Chapter Two: Survey Design 15
2.2 Community Water Systems
Community water systems are defined under the Safe Drinking Water Act and its regulations as
systems providing piped drinking water with at least 15 connections and/or serving a population of at least 25
persons who are permanent residents of the service area. For the purposes of the NFS, a CWS was further
defined as a system with at least one working well used to obtain drinking water. All such systems are listed
on the Federal Reporting Data System (FRDS) maintained by EPA's Office of Drinking Water. Only
community systems listed on FRDS for the period July 1984 through June 1985 were eligible to be included
in the Survey. Systems that obtain water exclusively from surface water were excluded, because a community
system had to be listed on FRDS as an "active" system, with at least one well under the operational control
of the system.
A two-stage statistical design was adopted for community water systems, based on the pilot study
conducted in 1987 and the recommendations of the SAP subpanel. A summary of the community well
selection process is illustrated in Exhibit 2-4 on page 16.
2.2.1 First-Stage Stratification
As Section 2.1 described, the sample frame for the CWS first-stage stratification was the FRDS.
Records in FRDS indicated that about 51,000 CWS using ground water as a source of drinking water existed
in the country, and EPA extracted information on those systems from FRDS in December 1985. A sample
of 7,083 systems was then randomly selected, stratified by the pesticide use (high, moderate, low, or
uncommon) and ground-water vulnerability (high, moderate, or low) of the county in which each CWS was
located. To improve the reliability of estimates made for the subset (domain) of CWS wells in the high
vulnerability strata, the Survey was designed to collect higher percentages of water samples from these strata
(i.e., to "oversample"). The Survey was not designed to oversample CWS wells in areas of high agricultural
pesticide use, because CWS wells, which are frequently located in urban areas, were expected to be less likely
to be impacted by agricultural use. Distribution of the first stage CWS samples across strata is described in
Exhibit 2-5.
Exhibit 2-5
CWS First-Stage Stratification Results
Stratum
1
2
3
4
5
6
7
8
Q
10
11
12
Pesticide
Usage
High
High
High
Medium
Medium
Medium
Low
Low
Low
Uncommon
Uncommon
Uncommon
DRASTIC
Score
High
Medium
Low
High
Medium
Low
High
Medium
Low
High
Medium
Low
Universe
Size*
1,310
2,669
1,432
2,183
3,352
2,559
3,578
7,199
5,545
3,472
9,511
7.961
50,771
Sample Rate
.22
.10
.10
.25
.10
.10
.24
.10
.12
.22
.11
.13
Sample Size*
293
276
148
555
355
265
866
771
647
777
1,069
1.061
7,083
* Number of systems
National Pesticide Survey: Phase I Report
-------
16 Chapter Two: Survey Design
Exhibit 2-4
Community Well Selection Process
1st Stage
Stratify all U.S. Counties by:
County-level DRASTIC score
Pesticide usage
2nd Stage
Extract information on Community
Water System from FRDS
Select Community Water Systems, based on County-level Strata,
from Approximately 51,000 CWS in FRDS
Conduct Telephone Interviews for
CWS Screening Questionnaire
Screen Out Ineligible Community
Water Systems
Randomly Select Sample of
Eligible Community Water
Systems
Randomly Select CWS Well for
Sampling and Data Collection
National Pesticide Survey: Phase I Report
-------
Chapter Two: Survey Design 17
2.2.2 Second-Stage Stratification
The pilot study conducted in 1987 at 10 community water systems in three of the States selected for
the nationwide sample identified a number of issues affecting the sample selection. Although FRDS
represented the best and most complete list of community water systems nationally, the pilot study indicated
that certain data elements in FRDS were not always complete or consistent In California, one of the States
included in the pilot study, a large proportion of the sample was non-community rather than community water
systems. This problem was believed to be limited to California, and was resolved by drawing a compensating,
augmented national sample. A second, more widespread, problem was that the FRDS records were sometimes
a mixture of individual systems and parent companies operating several systems at different locations. The
mixture of individual systems and parent companies in FRDS suggested that the sampling frame contained
multiplicities (i.e., a single system could be associated with more than one member of the frame). Finally, the
pilot study developed information suggesting that the selected systems had more wells than were shown on
FRDS or anticipated in the survey design. The Survey had been planned to sample each well in each selected
system, based on an estimate of an average of 1.5 wells per system, and on that basis to estimate the number
of community water systems in the country that have at least one well containing pesticides. Instead, however,
the pilot study results indicated that approximately 5.75 wells would have to be sampled per system, with
significant additional data collection costs.
These problems were addressed in two ways. First, questions were added to the CWS Screening
Questionnaire to assist in determining the number of times each system in the sample might have occurred
in the group from which the sample was chosen. Questions were included asking the CWS to identify its
parent company, if any, and procedures were developed to determine if the parent company was also part of
the frame. The results of this screening allowed more accurate estimates of the number of CWS and CWS
wells in the country to be made. Survey results are representative of the wells used by an estimated 38,300
CWS. Secondly, in the second-stage stratification, community water systems were selected with probabilities
of selection proportional to size. The measurement of size was determined by the number of wells in a system.
Community water systems with very large numbers of wells were automatically included in the Survey, and
given the chance to be selected numerous times. (Ultimately, more than one well was expected to be sampled
at 23 systems.) Each time one of these large systems was selected, a new one of its wells was randomly
selected for the Survey (i.e., if a system was selected three times, three of its wells were selected).
Telephone interviews using the CWS Screening Questionnaire were held with the first-stage sample
of 7,083 CWS. The names and addresses of all of these systems were reviewed by State and EPA Regional
staff before any screening interviews were conducted. The purposes of the screening survey were to establish
initial contact with the system owner/operator, verify eligibility, obtain cooperation with the Survey, and verify
the number of wells at the CWS. The results of the screening produced a frame of 5,660 eligible systems from
which to select the wells from which water samples and other data would be collected.
Distribution of the CWS samples across strata following second-stage stratification and sample
selection is described in Exhibit 2-6. A sample of 599 eligible systems was selected with probabilities
proportional to their number of wells. During the course of the Survey problems were encountered that made
it impossible to include certain CWS selected earlier. These problems included:
Changes in CWS operations, so that at least one eligible well was no longer under the
operational control of the system;
An incorrect number of operating wells reported by the CWS owner/operator during the
screening survey; or
Inability to sample a well prior to the point of treatment of the water.
National Pesticide Survey: Phase I Report
-------
18 Chapter Two: Survey Design
Therefore, an additional 68 CWS were randomly selected in August 1989 from the frame of eligible systems
to ensure that the Survey met the minimum sample size criteria, while minimizing the chance of selecting a
member of the original sample.
Exhibit 2-6
CWS Second-Stage Stratification Results
Stratum
1
2
3
4
5
6
7
8
9
10
11
12
Pesticide Usage
High
High
High
Medium
Medium
Medium
Low
Low
Low
Uncommon
Uncommon
Uncommon
DRASTIC Score
High
Medium
Low
High
Medium
Low
High
Medium
Low
High
Medium
Low
Design Sample Size*
27
26
14
45
32
25
73
68
53
71
90
75
599
* Number of systems selected, including allowances for inability to sample prior to treatment.
Minimum desired sample size after dropout was 564.
2.3 Rural Domestic Wells
Rural domestic wells were defined as private water wells supplying occupied housing units located in
rural areas in the U.S., except government reservations. No comprehensive listing of domestic water wells,
similar to FRDS for CWS, existed to serve as a sampling frame. EPA therefore developed a sampling method
to randomly select rural domestic wells that emphasized geographic areas that have high pesticide usage and
ground-water vulnerability. A summary of the domestic well selection process is illustrated in Exhibit 2-7.
2.3.1 First-Stage Stratification
As Section 2.1 described, in the first stage EPA stratified all of the counties and county equivalents
in the U.S. by pesticide use and ground-water vulnerability. Ninety counties then were selected as the primary
sampling units in the first stage. The selection of 90 Survey counties was based on the stratification rules and
Census Bureau estimates of the number of households with wells per county.
2.3.2 Second-Stage Stratification
In the second-stage stratification, DRASTIC vulnerability scores were assigned to subregions of each
of the 90 Survey counties, to divide them into regions with common levels of ground-water vulnerability. The
output of the second-stage DRASTIC scoring process was a map for each of the 90 counties, delineating the
hydrogeologic settings and associated vulnerability scores, accompanied by a list of reference materials for each
county. DRASTIC scoring for second-stage stratification was done using USGS 7.5-minute quadrangle maps
and geologic maps to identify boundaries between cropping areas. The boundaries were drawn in by hand on
USGS maps and quantified using a digitizing table and computer system with graphics software.
National Pesticide Survey: Phase I Report
-------
Chapter Two: Survey Design 19
Exhibit 2-7
Domestic Well Selection Process
1st Stage
Stratify all U.S. Counties by:
County-level DRASTIC score
Pesticide usage
Select 90 Domestic Well Counties
2nd Stage
Score Subcounty Areas
Using DRASTIC
Determine Cropped/
Non-cropped Status of
Subcounty Areas
I
Stratify SubCounty Areas:
Stratum 1: Stratum 2:
Cropped/Vulnerable Areas Non-cropped/Non-vulnerable Areas
1
Map Rural SubCounty Areas
3rd Stage
I
Select Telephone Number Clusters
in Each Rural Area
Screen Telephone Numbers to
Identify Active Non-Business
Numbers
Select Rural Households With
Wells for Sampling and Data
Collection
National Pesticide Survey: Phase I Report
-------
20 Chapter Two: Survey Design
Personal interviews were held with county agricultural extension agents to determine cropping
intensity, which was used as a surrogate measure for pesticide use intensity. The County Agent Questionnaire
also provided information for specifying the amount of pesticide use in each quadrangle relative to the rest
of the county (i.e., higher, lower, or about the same). Information from the questionnaire was used to classify
quadrants (USGS 7.5-minute quadrangle maps) of each of the 90 counties into one of three agriculture
cropping categories: (1) more than 50 percent of the total area of the county cropped; (2) 25 to 50 percent
of the area cropped; and (3) 25 percent or less of the area cropped.
EPA next partitioned the total area of each county into two strata, by combining cropping category
maps with DRASTIC vulnerability maps. One stratum contained the more heavily cropped and more
vulnerable part of the county; the remainder of the county was designated as the second stratum. The cropped
and vulnerable definition is illustrated in Exhibit 2-8, with the cropped and vulnerable stratum shaded.
Composite maps showing the cropped and vulnerable areas of each county were prepared using computer-
assisted mapping techniques. These maps were used to define areas from which a larger proportion of wells
would be selected. Exhibit 2-9 provides an example of a second-stage "cropped and vulnerable" map.
Exhibit 2-8
Cropped and Vulnerable Stratum for Subcounty Regions
D
r
a
s
t
i
c
S
c
0
r
i
n
9
Cropping Intensity by Land Area
High
>50%
Medium
25-50%
Low
<25%
High
>180
Medium
> 120
Low
< 120
National Pesticide Survey: Phase I Report
-------
Chapter Two: Survey Design 21
Exhibit 2-9
Example of Second-Stage "Cropped and Vulnerable" Map
HIWi
uwl
SCAII
-------
22 Chapter Two: Survey Design
2.3.3 Third-Stage Sample Selection
In addition to defining the cropped and vulnerable parts of each county, EPA also determined the
boundaries of all urban parts of each of the 90 counties. Two types of urban areas were defined from the 1980
census. Any incorporated community of 2,500 or more people was considered urban. The U.S. Bureau of the
Census also defines "urbanized areas" as the densely populated region stretching out from urban areas. Those
parts of a county that were neither designated as "urban areas" or "urbanized areas" were designated as rural
and contain all wells eligible for inclusion in the NFS.
Once these "urbanTrural" boundaries were determined, recognizable landmarks were identified that
best matched these boundaries. This was accomplished by combining information from census maps, county
maps of all kinds, zip code atlases, and post offices, and then overlaying this information on the cropped and
vulnerable maps.
In the third stage, NPS contractor staff conducted telephone screening interviews with households in
the Survey counties. The households were contacted through a statistical selection process known as Random
Digit Dialing (RDD) in which clusters of telephone numbers are defined and calls are made randomly to
numbers within the cluster. The telephone screening interview determined eligibility by inquiring if the
potential respondent lived in a rural area and had a private drinking water well, and was willing to participate.
The RDD screener also asked a series of geographic location questions to determine in which second-stage
stratum the respondent resided. In order to oversample areas defined as "cropped and vulnerable," only one-
third of all eligible well owners in the second stratum (i.e., not "cropped and vulnerable") were selected. The
results of the sample selection are described in Exhibit 2-10.
Finally, during the course of the Survey in August 1989, the sample size of the remaining counties in
each stratum was increased by one well to account for cancellations and refusals to participate and the inability
to sample wells prior to the point of treatment of the water.
Exhibit 2-10
Domestic Well Sample Selection Results
Stratum
1
2
3
4
5
6
7
8
9
10
11
12
Pesticide Usage
High
High
High
Medium
Medium
Medium
Low
Low
Low
Uncommon
Uncommon
Uncommon
DRASTIC Score
High
Medium
Low
High
Medium
Low
High
Medium
Low
High
Medium
Low
Design Sample Size*
91
93
48
50
72
31
77
112
53
67
88
45
827
* Number of wells selected, including allowances for cancellations, refusals to participate, and
inability to sample prior to treatment. Minimum desired sample size after dropout was 734.
National Pesticide Survey: Phase I Report
-------
Chapter Three: Implementation of the Survey
Introduction
From April 1988 through February 1990, NPS staff collected water samples and data from 566
community water systems and 783 rural domestic wells with the cooperation of Federal, State, and local
officials and private citizens. This chapter describes the roles of key project participants who assisted in
implementing the Survey and provides an overview of Survey implementation activities. It describes the NPS
staff training program, scheduling requirements, water sampling techniques and procedures, and the
administration of questionnaires developed for the Survey. Water sampling and data collection were supported
by an extensive network of operations management, data management, and communications systems used to
plan, organize, perform, and track NPS implementation activities. The roles of these systems in implementing
the NPS also are described in this chapter. Appendix C provides additional details and expanded discussions
of these topics.
3.1 Key Participants
The National Pesticide Survey was carried out by hundreds of EPA, State, and contractor personnel
with the cooperation of county officials, community water system owners/operators, and private citizens who
rely on domestic wells for their drinking water. Many of these participants, who came from every State, were
volunteers. The roles of key project participants are described below:
EPA Headquarters. EPA NPS staff in Washington, D.C. directed all Survey design and
implementation activities. As the focal point of all activities, the NPS Director was
responsible for overseeing all Survey activities and facilitating communication among the
various NPS participants.
EPA Laboratories. Two EPA laboratories (EPA/Environmental Chemistry Section and
EPA/Technical Support Division) participated in the survey design, were responsible for
managing analytic methods development, managed contracts for analytical work, reviewed
laboratory operations, and performed referee analyses using the appropriate analytical
methods. A third EPA laboratory (EPA/Risk Reduction Engineering) performed chemical
analyses.
EPA Regional Offices. EPA Regional staff conducted sampling in some States for the CWS
program and assisted in communicating information to other Regional personnel, State
participants, other State agencies, county officials, the public, and the media.
State Personnel. State participants conducted CWS sampling and data collection activities,
participated in the design of survey outreach activities, assisted EPA in distributing sampling
results to CWS owners/operators and domestic well owners/residents, and coordinated
participation by county officials.
County Personnel. County agriculture extension agents provided information about local
soils, hydrogeologic characteristics, and agricultural practices, such as pesticide usage and
cropping. County health and law enforcement officials assisted in facilitating Survey
participation by responding to public inquiries about the Survey and ensuring citizens that
the NPS was a bona-fide project of the U.S. EPA
Matlnnal Poctlrirlp Siiruev Phaco
-------
24 Chapter Three: Implementation of the Survey
CWS Owners/Operators. CWS owners/operators provided additional data about well
construction and non-farm and farm pesticide usage on the well property to NFS State
sampling teams who collected drinking water well samples and conducted interviews to collect
additional data.
Rural Domestic Well Owners, Residents, and Farmers. Private citizens throughout the
country voluntarily participated in the Survey by allowing EPA to sample their drinking water
wells and answering Survey questions.
Implementation and Laboratory Contractors. Contractors assisted EPA in designing and
implementing the Survey, performed chemical analyses of well water samples, and managed
day-to-day NPS activities. Research Triangle Institute assisted in the design of the Survey and
conducted the pilot survey. ICF Incorporated provided support to EPA by assisting in
designing the Survey, and implementing all aspects of the NPS (questionnaire design, staff
training, water sampling and data collection, communications, operations management,
database management, and data analysis and reporting). Westat Incorporated, ICFs
subcontractor, assisted EPA in designing the Survey, selecting the samples of rural domestic
wells and CWS wells, and developing the NPS questionnaires. Westat also designed and
conducted NPS staff training on questionnaires, conducted domestic well interviews with well
owners, residents and farmers, and compiled questionnaire data. Technology Applications
Incorporated was responsible for overall quality assurance/quality control coordination and
assessment during the Survey. Stretton Associates assisted with communications and
reporting. Five private contract laboratories (James Montgomery Laboratories, Alliance
Technologies/Clean Harbors Analytical Services, Radian Incorporated, Environmental Science
and Engineering Incorporated, and Battelle Columbus Division Laboratory) were responsible
for conducting chemical analyses of the drinking water samples.
3.2 Water Sampling and Data Collection
The two major components of Survey implementation were:
1. obtaining water samples from the selected wells to identify the amounts of possible pesticide
and nitrate concentrations; and
2. collecting key information through questionnaires and field documentation, to obtain data
on additional factors (i.e., well construction and non-farm and farm pesticide usage on the
well property, cropping activities, and industry locations in proximity to the well site)
potentially affecting pesticides and nitrate concentrations.
NPS water sampling and data collection protocols involved precise methods and procedures. Sample
preparation and collection techniques and handling procedures were conducted in accordance with strict
procedures and techniques and a specific schedule to ensure that samples were collected and analyzed within
quality control (QC) limits. To ensure adherence to these NPS-designed requirements and procedures, EPA
developed a training program to provide NPS sampling team personnel with the knowledge and skills necessary
to perform water sampling and data collection activities and to ensure that standard methods and procedures
were applied for all sampling activities.
3.2.1 NPS Training Program
The NPS training staff conducted 54 CWS training courses in 45 States to instruct EPA and State
personnel in CWS sampling and interviewing protocols. For rural domestic well sampling, 11 training courses
were held to instruct 30 regionally-based NPS contractor personnel on sampling protocols; three training
National Doctir-iHa Ciiruaw Dhaca I Dannrl
-------
Chapter Three: Implementation of the Survey 25
interviewing methods. NPS training workshops were held from May 1988 until April 1989; refresher training
courses were held until January 1990.
The NPS training program oriented CWS State sampling teams and rural domestic well contractor
sampling teams in all aspects of NPS sampling, questionnaire administration, and communications in advance
of conducting field sampling activities. The training program emphasized the following elements of NPS
sampling and data collection:
NPS scheduling and operations management procedures;
Unique NPS water sampling and data collection requirements, techniques, methods, and
procedures, including hands-on exercises to practice techniques;
t Quality assurance/quality control procedures; and
NPS communications protocols.
The NPS training program included three units:
1. Sampling Techniques and Procedures. Trainees learned proper sampling techniques and
procedures for all sampling collection activities, from preparatory requirements conducted
prior to field sampling to activities performed after well samples were coDected.
2. Questionnaire Administration. State personnel learned interviewing techniques and became
familiar with the questionnaire contents and proper use of the data collection forms. State
personnel learned how to interview community water system owners/operators and county
agricultural extension agents; contractor staff learned to use the rural domestic well question-
naires to interview private well owners/residents, fanners, and county agricultural extension
agents. State and contractor personnel both participated in role playing exercises to practice
the administration of questionnaires.
3. Communications Protocols. Survey personnel learned effective communications practices
to ensure that the confidentiality rights of private citizens who agreed to participate in the
Survey were honored, support the consistency and quality of the Survey results, uphold high
Survey participation rates, and develop a positive public perception of EPA and the NPS.
Survey personnel participated in role playing exercises to practice responding to citizens'
questions in non-technical language and to become conditioned to expecting unanticipated
circumstances.
3.2.2 Sampling and Data Collection Scheduling
During the period of sampling and data collection, from April 1988 through February 1990, State and
contractor NPS sampling teams collected drinking water samples from 566 community water system wells and
783 rural domestic wells. In addition, they gathered relevant data by interviewing county agricultural extension
agents, CWS owners or operators and rural domestic well owners, residents and farmers.
Water sampling of community water systems and domestic wells was scheduled in two-week periods
over the entire two-year implementation phase of the Survey. NPS scheduling accounted for the laboratories'
capacity to analyze the drinking water well samples and the availability of staff and resources to carry out the
sampling and data collection activities.
-------
26 Chapter Three: Implementation of the Survey
3.2.3 CWS Well Sampling
CWS samplers were assigned a two-week period from July 1988 through December 1989 in which to
conduct field sampling activities for each community water system well site. State CWS sampling teams
selected specific sampling dates that were convenient for CWS owners/operators within this two-week sampling
period. The NFS Communications Manager and NPS Sample Kit Preparation staff closely monitored the
actual sampling date to ensure that sample and data collection activities were conducted as scheduled. A
limited number of CWS scheduled sampling dates were revised to accommodate State sampling teams'
scheduling conflicts and logistical and budgetary constraints.
3.2.4 Rural Domestic Well Sampling
Rural domestic well sampling also was carried out in scheduled two-week intervals from April 1988
through February 1990. After a two-week period was randomly selected for sampling in a particular county,
the NPS contractor screened randomly selected households by telephone to select a representative sample of
eligible rural domestic wells. Whenever practical, field sampling and interviewing appointments were
scheduled during the screening interview. Non-resident well owners and farmers of the property where the
well was located, who were identified during the screening interview, were also contacted to schedule separate
interviews during the two-week sampling period. The NPS communications staff recontacted domestic well
owners/residents by mail and telephone to provide them with additional information and answer questions
about the Survey, schedule and confirm sampling dates, and determine detailed directions for traveling to the
well location.
The NPS Communications Manager tracked all sampling schedules to ensure that rural domestic well
sampling was conducted during the scheduled two-week sampling period. The NPS communications staff
completed sampling schedules for domestic well sampling teams, providing them with resident/owner names
and phone numbers, sampling dates and times, directions to well sites, and interview dates and times. Any
schedule changes reported from sampling teams in the field were carefully documented and scheduling
adjustments were made when practical.
During interviews additional farmers and non-resident well owners were sometimes identified. Field
teams would then contact these people and schedule interviews to obtain additional information concerning
the well and activities on the land surrounding the well.
The NPS field supervisor contacted county agricultural extension agents in advance of the two-week
sampling period and scheduled a time for interviewing. Field interviewers then interviewed the county
agricultural extension agent and collected information on the area surrounding each sampled well.
3.2.5 Scheduling Database Management
A computerized information management system, the National Pesticide Survey Information System
(NPSIS), was developed to monitor and track the CWS and domestic well sampling operations. NPSIS
generated tracking reports that documented the designated two-week sampling periods, scheduled sampling
dates, and sample receipt dates by NPS laboratories. A more detailed description of NPSIS and the operations
management reports it generated is provided in the Database Management Section of this chapter and
Appendix C.
3.2.6 Sampling and Data Collection Techniques and Procedures
All CWS and rural domestic well sampling activities were carried out using standard operating
procedures (SOPs), including quality assurance/quality control (QA/QC) checks, developed for the NPS. The
standard steps to conduct field sampling included collection of well water; completion of questionnaires and
-------
Chapter Three: Implementation of the Survey 27
other records to obtain supporting data; and sample collection follow-up activities, including shipment of
samples to laboratories and the completion and return of sampling logs, records, and questionnaires. These
steps are described below.
Sample Collection
Approval for Sampling. Upon arrival of the sampling team at a well site, the Field Team
Leader briefly described the sampling process to the well owner, obtained written permission
to sample from private well owners or residents, and inquired about water treatment and the
location of the well tap.
Sample Collection Procedures. Sample kits were prepared in advance to ensure that all
sample collection materials were on hand. Sample collection procedures included measuring
air temperature near the well tap; purging the well until temperature, conductivity, and pH
of the well water were stabilized to ensure that fresh ground water was being sampled; and
filling sample bottles with well water. Each sample was iced immediately after being taken
and placed in an insulated cooler for shipment to the analytic laboratories. NFS sampling
teams collected at least 17 water samples at each well site to allow one sample plus one
backup for each of the eight analytical methods. Additional samples were taken, as required,
for sample stability studies. Additional samples were taken from 5 to 10 percent of all wells
sampled for QA/QC review by two EPA laboratories, as described in Chapter 5.
Recordkeeping. Sampling team members recorded the date, time, and sampler's initials on
all sample bottles, and the team leader reviewed the Sample Tracking Form (see Appendix C)
to ensure that all samples were collected and to begin sample tracking. The team leader also
completed the Well Observation Record (see Appendix D) to collect information regarding
the well area, including topography, soils, water treatment and storage systems, well housing,
and additional observations deemed noteworthy because of their potential influence on
pesticide migration to the well being sampled. The team leader also prepared well-specific
notes, including a sketch of the well system indicating the sampling port or tap and a sketch
of local land features around the well, in the Well Observation Record.
Data Collection
Questionnaire Administration. The Survey research team member interviewed the CWS
well owner or operator, the rural domestic well owner, resident, and fanners when
appropriate and the county agriculture extension agent to collect information about well
construction, non-farm and farm pesticide usage on the well property, cropping activities, and
industry locations in proximity to the well site.
3.2.7 Activities Following Sample Collection
The sampling team completed a Field Logbook for each well site. The logbook included a Final
Domestic Well Checklist and a Well Sampling Information Sheet to document well location, well
owner/operator identification, sample kit identification, sampling date, shipment information, and the
responsible sampling team leader.
The sampling team shipped sample kits by express, overnight delivery to designated laboratories. The
sampling team also shipped completed questionnaires, sample tracking forms, airbills, and field logbooks to
NPS contractor staff, and reported shipment of water samples and field documentation via the NFS Hotline.
-------
28 Chapter Three: Implementation of the Survey
3.3 Questionnaires
NFS questionnaires were developed to collect data necessary for the interpretation of NFS analytical
results. (The NFS questionnaires are provided in Appendix D; a brief description of the NPS questionnaires
is provided in Exhibit 3-1.) Questionnaires were administered by the NPS to determine whether randomly
selected community water systems and domestic wells were eligible for NPS water sampling; record
observations about the well location and the surrounding area; collect information from the owner/operator
about the use and construction of the well sampled and use of non-farm and farm pesticides (e.g., volume of
and location of application in relation to the well) on the property where the well was located; and gather
information from local area experts about cropping, pesticide use, and industry locations within one-half mile
of the well.
Questionnaires were administered by NPS State, Region, or contractor sampling team personnel for
the CWS program, and by survey research contractors for the domestic well program. As part of both the
CWS and domestic well sampling collection activities, sampling team personnel administered two
questionnaires to the well owner/operator and another questionnaire to the local county agricultural extension
agent. If the domestic well owner or farmer(s) were different from the resident, each was interviewed using
the appropriate part of the questionnaires. NPS staff trained and periodically audited team leaders to ensure
that they were accurately recording field responses. Information gathered during these interviews was merged
with the analytical results to provide a data base for relational analyses.
3.3.1 Incorrect or Missing Data Items
Incomplete or incorrect information occurred in questionnaires either because a field interviewer failed
to record a response where one should have been recorded or because the recorded responses failed a quality
control check. NPS survey research staff instituted data retrieval efforts and imputation procedures to account
for incomplete or incorrect questionnaire items. NPS staff made follow-up calls to questionnaire respondents
to complete certain missing information or to confirm questionnaire responses identified as questionable by
range or logic checks.
3.4 Survey Operations Management
An extensive operations management system monitored field activities and maintained accurate records
of sampling and data collection activities. Operations management included determining the equipment and
supply needs of the Survey, establishing systems for tracking the inventory of supplies and equipment,
conducting logistics operations to support well sampling, mailing and tracking questionnaires, and planning
and processing travel arrangements for interviewers.
3.4.1 Equipment and Supplies
The NPS required large amounts of equipment and supplies to obtain and transport thousands of
water samples from over 1,300 wells. Equipment needs were projected at the beginning of the Survey and
adjusted, based on actual field sampling experience and to account for modifications in the number of wells
being sampled. NPS staff developed custom-made styrofoam water sample container kits to facilitate sample
handling and minimize bottle breakage. It was necessary to design and prepare custom-made sample kits
because commercially available coolers did not meet NPS sample bottle dimension specifications. The NPSIS
Equipment Tracking System was developed to monitor the inventory of equipment supply kits: temperature,
conductivity, and pH meters, and other field sampling tools and track equipment damage.
-------
Chapter Three: Implementation of the Survey 29
Exhibit 3-1
National Pesticide Survey Questionnaires
The Screening Survey Questionnaires were developed to identify a national sample of both
community water systems and domestic wells that met specific NPS eligibility criteria. NFS survey
research contractors conducted the CWS and domestic well screening interviews by telephone.
The domestic well questionnaire was administered using Computer Assisted Telephone Interviews
(CAT)), which automatically tracked the questions as responses were entered, reducing the
incidence of mis-keyed information, and used the correct unique set of geographic location
questions for each county.
The sampling team leader completed the Team Leader Introduction and Well Observation
Record by interviewing the respondent and recording on-site observations about the well and the
surrounding area, deemed important because of their potential influence on pesticide contaminant
migration into the sampled well.
The CWS and Domestic Well Main Questionnaires collected information regarding water use,
water treatment, well construction, and non-farm and farm pesticide usage on the property where
the well is located. Trained EPA Regional and State sampling team personnel conducted the CWS
interviews at the time of sampling; trained survey research contractors administered the domestic
well questionnaire on-site at the time of sampling.
The Domestic Well Second-Stage County Agent Questionnaire gathered cropping and
pesticide use information to determine what pesticides were used within the county for relational
analyses. Agricultural specialists, trained in administering the County Agent Questionnaire,
conducted the second-stage county agent interviews on-site. If necessary, NPS staff conducted
follow-up phone interviews, within one month of completing each County Agent Questionnaire, to
obtain any missing information.
The Local Area Expert Questionnaire was used to obtain additional information about local
conditions and usage of land and pesticides within one-haff mile of the well. Trained CWS EPA
Regional and State sampling team personnel and domestic well research contractors interviewed
local area experts at the time of sampling.
3.4.2 NPS Sampling Kit Preparation
All NPS water sampling kits were distributed from the NPS Preparation Room, located in the office
building of the implementation contractor in Fairfax, Virginia. NPS staff followed standard operating
procedures for preparing water sampling kits and tracking the shipment of water sampling kits to field
sampling teams and laboratories. A color coded system was developed to categorize the sample kits for the
contract laboratories and EPA referee laboratories.
An equipment kit, with an assigned serial number, was prepared for each well site, containing tools
and equipment required for sampling in the field. The appropriate sampling forms and bottles were placed
in sample kit containers according to the particular sampling scenario that was anticipated. The Preparation
Room Manager ensured that the standards for meter calibration were maintained, including keeping accurate
records of all completed calibrations.
Preparation Room personnel also prepared a Field Logbook for each CWS and rural domestic well
site sampled and shipped logbooks to the sampling teams approximately five weeks prior to the scheduled
sampling date.
-------
30 Chapter Three: Implementation of the Survey
Sampling teams shipped their completed sample kits to the NFS laboratories via express, overnight
delivery so that sample holding time requirements were observed. NFS laboratories returned used sample kits,
so that the kits (but not sample bottles) could be recycled for future use.
3.4.3 Mailing and Tracking Questionnaires
Questionnaires were mailed to NFS field staff for each well to be sampled. The status of each
questionnaire as it was mailed, received, and processed was recorded in separate CWS and rural domestic well
receipt control/tracking systems. For community well systems, the questionnaires were mailed first to the NFS
Preparation Room, where they were included with the field log books and then mailed to the field team leader.
Upon completion of all three field instruments, the field team leader would then return the questionnaires
with the log book.
For rural domestic wells, questionnaires for all wells in a county were mailed together to the field
interviewer. Upon completing all forms for a county, the interviewer mailed the entire set to the contractor
for processing.
3.5 Data Management
3.5.1 NPSIS
The National Pesticide Survey Information System (NPSIS) was designed to serve as a computerized
management information system for all aspects of the Survey. NPSIS was used to automate NFS sampling
implementation and analytical results information, with the exception of questionnaire data management. For
sampling implementation operations management, NPSIS was used to generate numerous management reports
to develop and monitor sampling schedules; track the preparation, shipment, and receipt of sample kits; record
sampling data, including problems encountered; and serve as a communications network between the NFS Prep
Room and laboratory staffs.
The objective of the data control methods adopted for the NFS was to collect and store data in an
efficient, user-friendly manner, while maintaining high data quality standards. NPSIS data control methods
included automating:
Data entry of Survey results and relevant operating information and manipulation of this data
using computers;
QA/QC checks for data entry via computer software; and
The backup of computerized data to ensure against loss.
3.5.2 Implementation Database Management
The implementation component of the database management system consisted of the Sampling
Control System (SCS), Communications Control System (CCS), Document Control System (DCS), and the
NFS Problem File. The functions of each of these database systems are as follows:
SCS was designed to automate information collection, storage, and retrieval for various
sampling tasks;
CCS was developed and integrated with the SCS to assist the NFS Communications Staff to
track DWS sampling schedules, communications information needed to secure voluntary
participation from domestic well owners, and participation rates;
National Pesticide Survev: Phase I Report
-------
Chapter Three: Implementation of the Survey 31
DCS was created to provide an efficient way to store and retrieve project records; and
The NPS Problem File was produced to document and help resolve problems associated with
sampling that could affect the validity of the analytic results.
3.5.3 Analytic Results Data Management
NPSIS was a geographically distributed personal computer network that linked sample delivery with
the contract laboratories and EPA laboratories. Using NPSIS, the Preparation Room staff were able to ensure
that the appropriate sample container kits arrived in the field on schedule and that all completed field sample
bottles were delivered to the designated laboratories in appropriate time and condition (i.e., iced to maintain
adequate temperature). NPSIS was used to inform NPS laboratories of anticipated sample arrivals and allowed
the laboratories direct computer access, to acknowledge sample and receipt conditions.
Contract laboratories analyzed samples, prepared preliminary analytical results, and submitted data
to EPA referee labs. The referee labs performed QA/QC checks. After extensive data review, analytical
results for each well were sent to communications staff for distribution to well owners/operators.
3.5.4 Questionnaire Results Data Management
Once questionnaires were returned to the contractor they were coded and data were entered. Missing,
out of range, or illogical data were identified (using computerized edits) for data retrieval or imputation, and
the revised data were reviewed for compliance with all range, logic, and consistency checks. CWS Team
Leader questionnaires underwent an additional initial data retrieval to clarify the identity of the well that was
sampled. A computerized tracking system kept track of each questionnaire at every stage. This system was
used to identify overdue cases (i.e., questionnaires that had not been returned on schedule), progress of data
preparation tasks, and the location of the hard copy of the questionnaires for review, as needed.
3.6 Communications
Communications operations were integral to the success of Survey implementation. The NPS involved
participation and cooperation among EPA Headquarters personnel, ten EPA Regional Offices, numerous
Federal agencies, State officials in all 50 States, community water system owners/operators and county officials,
private citizens, as well as implementation, survey research, and laboratory contractors. The large scale and
complex design of the NPS required communications activities that effectively managed information among
the many, diverse participants and interested panics throughout the country. This section provides a brief
description of NPS communications activities.
3.6.1 Internal Communications Activities
A communications network was established to facilitate the exchange of information between NPS
Headquarters in Washington, D.C. and all project participants about sampling schedules, results, and other
important information. EPA conducted regular NPS Work Group meetings, weekly staff meetings, monthly
conference calls with EPA Regional staff, and periodic meetings with State participants and other State
officials and EPA Regional staff in EPA Regional offices, to promote consistent communications among all
NPS participants. In addition, EPA established a temporary NPS Hotline - a toll-free telephone information
line ~ to facilitate communication among sampling team personnel and Headquarters staff. The hotline was
used to solve sampling problems while the sampling team was at the well site and to provide well owners and
State participants with easy access to the NPS information. The NPS communications staff operated the
hotline and responded to inquiries on Survey activities, including scheduling, field sampling, and the status of
well sampling results.
-------
32 Chapter Three: Implementation of the Survey
3.6.2 External Communications Activities
EPA kept a variety of interested parties (e.g., well owners, pesticide industry representatives, trade
associations, government agencies and associations, county and State officials, and Congressional staff,
agricultural groups, environmental and public interest groups, universities and professional associations, and
the general public) informed of the NFS survey design, sampling procedures, scheduling, and analytical
methods and results. EPA maintained an extensive NPS mailing list and periodically published newsletters,
the "NPS Project Update," to keep all interested parties informed of the Survey's progress. The EPA Press
Office released press updates to notify the public of current NPS activities. EPA established strict procedures
to ensure the proper dissemination of well sampling results. The NPS communications staff notified well
owners and State officials of well sampling results by instituting these procedures to ensure that the privacy
of domestic well residents/owners was protected and to comply with Freedom of Information Act regulations.
Results from the sampling of community water system wells were made public as soon as the State and the
CWS had been notified. Sampling results relating to the wells of private individuals were protected to ensure
individual privacy, while still permitting disclosure of sampling results. Domestic well sampling records were
not accessible by the name or address of any individual but by an anonymous well code. States were asked
to provide written statements of their ability to maintain confidentiality. Results of domestic well sampling
and other information will be released only in appropriately aggregated form or by anonymous well code.
3.6.3 Implementation Communications Materials
EPA developed and distributed a variety of outreach materials and implemented numerous
communications activities to ensure that all NPS participants and interested parties were informed. A
description of these materials and activities is provided below.
NPS Pamphlets. EPA developed a CWS pamphlet and a companion rural domestic well
pamphlet for CWS owner/operators and owners of domestic wells to provide information
about the purpose of the Survey and explain what was required to participate.
NPS Project Updates. EPA published periodic newsletters to inform project participants
and interested parties of the progress of the National Pesticide Survey implementation efforts.
EPA maintained an extensive NPS mailing list of Regional, State, and county officials,
industry representatives, public interest groups, and associations and periodically published
the "NPS Project Update" to keep all interested parties informed of the Survey progress.
Rural Domestic Well Sampling Team Communications Packets. NPS communications
staff prepared communications packets for contractor sampling teams for each county in
which rural domestic wells were sampled. The packets contained NPS pamphlets and
updates, the NPS confidentiality policy, a list of appropriate State and EPA contacts,
sampling schedules, and permission forms. Rural domestic well sampling teams used these
documents to respond to questions from well owners, local citizens and officials, and the
media and to provide information to well owners about the purpose of the NPS, sampling
procedures, the notification of results, and confidentiality issues.
Media Relations. Pesticide use and ground-water quality have received a great deal of media
coverage in the past several years. EPA Headquarter's Press Office issued press advisories
and news releases on NPS activities, and EPA Regional contacts responded to local media
inquiries.
Matlnnal Doctiflsta Ciin/ou- Dhaee I Ronnrt
-------
Chapter Three: Implementation of the Survey 33
Health Advisories. EPA developed 99 technical Health Advisories for pesticides and nitrate
using information collected on physicochemjcal properties, uses, chemical fete, health effects,
treatment, and existing criteria and guidelines. These Health Advisories are scientific
guidance documents that were used, to evaluate the results of well sampling and to determine
whether the contamination levels found warranted further action.
Health Advisory Summaries. As pan of EPA's effort to improve risk communication, the
Agency prepared non-technical Health Advisory Summaries explaining the health effects of
exposure to pesticides and nitrate in simple language; the summaries were distributed to the
owners, residents, and operators of sampled wells where pesticides and nitrate were found.
Notification of Results Packets. EPA established strict procedures to ensure the proper
dissemination of well sampling results. The NFS communications staff notified well owners
and State officials of well sampling results by instituting these procedures to ensure that the
privacy of domestic well owners was protected and to comply with Freedom of Information
Act regulations. Owners and residents of all NPS sampled wells in which detections of
pesticides and nitrate were found were notified by letter and provided appropriate Health
Advisory Summaries and information to assist them in contacting State and county officials
for assistance.
3.6.4 NPS Communications Strategy
A communications strategy was prepared in late 1988 to establish communications protocols and
provide guidance for communications activities throughout the Survey, from pre-survey development to release
of the Phase II Report.
EPA held meetings with key Federal, State, and local agencies to gather input about the Survey design
and to promote cooperation from these groups in implementing the Survey. EPA met with the Department
of Agriculture (USDA), the Bureau of the Census, and the U.S. Geological Survey (USGS) to discuss the
survey design and implementation issues.
The NPS Work Group was established in the fall of 1986 to review NPS project plans, solicit
comments and advice from participating members, and establish communications procedures for participating
States and counties. The Work Group consisted of EPA Headquarters and Regional staff, State personnel,
and other involved persons. The Work Group met periodically to advise on implementation issues such as
confidentiality, notification of results, and communication materials. These early coordination efforts were
important to the long term success of the Survey.
3.6.5 Release of NPS Results
The following activities are planned to release NPS results to interested parties and the general public.
Briefings/Presentations. EPA Headquarters will conduct briefings for EPA management
and Regional EPA offices, other Federal ageqcies, State officials, and professional and trade
associations. Topics include Survey implementation and Survey results.
Fact Sheets/Outreach Materials. EPA has prepared a series of fact sheets and outreach
materials to disseminate the background information about the NPS and the findings of the
Survey in a concise, easy to understand manner.
National Pesticide Survey: Phase I Report
-------
34 Chapter Three: Implementation of the Survey
Dissemination of Materials. EPA has compiled a contact list from existing NFS and master
mailing lists that will be used to distribute copies of the Phase I and Phase II reports and
notices of their availability, pamphlets and outreach materials, and press kits.
Reports and Information Materials. A list of key NFS reports, supplemental materials, and
other information materials is provided in Exhibit 3-2.
Exhibit 3-2
NPS Reports and Information Materials
Pre-Survey Reports:
Final Report on the Management Systems Review of the Pilot Survey of Pesticide Contamination in
Groundwater, 1987, Quality Assurance Management Staff, USEPA.
National Survey of Pesticides in Drinking Water Wells, A Review of the Planning Process and the Data
Quality Objectives, 1987, Monica Nees and Cynthia Salmons, RTV7801/08/01F.
National Pesticide Survey Data Quality Objectives, 1988, ICF Incorporated.
National Pesticide Survey Pilot Study Evaluation Summary Report, 1987, Office of Drinking Water and
Office of Pesticide Programs, USEPA.
National Pesticide Survey Pilot Evaluation Technical Report, 1987, R.E. Mason, LL Piper, W.J.
Alexander, R.W. Pratt, S.K Liddle, J.T. Lessler, M.C. Ganley, D.J. Munch, and G. Langner, RTI/7801/06-
02F.
National Pesticide Survey Communications Strategy, 1986, ICF Incorporated.
Ground-water Vulnerability Assessment In Support Of The First Stage Of The National Pesticide Survey,
1986, W.J. Alexander, S.K. Liddle, R.E. Mason, and W.B. Yeager, RTI.
National Survey Of Pesticide Residues In Community System And Rural Domestic Wells Sample
Allocation Report, 1988, R.E. Mason and R.M. Lucas, RTI/7801/04-04F.
Scientific Advisory Panel Reports:
Technical Review of the Preliminary Design of the National Survey of Pesticides in Ground Water -
Report of Subpanel Recommendations, October 9, 1985, Federal Insecticide, Fungicide, and
Rodenticide Act Scientific Advisory Panel Subpanel.
A Set of Scientific Issues Being Considered by the Agency in Connection with the National Pesticide
Survey Pilot Study, October 9, 1987, Federal Insecticide, Fungicide, and Rodenticide Act, Scientific
Advisory Panel Subpanel.
Quality Assurance Project Plans:
The National Survey of Pesticides in Drinking Water Wells Quality Assurance Project Plan for the Pilot
Study, 1987, Shri Kulkarni, Frank Smith, Cindi Salmons, and Susan Coffey, RTI/7801/08/01.
Final Quality Assurance Project Plans are listed in Exhibit 5-2.
Project Updates
Health Advisories
Health Advisory Summaries
NPS Fact Sheets
Brochures
National Pesticide Survey: Phase I Report
-------
Chapter Four: Analytical Chemistry
introduction
This chapter describes how EPA chose the pesticides and pesticide degradates (the analytes) to be
tested for in NFS well water samples. It identifies and describes the development of the testing methods that
were used. Finally, it identifies and describes the activities of the analytical laboratories that conducted the
testing.
4.1 Analyte Selection and Methods Development
Preliminary development of a list of potential analytes for the NPS began in 1984, when an NPS
Analyte Selection Task Group was formed. The Task Group was composed of EPA personnel from the Office
of Pesticide Programs and Office of Drinking Water.
The Task Group began with a list of approximately 600 pesticide active ingredients. From that list,
the Task Group identified pesticides and pesticide degradates as potential Survey analytes by choosing those
compounds that met several criteria. First, pesticides and pesticide degradates previously detected in ground
water, as well as pesticides regulated under the Safe Drinking Water Act, were automatically included in the
list of potential Survey analytes. In addition, pesticides and pesticide degradates were included if:
the pesticide's chemical/physical parameters, including water solubility, partition coefficients,
field half-life, hydrolysis half-life, and others, indicated a potential to leach to ground water;
and
at least one million pounds of the pesticide were estimated to have been used nationwide in
1982.
By October 1986 a list of 100 pesticides and pesticide degradates was identified as potential NPS
analytes. Additional analytes were added to this list based on the studies of analytic methods conducted
simultaneously with analyte selection. The additional analyles were chemicals that the NPS methods would
identify and quantify for little or no added expense while the multi-residue analyses were being carried out.
Ultimately, 127 analytes were chosen. Detailed information about the NPS analytes is provided in Exhibit E-l
in Appendix E.
Work to identify appropriate analytical methods began with the development of the first list of
potential analytes in mid 1984. An NPS Analytical Methods Task Group identified analytic methods through
literature searches, consultations with pesticide analysis experts, and laboratory testing of methods.
An important goal was identifying or developing analytic methods that could identify more than one
analyte ("multi-residue methods") so that a small number of methods covered as many pesticides as possible.
Whenever a published analytical method was not available for a potential analyte, the importance of
developing such a method was evaluated, based on the importance of the pesticide and the cost and time
necessary to develop an analytic method. Research into multi-residue analytic methods for NPS analytes was
begun in early 1985 at the Battelle Laboratories, Columbus Division, under the joint direction of the Technical
Support Division of ODW and the Environmental Monitoring and Systems Laboratory - Cincinnati. Reviews
of the available literature indicated that analytical procedures existed for most of the analytes of particular
interest to EPA However, these methods varied widely in procedure and application, and were applicable to
single analytes. The primary goal of the methods development study therefore was the consolidation of these
many analytical procedures into a smaller group of methods capable of evaluating multiple analytes. Efforts
were directed towards retaining the proven aspects of previously published methods, while developing the
-------
36 Chapter Four: Analytical Chemistry
procedures necessary to ensure the quality of the analyses. In this manner, the original published methods
were transformed into methods amenable to use in the NPS as well as in future analysis of pesticides in water.
The major work of determining proper sample preparation, separation, and detection techniques was accom-
plished during the course of 1985, but continued up to the beginning of implementation of the full Survey.
Six new multi-residue analytic methods were developed for the Survey (Methods No. 1-6). In addition,
two existing EPA methods (No. 7 and 9) also were used. One method (No. 8), for volatile organic compounds,
was dropped after extensive evaluation because of problems with sample aeration under Survey conditions and
because of cost considerations. These methods are described briefly in Exhibit 4-1.
4.2 Analytical Laboratories
The chemical analyses of NPS water samples were performed at five contract laboratories and three
EPA laboratories.4 The contract laboratories were responsible for analyzing water samples collected at all
of the field sites by one or more analytic methods. In addition, analytical precision and accuracy and analyte
stability determinations were made using laboratory spiked well water and reagent water samples analyzed at
the contract laboratories. The EPA laboratories analyzed duplicate water samples collected at 5 to 10 percent
of the field sites (the rate varied with each NPS method) as one measure of the quality of the analyses being
performed at the primary laboratories, and provided back-up capacity. (See Sections 4.4 and 4.5 for additional
information on this topic.)
Personnel at the EPA laboratories were responsible for the technical and contractual management
of the contract laboratories and for performing referee analyses of duplicate water samples using the
appropriate analytical methods. Exhibit 4-2 identifies the contract and EPA laboratories and the analytic
methods performed at each laboratory.
4.3 Minimum Quantification and Reporting Limits
The NPS analytic methods were designed to identify both the presence and concentration of analytes
in water samples, whenever possible. For 112 of the 127 analytes, minimum reporting limits were established,
above which the analyte was reported as having been detected; in addition, minimum quantification limits were
established above which the concentration as well as the detection was reported. For 15 analytes which were
either unstable or lacked reliable method performance, no reporting limit was established. The presence of
these analytes is reported as a "positive detection" without quantification. Exhibit 4-3, which identifies the
analytic method for each analyte, also indicates which analytes did not have quantification limits.
The minimum quantification limit (MQL) was established through a two-step process. First, an
estimated detection limit (EDL, i.e., the minimum concentration of the analyte that can be measured and
reported with confidence that the analyte concentration is greater than zero as determined from the analysis
of a sample in a given matrix containing the analyte) was determined for each analyte. The EDL was
calculated from results of seven or more replicate analyses of a reagent water sample fortified with the
concentration of the analyte that would yield a chromatographic peak with a signal to noise ratio of
approximately five to one. The standard deviation of these data was then multiplied by the students' t value
appropriate for a 99 percent confidence level and a standard deviation estimate with n-1 degrees of freedom
where n equalled the number of replicate analyses. The EDL was defined as either this calculated
concentration or the concentration of analyte yielding a five to one signal to noise ratio, whichever was greater.
4 The EPA Environmental Chemistry Section and the EPA Technical Support Division also managed the contracts with
other laboratories for analytical services. Method 9 was performed by the Risk Reduction Engineering Laboratory, Office
of Research and Development, U.S. EPA, 26 W. Martin Luther King Drive, Cincinnati, Ohio, but this EPA laboratory did
not manage contract laboratories.
Natirmal Postiriri* Survew Phacp I Rpnnrt
-------
Chapter Four: Analytical Chemistry 37
Exhibit 4-1
NPS Analytic Methods
Method
Number
Analytes
Number of
Analytes
Method
NPS#1
Nitrogen-Phosphorous containing
(EPA #507) Pesticides
NPS #2* Chlorinated Pesticides
(EPA #508)
NPS #3* Chlorinated Acid Pesticides
(EPA #515.1)
NPS #4
NPS #5
(EPA #531.1)
NPS #6*
NPS #7
(EPA #504)
NPS #8
(EPA #524)
NPS #9
(EPA #353.2)
Pesticides and Pesticide
Degradates (Carbonates and
related compounds)
N-methyl Carbamoyloximes and N-
methyl Carbamates
Ethylene thiourea
1,2 - dichloropropane;
Dibromochloropropane (DBPC);
Ethylene dibromide (EDB); cis -1,3
- dichloropropene; trans -1,3
- dichloropropene
Nitrate/Nitrite
46
29
17
18
10
Total 127
Gas Chromatography with a
Nitrogen-Phosphorous De-
tector
Gas Chromatography with
an Electron Capture De-
tector
Gas Chromatography with
an Electron Capture De-
tector
High Performance Liquid
Chromatography with an
Ultraviolet Detector
Direct Aqueous Injection
HPLC with Post-Column
Derivatization
Gas Chromatography with a
Nitrogen-Phosphorous
Detector
Microextraction and Gas
Chromatography
Automated Cadmium
Reduction and Colorimetric
Detection
*
**
New method developed for NPS.
The EPA method listed (e.g., 'EPA #507") is the NPS method adopted by EPA as an
official Agency method and currently referenced under the designated number. For de-
tails on each NPS Method, see the Quality Assurance Project Plans listed in Exhibit 5-2 in
the following chapter.
Method #8 (EPA Method #524 for volatile halogenated, aromatic, and unsaturated organic
compounds in water by purge and trap gas Chromatography) was initially considered and
evaluated for use but was dropped.
National Pesticide Survev: Phase I Reoort
-------
38 Chapter Four: Analytical Chemistry
Exhibit 4-2
NPS Contract Laboratories and EPA Laboratories for the Full Survey
Contract Laboratories
Methods
James Montgomery Laboratories
555 East Walnut Street
Pasadena, California 91109
Alliance Technologies/Clean Harbors Analytical Services
213 Burlington Road
Bedford, Massachusetts 01730
Radian, Inc.
8501 Mo-Pac Boulevard
Austin, Texas 78766
Environmental Science and Engineering, Inc.
14220 West Newbury Road
Gainesville, Florida 32607
Battelle Columbus Division
505 King Avenue
Columbus, Ohio 43201
1,3,9
5,7
EPA Laboratories
Methods
Environmental Chemistry Section
Office of Pesticide Programs
U.S. EPA
Stennis Space Center
Bay St. Louis, Mississippi 39529-6000
Technical Support Division
Office of Drinking Water
U.S. EPA
26 West Martin Luther King Drive
Cincinnati, Ohio 45268
Risk Reduction Engineering Laboratory
Office of Research & Development
U.S. EPA
26 West Martin Luther King Drive
Cincinnati, Ohio 45268
1,3,6
2, 4, 5, 7, 9
National Pesticide Survey: Phase I Report
-------
Chapter Four: Analytical Chemistry 39
Exhibit 4-3
Survey Analytes and Analytic Methods
NPS METHOD 1: Gas Chromatography with a Nitrogen-Phosphorous Detector
Alachlor Diphenamid Methyl paraoxon
Ametryn Disulfoton* Metolachlor
Atraton Disulfoton sulfone* Metribuzin
Atrazine Disulfoton sutfoxide* Mevinphos
Bromacil EPIC Molinate
Butachlor Ethoprop Napropamide
Butylate Fenamiphos Norflurazon
Carboxin Fenarimol Pebulate
Chlorpropham Fluridone Prometon
Cycloate Hexazinone Prometryn
Diazinon* MGK 264 Pronamide*
Dichlorvos Merphos* Propazine
NPS METHOD 2: Gas Chromatography with an Electron Capture Detector
4,4-DDD Dieldrin Heptachlor epoxide
4,4-DDE Endosutfan I Hexachlorobenzene
4,4-DDT Endosutfan II Methoxychlor
Aldrin Endosutfan sulfate Propachlor
Chlorobenzilate* Endrin Trifluralin
Chloroneb Endrin aldehyde alpha - HCH
Chlorothalonil Etridiazole beta - HCH
DCPA Heptachlor delta - HCH*
NPS METHOD 3: Gas Chromatography wfth an Electron Capture Detector
2,4-D 4-Nitrophenol* Dalapon*
2,4-DB Acifluorfen* Dicamba
2,4,5-TP Bentazon Dicamba, 5-hydroxy-
2,4,5-T Chloramben* Dichlorprop
3,5-Dichlorobenzoic acid DCPA acid metabolites Dinoseb
NPS METHOD 4: High Performance Liquid Chromatography with an Ultraviolet Detector
Atrazine, deethylated Diuron Metribuzin DA
Barban Fenamiphos sulfone Metribuzin DADK*
Carbofuran, phenol-3-keto- Fenamiphos sulfoxide Metribuzin DK*
Carbofuran, phenol Fluometuron Neburon
Cyanazine Linuron Pronarnide metabolite
NPS METHOD 5: Direct Aqueous Injection HPLC with Post-Column Derivatization
Aldicarb Baygon Carbofuran, 3-hydroxy-
Aldicarb sulfone Carbaryl Methiocarb
Aldicarb sulfoxide Carbofuran Methomyl
NPS METHOD 6: Gas Chromatography with a Nitrogen-Phosphorous Detector
Ethylene thiourea (ETU)
NPS METHOD 7: Microextractlon and Gas Chromatography
Ethylene dibromide (EDB) 1,2 - dichloropropane** trans -1,3-
Dibromochloropropane (DBCP) cis - 1,3 - dichloropropene** dichloropropene**
NPS METHOD 9: Automated Cadmium Reduction and Colorlmetrlc Detection
Nitrate and nitrite measured as nitrogen (N)
* Qualitative only.
** Method 8 dropped. Analytes previously included in Method 8 also detectable by Method 7.
(46 Analytes)
Simazine
Simetryn
Stirofos
Tebuthiuron
Terbacil
Terbufos*
Terbutryn
Triademefon
Tricyclazole
Vernolate
(29 Analytes)
gamma - HCH
alpha-Chlordane
gamma-Chlordane
cis - Permethrin
trans - Permethrin
(17 Analytes)
Pentachlorophenol (PCP)
Picloram
(18 Analytes)
Propanil
Propham
Swep
(10 Anatytes)
Oxamyl
(1 Anafyte)
(5 Analytes)
(1 Analyte)
Pesticide Survew Phn«u» I Rennrl
-------
40 Chapter Four: Analytical Chemistry
This procedure was an adaptation of EPA's existing minimum detection limit procedure. [Glaser, et al., Trace
Analysis for Waste Waters," 15 Environmental Science and Technology 1426 (1981)].
The MQL then was determined by multiplying the EDL by a factor, based on the best professional
judgment of the EPA referee laboratory staff, determined in part by the degree to which the method is subject
to interferences. The factors were: Method #1 = 4, Methods #2-4 = 5, and Methods #5-7 and 9 = 3. The
purpose of this step was to determine a MQL that could be reasonably maintained throughout the course of
the Survey and take into account day-to-day variations. The MQLs for each analyte are provided in
Appendix E, Exhibit E-2.
The minimum reporting limit (MRL) was based on the MQL. Analytes detected above the MQL were
reported with quantification (i.e., at a specified concentration). Analytes detected at a concentration that was
estimated to be between one-half the established MQL and the MQL were reported as detected without
quantification, in order to maximize the sensitivity of the methods while maintaining reasonable minimum
reporting and quantification limits. EPA chose relatively stringent reporting limits. If the reporting limits
used in the NPS had been consistently lower, the number of reported detections might have been higher, but
such lower limits would be reflected in less confidence in the analytic results.
4.4 Confirmation of Positive Detections
The identity and concentration of all pesticide and pesticide-related analytes detected was confirmed
with a second, separate analysis on another chromatographic column. It was not necessary to confirm
detections of nitrate because of the extremely specific nature of the analysis.
The primary laboratories confirmed detections by reanalyzing all suspected positive extracts using a
second (different) capillary gas chromatographic (GC) column (Methods 1, 2, 3, 6, and 7) or a different
quantitative and qualitative high performance liquid chromatographic (HPLC) column (Methods 4 and 5).
These analyses provided both a preliminary qualitative confirmation for all GC determinations and the final
confirmation for HPLC analyses. The analytes detected using a GC-based method also were qualitatively
confirmed using gas chromatography/mass spectrometry (GC/MS).
Quantitative results were reported, if the concentration determined on the confirmational column
agreed within 25 percent of that determined on the primary column. If this requirement was met, the
quantitative results determined on the primary column were reported. If this requirement was not met, and
for analytes determined using a GC method, the presence of the analyte was confirmed by GC/MS, and the
detection of the analyte was reported without quantification.
4.5 Quality Control
To ensure the accuracy of calibration standards, verified 1-2 mg/mL stock standards (solutions
containing a standard that is a method analyte obtained from and certified by EPA's Pesticide and Industrial
Chemicals Repository) were provided in sealed glass ampules to all of the participating laboratories. These
standards were prepared by The Bionetics Corporation in Cincinnati, Ohio and verified by the EPA referee
laboratories prior to distribution to the primary laboratories.
A number of quality control procedures were implemented for all analyses. Control charts were
maintained tracking the recovery of laboratory control standards and surrogate standards. Quality control
measures included but were not limited to the use of the following:
Laboratory Control Standards (a solution of analytes prepared in the laboratory by dissolving
known amounts of pure analytes in a known amount of reagent water) to ensure that the
extraction processes maintained consistent analyte recovery (Methods 1-6);
National Pesticide Survey: Phase I Report
-------
Chapter Four: Analytical Chemistry 41
Instrument Control Standards (Methods 1-6) (a solution containing specified concentrations
of specified analytes that is analyzed each working day prior to the analysis of sample
extracts) to ensure that analytical instrumentation was properly calibrated and performing
properly prior to sample analysis;
Surrogate Standards (Methods 1-7) (pure compounds that are not sample components added
to a sample in a known amount and used to detect gross abnormalities during sample
preparation) to demonstrate proper analyte recovery in individual samples; and
Method Blanks (a portion of reagent water analyzed as if it were a water sample) (Methods
1-7 and 9) and Shipping Blanks (reagent water shipped to the field and returned to the lab
with the samples) (Method 7) were analyzed to help control false positives.
In addition to the quality control procedures employed in the laboratories, a series of quality control
measures, detailed in Chapters 3 and 5, were implemented for the sample collection and shipping operations.
Although in some cases the judgment of the EPA chemist performing final data review did override the formal
quality control procedures, generally data that failed to meet the established quality control limits were not
used in the analysis of Survey results. Exhibit E-3 in Appendix E presents the numbers of analyses deleted
due to failure to meet quality control guidelines.
4.6 Precision and Accuracy
To determine the precision and accuracy of chemical analyses, field samples were fortified with the
method analytes at the primary laboratories (spiked samples). Approximately 40 field samples were fortified
at each of three concentrations: two, five, and ten times the MQL for methods 1-7, and two times the MQL,
five and ten mg/L as nitrogen for method 9. The precision and accuracy determined from the analyses of field
samples fortified at the ten times MQL, methods 1-7, and 10 mg/L, method 9, concentrations indicate that
excellent precision and accuracy were obtained over the course of the Survey. Exhibit E-4 in Appendix E
presents these results for each analysis method and analyte.
Comparisons of the results of the analyses conducted at the primary laboratories of both fortified field
samples and laboratory control standards were used to indicate the effect of the sample matrix (i.e., the well
water samples) on the analytical precision and accuracy. These data indicate that there was virtually no impact
on analytical precision and accuracy due to the sample matrix.
The results of analyses of laboratory control standards, fortified at ten times the MQL, conducted at
both the primary and referee laboratories were used to assess the relative precision and accuracy obtained at
each laboratory. These results are presented in Exhibit E-5 in Appendix E for each analysis method and
analyte for both primary and referee laboratories. These data indicate that similar precision and accuracy were
obtained at both the primary and referee laboratories.
4.7 Analyte Stability Studies
A limited study was conducted during methods development and the pilot study to evaluate the
stability of the analytes in water samples over time. The purpose of the study was to ensure that the analytes
would be stable for the maximum time anticipated between collection and analysis. During methods
development, samples of central Ohio ground water were acquired for these experiments. First, this water was
analyzed using each of the six newly developed methods to determine if any of the analytes of interest or
analyte interferences were observed. Replicate samples were fortified with method analytes and then analyzed
after 0,14, and 28 days storage at 4°C. The extracts obtained during the analyses of the day 0 samples were
also analyzed after 14, and 28 days storage. Serious analyte degradation upon storage was observed for 24 of
the potential analytes studied, including 22 potential Method #1 analytes, and two potential Method #2
-------
42 Chapter Four: Analytical Chemistry
analytes. Eighteen of the potential Method #1 analytes, and two of the potential Method #2 analytes that
degraded during storage were deleted from the scope of these two methods. The remaining four potential
Method #1 analytes (Disulfoton sulfoxide, Prometon, Pronamide, and Terbufos) were of specific interest to
the Agency and were kept within the method scope, pending further studies.
Additional analyte stability studies ("time storage" studies) were conducted throughout the full Survey.
The data obtained during the full Survey confirmed those previously obtained, with the exception that no
instability problems were encountered for Prometon.
4.8 False Positives and False Negatives
Due to the use of method blanks, appropriate MQLs, and extensive confirmation of all positive results,
the risk of any possible false positive was negligible.
Duplicate samples were collected at 5-10 percent of the field sites for analysis at the referee
laboratories. Exhibit E-6 in Appendix E presents the results of the comparison of the referee laboratory
analyses of each analyte in samples in which this analyte was not observed by the corresponding primary
laboratory. Taking differences in MQL into account (referee laboratory MQLs were different from primary
laboratory MQLs for many analytes), an analyte detected in the referee laboratory analysis was not observed
at the primary laboratory in only two cases out of 200 samples. The largest false negative rate of these two
cases - atrazine - was only 0.98 percent. Thus, there was a maximum possible false negative rate of
approximately 1 percent or less for each analyte.
National Pesticide Survey: Phase I Report
-------
Chapter Five: Quality Assurance/Quality Control
Introduction
EPA designed and carried out an intensive quality assurance program for the NPS. The key elements
of that program -- expert review, pilot studies, training, specific standard operating procedures, detailed quality
assurance plans, and frequent audits - are recognized as a model for other EPA studies. This chapter
discusses the quality assurance/quality control (QA/QC) program for the Survey. In particular, it describes
the objectives, organization, and key components of quality assurance and quality control as they relate to each
of the major components of the Survey.
5.1 Overview of the Quality Assurance Program
The quality assurance (QA) program for the NPS was designed to encompass all NPS activities to
ensure the reliability and validity of data produced by the Survey. This system integrated the quality planning,
quality assessment, and quality improvement efforts of various groups in the organization to enable data
generators to meet data user requirements. The QA program addressed the activities of ten major
organizations across the country responsible for managing the various activities associated with the Survey,
including eight laboratories (five contract and three EPA) using eight analytical methods, and one prime
contractor and one subcontractor responsible for all non-analytical support activities such as mapping, site
selection, training, sampling, questionnaire administration, communications, and database management.
The overall objective of the quality assurance/quality control program was to provide program staff
in the Office of Pesticide Programs and the Office of Drinking Water with data of known and useful quality.
Using guidance developed by the Quality Assurance Management Staff for EPA quality assurance programs,
the NPS QA program monitored and controlled information-gathering processes so that data met design
criteria for accuracy, precision, representativeness, comparability, and completeness.
5.1.1 Organization
Responsibility for implementing the quality assurance program rested with a designated Survey Quality
Assurance Officer (QAO), a contract QA specialist who reported directly to the Director of the Survey. The
Quality Assurance Officers for the Office of Pesticide Programs and the Office of Drinking Water provided
independent oversight and guidance to contractors. The Quality Assurance Officer called upon technical staff
from organizational units within the Office of Pesticide Programs and the Office of Drinking Water as
necessary, for quality assurance support. Typically, the QA support staff had NPS line management
responsibilities. For example, the technical monitors who were responsible for overseeing each of the
analytical methods often were asked to assist with QA functions. Exhibit 5-1 displays the overall quality
assurance organization, including the oversight that was provided by individuals within each organization. As
Exhibit 5-1 indicates, overall responsibility for quality assurance rested with the NPS Quality Assurance
Officer, who reported to the NPS Director and was provided support by the OD W and OPP Quality Assurance
Officers and staff. Analytical coordinators from ODW and OPP assisted the NPS QAO, and supervised EPA
technical monitors, who were responsible for ensuring that particular analytic methods were carried out
properly by the contract and EPA laboratories performing these methods. An NPS implementation
coordinator was responsible for ensuring the reliability and validity of statistical, sampling, and hydrogeological
activities, acting with the assistance of EPA technical monitors for each of these activities. Each laboratory
and implementation contractor also carried out internal QA/QC functions.
-------
44 Chapter Five: Quality Assurance/Quality Control
Exhibit 5-1
NFS Quality Assurance Organization
8:
6
ex
a. c/3
7 0-
, "Z-
2 " 6 "
H <
5 a
Q
L o
^
O'
1
!
-
[>lemenlalion
dinator - NFS
S,8
U
1
? *
i1 «
4B
^S
0
,B
3
a te
s I
H
O
1
4
$
1
_
-
Technical
Monitor -
Technical
Monitor -
1 5-i
41^
"« '
3 2
i '§
4l:
"M J. '
.H O '
IJ;
ODW Lab
Coordinator
.§ S
H S
|i;
H S ,
Technical |
Monitor - |
s
\
i
f
n
n
i
3
3
o
I
M
Q
i
j
*
2
n
!;
i
N
I
u
>
§
a
Implemenlatioi
Contractor
Project Direcio
andQAO
l .. ^ ,
no
III 1
i|f 1
M S -s
t i
^cSo ««
s s 1 o ;
J £ '
t^ " ° 6 O
o Ojs S?
2<-> j g
22 S 1
8 I1S.5
? S » S? 2/
isjr
aE^ 1 s
1i? li
(Ssl Is
.0 «
5-3 - s
lifr^
Su^l
;J § s
is IE?
^ 6 Ǥ S ^
SJ£|
JS TJ
SJ i s
1 alE?
l!!J*
Inyarogeoiogy
Group Group
Leader and
QAC
53
|
3 & So
M *J J? 5
j « Q O'
2 1 « i
s S-£S
" £
1 2
1J i
*
" m
"3§
u o fi«
2*
;S^
fS
!l
*3i
2|£
JS
i^s
1 «}
5 i*
2 j
1 Method 2
Referee Lab
1 ProRram
^
5
N
s
s
?n^
jPS
2
a
&^
« ry
s a
^
^^
i0'
z
^^^B
(Manager and
QAC
o
§
^ <
11
O
8 8
5 c
I i
>v >>
II
II II
ou
* *
^3-5 ^^cl °"J6S
iifiiiilpiilp
SJ I |3*| ** I
Na»
^*
3
i
4
-------
Chapter Five: Quality Assurance/Quality Control 45
5.1.2 Key Components
The key components of the QA program were planning, program oversight, project oversight, and
oversight at the operations level by organizations conducting NFS activities. Each of these components is
described in the following sections.
Planning. In order to design the NFS so that the end product would be acceptable to all concerned
parties, policy-makers and scientists agreed upon quantitative levels of precision required of the national
estimates for pesticide frequency and occurrence that also could be achieved within budgetary and technical
constraints. In October 1984, EPA established formal procedures, known as the Data Quality Objective
(DQO) process, for reaching consensus on acceptable levels of data quality. NFS planning followed the EPA
DQO process, and a DQO document was written in 1987 (see Exhibit 3-2).
Once objectives were clearly stated in the DQO document and the research design was agreed upon,
the NFS pilot study was conducted to test the analytical and implementation procedures that were planned
for the full Survey. From a quality assurance perspective, the pilot study was extremely useful. Problems with
logistics were identified and resolved, and an approach to analytical quality control was selected based on
direct comparison of different quality control criteria. The quality assurance program also was refined for the
full NFS and the roles of Quality Assurance Officers and the number and types of necessary audits were
clarified.
Program Oversight. Program oversight was necessary for coordinating the activities of all
participating organizations, for ensuring accuracy and compatibility of data, and for monitoring compliance.
Three tools were used to accomplish these goals. The first was the development of a quality assurance
program plan, which described NFS policies and procedures for achieving quality. The second was auditing,
which was used to communicate Survey expectations with respect to quality to participating organizations and
their personnel, to identify potential problems, and to assist in improving the implementation process to
ensure high quality data. The third was frequent reporting to Survey management on the status of quality
assurance efforts.
Project Oversight. Project oversight was necessary to ensure that technically correct procedures were
being employed to generate Survey data. This goal was accomplished by requiring that each discrete Survey
activity be described in a quality assurance project plan, in which an operations level internal quality assurance
program had to be described, that included all relevant standard operating procedures (SOPs) for that activity.
Project plans were approved by the Survey Director and the Quality Assurance Officers for the Office of
Drinking Water and the Office of Pesticide Programs. Audits then were conducted using the plans as
evaluation criteria to verify that data collection and QA/QC efforts were implemented as described. The
technical monitors were primarily responsible for providing oversight at the project level.
Operations Oversight. Each of the organizations, including both EPA and contractors, that
participated in the planning and implementation of the Survey carried out internal oversight of their own
activities. The personnel who performed this internal oversight function were required to be independent of
the data generation processes, yet familiar with the technical aspects of efforts within their organizations in
order to perform audits. Each organization's management was then accountable for responding to
recommendations made by their own quality assurance staff to ensure that Survey products were delivered at
the agreed-upon levels of quality.
Specific information on the QA/QC requirements for each of the analytical and implementation
activities can be found in individual quality assurance project plans, available from the National Technical
Information Service (NTIS) in Springfield, Virginia. Exhibit 5-2 lists the titles of all NFS quality assurance
documents.
-------
46 Chapter Five: Quality Assurance/Quality Control
Exhibit 5-2
National Pesticide Survey Quality Assurance Plans
OVERALL QUALITY ASSURANCE PROGRAM PLAN (QAPP)
1. Quality Assurance Program Plan for the National Pesticide Survey of Drinking Water
Wells, 1988, ICF Incorporated.
ANALYTICAL QUALITY ASSURANCE PROJECT PLANS (QAPjPs)
2. Quality Assurance Project Plan for the National Pesticide Survey of Drinking Water
Wells: Analytical Method 1 - Phosphorus Pesticides and Analytical Method 3 -
Chlorinated Acid Herbicides, Montgomery Laboratories, September 1990.
3. Quality Assurance Project Plan for the National Pesticide Survey of Drinking Water
Wells: Analytical Method 2 ~ Organochlorine Pesticides, Clean Harbors
Incorporated, November 1989.
4. Quality Assurance Project Plan for the National Pesticide Survey of Drinking Water
Wells: Analytical Method 4 - Carbamates, Radian Corporation (no date).
5. Quality Assurance Project Plan for the National Pesticide Survey of Drinking Water
Wells: Analytical Method 5 - Methylcarbamates, ES&E, July 1990.
6. Quality Assurance Project Plan for the National Pesticide Survey of Drinking Water
Wells: Analytical Method 6 - ETU, Battelle - Columbus, April 1990,
7. Quality Assurance Project Plan for the National Pesticide Survey of Drinking Water
Wells: Analytical Method 7 - Fumigants, ES&E, July 1990.
8. Quality Assurance Project Plan for the National Pesticide Survey of Drinking Water
Wells: Analytical Method 9 - Nitrate/Nitrite, Montgomery Laboratories, May 1990.
9. Quality Assurance Project Plan for the National Pesticide Survey of Drinking Water
Wells: Referee Analyses for Analytical Method 1 - Phosphorus Pesticides,
Environmental Chemistry Laboratory of the Office of Pesticide Programs - USEPA,
December 1989.
10. Quality Assurance Project Plan for the National Pesticide Survey of Drinking Water
Wells: Referee Analyses for Analytical Method 3 - Chlorinated Acid Herbicides,
Environmental Chemistry Section of the Office of Pesticide Programs - USEPA,
December 1989.
11. Quality Assurance Project Plan for the National Pesticide Survey of Drinking Water
Wells: Referee Analyses for Analytical Method 6 - ETU, Environmental Chemistry
Section of the Office of Pesticide Programs - USEPA, December 1989.
12. Quality Assurance Project Plan for the National Pesticide Survey of Drinking Water
Wells: Referee Analyses for Method 2 - Organochlorine Pesticides, Method 4 -
Carbamates, Method 5 - Methylcarbamates, Method 7 - Fumigants, and
Method 9 - Nitrate/Nitrite, Technical Support Division, Office of Drinking Water and
Risk Reduction Engineering Lab, Office of Research and Development, USEPA.
National Pesticide Survey: Phase I Report
-------
Chapter Five: Quality Assurance/Quality Control 47
Exhibit 5-2 (continued)
National Pesticide Survey Quality Assurance Plans
IMPLEMENTATION QUALITY ASSURANCE PROJECT PLANS
13. Quality Assurance Project Plan for the National Pesticide Survey of Drinking Water
Wells: Hydrogeologic Characterization and Second-Stage Stratification Activities,
1988, rev. 1990, ICF Incorporated.
14. Quality Assurance Project Plan for the National Pesticide Survey of Drinking Water
Wells: Well Sampling, Data Collection, and Processing, 1988, rev. 1990, ICF
Incorporated.
15. Quality Assurance Project Plan for the National Pesticide Survey of Drinking Water
Wells: Survey Statistics, Data Collection and Processing, 1990, ICF Incorporated
and Westat Incorporated.
5.2 Quality Control
EPA defines quality control (QC) as the overall system of activities whose purpose is to control the
quality of a product so that it meets the needs of its users. Quality assurance is distinguished from QC in that
QA is an oversight function and therefore examines and monitors the implementation and effectiveness of
quality control. For the NFS, information was generated through physical or chemical measurements,
collection of opinions/facts through the use of questionnaires, and from the manipulation of electronically
stored data. Quality control activities were used in the NFS to set up, monitor, and adjust information-
generating processes to maintain or improve their quality. Quality control took place while the activities were
occurring and was performed by those individuals who ran the processes. Both analytical and implementation
activities were subject to stringent quality control procedures.
5.2.1 Analytical QC
Depending on the method, each laboratory had to meet specific quality control criteria with respect
to initial demonstration of capabilities, method blanks, calibrations, surrogate recoveries, internal standard
responses, instrument control standards, lab control standards, control charts, and lab spike samples. They also
were required to obtain standard materials from NFS, report in a standardized format on magnetic media, and
pass an audit for compliance with required QC criteria for all data sets.
5.2.2 Implementation QC
For the purposes of QC, implementation activities were placed into three categories: well sampling,
second stage stratification, and statistics, including questionnaires and data processing. A number of quality
control tools were applied to all three categories, including training, development of consistent procedures,
expert evaluation of intermediate Survey products (questionnaires, maps, telephone screeners), and monitoring
of the progress of the activities to identify and resolve any quality-related problems.
5.3 Audits
The NFS quality assurance program relied very heavily on auditing during implementation. When
audits identified deficiencies, corrective action was either taken immediately or coordinated with the technical
monitors, depending on the nature of the problem. Implementation of corrective action was monitored and
National Pesticide Survev: Phase I Reoort
-------
48 Chapter Five: Quality Assurance/Quality Control
then verified at the next audit. Five types of audits were conducted: Technical System Audits (TSAs), Audits
of Data Quality (ADQs), Performance Evaluation (PE) Studies, Close-out Audits, and Internal Audits. Each
type of audit is described in this section.
Technical System Audits (TSAs). TSAs were conducted for field, laboratory, and office activities,
using the quality assurance project plan for that activity or process as the standard against which the activities
were audited. The auditors reviewed each organization's internally-managed quality assurance efforts in
addition to their facilities and management systems for the project, data, and documentation. The laboratories
also were evaluated against generally accepted good laboratory practices. A team of auditors was used
whenever possible. Typically the audit team was composed of QAO(s) and technical monitors. Following each
audit, a draft report was written, reviewed by audit participants (including a representative of the audited
organization), revised according to review comments, and then forwarded to the Survey Director. Technical
system audits were scheduled every six months, or more frequently, if necessary.
Audits of Data Quality (AOQs). ADQs were conducted primarily on laboratory data. The procedure
involved tracking a water sample from the time it was received by the laboratory, through analysis, and finally
to the reporting of the analytical result to the Survey database manager. All QC criteria were checked and
calculations were verified. These audits were performed at the same time as the technical system audits (i.e.,
every six months).
Performance Evaluation (PE) Studies. Both the contract analytical laboratories and the EPA
laboratories participated in performance evaluation studies. Standards, consisting of a mixture of analytes or
of single analytes, depending on the method, were prepared in an appropriate solvent, verified, and shipped
to the laboratories in sealed glass ampules with directions for preparation as an aqueous sample. Laboratories
then were checked for correct identification and quantification of the compounds in the standard.
Performance evaluation studies were conducted on a quarterly basis and were carried out for all analytical
methods.
Close-out Audits. Close-out audits were required of every organization that participated in NPS as
a check that all work had been completed and all technical supporting materials had been organized and filed
in a manner consistent with the Survey policy on archival records (i.e., materials must be managed so that they
are readily accessible for reference purposes). Close-out audits were performed following the cessation of
NPS-sponsored activities.
Internal Audits. Internal audits were audits of operations conducted by QA staff employed by the
organizations actually performing the work. The auditors therefore were more familiar with procedures,
policies and problems inherent to organizations than could be the team of auditors led by the NPS QAO.
Major responsibility for quality assurance was thus delegated to the lowest organizational level possible. The
frequency with which these audits were held was identified in each of the quality assurance project plans and
varied among organizations according to the extent of their role in NPS and available resources. Results of
the audits were summarized in monthly progress reports and discussed during NPS systems audits.
5.4 Corrective Action Summary
Audits typically identified discrepancies that required corrective actions, such as a process
improvement, rather than a need to stop work due to an out-of-control situation. On several occasions,
problems were identified that required Survey management to intervene to correct deficiencies. Ultimately,
corrective action was not used to assign blame but rather was viewed as an effort to improve quality. The
balance of this section describes the types of discrepancies and corrective actions identified in the audits. The
types of discrepancies identified in Technical System Audits ranged in seriousness from those that had already
compromised data quality to those that were simply questionable procedures. The number of problems in the
first category were very few and consisted of issues such as exceeded sample holding times, improper sample
National Pesticide Survey: Phase I Report
-------
Chapter Five: Quality Assurance/Quality Control 49
storage, and incorrect interpretation of questionnaire terms. Much more numerous were those problems that
had the potential to result in the loss of data, such as computer files that had not been archived, a calibration
standard that was no longer accurate, and a vendor supply problem with teflon-lined caps for sample bottles.
These situations were identified as potential problems and corrected before they could result in data loss.
Data audits, because they were an intensive review of the result generation process, were very
productive for verifying that procedures were followed exactly as the procedures were described in the quality
assurance project plan or were specified during the Technical Systems Audit. Typically, the laboratories were
found to be following their procedures as described, and the problems that were identified were very specific,
such as incorrect interpretation of the time storage analysis scheme or the lack of analysis of a required quality
control sample.
PE studies were useful for demonstrating laboratory capabilities with respect to identifying and
quantifying NFS analytes and proved to be an excellent tool for identifying problems that had previously gone
undetected. Over the course of six PE studies, the laboratories successfully identified more than 98 percent
of the analytes, as shown in Exhibit 5-3. Also shown in Exhibit 5-3 are the number of PE observations that
met study criteria for quantification. Overall, 93 percent of the laboratory results met study criteria, a slightly
lower percentage than for correct identification.
An example of a problem that was detected through the performance evaluation studies was a
misidentification of the elution order for two compounds on the Method 2 confirmatory column. On another
occasion, a problem existed with consistently quantifying a Method 3 analyte accurately, even though all the
quality control indicators for the method met the criteria. The primary laboratories worked cooperatively with
the technical monitors and EPA laboratories to resolve these problems.
Close-out audits were a powerful tool for enforcing requirements for long-term management of Survey
data. Unlike all previous audits, which were conducted with an end goal of process improvement, the close-
out audits were a mechanism to ensure that organizations pulled together all materials they had generated for
NPS, filed them in an organized fashion, and maintained their safety and security.
Many of the deficiencies discovered by the auditors during the close-out audits were corrected at that
time. Any remaining issues were corrected later and verification of the correction was made by the on-site
QA personnel. A typical example of a deficiency that was identified on these audits was the absence of a
complete record. Corrective action, therefore, entailed locating the files and physically placing them with all
other NPS materials. An index to the records was required, and often this was not generated until the auditor
made the request in person. Given the follow-up that typically occurred subsequent to the close-out audits,
these audits were extremely successful for achieving the desired end result, i.e., that a complete record be
maintained in an organized fashion for several years under safe and secure conditions.
-------
50 Chapter Five: Quality Assurance/Quality Control
Exhibit 5-3
Summary of Performance Evaluation Results
Performance Evaluation Study
1988 1989 1990
Number of Analytes* 18 26 28 27 20 22
Number of Observations** 36 52 51 46 34 38
Number within Criteria for:
Identification 36 52 48 45 34 38
Quantification 34 51 48 38 34 35
Overall percentage correctly identified: 98 percent
Overall percentage correctly quantified: 93 percent
PE studies cover all analytic methods.
Number of observations includes results from analyses performed at both the contract and EPA referee labs.
National Pesticide Survey: Phase I Report
-------
Chapter Six: Findings and Results
Introduction
This chapter presents Survey findings and results based upon the data obtained from the analysis of
well water samples and the data obtained from Survey questionnaires collected and analyzed by the NFS.
These results are stated in terms of the estimated number or proportion of community water system wells and
rural domestic wells in the United States with a particular characteristic. These findings have been calculated
from the results of the analysis of water samples obtained from 540 CWS wells and 752 rural domestic wells.
Because the Survey's national findings depend on a multi-stage sampling design, weighted estimates of the
number and percent of wells are presented, along with measures of the estimates' accuracy.
Results are presented in the order of EPA's objectives and Survey design specifications. These
objectives and specifications are also crucial to interpreting the Survey's key findings correctly. EPA's first
objective was to estimate the frequency and concentration of pesticides and nitrate in drinking water wells on
a national level. EPA's second objective was to collect sufficient data to examine the relationships between
the presence of pesticides and nitrate in drinking water wells and patterns of pesticide use and ground-water
vulnerability. EPA established precision criteria for national level estimates for rural domestic wells and
community water system wells to meet the first objective. The 127 NPS analytes were selected to provide a
national-level assessment of the presence in drinking water wells of those chemicals that had been detected
in local monitoring studies, were suspected of leaching to ground water, or had major agricultural use patterns.
EPA, however, did not design the Survey to target areas where specific pesticides are used or "hot spots" where
specific pesticides had previously been detected in ground water.
National estimates in this chapter are presented in terms of an estimate with upper and lower limits
for analytical data; such limits are also presented in Appendix D for questionnaire data. These limits define
the "confidence interval" and reflect the relative certainty of population estimates that are based on data from
the wells sampled by the Survey. They indicate the upper and lower limits of the range that EPA is reasonably
confident contains the actual detection rate for wells nationally. Estimates in this chapter are presented with
a 95 percent confidence interval. Confidence intervals for numbers and percents of population estimates are
not always symmetrical (i.e., they are not an equal amount above and below the estimate) because of their
underlying statistical distribution.
The NPS provides estimates for five important subgroups of U.S. wells, in addition to estimating the
rates of detection of pesticides in U.S. wells nationwide. EPA defined these subgroups of wells (domains of
interest) in areas of high pesticide use, vulnerable ground water, and intensive agricultural activity. The Survey
design also provided that if specific analytes occurred frequently, estimates of the occurrence and concentration
of these individual chemicals could be provided for the nation. EPA did not design the Survey, however, to
estimate the frequency and concentration of individual analytes.
In contrast to the NPS, other surveys have been designed to produce estimates of the presence of one
or more specific pesticides in particular regions of the country. For example, Monsanto and the Research
Triangle Institute, with the active participation of EPA, designed the National Alachlor Well Water Survey
(NAWWS) to estimate the proportion of private, rural domestic wells with detectable concentrations of the
herbicide alachlor. The NAWWS was designed to estimate contamination in counties where alachlor was used
rather than in the nation as a whole. The Iowa State-Wide Rural Well-Water Survey was designed to provide
statistically valid estimates, both State-wide and within regions of the State, of the quality of water supplies.
-------
52 Chapter Six: Findings and Results
This chapter is divided into six sections. Section 6.1 presents estimates of the number of CWS wells
and the number of rural domestic wells in the United States containing nitrate or at least one pesticide. Well
population estimates and proportions of wells are based upon detections of any pesticide and nitrate in the
community water system wells and rural domestic wells included in the Survey.
Section 6.2 presents national estimates for specific NPS analytes. Section 6.2.1 presents estimates for
the three chemicals detected most frequently in the Survey -- nitrate, the pesticide degradate DCPA acid
metabolites, and the pesticide atrazine. Section 6.2.2 presents national estimates for ten additional pesticides
that were detected in the Survey above Survey Minimum Reporting Limits (MRLs), but which were detected
less frequently than nitrate, DCPA acid metabolites, and atrazine. Finally, Section 6.2.3 presents national
estimates for the presence in drinking water wells of analytes that were not detected in the Survey. Wells
nationwide may contain other NPS analytes not detected during the Survey for several reasons. The Survey
used a stringent series of minimum reporting limits and laboratory confirmation requirements to ensure a high
degree of certainty about the detections that were reported. Additional NPS analytes may have been present
in the wells sampled at levels too low to be reported as detections by the Survey. In addition, the Survey
sampled a limited number of wells in randomly selected geographic areas of the country. Public and private
wells in local areas with high levels of NPS analytes may not be included in the Survey due to random
selection. EPA will conduct extensive analysis of the Survey results to understand if these factors affected
Survey findings. Research plans for these additional analyses are discussed in Section 6.6.
Section 6.3 presents data on the concentrations of chemicals detected most frequently in water
samples, and compares those concentrations when possible to regulatory and health standards established by
EPA to protect human health. Two such standards, Maximum Contaminant Levels (MCLs) and Lifetime
Health Advisory Levels (HALs), are discussed. Although MCLs apply by law only to community water
systems, they also are used in this chapter as a standard of quality for source drinking water for rural domestic
wells. It also is important to note that while these standards pertain to water after treatment, the water
samples collected and analyzed in the Survey were obtained from CWS wells and rural domestic wells before
water from these wells had undergone any treatment. NPS field teams purged wells from 10 to 30 minutes
before taking samples, to ensure that samples were not stagnant water from the piping system or treated water
that had back-siphoned into the well. EPA also conducted intensive monitoring of Survey field teams and
sampling records to ensure that samples were collected prior to water treatment. Sample results indicating
any violations of sampling procedures were deleted from the final data base used to prepare the estimates
presented in Chapter 6. Although nitrate, DCPA acid metabolites, and atrazine were all detected with
sufficient frequency to enable statistically valid analysis of concentration levels to be completed for them, only
nitrate was detected with sufficient frequency to provide enough cases for comparing its measured
concentration with regulatory and health standards. Section 6.3.1 presents concentration data for nitrate, and
compares nitrate concentration to EPA health standards. Section 6.3.2 presents concentration data for DCPA
acid metabolites and atrazine. Section 6.3.2 also presents a consolidated national estimate for the presence
of Survey analytes in drinking water wells above the health-based limits. Comparisons of these chemicals with
EPA health-based standards are based on the upper bound of a 95% confidence interval.
Section 6.4 presents a summary of survey-derived information about the construction and condition
of CWS wells and rural domestic wells, and about pesticide use, farming activities, and other characteristics
of the areas surrounding the wells. This information was obtained from the Survey questionnaires
administered to well owners or operators and local area experts and is presented on a national basis.
Section 6.5 describes Survey results for certain subgroups of the population (domains of interest) of
drinking water wells that were identified as being of special interest to the Survey. Five subgroups are
described: community water system wells in counties with high ground-water vulnerability, rural domestic wells
in counties with high ground-water vulnerability, rural domestic wells in counties with high pesticide use, rural
domestic wells in counties with high pesticide use and high ground-water vulnerability, and rural domestic wells
in "cropped and vulnerable" portions of counties. Section 6.5 presents data on these subgroups of wells that
are derived both from the analysis of well water samples and from questionnaires.
National Pesticide Survey: Phase I Report
-------
Chapter Six: Findings and Results 53
Section 6.6 describes the analyses that EPA is planning to conduct in Phase II of the Survey of the
relationships between data on pesticide and nitrate detections and various categories of information about
wells gathered during the Survey, such as pesticide use, well depth, and geologic characteristics.
6.1 National Estimates
One of the two major objectives of the National Pesticide Survey was to estimate the number of
drinking water wells in the nation that contain at least one pesticide or pesticide degradate. The Survey tested
approximately 1300 wells across the country for the presence of 101 pesticides, 25 pesticide degradates, and
nitrate1 (127 analytes). The Survey was designed to provide statistical estimates, using the results obtained
from those sampled wells, of the approximate number of wells in the United States containing at least one
pesticide or nitrate. This section presents national estimates of the frequency with which drinking water wells
contain any pesticide or nitrate. It also summarizes the information obtained by the Survey about the
occurrence of specific pesticides and pesticide degradates.
Section 6.1.1 provides summary national estimates of the number of drinking water wells nationwide,
in order to establish the baseline for the other estimates presented in this chapter. Section 6.1.2 presents esti-
mates for the number and proportion of community water system wells and rural domestic wells that contain
at least one Survey analyte. These estimates are based on survey data for the 112 quantifiable NPS analytes.
(The remaining 15 analytes could not be reliably measured and therefore are described as qualitative analytes.)
The analytical results presented in this report reflect the detection limits of the contract laboratories
that conducted the analytical determinations for the Survey. In all, 17 analytes were detected. Thirteen were
detected at levels above the NPS MRLs used by the contract laboratories that performed the initial analysis
of samples; three were detected by EPA laboratories at concentrations lower than the respective contract
laboratory MRLs. Because EPA laboratories analyzed only about 10% of all sampled wells, such detections
are not included for statistical analysis in this report, but will be discussed further in the NPS Phase II Report.
In several cases pesticide detections were used from the EPA laboratories to replace the contract laboratory
results. In these cases, the EPA laboratory detection limits were slightly higher than the contract laboratory
limits. One detected chemical could not be measured, and that detection also is not included for statistical
analysis. In the balance of this chapter, the term "detection" is used to mean a detection at concentrations at
or above the contract laboratory MRL. The MRL is defined as one-half of the Minimum Quantification Limit
(MQL) (see Section 4.3).
6.1.1 National Estimates of Wells
At the commencement of the design of the Survey, EPA estimated that in the United States there
were approximately 51,000 community water systems using ground water and, based on the 1980 census,
estimated that there were approximately 13 million rural domestic wells. Based on the Survey results, EPA
now estimates that there are approximately SS^OO2 community water systems with ground-water wells in the
United States, which together operate 94,600 wells (Exhibit 6-1). The number of CWS wells nationally could
be as high as 98300 or as low as 90,900, as indicated by the associated 95 percent confidence interval. The
number of rural domestic wells nationally is now estimated to be 10.5 million from data obtained from the
1987 American Housing Survey (AHS) conducted by the U.S. Bureau of the Census. Given the change in
rural populations since 1980, the 1987 AHS estimates provide a more accurate basis for estimating NPS results
1 Samples were analyzed for the combined presence of nitrate and nitrite, and reported as a single concentration in
milligrams per liter (mg/L) as nitrogen.
2 Estimates over 1,000 are rounded to three significant digits; estimates under 1,000 are rounded to two significant digits.
Percents are rounded to one decimal place.
National Pesticide Survey: Phase I Report
-------
54 Chapter Six: Findings and Results
Exhibit 6-1
Estimated Number of Drinking Water Wells in the United States
Domain Description
Community water system wells
Rural domestic wells
Estimated Number
94,600
10,500,000
95% Confidence Interval
(lower - upper)
(90,900 - 98,300)
than 1980 Census estimates. The AHS estimate is treated as a constant for the NFS, and therefore does not
have an associated error bound (95% confidence interval) to measure its precision. A detailed explanation
of these adjustments to the Survey's initial population estimates is presented in Appendix B.
6.1.2 National Estimates for Pesticides and Nitrate
The NFS obtained water samples from 566 CWS wells and 783 rural domestic wells. After technical
review of individual well plumbing systems and sample collection procedures, however, not all water samples
met NPS quality control standards. Therefore, this chapter presents summary results based on samples from
the 540 CWS wells and 752 rural domestic wells whose samples met the Survey's quality assurance/quality
control criteria.
The national estimates in this chapter of the number and proportion of all CWS wells and rural
domestic wells that contain Survey analytes are derived from the Survey data using appropriate weights. That
is, the Survey was designed so that the number of detections of analytes in the 540 CWS wells and 752 rural
domestic wells are used as the basis of a calculation that weights them according to precise factors that account
for the multi-stage Survey design and provides estimates representative of the 94,600 CWS wells and 10.5
million rural domestic wells in the United States.3
A graphic summary of the weighted estimates of the percent of CWS wells and rural domestic wells
with nitrate and pesticides above Survey minimum reporting limits is presented in Exhibit 6-2. This exhibit
shows the estimated percent of wells nationally that contain detections (above the MRLs) of pesticides only,
nitrate only, and nitrate plus pesticides. The percent indicating wells with nitrate and pesticides represents
the national estimate of wells containing both pesticides and nitrate.
3 The "weight" of a survey well corresponds loosely to the number of wells it represents in the population and is equal
to the inverse of its probability of selection for the Survey. A more detailed discussion of the actual calculations is given in
Appendix B.
Klofinnol Daet!/*!HA Ciirwaw Dhaca I Don/ift
-------
Chapter Six: Findings and Results 55
Exhibit 6-2
Estimated Percent of Community Water System Wells and
Rural Domestic Wells in the United States with Nitrate and/or Pesticides
Community Water System Wells
Rural Domestic Wells
Without Nitrate or
Pesticide* (44.6%)
Wlin Nitrate Only
(46.0%)
Without Mitral* or
Pesticide* (42.0%)
With Nitrate Only
(53.8%)
With Pesticides
Only (3.3%)
With Nitrate and
Pesticides (7.1%)
With Pesticides
Only (1.0%) with Nitrate and
Pesticides (3.2%)
National Estimates for Specific Pesticides
This section provides national estimates for the number of wells in the United States that contain
pesticides, based on the pesticide detections in the Survey. In order to provide national estimates for the
presence of pesticides in drinking water wells nationwide that are both statistically valid and developed on a
standard basis, only those pesticide detections measured by the five contract laboratories are included in the
calculation of national estimates. Survey results shown in Exhibit 6-3 indicate that:
About 9,850 (10.4%) of the CWS wells in the United States are estimated to contain
concentrations above the MRL for at least one pesticide. Considering the related confidence
interval,4 this estimate could be as low as 6330 or as high as 13,400; and
About 446,000 (4.2%) of the rural domestic wells in the United States are estimated to
contain concentrations above the MRL for at least one pesticide; this estimate could be as
low as 246,000 or as high as 647,000.
4 For most of the estimates presented io this chapter, the underlying statistical distribution is binomial (e.g., a well water
sample either does or does not test positive for anaJyte detection). When the sample size is large enough and the estimated
percent is not near zero or 100, confidence intervals are based on the normal distribution approximation to the binomial
distribution. When the sample size is not large enough to justify the normal approximation, however, confidence intervals
for the analytical results are based on the binomial distribution instead of the normal approximation. Confidence intervals
based on the binomial distribution generally are not symmetric. A more detailed discussion of the actual calculations is given
in Appendix B.
-------
56 Chapter Six: Findings and Results
Exhibit 6-3
Estimated Number and Percent of Drinking Water Wells in the
United States Containing at Least One Pesticide
Domain Description
Community water system wells
Rural domestic wells
95%
Confidence
Estimated Interval
Number (lower - upper)
9,850 (6,330 - 13,400)
446,000 (246,000 - 647,000)
95%
Confidence
Estimated Interval
Percent (lower upper)
10.4 (6.8 - 14.1)
4.2 (2.3 - 6.2)
National Estimate for Nitrate
Nitrate is the analyte most frequently detected in both CWS wells and rural domestic wells, and was
detected at a much higher frequency than specific pesticides and pesticide degradates. Survey results indicate
that over half of the CWS wells and rural domestic wells in the United States contain levels of nitrate at or
above the survey's MRL of 0.15 mg/L (Exhibit 6-4).
Exhibit 6-4
Estimated Number and Percent of Wells
in the United States Containing Nitrate
Domain Description
Community water system wells
Rural domestic wells
95%
Confidence
Estimated Interval
Number (lower - upper)
49,300 (45,300 - 53,300)
5,990,000 (5,280,000 - 6,700,000)
95%
Confidence
Estimated Interval
Percent (lower - upper)
52.1 (48.0 - 56.3)
57.0 (50.3 - 63.8)
National estimates for nitrate are presented separately because of the differences between nitrate and
the other analytes. Nitrate (NO3") is not regulated under the Federal Insecticide, Fungicide, and Rodenticide
Act (FTFRA). Nitrate can occur naturally or as a result of human activity. It is found in some fertilizers or
as a result of nitrification of ammonium fertilizers. Nitrate is produced in soil through nitrification and
mineralization. Nitrogen is present in discharges from septic systems, animal feed lots, industrial wastewaters,
and sanitary landfills. Nitrite (NO2") is an intermediate product of the conversion of ammonium (NH4+) to
nitrate or as the intermediate product of the conversion of nitrate to ammonium. Nitrate and nitrite were not
separately distinguished in the Survey laboratory method and testing, and the results presented reflect both
nitrate and nitrite, measured as nitrogen (N) in mg/L.
Exhibits 6-5 and 6-6 present graphic summaries of the national estimates for drinking water wells
containing at least one detected pesticide and for wells containing nitrate.
National Pesticide Survey: Phase I Report
-------
Chapter Six: Findings and Results 57
Exhibit 6-5
Estimated Percent of Community Water System Wells and
Rural Domestic Wells in the United States Containing at Least One Pesticide
Community Water System Wells
With At Least One Pesticide
(10.4%)
Without Pesticides
(89.6%)
Rural Domestic Wells
With At Least One Pesticide
(4.2%)
Without Pesticides
(95.8%)
Exhibit 6-6
Estimated Percent of Community Water System Wells and
Rural Domestic Wells in the United States Containing Nitrate"
Community Water System Well*
Rural Domestic Wells
Without Nitrate
(47.9%)
Without Nitrate
(43.0%)
With Nltrtt*
(52.1%)
With Nltrmt*
(57.0%)
* The minimum reporting limit for nitrate is 0.15 mg/L
-------
58 Chapter Six: Findings and Results
6.2 National Estimates for Pesticides
This section presents national estimates of the frequency of occurrence in drinking water wells in the
United States of the pesticides and pesticide degradates detected in the Survey and for all other pesticide
analytes. Section 6.2.1 provides estimates for the two most frequently detected pesticides, DCPA acid
metabolites and atrazine. Section 6.2.2 provides estimates for 10 other pesticides or pesticide degradates
detected at concentrations above their respective MRLs. Finally, Section 6.2.3 presents national estimates
(using the upper bound of a 95% confidence interval) for those pesticide analytes that were not detected.
6.2.1 National Estimates for Two Most Frequently Detected Pesticides: DCPA
Acid Metabolites and Atrazine
Two analytes - DCPA acid metabolites (degradates of DCPA) and the pesticide atrazine - were the
most frequently detected pesticide analytes. This section presents estimates of the national frequency of
occurrence of these two analytes in drinking water wells.
As Exhibit 6-7 indicates, approximately 6,010 CWS wells and 264,000 rural domestic wells are
estimated to contain DCPA acid metabolites. These numbers could be as high as 8,840 and 477,000 or as low
as 3,170 and 129,000. Approximately 1,570 CWS wells and 70,800 rural domestic wells are estimated to
contain atrazine. These numbers could be as high as 2,710 and 214,000 or as low as 420 and 13300 for CWS
and rural domestic wells.
Exhibit 6-7
Estimated Number and Percent of Wells in the United States
Containing DCPA Acid Metabolites or Atrazine8
Domain Description
Community water system wells:
DCPA acid metabolites
Atrazine
Rural domestic wells:
DCPA acid metabolites
Atrazine
Estimated
Number
6,010
1,570
264,000
70,800
95%
Confidence
Interval
(lower - upper)
(3,170-8,840)
(420-2,710)
(129,000 - 477,000)
(13,300 - 214,000)
Estimated
Percent
6.4
1.7
2.5
0.7
95%
Confidence
Interval
(lower - upper)
(3.4 - 9.3)
(0.5 - 2.9)
(1.2-4.5)
(0.1 - 2.0)
* The minimum reporting limit for DCPA acid metabolites is 0.10 /ig/L and the minimum reporting limit for atrazine is 0.12 /tg/L
Dimethyl tetrachloroterephthalate (DCPA), a selective herbicide registered in 1958, is the parent
compound of DCPA acid metabolites. DCPA, which was not detected in the Survey, is sold under the trade
name Dacthal W-75 and has been sold under the discontinued trade names of DAC-893 and Fatal. The
primary use of DCPA, more than 80%, is on turf such as golf courses and home lawns. It is used to combat
smooth and hairy crabgrass, witchgrass, green and yellow foxtails, fall panicum, and other annual grasses. It
is also used against certain broadleaf weeds such as carpet weed, dodder, purslane, and common chickweed,
and is tolerated by many crop plants. It is presently approved for use on turf, ornamentals, strawberries,
collards, cucumbers, eggplant, garlic, horseradish, kale, mustard greens, onions, peppers, potatoes, radish,
seeded melons, squash, sweet potatoes, tomatoes, turnips, and soybeans. DCPA acid metabolites have a high
potential to migrate through soil due to their high water solubility.
National Pesticide Survey: Phase I Report
-------
Chapter Six: Findings and Results 59
Atrazine is the common name of an herbicide which is a member of the chemical family of triazines.
Atrazine was registered in the late 1950's. It has been sold under the trade names of Atrazine, AAtrex,
Atratol, Gesaprim, and Zeaphos. Atrazine is a component of other herbicides such as Bicep, Bullet,
Extrazine, Lariat, Marksman, Prozine, and Sutazine. Atrazine is used to control many annual broadleaf weeds
and certain grasses in corn, sorghum, sugarcane, macadamias, and subtropical tree fruits such as guavas and
pineapples. It is also used for general weed control on non-cropped industrial land, selective weed control
in conifer restoration and Christmas tree plantations, and non-selective control of vegetation on fallow land.
Atrazine has a high potential to migrate through soil due to its water solubility and relatively long half-life.
Exhibits 6-8 and 6-9 present graphic representations of the proportion of CWS wells and rural domestic wells
nationally containing DCPA acid metabolites, and the proportion of drinking water wells containing atrazine.
6.2.2 National Estimates for All Analytes Detected
In addition to DCPA acid metabolites and atrazine, 10 other pesticides or pesticide degradates were
detected above NFS minimum reporting limits. With some exceptions, these analytes were detected at both
CWS wells and rural domestic wells. The following lists identify all analytes detected above minimum
reporting limits and whether they were detected at a CWS well or rural domestic well:
CWS wells: atrazine, DCPA acid metabolites, dibromochloropropane, dinoseb,
hexachlorobenzene, nitrate, prometon, and simazine; and
Rural domestic wells: alachlor, atrazine, bentazon, DCPA acid metabolites,
dibromochloropropane, ethylene dibromide, ethylene thiourea, gamma-HCH, nitrate,
prometon, and simazine.
The following exhibits (Exhibits 6-10 and 6-11) provide national estimates of the number and
proportion of the 94,600 community water system wells and 10.5 million rural domestic wells containing these
pesticides (national estimates for nitrate, DCPA acid metabolites, and atrazine also are included for ease of
comparison).
In addition to the pesticides discussed in this section, for which estimates are shown in Exhibits 6-10
and 6-11, three analytes (alpha-chlordane, gamma-chlordane, and beta-HCH) were detected by the EPA
laboratories at concentrations much lower than the respective MRLs used by the contract laboratories. The
EPA laboratories had the task of verifying the results of the contract laboratories, and in many cases were able
to detect analytes at concentrations lower than the MRLs established for the Survey. National estimates
cannot be produced for these analytes because the EPA laboratories did not analyze water samples from every
well. In addition, one analyte, 4-nitrophenol, was detected, but national estimates could not be produced due
to limitations of the laboratory analytic method for this chemical.
Exhibit 6-12 provides information regarding the predominate uses of the pesticides and pesticide
degradates detected in the Survey and their behavior in the environment. Seven of the 14 analytes have a high
potential to migrate in soil. Of these 14 pesticides, 9 have an MCL and 8 have a Lifetime HAL
National Pesticide Survey: Phase I Report
-------
60 Chapter Six: Findings and Results
Exhibit 6-8
Estimated Percent of Community Water System Wells and
Rural Domestic Wells Containing DCPA Acid Metabolites3
Community Water System Wells
With DCPA Acid Metabolites
(6.4%)
Without DCPA Acid Metabolites
(93.6%)
Rural Domestic Wells
With DCPA Add Metabolites
(2.5%)
Without DCPA Acid Metabolites
(97.5%)
* Minimum reporting limit = 0.10 jig/L
Exhibit 6-9
Estimated Percent of Community Water System Wells and
Rural Domestic Wells Containing Atrazine3
Community Water System Wells
With Auazine
(1.7%)
Without Atrazine
(98.3%)
Rural Domestic Wells
With Atrazine
(0.7%)
Without Atrazine
(99.3%)
* Minimum reporting limit = 0.12 jig/L.
National Pesticide Survey: Phase I Report
-------
Chapter Six: Findings and Results 61
Exhibit 6-10
Estimated Number and Percent of Community Water System Wells
Containing NFS Analytes
Analyte
Nitrate
DCPA acid metabolites
Atrazine
Simazine
Prometon
Hexachlorobenzene8
Dibromochloropropane (DBCP)8
Dinoseb*
Estimated
Number
49,300
6,010
1,570
1,080
520
470
370
25
95% Confidence
Interval
(lower - upper)
(45,300 - 53,300)
(3,170-8,840)
(420-2,710)
(350 - 2,540)
(78-1,710)
(61 - 1,630)
(33 - 1,480)
(1 - 870)
Estimated
Percent11
52.1
6.4
1.7
1.1
0.5
0.5
0.4
95%
Confidence
Interval
(lower - tipper)
(50.0 - 56.3)
(3.4 - 9.3)
(0.5 - 2.9)
(0.4 - 2.7)
(0.1 - 1.8)
(0.1-1.7)
(<0.1 - 1.6)
<0.1 (<0.1 - 0.9)
a Registration cancelled by EPA.
b Numbers between zero and 0.05 are reported as less than 0.1 (<0.1).
Exhibit 6-11
Estimated Number and Percent of Rural Domestic Wells
Containing NPS Analytes
Analyte
Nitrate
DCPA acid metabolites
Atrazine
Dibromochloropropane
(DBCP)8
Prometon
Simazine
Ethylene dibromide8
gamma-HCH (Lindane)
Ethylene thiourea (ETU)
Bentazon
Alachlor
Estimated
Number
5,990,000
264,000
70,800
38,400
25,600
25,100
19,200
13,100
8,470
7,160
3,140
95% Confidence
Interval
(lower - upper)
(5,280,000 - 6,700,000)
(129,000 - 477,000)
(13,300 - 214,000)
(2,740 - 164,000)
(640 - 142,000)
(590-141,000)
(160 - 131,000)
(14 - 120,000)
(1 -111,000)
(1 - 109,000)
(1 - 101,000)
Estimated
Percent"
57.0
2.5
0.7
0.4
0.2
0.2
0.2
0.1
0.1
0.1
95%
Confidence
Interval
(lower
upper)
(50.3 - 63.8)
(1.2-4.5)
(0.1 - 2.0)
(<0.1 - 1.6)
(<0.1 -1.4)
(<0.1 -1.3)
(<0.1 -1.2)
(<0.1 -1.1)
(<0.1 -1.1)
(<0.1 -1.0)
<0.1 (<0.1 - 1.0)
a Registration cancelled by EPA.
b Numbers between zero and 0.05 are reported as less than 0.1 (<0.1).
National Pesticide Survev: Phase I Reoort
-------
62 Chapter Six: Findings and Results
Exhibit 6-12
Key Characteristics and Uses of Detected Pesticides
Analyte
Alachlor
Atrazine
Bentazon
Chlordane
DCPA acid metabolites
Dibromochloropropane (DBCP)b
Dinoseb0
Ethylene Dibromide (EDB)d
Ethylene thiourea (ETU)e
Hexachlorobenzeneb|t
Gamma-HCH (Lindane)
4-Nrtrophenol
Prometon
Simazine
Type
Herbicide
Herbicide
Herbicide
Insecticide
Herbicide
(degradate)
Herbicide
Herbicide
Insecticide
Fungicide
(degradate)
Fungicide
Insecticide
Insecticide
(degradate)
Herbicide
Herbicide
Potential
to Migrate
In Soil*
Medium
High
High
Low
High
High
Medium
High
Medium
Medium
Medium
High
High
Medium
Keyltoe
Controls annual grasses and weeds in
corn, soybeans, and peanuts
Controls certain weeds in com, sorghum,
sugarcane, pineapple, and citrus fruits
Controls broadleaf weeds in soybeans, rice,
corn, peanuts, beans, peas, and mint
Controls termites in homes and other
insects such as fire ants around
underground cables
Controls annual grasses in turf
ornamentals, fruits, and vegetables
Controls nematodes on berries, citrus,
melons, and nuts (cancelled)
Controls potato vines and desiccating seed
crops (cancelled)
Controls insects in soil (cancelled) and is
an additive in leaded gasolines
Breakdown product of EBDC fungicides
(cancelled)
Controls fungus on wheat (cancelled)
Controls leafhoppers in lowland rice and
wood inhabiting beetles
Degradate of parathion pesticides which
control a variety of insects such as aphids
and mosquitoes on pears and apples
Controls perennials, broadleaf weeds, and
grasses in non-crop areas
Controls annual grasses and weeds in
crops especially corn and fruit such as
citrus, asparagus, and nuts
a Migration potential based on an assessment of available data on physical and chemical properties, including water
solubility, soil adsorption coefficients, and half-life data.
b Cancelled.
c Commercial use suspended in 1986.
d Cancelled in 1984.
e EPA proposed the cancellation of many of the uses of the parent compounds in 1989.
f Also a contaminant of several chlorinated pesticides.
National Pesticide Survey: Phase I Report
-------
Chapter Six: Findings and Results 63
6.2.3 National Estimates for Analytes Not Detected
Many of the 127 Survey analytes were not detected in water samples collected in the Survey. These
analytes, however, may or may not be present in drinking water wells in the United States. In the NPS,
because only a relatively small group of wells in the population were sampled, wells containing specific
pesticides might not have been selected for participation in the Survey. Estimates for these analytes were
developed based on the upper bound of a 95% confidence interval. If EPA had sampled every well in the
population (i.e., conducted a census), the presence of some pesticides or nitrate above an MRL would probably
have been detected.
Exhibit 6-13 presents estimates for the maximum numbers and percents of CWS wells and rural
domestic wells potentially containing at least one pesticide that was not detected in the Survey. The estimates
in this exhibit suggest, for example, that no more than 750 (0.8%) CWS wells in the United States contain
pesticides not detected in the Survey. The estimate of 0.8% is a 95% upper confidence bound for the
proportion of wells containing at least one pesticide that was not detected in the Survey.5
Exhibit 6-13
Estimates of the Maximum Number of Drinking Water Wells
Containing Pesticides Not Found in the Survey
Domain Description
Community water system wells
Rural domestic wells
Estimated Number
750
83,100
Estimated Percent
0.8
0.8
6.3 Estimated Concentrations of Frequently Detected Analytes and Comparison
to Regulatory and Health Limits
This section presents concentration levels of those analytes that were detected in sampled well water
with sufficient frequency so that the statistical measures for the concentrations can be provided. It presents
four measures of concentration: the minimum concentration, which corresponds to the minimum reporting
limit for the analyte; the maximum concentration detected; the median concentration, above and below which
there are an equal number of observations; and the weighted geometric mean and its confidence interval. The
geometric mean is an appropriate statistic for data where the observations tend to cluster at the lower values
(i.e., are positively skewed).
Section 6.3.1 presents concentration estimates for nitrate, the most frequently detected analyte. DCPA
acid metabolites and atrazine were detected with sufficient frequency in the Survey so that statistically reliable
analyses of concentration levels of these chemicals in drinking water wells in the United States could be
completed. Section 6.3.2 presents estimated concentration levels for the two analytes.
For additional details on the calculation, see Section B6.4 of Appendix B.
-------
64 Chapter Six: Findings and Results
This section also presents estimates of the number of wells in the United States containing nitrate at
concentration levels that exceed certain regulatory or health-based levels for nitrate. Nitrate was the only
analyte that was detected often enough to produce an estimate of the relationship of detected concentrations
to regulatory and health levels for that analyte. DCPA acid metabolites were not detected at concentrations
above the HAL of 4,000 /*g/L (ppb) and atrazine was not detected often enough at concentrations above its
MCL and HAL of 3 /*g/L to provide reliable estimates. Estimates of the numbers of wells in the United States
with at least one pesticide above health-based standards also are presented. This topic will be investigated in
the Phase II Report.
6.3.1 Estimated Nitrate Concentrations
As Exhibit 6-14 shows, the median concentration for nitrate in the wells in which it was detected is
1.6 mg/L for both CWS wells and rural domestic wells. The geometric means for nitrate concentrations in
CWS and rural domestic wells also were estimated for the United States. The geometric mean concentration
of nitrate in CWS wells and rural domestic wells is 1.3 and 1.4 mg/L, respectively. The 95% confidence
interval for these estimates indicates that the geometric means could be at any point between 1.1 to 1.5 mg/L
and 1.1 to 1.8 mg/L, respectively. As Exhibit 6-14 shows, both the geometric mean and the median are below
the health-based standard of 10 mg/L for nitrate. The geometric mean and confidence intervals are not based
on the normal distribution (see Appendix B for additional discussion).
Exhibit 6-14
Concentrations of Nitrate Detected in Wells Sampled by NFS
Domain Description
Community water system wells
Rural domestic wells
Concentration (mg/L)a
Minimum
Reporting
Limit
0.15
0.15
95%
Geometric Confidence
Mean Interval
1.3 (1.1-1.5)
1.4 (1.1 - 1.8)
Median
1.6
1.6
Maximum
13
120
a The geometric mean estimate is based on weighted national data, and the median and maximum are based on sample
data only.
It is important to note, with respect to the concentrations reported in Exhibit 6-14, that although
detections were not publicly reported with a measured concentration between the MRL and the MQL of an
analyte, these concentrations were used in the calculation of the values presented in Exhibit 6-14. TTie median
and maximum are based on sample data only. In contrast, the geometric mean estimate is based oriVeighted
national data. Details of calculations of the geometric means and confidence intervals are provided in
Appendix B.
Exhibit 6-15 presents the estimated frequency distributions for nitrate concentrations for CWS wells\
and rural domestic wells in the United States. The frequencies, which are based on weighted survey results, \
are national estimates for each concentration interval or group presented. For instance, EPA estimates that
for CWS wells nationally that contain nitrate above the Survey detection level of 0.15 mg/L, 42.8% contain
nitrate at concentrations between 0.15 and 1.15 mg/L (centered at 0.65 mg/L) and for rural domestic wells
nationally that contain nitrate above the Survey detection level, 83.5% contain nitrate at concentrations
between 0.15 and 5.15 mg/L (centered at 2.65 mg/L). For both CWS wells and rural domestic wells nationally,
most observations are at concentrations below the MCL/HAL for nitrate of 10 mg/L.
National Pesticide Survey: Phase I Report
-------
Chapter Six: Findings and Results 65
Exhibit 6-15
Estimated Frequency Distribution of Nitrate Concentrations
of Community Water System Wells and Rural Domestic Wells Sampled by NPS"
C/3
o
*->
c
50-
40-
0)
$ 30-
20-
10-
0-
42.8
Community Water
System Wells
20.8
12.3
5.7
4.2
5.1
I1IIII11I I T ~TT
0.65 1.65 2.65 3.65 4.65 5.65 6.65 7.65 8.65 9.65 10.65 11.65 12.65
Nitrate Concentrations in milligrams per liter
100
00
§ 40-|
I
0
83.5
Rural Domestic
Wells
12.5
L6 L6 0.1 0.5 0.1 0.1
2.65 7.65 12.65 17.65 22.65 27.65 32.65
Nitrate Concentrations in milligrams per liter
> 35.15
Percent of wells based on water samples containing nitrate levels greater than or equal to the minimum reporting
limit of 0.15 mg/L Numbers presented for nitrate concentrations are mid-points of the frequency intervals. For
example, concentrations between 0.15 and 1.15 mg/L are centered at 0.65 mg/L
._.**... m.-_-
-------
66 Chapter Six: Findings and Results
Nitrate in ground water can originate from sources such as rain (as well as acid rain) and normal
biological activities. The major source of nitrate in cultivated soils is from inorganic fertilizers containing
nitrogen. Nitrogen fertilizers are applied to soils to enhance plant growth and nitrogen is necessary in the
synthesis of plant proteins. Other sources of nitrate in soil and water include septic systems, animal wastes,
plant residues, and fixation from the atmosphere. Exhibit 6-16 compares nitrate concentration to the
regulatory standard established under the Safe Drinking Water Act (SDWA) for nitrate. The standard used
as the basis of comparison is the Maximum Contaminant Limit (MCL) of 10 mg/L, which in the case of nitrate
equals the Health Advisory Level (HAL).6 The MCL is a regulatory standard, legally applicable under the
SDWA only to public water systems. It represents the maximum level of a contaminant that may be contained
in water delivered by a piped system of supply serving 15 or more connections or a population of at least 25
persons. MCLs apply to water at the tap, following treatment, and are not applicable to surface or ground-
water sources of drinking water. Although the MCL is not legally applicable to rural domestic wells, it was
used for this analysis as a standard of quality for source drinking water. HALs represent the maximum
concentration of a contaminant in water that may be consumed safely over an average human lifetime. EPA
calculated national estimates of the number of wells containing concentrations of nitrate based on the sample
data and national estimates of nitrate detections at levels above the health-based standard of 10 mg/L. The
health effects caused by these chemicals are described in Health Advisory Summaries prepared by EPA.
Exhibit 6-16
Estimated Number and Percent of Wells with Nitrate Contamination in the
United States Above the Health-Based Standard
Domain Description
Community water system wells
Rural domestic wells
Nitrate Estimate Above MCL/HAL*
Estimated
Number
1,130
254,000
95%
Confidence
Interval
(lower - upper)
(370 - 2,600)
(128,000-464,000)
95%
Estimated Confidence
Percent Interval
(lower - upper)
1.2
2.4
(0.4 - 2.7)
(1.2-4.4)
8 The MCL and HAL for nitrate are both 10 mg/L
Survey results include several important findings concerning the frequency with which the
concentrations of nitrate in drinking water wells may exceed health-based standards. EPA estimates that:
More than half of the CWS wells in the United States (about 52.1% or 49,300 wells) contain
detectable levels of nitrate. About 1.2% of all CWS wells (approximately 1,130 wells) contain
nitrate concentrations that exceed the MCL/HAL of 10 mg/L; and
6 The Health Advisory Level (HAL) is defined as the level of exposure to contaminants at which adverse health effects
would not be expected to occur in most members of the population. Unlike MCLs, which are enforceable standards, HALs
are non-enforceable health-based guidelines. HALs are calculated for various periods of exposure, including one day, 10 days,
and longer term (e.g., few months to lifetime); lifetime HALs are used in this report. HALs are developed using data related
to non-carcinogenic toxic effects. Some analytes have Risk Specific Doses (RSDs) instead of or in addition to HALs. The
RSD represents the concentration of a chemical in drinking water that, if consumed for a lifetime, would result in an excess
lifetime cancer risk of one in one million.
National Pesticide Survey: Phase I Report
-------
Chapter Six: Findings and Results 67
About 57.0% of rural domestic wells in the United States (approximately 5,990,000 wells)
contain detectable levels of nitrate; about 2.4% of all rural domestic wells (about 254,000
wells) contain nitrate levels that exceed the MCL/HAL.
From the measured concentration data, EPA found that:
The maximum concentration of nitrate detected in CWS wells was approximately 13 mg/L;
and
The maximum concentration of nitrate detected in rural domestic wells was approximately
120 mg/L.
6.3.2 Estimated Pesticide Concentrations
DCPA acid metabolites and atrazine were the only pesticides detected with sufficient frequency for
reliable concentration estimates to be developed based on their detections. Exhibit 6-17 presents those
estimates. Although detections were not publicly reported with a measured concentration between the MRL
and the MQL for DCPA acid metabolites and atrazine, these concentrations were used in the calculation of
the geometric mean, and the median and the maximum. In contrast to the median and maximum, which are
calculated from sample data, the geometric mean is calculated from weighted national data.
As Exhibit 6-17 indicates:
In sampled CWS wells, maximum concentrations of DCPA acid metabolites and atrazine were
approximately 7.2 and 0.92 /xg/L, respectively.
In sampled rural domestic wells, maximum concentrations of DCPA acid metabolites and
atrazine were approximately 2.4 and 7.0 ^g/L, respectively.
Exhibit 6-17
Concentrations of DCPA Acid Metabolites and
Atrazine Detected in Wells Sampled by NPS
Domain Description
Community water system
wells:
DCPA acid metabolites
Atra2ine
Rural domestic wells:
DCPA acid metabolites
Atrazine
Concentration (jug/L)a
Minimum
Reporting
Limit
0.10
0.12
0.10
0.12
95%
Geometric Confidence
Mean Interval
0.56 (0.35 - 0.89)
0.40 (0.20 - 0.81)
0.37 (0.22 - 0.63)
0.44 (0.18-1.04)
Median
0.34
0.26
0.38
0.28
Maximum
7.2
0.92
2.4
7.0
a The geometric mean estimate is based on weighted national data, and the medium and maximum are based on
sample data only.
National Pesticide Survey: Phase I Report
-------
68 Chapter Six: Findings and Results
As Exhibit 6-17 indicates, the geometric mean concentrations of DCPA acid metabolites in CWS wells
and rural domestic wells throughout the United States are 0.56 and 0.37 /*g/L, respectively. The geometric
mean for CWS wells could be as low as 0.35 fig/L or as high as 0.89 /*g/L, as indicated by the 95% confidence
interval. Similarly, the geometric mean for rural domestic wells could be as low as 0.22 /ig/L or as high as 0.63
/xg/L. Fifty percent of DCPA acid metabolite detections in CWS wells and rural domestic wells were less than
the median values of 0.34 and 0.38 /tg/L, respectively.
The geometric mean concentrations of atrazine detections in CWS wells and rural domestic wells are
0.40 and 0.44 figfL, respectively. The geometric mean for CWS wells could be as low as 0.20 /itg/L or as high
as 0.81 fjig/L, as indicated by the 95% confidence interval. Similarly, the geometric mean for rural domestic
wells could be as low as 0.18 jtg/L or as high as 1.04 /ig/L. Fifty percent of atrazine detections in CWS wells
and rural domestic wells were at concentrations less than the median values of 0.26 and 0.28 pg/L, respectively.
The confidence intervals for the geometric means are calculated for each well population individually;
they do not account for the relationship between CWS and rural domestic well detections. Statistical tests to
determine the extent of the differences between geometric means for CWS wells and rural domestic wells will
be performed in Phase II.
Finally, EPA calculated estimates of the number and percent of community water system wells and
rural domestic wells in the United States with at least one pesticide detection above health-based standards.
Exhibit 6-18 presents these estimates.7
Exhibit 6-18
Estimated Number and Percent of Wells in the United States with at Least
One Pesticide Detection Above Health-Based Standard3
Domain Description
CWS wells above MCL/HAL
Rural domestic wells above
MCL7HAL
95%
Confidence
Estimated Interval
Number (lower - upper)
0 (0 - 750)
60,900 (9,430-199,000)
95%
Confidence
Estimated Interval
Percent (lower - upper)
0.0 (0.0 - 0.8)
0.6 (0.1 - 1.9)
* MCLs are not legally applicable to rural domestic wells.
As Exhibit 6-18 illustrates:
At most 0.8% (750 wells) of CWS wells in the United States contain at least one pesticide
above health based standards; and
Of rural domestic wells in the United States, approximately 0.6% (60,900) have
concentrations above their respective MCLs8 and HALs. This percent could be as low as
0.1% or as high as 1.9%, as indicated by the associated 95% confidence interval.
7 These calculations assume that the probability of detection follows a binomial distribution, and takes into account the
complex sample design by using appropriate weights. A more detailed discussion of the actual calculations is given in
Appendix B.
8 This estimate is based on those pesticides that have an associated MCL or HAL (see Appendix E, Exhibit E-l).
National Pesticide Survey: Phase I Report
-------
Chapter Six: Findings and Results 69
6.4 Characteristics of CWS Wells and Rural Domestic Wells
The second major purpose of the NPS is to investigate the relationship between the presence of
pesticides and nitrate in drinking water wells and a number of factors potentially affecting that presence, such
as pesticide use, ground-water vulnerability, well construction and maintenance, and farming activities
conducted in the vicinity of wells. In order to gather data on these factors, several detailed questionnaires
were administered to the owners or operators of wells, householders using rural domestic wells, farmers, and
local area experts such as county agricultural extension agents. This section presents national estimates of the
characteristics of CWS wells and rural domestic wells, based on the data compiled from the field
questionnaires. Copies of these questionnaires are provided in Appendix D. In addition, detailed compilations
of the results for key questionnaire items are provided in Appendix D, with indications of the particular
questionnaires and questions from which the data are compiled.9 Exhibits in Appendix D also include the
95% confidence intervals for these data.
Section 6.4.1 discusses characteristics of CWS wells;
Section 6.4.2 discusses characteristics of rural domestic wells; and
Section 6.4.3 contrasts CWS well and rural domestic well characteristics.
These characteristics are described to help interpret the analytical results of the Survey. Phase II of
the NPS will investigate further the association among nitrate detections, pesticide detections, pesticide use,
well depth, age of well, and geologic characteristics. The Phase I Report is not intended to determine
statistical differences between related items. The confidence intervals presented are for single characteristics
only and do not take into account the relationship between variables.
6.4.1 Characteristics of CWS Wells
National estimates for CWS wells based on the responses to key questionnaire items are presented
in Appendix D,10 which presents descriptive statistics in three broad categories: (1) characteristics of the
well and surrounding area; (2) fanning and land use; and (3) pesticide use. This section presents highlights
of the descriptive information collected in that appendix.
Questionnaire responses show that about 15,900 CWS wells, which is equivalent to about 16.8% of
CWS wells nationally, belong to systems with one working well (see Item 1.1 in Exhibit D-l in
Appendix D).11 Some additional statistics of interest in Exhibit D-l include the following:
9 The data are compiled from the CWS and Domestic Well Team Leader Introduction (TU) and Well Observation
Record (WOR), The CWS and Domestic Well Main Questionnaires (CWS Main and DWS Main), and the CWS and
Domestic Local Area Questionnaires (LAQ).
10 In Appendix D, for those questions where the possible Survey responses are mutually exclusive and exhaustive, the
number and percent estimates sum exactly to the domain population size and one hundred percent, respectively. Rounding
errors presented with the questionnaire responses can cause the summations to appear inexact. A few questionnaire items
do not contain a mutually exclusive or exhaustive list of possible responses. The sum of estimates for these questionnaire
responses should not be expected to correspond to the domain population sizes or one hundred percent. All confidence
intervals in Appendix D are based on the normal approximation to the binomial distribution.
11 Numbers corresponding to responses have been included for reviewing Appendix D for ease of comparison between
the domains of interest. These numbers do not, however, correspond to numbering schemes used in any Survey
questionnaires. Some questionnaire items contain skip patterns (i.e., instructions to skip particular questions based on
previous answers). A questionnaire item is presented as a single item, including sub-items resulting from skip patterns in the
questionnaires which are separated with a blank line.
National DoctiHriA Survew PhflftB I Ranrkrt
-------
70 Chapter Six: Findings and Results
Almost half (44.5%) of all CWS wells are in systems with only one or two wells (see Item
i.i);
About 35.6% of the CWS wells nationally are more than 20 years old (see Item 1.6);
For about 72.4% of the wells, the land area within 300 feet of the well is primarily exposed
or vegetated soil (e.g., permeable to precipitation) (see Item 1.4);
Most of the wells (83.9%) are capped at the ground surface (see Item 1.10); concrete pads
(about 78.0%) and sanitary or grouted seals (about 69.2%) are used alone or in combination
for ground surface protection at the majority of the well sites (see Item 1.5); and most wells
(about 99.0%) are cased, about 46.2% of cased wells are cased to their total depth (see Item
1.9), and about 69.6% of cased wells are grouted (see Item I.9);12
Approximately 56.2% of the wells are deeper than 200 feet, whereas about 38.4% of the wells
are less than 200 feet deep (see Item 1.8). For the remaining wells the response to the
question was "don't know" (see footnote 12); and
Approximately 43.5% of the wells tap confined aquifers (see Item 1.12).
Farming and Land Use Near CWS Wells
Survey questionnaire data indicate that:
About 7.4% of the properties where CWS wells are located were fanned in the last 5 years
(see Item 1.18);
About 40.3% of the wells are located within 1/2 mile of land on which crops have been
farmed within the last 3 years (see Item 1.22); approximately 41.0% of the wells are located
within 1/2 mile of land used as pasture within the last 3 years (see Item 1.23);
About 9.5% of the wells are located within 1/2 mile of a golf course; most (about 83.3%) are
located within 1/2 mile of a surface water body (see Items 1.27 and 1.28); and
Approximately 25.0% of the wells are located within 1/2 mile of irrigated land (see Item 1.29).
Pesticide Use Near CWS Wells
Key questionnaire results on pesticide use include:
About 7.4% of all CWS wells are located on farm property; pesticides are used at
approximately 34.4% of these farm properties (see Item 1.18);
About 41.0% of CWS wells are located within 1A mile of pasture lands; pesticides were used
within the last 3 years on approximately 57.2% of these pasture lands (see Item 1.23). This
corresponds to about 23.5% CWS wells in the United Stales (confidence intervals for
calculation of national estimates for such "nested" data will be provided in Phase II);
12 Many of the questionnaire items listed in the exhibits in Appendix D contain "don't know" as a possible response.
Summary results presented in the text treat this response as being exclusive of all other possible responses. For instance, the
statement that "approximately 69.6% of cased CWS wells are grouted" does not allow for the approximately 18.9% "don't
know" response. The "don't know" responses affect Survey estimates. In this example, since a well is either grouted or not
grouted, then considering this form of measurement error only, the percent of cased CWS wells that are grouted could lie
between 69.6% and 88.5%. The questionnaire item responses as presented portray measurement error through the "don't
know" response and sampling error through the 95% confidence interval.
National Pesticide Survey: Phase I Report
-------
Chapter Six: Findings and Results 71
Approximately 25.5% of CWS wells nationally are located on properties where non-farm
pesticides have been used within the last 3 years; at about 73.5% of these properties,
pesticides have been used within 100 feet of the well (see Item 1.14). This corresponds to
about 18.7% CWS wells in the United States (confidence intervals for calculation of national
estimates for such "nested" data will be provided in Phase II); and
At about 19.8% of the CWS wells, major non-farm pesticides13 (e.g., spraying for gypsy
moth or mosquitoes) were applied within 1/2 mile of the well in the last 3 years (see
Item 1.24).
6.4.2 Characteristics of Rural Domestic Wells
Statistics for questionnaire variables for rural domestic wells nationally also are presented in
Appendix D, Exhibit D-3. Rural domestic well characteristics, farming and land use, and pesticide use are
highlighted below:
For most rural domestic wells nationally (about 91.1%), the land area within 300 feet of the
well is primarily exposed or vegetated soil (e.g., permeable to precipitation) (see Item 1.3);
Most (79.8%) rural domestic wells are closed (i.e., capped) at the ground surface (see Item
1.9); about 85.8% are cased, and about 47.5% of cased wells are cased to their total depth
(see Item 1.8);
Most wells (87.7%) have not been redrilled (see item 1.6); and
Approximately 22.4% of the wells are deeper than 200 feet, whereas about 62.6% of the wells
are less than 200 feet deep (see Item 1.7). For the remaining wells the response to the
question was "don't know" (see footnote 12).
Farming and Land Use Near Rural Domestic Wells
Survey questionnaire data indicate that:
About 11.7% of rural domestic wells nationally are located on properties that are farmed (see
Item LIT);14
Approximately 66.2% of the wells are located within 1/2 mile of land where crops have been
fanned within the last 3 years (see Item 1.21); about 65.4% of the wells are located within 1/2
mile of land used as pasture within the last 3 years (see Item 1.22);
About 5.0% of rural domestic wells nationally are located within 1/2 mile of a golf course;
approximately 93.8% are located within 1/2 mile of a surface water body (see Items 1.27
and 1.28); and
About 18.1% of the wells are located within 1/2 mile of irrigated land (see Item 1.29).
13 The NFS defined farm pesticides as all chemicals used to control insects, plants, fungi, nematodes, and rodents.
Farmed property is a place where the sale of crops or animals, exceeds an amount of $1,000 and includes land in a land
bank if the farmer could have earned $1,000 or more.
14 EPA estimated the number of rural domestic wells located on form property in the United States (1,220,000)
from the 1987 American Housing Survey.
National Pesticide Survey: Phase I Reoort
-------
72 Chapter Six: Findings and Results
Pesticide Use Near Rural Domestic Wells
Key questionnaire results on pesticide use include:
Farm pesticides have been used in the last five years at about 83.6% of fanned properties that
have well sites (see Item 1.18);
Pesticides have been applied in the last three years to 41.1% of the pasture land located
within Yi mile of wells (see Item 1.23);
Non-farm pesticide use on the properties on which rural domestic wells are located includes
use inside the home (about 75.7%), use on lawns (about 40.0%), and garden use (about
41.9%) (see Items 1.12,1.13, and 1.14);
At about two thirds of the rural domestic well sites (67.7%), non-farm pesticides have been
stored on the property for at least one month in the last three years; in 72.0% of these wells
they were stored within 100 feet of the well (see Item 1.15); and
At about 13.9% of the wells, major non-farm pesticides (e.g., spraying for gypsy moth or
mosquitoes) were applied within 1/2 mile of the well in the last 3 years (see Item 1.24).
6.4.3 Summary of CWS and Rural Domestic Well Characteristics
One of the objectives of the Survey is to explore the relationships among variables for CWS wells and
rural domestic wells. These two populations of drinking water wells have certain common characteristics; a
number of differences between them, however, were captured by some of the questionnaire data that were
collected about both CWS wells and rural domestic wells. The following exhibits organize data in three
categories - well setting, well construction and condition, and land and pesticide use near wells. Exhibits 6-19,
6-20, and 6-21 provide graphic comparisons showing estimated percents of specific questionnaire responses
on a national basis for these three categories. The confidence intervals presented in Appendix D for CWS
wells and rural domestic wells are calculated for each well characteristic individually; they do not account for
the relationship among variables. Caution should be exercised in drawing comparisons between related items
until appropriate statistical tests are performed during Phase II. Responses to questionnaire items frequently
included "don't knows." "Don't know" responses are not included in population estimates reporting a given
characteristic. Consequently, these comparisons should be reviewed with caution.
National Pesticide Survey: Phase I Report
-------
Chapter Six: Findings and Results 73
Key results illustrated by these exhibits include the following:
Well Setting
Rural domestic wells are approximately four times as likely as community water systems to
have septic units on the property where the well is located (Exhibit 6-19); and
Land area within three hundred feet of both CWS wells and rural domestic wells is
predominantly exposed or vegetated soil (Exhibit 6-19).
Exhibit 6-19
Comparison of Well Setting
Septic Units on Property
100%
80%-
c 60%-
u
£ 40%-
20%-
0%
m_
??%
O Septic Tink
D Septic Field
0 Geupoal
Community Water
System Wells
Rural Domestic
Wells
Land Area Within 300 Feet of Well
ioo%-
80% -
60%-
s. *>%
20%-
0%
D RooCed
D Pived
Q Gnvetod
D Expottd/Vc(. SoU
Otter
Community Water
System Well*
Rural Domestic
Wells
National Pesticide Survey: Phase I Report
-------
74 Chapter Six: Findings and Results
Well Construction and Condition
Community Water System wells tend to be older than rural domestic wells; approximately
35.6% of CWS wells are more than 20 years old, while about 24.0% of rural domestic wells
are more than 20 years old (Exhibit 6-20); and
CWS wells tend to be deeper than rural domestic wells; about 22.4% of rural domestic wells
are deeper than 200 feet whereas about 56.2% of CWS wells are deeper than 200 feet
(Exhibit 6-20).
Exhibit 6-20
Comparison of Well Construction and Condition
Age of Well
30%-
c
£ 20%-
o.
10%-
J7
y,
t.
t
rl
£
1:
&
&
;!:
.ft'
=
~
'/f
V
=55
SI
1
1;
1
s
q
30%-
^
c
n 20ycvs
B Dart Know
I
Well Depth
5 $
\\
%
n
/
:!
<\
*
/
;/
i /
. /
?
9
D 500 ft
5 Don't Know
Community Water Rural Domestic Community Water Rural Domestic
System Wells
Wells
System Wells
Wells
Land and Pesticide Use
Approximately 7.4% of CWS properties were used for farming in the last 5 years, compared
to about 11.7% for rural domestic well properties (Exhibit 6-21);
Pesticides were used more frequently on fanned properties where rural domestic wells are
located during the past 5 years than on farmed properties where CWS wells are located
(Exhibit 6-21);
Non-farm pesticides are stored on CWS properties much less frequently than on rural
domestic well properties (Exhibit 6-21);
Farming within 1/2 mile occurs much more frequently near rural domestic wells than CWS
wells (Exhibit 6-21);
Approximately 9.5% of CWS wells and 5.0% of rural domestic wells are located within 1/2
mile of a golf course (Exhibit 6-21); and
Approximately 83.3% of CWS wells and 93.8% of rural domestic wells are located within 1/2
mile of a body of water (river, canal, bay, pond) (Exhibit 6-21).
National Pesticide Survey: Phase I Report
-------
Chapter Six: Findings and Results 75
Exhibit 6-21
Comparison of Land and Pesticide Use
Property Farmed in the Last 5 Years
100%-
80%-
5 60%
S. 40%
20%-
0%
Pesticides Used on Farmed Property
in the Last 5 Years
(reported percent of firmed property only)
100%-
Q Finned
Q Not rvimj
B Boot Know
80%-
e 60%-
0- 40%-
20%-
0%
I
"1
O Itod
O NOUnt
O DotftKnow
Community Water
System Wells
Rural Domestic
Wells
Community Water
System Wells
Rural Domestic
Wells
Non-farm Pesticides Stored for
More Than 1 Month in the Last 3 Years
100%-
80%-
60%-
£ 40% -j
D Stored
20% H D Not Stored
DonlKnou
Crops Farmed Within 1/2 Mile of Well in
the Last 3 Years
Q.
100%
80%
60%'
40%'
20%-
0%'
D Yes
D No
B Dent Know
Commumty Waiar
System Wells
Rural Domestic
Wells
Community Water
System Wells
Rural Domestic
Wells
Presence of Golf Course
Within 1/2 Mile of Well
Presence of Water Bodies
Within 1/2 Mile of Well
100%
80%
J 60%-
S. 40%
20%
0%
D Ye»
Q No
S DcotKao*
100%
80%
| 60%'
o
a 40%
D Yw
D No
B DcatKBDv
Community Water
System Wells
Rural Domestic
Wells
Community Water
System Wells
Rural Domestic
Wells
-------
76 Chapter Six: Findings and Results
6.5 Populations of Domains of Special Interest
The Survey findings and results presented in this chapter to this point have been for all CWS and
rural domestic wells in the United States. In addition to national estimates for all wells, however, EPA also
designed the NFS to provide summary statistics for five domains (subgroups) of the entire population of
drinking water wells that are of special interest to EPA These domains were selected to provide data to
better understand how the presence of pesticides or nitrate in drinking water wells is associated with patterns
of pesticide use and ground-water vulnerability. The five domains are:
CWS wells in counties with high ground-water vulnerability as measured by DRASTIC (first
stage strata 1, 4, 7, and 10, see Chapter 2);
Rural domestic wells in counties with high ground-water vulnerability as measured by
DRASTIC (first stage strata 1, 4, 7, and 10, see Chapter 2);
Rural domestic wells in counties with high pesticide use (first stage strata 1, 2, and 3);
Rural domestic wells in counties with high pesticide use and high ground-water vulnerability
as measured by DRASTIC (first stage strata 1); and
Rural domestic wells in "cropped and vulnerable" portions of counties (second stage
stratification).
The categories of results reported in this section parallel the categories used to report national
estimates. First, Section 6.5.1 provides estimates for the numbers of wells nationwide, that belong to each of
the five domains. Section 6.5.2 provides estimates for the number of wells in the United States in each domain
containing at least one detected analyte. Finally, Section 6.5.3 provides information on well characteristics.
6.5.1 Well Populations of Domains of Interest
EPA prepared estimates of the five domains of interest (subgroups) of CWS wells and rural domestic
wells based on the national estimates for all wells. (Because the five domains are not mutually exclusive, and
because they do not exhaust all categories of wells in the United States, the domain estimates in Exhibits 6-22
through 6-28 do not sum to the national estimates.) The breakdown of CWS wells and rural domestic wells
by the Survey's domains of interest is shown in Exhibit 6-22, which indicates that the following percentages
of rural domestic wells were located in the listed domains:
about 11.0% (1,160,000) are located in counties with high pesticide use:
approximately 27.3% (2,870,000) are located in counties with high ground-water vulnerability;
about 35.1% (3,690,000) are located in "cropped and vulnerable" county subregions; and
about 3.3% (342,000) are located in counties having high pesticide use and ground-water
vulnerability.
These estimates are the basis for preparing the results of the five domain estimates for detected
analytes and well characteristics discussed below. Use caution when trying to compare drinking water wells
from the different domains. The confidence intervals do not account for possible relationships among
variables or domains and in addition the domains are not independent. Therefore, trying to conclude that
estimates in different domains are different is very difficult without conducting additional statistical tests.
Statistical tests that account for relationships among variables will be conducted in the Phase II report.
National Pesticide Survey: Phase I Report
-------
Chapter Six: Findings and Results 77
Exhibit 6-22
Estimated Number and Percent of Community Water System Wells
and Rural Domestic Wells Belonging to Domains of Interest9
Domain Description
Community water system wells:
In counties having the
highest average vulnerability
Rural domestic wells:
In counties with highest
average pesticide use
In counties with highest
average vulnerability
In "cropped and vulnerable"
county subregions
In counties having the
highest average pesticide
use and ground-water
vulnerability
Estimated
Number
94,600
20,800
95%
Confidence
Interval
(lower - upper)
(90,900 - 98,300)
(19,800-21,800)
10,500,000
1,160,000
2,870,000
3,690,000
342,000
(831,000-1,480,000)
(2,380,000 - 3,370,000)
(2.890,000 - 4,490,000)
(223,000-461,000)
Estimated
Percent
100.0
22.0
95%
Confidence
Interval
(lower - upper)
N/A
(20.8 - 23.2)
100
11.0
27.3
35.1
3.3
(7.9 - 14.1)
(22.7 - 32.0)
(27.5 - 42.8)
(2.1 - 4.4)
a The five domains are not mutually exclusive, and because they do not exhaust all categories of wells, the domain
estimates do not sum to the national estimates. Use caution when trying to compare drinking water wells from the
different domains. The confidence intervals do not account for possible relationships among variables or domains
and in addition the domains are not independent.
6.5.2 Estimates for Detected Analytes
This section presents estimates for the various domains (subgroups) of special interest in the Survey.
National estimates for community water system wells and rural domestic wells are presented for ease of
comparison. Exhibit 6-23, which shows the estimated number and percent of wells by domain containing
concentrations of nitrate above health-based standards, indicates that in general lower proportions of rural
domestic wells containing nitrate are found in domains with high pesticide use. The following exhibits provide
information about the subgroup of interest:
Estimated number and percent of wells by domain containing concentrations above MRLs
of at least one pesticide (Exhibit 6-24);
Estimated number and percent of wells by domain containing concentrations above health-
based standards of at least one pesticide (Exhibit 6-25);
Estimated number and percent of wells by domain containing DCPA acid metabolites
(Exhibit 6-26);
Estimated number and percent of wells by domain containing atrazine (Exhibit 6-27); and
Estimates, given by the 95% upper confidence bound of the maximum number of wells by
domain containing concentrations of pesticides above the MRLs for pesticides not found in
the Survey (Exhibit 6-28).
National Pesticide Survey: Phase I Report
-------
78 Chapter Six: Findings and Results
o
O
i^
| 5.
1
5
j
i
!*
1
*
i
2
i
i
$
5
_j
i
-1
6 9
$ fc
i
ii
i2
e-
6 f
i
0
1|
C ?
UJZ
1
S a"
* il
5 i
11
u
O 9
i
d
|l
E E
5 9
UJ
Domain Description
co
"f
WS wells nationally:
U
CO
^
»
in
6
s-
eg
o
S.
o
d
I
S.
0
if
5
^.
&
8
to
to'
§
£
§
N."
In counties having
the highest average
vulnerability
S
ca
r-
n
0
9
+
ft
*'
cvi
|
1
§
§
§
$
to
S
co
p
in
|
ca*
§
I
C.
§
in
lural domestic wells
atlonalry:
n c
S
,
ri
in
o
P
b
p
m
|
5
1
8
i
sr
?
o>
p
~
s
I
s
§
co
n
In counties with
highest average
pesticide use
8
N
*
in
b
^^
<6
v
e.
|
g"
8
w"
1
S
S'
S
o
a
CD
ID
g-
|
i
8
t.
§
u5
fj
"-""
In counties wtth
highest average ;
vulnerability
8
N
oi
m
6
CO
CO
"*
p
1
i
§
a
§
V
So-
S
o
iri
2,
CO
si
|
CN*
§
1
i
§
W
CD
*"
In 'cropped and
vulnerable* county
subreglons
8
n
ri
in
b
&
t
co
K
N
g-
CO
§
&
§
5
P
8
CO
c\i
K
o>
CO
s?
CM
g
0>*
§
JO
^.
In counties having
the highest average
pesticide use and
ground-water
vulnerability
I
I
1
.s
o
o
g
National Pesticide Survey: Phase I Report
-------
Chapter Six: Findings and Results 79
Exhibit 6-24
Estimated Number and Percent of Wells By Domain
Containing at Least One Pesticide
Domain Description
Community water system wells:
In counties having the highest
average vulnerability
Rural domestic wells:
In counties with highest
average pesticide use
In counties with highest
average vulnerability
In "cropped and vulnerable"
county subregions
In counties having the highest
average pesticide use and
ground-water vulnerability
Estimated
Number
9,850
1,930
446,000
41,100
79,700
202,000
4,820
95% Confidence
Interval
(lower - upper)
(6,330 - 13,400)
(1,200-2,670)
(246,000 - 647,000)
(2,980 - 79,200)
(9,260- 150,000)
(37,600 - 367,000)
(59 - 29,900)
Estimated
Percent
10.4
9.3
4.2
3.6
2.8
5.5
1.4
95% Confidence
Interval
(lower - upper)b
(6.8 - 14.1)
(5.8 - 12.9)
(2.3 - 6.2)
(0.2 - 6.9)
(0.4 - 5.1)
(0.9-10.1)
(<0.1 - 8.8)
a The five domains are not mutually exclusive and because they do not exhaust all categories of wells in the United States
the domain estimates do not sum to the national estimates. Use caution when trying to compare drinking water wells from
the different domains. The confidence intervals do not account for possible relationships among variables or domains and
in addition the domains are not independent.
b Numbers between zero and 0.05 are reported as less than 0.1 (<0.1).
National Pesticide Survey: Phase I Report
-------
80 Chapter Six: Findings and Results
Exhibit 6-25
Estimated Number and Percent of Wells By Domain
Containing Concentrations Above Health-Based Standards
of at Least One Pesticide8
Domain Description
Community water system wells:
In counties having the highest
average vulnerability
Rural domestic wells:
In counties with highest
average pesticide use
In counties with highest
average vulnerability
In 'cropped and vulnerable"
county subregions
In counties having the highest
average pesticide use and
ground-water vulnerability
Pesticides Above MCLs/HALs
Estimated
Number
0
0
60,900
3,140
6,280
3,140
0
95% Confidence
Interval
(lower - upper)
(0 - 750)
(0 - 430)
(9,430-199,000)
(1 - 40,400)
(1 -88,100)
(1 - 87,300)
(0-21,600)
Estimated
Percent
0.0
0.0
0.6
0.3
0.2
0.1
0.0
95% Confidence
Interval
(lower - upper) b
(0.0 - 0.8)
(0.0 - 2.0)
(0.1 - 1.9)
(<0.1 - 3.5)
(<0.1 -3.1)
(<0.1 - 2.4)
(0.0 - 6.3)
a The five domains are not mutually exclusive and because they do not exhaust all categories of wells in the United States
the domain estimates do not sum to the national estimates. Use caution when trying to compare drinking water wells from
the different domains. The confidence intervals do not account for possible relationships among variables or domains and
in addition the domains are not independent. This estimate is based only on the pesticides that have an associated MCL
or HAL
b Numbers between zero and 0.05 are reported as less than 0.1 (<0.1).
National Pesticide Survey: Phase I Report
-------
Chapter Six: Findings and Results 81
Exhibit 6-26
Estimated Number and Percent of Wells By Domain
Containing DCPA Acid Metabolites8
Domain Description
Community water system wells:
In counties having the highest
average vulnerability
Rural domestic wells:
In counties with highest
average pesticide use
In counties with highest
average vulnerability
In "cropped and vulnerable'
county subregions
In counties having the highest
average pesticide use and
ground-water vulnerability
Detected
Estimated
Number
6,010
1,390
264,000
9,150
42,700
145,000
0
95% Confidence
Interval
(lower - upper)
(3,170-8,840)
(790 - 19,800)
(129,000 - 477,000)
(210-51,000)
(5,420 - 149,000)
(56,700 - 300,000)
(0-21,600)
Estimated
Percent
6.4
6.7
2.5
0.8
1.5
3.9
0.0
95% Confidence
Interval
(lower - upper) b
(3.4 - 9.3)
(3.8 - 9.6)
(1.2-4.5)
(<0.1 - 4.4)
(0.2 - 5.2)
(1.5-8.1)
(0.0 - 6.3)
a The five domains are not mutually exclusive and because they do not exhaust all categories of wells in the United States,
the domain estimates do not sum to the national estimates. Use caution when trying to compare drinking water wells from
the different domains. The confidence intervals do not account for possible relationships among variables or domains and
in addition the domains are not independent.
b Numbers between zero and 0.05 are reported as less than 0.1 (<0.1).
National Pesticide Survey: Phase I Report
-------
82 Chapter Six: Findings and Results
Exhibit 6-27
Estimated Number and Percent of Wells By Domain
Containing Atrazine8
Domain Description
Community water system wells:
In counties having the highest
average vulnerability
Rural domestic wells:
In counties with highest
average pesticide use
In counties with highest
average vulnerability
In "cropped and vulnerable1
county subregions
In counties having the highest
average pesticide use and
ground-water vulnerability
Detected
Estimated
Number
1,570
83
70,800
21,600
19,500
55,400
4,820
95% Confidence
Interval
(lower upper)
(420-2,710)
(1 -610)
(17,900-189,000)
(3,220 - 70,900)
(400-112,000)
(6,120-105,000)
(59 - 29,900)
Estimated
Percent
1.7
0.4
0.7
1.9
0.7
1.5
1.4
95% Confidence
Interval
(lower upper)b
(0.5 - 2.9)
(<0.1 - 2.9)
(0.2-1.8)
(0.3 - 6.1)
(<0.1 - 3.9)
(0.2 - 7.8)
(<0.1 - 8.8)
a The five domains are not mutually exclusive and because they do not exhaust all categories of wells in the United States, the
domain estimates do not sum to the national estimates. Use caution when trying to compare drinking water wells from the
different domains. The confidence intervals do not account for possible relationships among variables or domains and in
addition the domains are not independent.
b Numbers between zero and 0.05 are reported as less than 0.1 (<0.1).
National Pesticide Survey: Phase I Report
-------
Chapter Six: Findings and Results 83
Exhibit 6-28
Upper Bound Estimates of the Number of Wells By Domain Containing
Pesticides Not Found in the Survey8
Domain Description
Community water system wells:
In counties having the highest
average vulnerability
Rural domestic wells:
In counties with highest average
pesticide use
In counties with highest average
vulnerability
In "cropped and vulnerable' county
subregions
In counties having the highest
average pesticide use and ground-
water vulnerability
Estimated Number
750
430
83,100
31,600
69,100
74,000
21,600
Estimated Percent
0.8
2.0
0.8
2.7
2.4
2.0
6.3
a The five domains are not mutually exclusive and because they do not exhaust all categories of wells in the United States, the
domain estimates do not sum to the national estimates. The upper bound estimates presented are 95% upper confidence bounds
based on the binomial distribution where the estimated proportion is zero. For further details see Section B6.4 of Appendix B.
Some items of interest for these exhibits include:
About 36.7% of CWS wells located in counties with high ground-water vulnerability contain
detectable levels of nitrate (Exhibit 6-23);
About 57.0% of rural domestic wells nationally contain detectable levels of nitrate; about
27.0% of rural domestic wells in counties with high pesticide use recorded nitrate detections
(Exhibit 6-23);
Approximately 5.5% of rural domestic wells in "cropped and vulnerable" county subregions
contain concentrations of at least one pesticide; about 1.4% in counties with high pesticide
use and ground-water vulnerability (Exhibit 6-24);
No more than 0.3% of rural domestic wells in the four domains of special interest contain
pesticide concentrations above health-based standards for those with MCLs and HALs
(Exhibit 6-25);
The estimated percent of rural domestic wells containing DCPA acid metabolites was no
more than 2.5% in all domains except in the "cropped and vulnerable" county subregions
domain (3.9%) (Exhibit 6-26);
The estimated percent of wells containing atrazine was less than 2% in all domains
(Exhibit 6-27); and
National Pesticide Survey: Phase I Reoort
-------
84 Chapter Six: Findings and Results
For analytes not detected by the NFS, the maximum estimated percent of wells containing
pesticides above the analytical detection limits set in the Survey ranged from 2.0% to 6.3%
for the five domains of special interest (Exhibit 6-28).
Well Characteristics
The Survey sought data for CWS wells in counties with high ground-water vulnerability, about 22.0%
of CWS wells are located in counties with high ground-water vulnerability. Detailed compilations of the
questionnaire results are provided in Appendix D, with indications of the particular questionnaire(s) and
question (s) from which the data are compiled. Estimates for well characteristics in counties with high ground-
water vulnerability were similar to the national estimates for the CWS survey.
The Survey also identified four domains of rural domestic wells of particular interest These additional
domains included rural domestic wells in counties: with highest pesticide use; highest average ground-water
vulnerability, highest average pesticide use and ground-water vulnerability, and "cropped and vulnerable"
county subregions. For wells in counties with high pesticide use, questionnaire items relevant to that domain
which differ substantially from the national estimates include "property farmed" (about 24.8% in high pesticide
use counties versus about 11.7% nationally); and "crops farmed within 1/2 mile of the well in the last 3 years"
(about 89.8% versus about 66.2% nationally).
For the domains corresponding to counties with high vulnerability and "cropped and vulnerable"
county subregions, estimates of parameters related to vulnerability were similar to the national estimates
presented in Section 6.4.2, although 79.7% of wells in "cropped and vulnerable" county subregions had crops
farmed within Vi mile of the well compared to 66.2% nationally.
For wells in the high ground-water vulnerability and high pesticide use counties, factors showing
important differences from all rural domestic wells may exist, but the large confidence intervals for the data
make comparisons too uncertain to be presented.
6.6 Additional Analyses to Appear in Phase II Report
The focus of Chapter 6 has been on presenting estimates of the number and proportion of wells in
the United States and in the five domains of special interest that contain detectable levels of nitrate and
pesticides. In addition, this chapter discussed summary statistics on key questionnaire items such as farm and
non-farm pesticide use on the property and well characteristics. The purpose of the NFS Phase II Report is
to present the findings of studies of interrelationships among data gathered in the questionnaires that may
affect detection rates, concentrations (e.g., well depth, well age, and well construction), and analyte detections.
Section 6.6.1 describes some of the relational analyses that will be conducted. In addition, the Phase II report
will expand upon results of the analysis EPA conducted to determine if the Survey's results were affected by
seasonal variations in the time when the water samples were obtained. Section 6.6.2 describes a preliminary
analysis of seasonal variation.
6.6.1 Relational Analyses
This section discusses the statistical studies that are planned for the Phase II Report. Numerous
computerized databases have been compiled from the questionnaires and other sources for the Survey to deter-
mine the association of well contamination with such factors as fertilizer and pesticide use on the property,
within one-half mile of the well, and within the county; and ground-water vulnerability characteristics including
depth to ground water, recharge, aquifer media, soil media, topography, impact of unsaturated zone, and hy-
draulic conductivity. The Phase II Report will analyze the questionnaire databases, the first-stage and second-
stage DRASTIC stratification scoring results, nitrogen fertilizer and pesticide sales, and the Survey analytical
results, to investigate the potential causes and consequences of pesticide residues in drinking water wells.
National Pesticide Survey: Phase I Report
-------
Chapter Six: Findings and Results 85
The Phase II Report will investigate possible reasons why the Survey detected the presence of no more
than 17 chemicals of the 127 studied. The Survey was designed to test approximately 560 CWS wells and 750
rural domestic wells for one or more pesticides. A different design with a larger sample size and different
analytical methods for each pesticide would be necessary to assure a greater chance of detection of any given
chemical. The Survey results may be a reflection of the detection limits associated with the Survey's analytical
methods. The NFS Minimum Reporting Limits were established at relatively high levels above instrument
detection limits, and were backed up by stringent confirmation requirements. These procedures reduced the
number of "false positives" (i.e., reports of analyte detections that were not confirmed) and reduced the cost
of laboratory analysis, but may also have led to a lower number of detections for specific NFS analytes.
Second, certain analytes may have been present in the wells that were sampled, but were not detected because
of either their instability (6 analytes) or the analytical methods were unreliable (9 analytes). That is, they may
have transformed or deteriorated so rapidly that their presence was not detectable by the analytic methods
used. As discussed in Chapter 4, fifteen analytes exhibit these characteristics. EPA conducted a number of
analyte stability studies over the course of the Survey whose results will be reported and analyzed in Phase II.
Finally, the process of survey selection for the wells to be sampled may have identified wells in locations where
certain pesticides are relatively uncommon. The Phase II analysis will investigate further the effects of these
factors.
Relational analyses will include a study of whether there is an association among questionnaire
responses or between questionnaire responses and water quality. The hypotheses that might be tested include
the following: Are there more analyte detections of chemicals associated with unconfined aquifers than with
confined aquifers? Are nitrate detections associated with septic units on the property? Are there more
analyte detections of chemicals associated with areas where irrigation is used? EPA is planning scientific
investigations to:
Determine the relationship between nitrate and pesticide detections;
Determine the association among detections and well characteristics;
Correlate DRASTIC subscores by individual factors (e.g., depth to ground water) with
detections in rural domestic wells;
Determine associations between pesticide use and pesticide detections in water samples;
Prepare tables showing pesticide detections by county pesticide use estimates for specific
analytes that were detected in the Survey;
Prepare summary statistics, such as frequencies, of analytes detected in areas where crops with
known associated pesticide use are grown;
Conduct regression modeling for analytes with sufficient detections. These analyses will seek
to identify significant relationships between variables that are not readily discernible from the
demographic analyses presented in the Phase I Report. Regression modeling will include:
Stepwise multivariate regression analyses using all the relevant variables evaluated
in the demographic analyses; and
Nonlinear (e.g., logistic) regression analyses on subsets of significant variables
identified in the multivariate regression.
National Pesticide Survey: Phase I Report
-------
86 Chapter Six: Findings and Results
Evaluate analytes with few detections by non-statistical review of all available data. Site-
specific parameters that may be related to water sample findings will be evaluated. Data
sources will include:
marginal comments from questionnaires made by interviewers to note special types
of information;
records of conversations with samplers and owner/operators; and
analytical instrumentation detection capabilities.
Compare NFS findings with other studies (e.g., Monsanto's National Alachlor Well Water
Survey, selected state surveys, pesticides in ground water database).
6.6.2 Evaluation of the Impact of Seasonal Variation on National Estimates
In designing the Survey, EPA minimized the impact of seasonal variability in analyte detections that
might occur as a function of the time of year or point in the agricultural growing season when the well water
samples were obtained, i.e., more frequent or higher levels of concentrations might occur during particular
seasons or months. The FIFRA Scientific Advisory Panel subpanel advised EPA in 1987 in its review of the
Survey design that certain of the large number of factors affecting the presence of pesticides in drinking water
wells could offset seasonal patterns of precipitation or pesticide use. The subpanel, therefore, encouraged
EPA to develop practical approaches for minimizing the effect of seasonal variability, and EPA did so. The
final Survey included allocation of individual well sites across the 22 month sampling period. Given the
constraints of contract laboratory capacity, cost, and project schedules, EPA selected a method to assign the
times of well water sampling randomly to specified two-week sampling intervals distributed across the full-scale
sampling period through February 1990. Using this method, EPA sought to obtain data concerning seasonal
variations in the presence of pesticides in drinking water wells, while minimizing the effect of the Survey's
sampling schedule on the NPS results (see Section B5 of Appendix B for further details).
EPA's initial temporal analysis focused on national nitrate detections for both CWS wells and rural
domestic wells. This approach was taken based on the large number of nitrate detections in comparison to
all other analytes. EPA compared the proportion of nitrate detections in each temporal period (e.g., a month,
or a season). Different seasonal groupings and different groupings of months across the two years of the
Survey were studied.
Initial analyses of temporal effects were performed using chi-square tests for independence of Survey
nitrate detections and various temporal variables. The temporal variables included monthly and various
seasonal groupings. Initial analyses suggest differing results for CWS wells and rural domestic wells.
Preliminary analyses suggest that there is no seasonal variation on a national basis for CWS wells. In other
words, nitrate detections in CWS wells do not appear to occur more frequently during a specific month or
season. In contrast, preliminary analyses for rural domestic wells suggest that more nitrate detections did
occur during specific months of the year for rural domestic wells, in particular the summer and fall months
of July through December. EPA subsequently performed similar temporal analyses for the detection of at least
one pesticide nationally. The results of these analyses were similar in direction to those for nitrates, but not
nearly as pronounced. Pesticide detections appear to have occurred at greater frequency in the latter months
of 1988 in rural domestic wells only (there was no apparent effect for CWS wells).
If these initial preliminary indications of a seasonal effect for rural domestic wells is supported by
more detailed scientific and statistical analysis in Phase II, a more complex analysis of the nitrate and pesticide
data may be required. It appears, however, that this early indication of seasonal variability in the number of
National Pesticide Survey: Phase I Report
-------
Chapter Six: Findings and Results 87
detections is not a function of EPA's sampling schedule, but is a result of a variety of factors that will require
more detailed analysis in Phase II. A number of key questions will be investigated:
Why is there an apparent temporal effect for rural domestic wells and not for CWS wells?
How does the nitrate temporal effect compare to temporal effects for other anaJytes?
Is the seasonal variation, if any, correlated with the stratification variables used in the Survey?
and
How does (any observed) seasonal variation affect conclusions that may be drawn about the
frequency or proportion of wells containing pesticides or nitrate.
Conclusion
The findings and results described in this chapter conclude several years of intense activity by EPA
to assess the problem of pesticides and nitrate in drinking water wells. Extensive statistical analysis, to be
conducted with NPS data during Phase II, will provide additional findings to enhance the Agency's
understanding of major factors that affect well water quality. Although more analysis is required, the Survey
has produced several important products, described in Chapter 7, and several key findings presented here.
Three key points can be identified from the NPS results. First, low levels of nitrate occur in more
than fifty percent of the nation's drinking water wells. Second, DCPA acid metabolites and atrazine were
detected consistently in both community water system wells and rural domestic wells. Third, although several
pesticides and nitrate were frequently detected in the nation's drinking water wells, most are below health-
based levels and do not pose a risk to public health. These Survey results show that substantial numbers of
wells, particularly rural domestic wells, could be affected by the presence of one or more pesticides or nitrate.
This evidence of widespread migration of chemicals indicates that there is a need for continued attention to
ground-water protection and additional analysis of the issue.
National Pesticide Survev: Phase I Renort
-------
Chapter Seven: Survey Products
Introduction
This chapter briefly describes products that were developed during the National Pesticide Survey.
These products were designed to facilitate the overall Survey planning and implementation activities and
include documents, techniques, and procedures in the areas of: risk communication, techniques for
characterizing pesticide use and ground water vulnerability, quality assurance, survey implementation
procedures, and analytical methods. Information on the availability of each products also is provided in this
chapter.
7.1 Survey Products
Hearth Advisories. A Health Advisory is a technical, scientific guidance document containing
information on physicochemical properties, uses, chemical fate, health effects, treatment, and existing
criteria and guidelines. EPA developed 99 technical Health Advisories for pesticides and nitrate. The
Health Advisories can be obtained from the National Pesticide Telecommunications Network at (800)
858-4791.
Health Advisory Summaries. As part of EPA's effort to explain the health effects of exposure to
pesticides and nitrate in non-technical language, the Agency prepared and distributed over 50 Health
Advisory Summaries. EPA distributed Health Advisory Summaries immediately to the well owners
and operators of sampled wells where pesticides or nitrate were found and to residents who might
consume water from such wells. Copies of these summaries can be obtained from the National
Pesticide Telecommunications Network at (800) 858-4791.
Pesticides in Drinking Water Wells' Brochure. EPA prepared a "Pesticides in Drinking Water
Wells" brochure, which provides an overview of the nature of pesticides and their possible effects in
drinking water wells. A copy of this brochure can be obtained by contacting the EPA Safe Drinking
Water Hotline at (800) 426-4791 or (202) 382-5533.
First Stage County Use and Vulnerability Scores. During the Survey design, EPA created data
files on tape of pesticide use and overall DRASTIC score ratings for all 3,137 counties in the United
States. The data tapes will be available through the Office of Pesticide Programs Docket (703) 557-
2805. A technical document that explains which files are on the tape and provides information on
how to access the data will accompany the tapes in the Docket.
Second Stage County Maps. Following the DRASTIC scoring process during the Survey design,
NPS staff mapped each of the 90 Survey counties selected for rural domestic well sampling delineating
the hydrogeologic settings, cropping areas, and associated vulnerability scores. An example of such
a "cropped and vulnerable" map appears in Exhibit 2-9 of this report The maps will be available for
review in the Office of Pesticide Programs Docket.
Quality Assurance Project Plans. During planning and implementation of the Survey detailed
plans were prepared outlining the oversight activities that would be undertaken to ensure that the
Survey produced data of a known and useful quality. These quality assurance project plans provide
specific information on the QA/QC requirements for each of the analytical and implementation
activities. These plans are available from the National Technical Information Service in Springfield,
Virginia, (703) 487-4650.
National Pesticide Survev: Phase I Reoort
-------
90 Chapter Seven: Survey Products
Pilot Study Reports. Several reports on the pilot study are available through the Office of Pesticide
Programs Docket, including:
1. Final Report on the Management Systems Review of the Pilot Survey of Pesticide
Contamination in Groundwater, Quality Assurance Management Staff, USEPA, 1987;
2. National Pesticide Survey Pilot Study Evaluation Summary Report, Office of Drinking Water
and Office of Pesticide Programs, USEPA, 1987; and
3. National Pesticide Survey Pilot Evaluation Technical Report, R.E. Mason, L.L. Piper, W.J.
Alexander, R.W. Pratt, S.K. Liddle, J.T. Lessler, M.C. Ganley, D.J. Munch, and G. Langner,
RTI/7801/06-02F, 1987.
Scientific Advisory Panel Reports. The Federal Insecticide, Fungicide, and Rodenticide Act,
Scientific Advisory Panel Subpanel prepared two reports, which are available for review at the Office
of Pesticide Programs Docket.
1. Technical Review of the Preliminary Design of the National Survey of Pesticides in Ground
Water - Report of Subpanel Recommendations: October 9, 1985, Federal Insecticide,
Fungicide, and Rodenticide Act Scientific Advisory Panel Subpanel; and
2. A Set of Scientific Issues Being Considered by the Agency in Connection with the National
Pesticide Survey Pilot Study, October 9,1987, Federal Insecticide, Fungicide, and Rodenticide
Act, Scientific Advisory Panel Subpanel.
Survey Questionnaires. EPA developed several questionnaires to collect data necessary for the
interpretation of NPS analytical results. The Agency administered questionnaires to determine
whether randomly selected community water systems and domestic wells were eligible for NPS water
sampling; record observations about the well location and the surrounding area; collect information
from the owner/operator about the use and construction of the well sampled and use of non-farm and
farm pesticide (e.g., volume of and location of application in relation to the well) on the property
where the well was located; and gather information from local area experts about cropping, pesticide
use, and industry locations within one-half mile of the well. The NPS questionnaires are provided in
Appendix D.
Training Manuals. EPA prepared training manuals to instruct EPA, State, and regionally-based
contractor personnel on sampling techniques and procedures, questionnaire administration, and
communication protocols. The manuals were distributed during training workshops that were held
from May 1988 through January 1990. The manuals are available for review at the EPA Office of
Pesticide Programs Docket.
Analytical Methods. The NPS Analytical Methods Task Force identified six new methods, five of
which are capable of evaluating multiple analytes (multi-residue methods) and two existing EPA
analytic methods for use in the NPS as well as in future analysis of pesticides in water. The methods
are listed in Exhibit 4-2, Chapter 4, and are available from the National Technical Information
Service.
National Pesticide Survey: Phase I Report
-------
Glossary and List of Acronyms
Glossary
Active Ingredient
Analyte
Analyte Stability
Study
The specific chemical in a product that is biologically active against pest organisms.
One of 127 chemicals that the Survey analytical methods were chosen or designed to
identify. The Survey analytes include 101 pesticides, 25 pesticide degradates, and nitrate.
Analytes were selected on the basis of expected leaching potential, occurrence in ground
water, volume of use, and other considerations.
An analysis of a laboratory sample in which a known quantity of an analyte has been
placed (a "spiked" sample). The sample is analyzed at a pre-determined time to assess the
stability and recovery of the analyte.
Analytical Method One of the eight chemical testing methods used to analyze Survey water samples. Six of
the analytic methods were specifically developed for the NFS.
Carcinogen
A substance that is either known or suspected to cause cancer.
Community Water A system of piped drinking water that either has at least 15 connections or serves at least
System (CWS) 25 permanent residents. To be eligible for the NFS, a system had to have at least one
operable well (at the time of sampling) supplying drinking water.
Confidence
Interval
Contaminant
Contamination
Contract
Laboratory
Cropped and
Vulnerable
Degrade te
The upper and lower limits around an estimated value within which the actual population
value is expected to fall. The confidence interval is stated as a specified level, such as
95%, of confidence. For the NFS, estimates for wells throughout the United States, based
on analysis of results for the NFS sample of wells, are given with a 95% confidence
interval, indicating the upper and lower limits that EPA is reasonably confident contain
the actual detection rate.
Any substance such as a chemical, ion, radionuclide, synthetic organic compound,
microorganism, waste, or other substance that occurs in water causing it to be impure.
Contaminated water does not necessarily pose a health risk if the concentration of the
contaminant does not exceed acceptable drinking water standards.
The direct or indirect introduction of any contaminant into ground water caused in whole
or in pan by human activities.
One of the five laboratories hired to analyze Survey water samples. The laboratories
performed different analytic methods. EPA laboratories performed referee and
confirmation analyses to verify the results of the contract laboratories.
Sub-county areas specified for the rural domestic well survey that have high ground-water
vulnerability and greater than 25 percent of the land area used for agricultural production
or medium ground-water vulnerability and greater than 50 percent of the land area used
for agricultural production.
The product of the chemical or biological breakdown of a complex compound into simpler
compounds. See Pesticide Degradate.
National Pesticide Survey: Phase I Report
-------
92 Glossary and List of Acronyms
Domains of
Interest
DRASTIC
Drinking Water
Well
EPA Laboratory
Estimated
Detection Limit
Federal
Insecticide,
Fungicide, and
Rodenticide Act
(FD7RA)
The seven subgroups of the entire population of drinking water wells about which the
Survey was particularly interested in obtaining data. The seven domains of interest are
(1) community water systems nationally; (2) community water systems in counties with the
highest average ground-water vulnerability; (3) rural domestic wells nationally; (4) rural
domestic wells in counties with the highest average pesticide use; (5) rural domestic wells
in counties with the highest average ground-water vulnerability; (6) rural domestic wells
in counties with the highest average pesticide use and ground-water vulnerability; and (7)
rural domestic wells in "cropped and vulnerable" areas of counties.
A classification system that attempts to provide a relative ranking of the vulnerability of
ground-water to contamination. The letters in DRASTIC stand for features of the area
around the well that may affect the movement of pesticides into ground water: (depth to
water, recharge, aquifer media, soil media, topography, impact of the unsaturated aqueous
zone between the soil media and water table, and conductivity of the aquifer). DRASTIC
does not consider sources of the contamination or population affected. A DRASTIC
score was computed for all of the counties in the United States as part of the Survey.
A rural domestic well or a community water system well whose water is used primarily for
human consumption (i.e., drinking, cooking, and bathing).
One of two EPA laboratories that were chosen to manage contracts for the analytical
laboratories, and perform referee analyses.
The minimum concentration of an analyte that can be measured and reported with
confidence that the analyte concentration is greater than zero.
Federal law first enacted in 1947 and administered by the EPA since 1970. Under FIFRA,
EPA registers pesticide products and ensures that they will not present unreasonable risks
to human health or the environment when used according to label directions.
Federal Reporting A data base maintained by EPA's Office of Drinking Water (ODW) that contains
Data System information on all public water supply systems in the United States. The FRDS was used
(FRDS) as the sampling frame for selecting community water system wells for NPS water sampling
and analysis.
Ground Water
Ground-Water
Vulnerability
Health Advisory
Water found beneath the earth's surface contained in the interconnected pores and
fractures of soils and geologic formations.
The degree to which ground water in the area surrounding a well is sensitive to pesticide
infiltration, based on an assessment of the hydrogeological makeup of an aquifer, such as
depth of the water table, soil properties, and ground-water recharge rates.
A technical and scientific guidance document prepared by EPA for many of the analytes
included in the Survey. Each Advisory contains information on the analyte's
physicochemical properties, uses, chemical fate, health effects, treatment, and existing
criteria and guidelines. EPA also prepared non-technical Health Advisory Summaries
based on the Health Advisories.
National Pesticide Survey: Phase I Report
-------
Glossary and List of Acronyms 93
Health Advisory
Level (HAL)
Health Advisory
Summary
Herbicide
Human Health
Risk
Hydrology
Insecticide
Leaching
Maximum
Contaminant
Level (MCL)
The maximum concentration of a contaminant in water that may safely be consumed over
a specific time period. EPA sets HALs for short-term exposures, such as one day and ten
days, and longer-term exposures of greater than ten days up to several years, and over a
lifetime. A pesticide's HAL is based on health effects (other than cancer) that were found
in humans or in animals given high doses of the pesticide in laboratory studies. For
pesticides believed to cause cancer, EPA does not calculate a Lifetime HAL. Instead,
EPA calculates the increased risks of cancer that are associated with different
concentrations and exposures to the pesticide.
A one-page summary prepared by EPA for many of the analytes included in the Survey.
These summaries provide concise information on possible health effects of these chemicals
and actions that may be taken to ensure a safe drinking water supply. EPA distributed
these summaries to the owners, residents, and operators of sampled Survey wells in which
analytes were detected.
A pesticide used to limit or inhibit plant growth.
The probability that a given exposure or series of exposures will damage the health of
individuals experiencing the exposures.
The science dealing with the properties, distribution, and circulation of water on the land
surface, in the soil, and in underlying rock formations.
A pesticide used to control insects.
The downward transport through the soil by percolating water of dissolved or suspended
minerals, fertilizers, pesticides, and other substances. A chemical's solubility, the soil
texture, and the amount and timing of water applied to the soil (as in irrigation, rainfall,
or heavy runoff) all contribute to the rate of leaching.
The maximum permissible level of a contaminant in water that is delivered to any user of
a public water system (established by the Safe Drinking Water Act (SDWA)).
Method Blank A portion of reagent water analyzed as if it were a sample.
Micrograms per
Liter
-------
94 Glossary and List of Acronyms
National Pesticide The preliminary study for the NFS conducted in 1987 in the States of California,
Survey Pilot Study Minnesota, and Mississippi. The pilot study determined whether any changes in the
proposed Survey design were necessary or advisable prior to the implementation of the
full-scale Survey. The pilot program sampled approximately 50 rural domestic and
community water system wells.
National Pesticide A toll-free information service that answers inquiries on the health effects of pesticides
Telecommuni- and pesticide poisonings. This service operates 24 hours a day and can be reached at 1-
cations Network 800-858-7378.
Nitrate (NO3')
Nitrite (NO2')
Office of Drinking
Water (ODW)
Office of Pesticide
Programs (OPP)
Organic
Compound
Parts per billion
(PPB)
Parts per million
(PPM)
Pesticide
Pesticide
Degradate
Pesticide
Metabolite
An oxidized form of nitrogen that is an important plant nutrient and inorganic fertilizer.
The major sources of nitrate are septic systems, animal feed lots, agricultural fertilizers,
manured fields, industrial waste waters, sanitary landfills, and garbage dumps. Nitrate also
occurs naturally in ground water. The Survey analyzed water samples for the combined
presence of nitrate and nitrite, measured as nitrogen (N).
A form of nitrogen that is less oxidized than nitrate. Nitrite is a relatively unstable
transitional form between nitrate and ammonium.
The EPA office, under the management of the Office of Water, that is primarily
responsible for implementing the Safe Drinking Water Act (SDWA). The Office of
Drinking Water jointly conducted the NPS with the Office of Pesticide Programs.
The EPA office, under the management of the Office of Pesticides and Toxic Substances,
that is primarily responsible for implementing the Federal Insecticide, Fungicide, and
Rodenticide Act (FTFRA). The Office of Pesticide Programs jointly conducted the NPS
with the Office of Drinking Water.
Natural or synthetic chemical containing carbon.
A measure that corresponds to one part pesticide in one billion parts of water.
PPB =
Weight of material
X 1,000,000,000
Total wt. of gas, liquid, or solid
A measure that corresponds to one part nitrate in one million parts of water.
PPM = Weight of material X 1,000,000
Total wt. of gas, liquid, or solid
Chemical substance used to destroy, control, or repel undesirable organisms which may
include plants, insects, fungi, nematodes, rodents, predators, or microorganisms.
A generic term that includes breakdown products of a pesticide active ingredient resulting
from biological processes (i.e., metabolites) and chemical processes (i.e., hydrolysis,
photolysis, photooxidation). See Degradate.
A product of biological processes (e.g., metabolism, or the chemical changes in living
cells) of a pesticide active ingredient by microorganisms, plants, or animals.
National Pesticide Survey: Phase I Report
-------
Glossary and List of Acronyms 95
Public Water The unique identifying number given to public water systems in the Federal Reporting
Supply Data System (FRDS) that was used when statistically selecting community water systems
Identification for the NPS.
Number (PWSID)
Quality Assurance Oversight activities performed by EPA and its contractors to maintain quality control
(QA) standards and to ensure compliance with the standard operating procedures for conducting
well sampling, laboratory analysis, and processing Survey questionnaires.
Quality Control
(QC)
Random Sample
Rapid Reporting
Level
Raw Water
Sample
Reagent Water
Relational
Analysis
Rural Domestic
Well
Safe Drinking
Water Act
(SDWA)
Safe Drinking
Water Hotline
Sample Blank
Monitoring activities performed by EPA and its contractors to ensure that sampling, data
collection, and laboratory procedures are properly conducted and meet specified
performance standards.
A method of statistical sampling that ensures each member of the population has the
same chance of being chosen.
A level of pesticide or nitrate presence detected by the Survey laboratories that warranted
immediate contact with the well owner/operator. Rapid reporting occurred if the
measured detection equalled or exceeded EPA's Lifetime Health Advisory Levels for the
detected analyte.
A water sample gathered prior to treatment of any kind.
Water used in the laboratory quality assurance/quality control procedures that is treated
to remove any contaminant so it will not be observed at or above the estimated detection
limit of any analyte.
Any of the various types of statistical methods used to estimate the correlation or
relationship between variables in a given population.
A drinking water well that supplies an occupied private household located in rural areas
of the United States, except for wells located on government reservations. The Census
Bureau defines rural areas as households outside of incorporated or unincorporated places
with a population of 2,500 or more and outside of urban areas. To be eligible for the
Survey, the NPS required that the well be used for human consumption (i.e., drinking,
cooking, bathing).
A law passed in 1974 and administered by EPA that establishes national standards for
drinking water to provide a safe and wholesome water supply from both surface and
ground-water sources.
A toll-free hotline established by EPA in July 1987 to handle requests for information on
drinking water issues, including the NPS. The hotline can be reached at 1-800-426-4791
(in Washington, D.C. at (202) 382-5533) and is operated Monday through Friday, 8:30 to
4:30 eastern standard time.
Water samples that were shipped to analytical laboratories for analysis and compared to
the analytical results of NPS well water samples to ensure that the analytical methods
were accurate.
National Pesticide Survey: Phase I Report
-------
96 Glossary and List of Acronyms
Scientific Advisory A panel of scientists created under the authority of FIFRA to advise the EPA on scientific
Panel (SAP) issues related to the assessment of risks posed by pesticides. A special subpanel of the
SAP reviewed the Survey design in 1985 and in 1987 reviewed technical issues such as data
collection options, well selection methods, temporal variation, and analyte instability.
Septic System
Spiked Sample
Surface Water
Synthetic Organic
Compound (SOC)
Temporal
Variability
Volatile Organic
Compound (VOC)
Water Table
Well Casing
A sewage system composed of both a septic tank and septic field. The septic tank is an
underground watertight container made of durable material through which sewage flows
very slowly and in which solids separate from the liquid to be decomposed or broken
down by bacterial action. The septic field is the area through which the sewage liquid
passes and in which it is cleaned through physical filtering by soils, biodegradation, and
evaporation.
A water sample to which a known quantity of a pesticide has been added so that the
accuracy of the laboratory analyses can be determined.
Water found on the land surface in streams, ponds, marshes, lakes, or other fresh water
sources.
Man-made organic compound, not naturally found in ground water.
Seasonal fluctuations in weather conditions and pesticide use that may affect the presence
of pesticides in ground water and may potentially bias Survey sampling. This problem was
minimized by randomly allocating selected wells to specified two-week periods within each
strata across the two year sampling period.
An organic compound that evaporates (volatilizes) readily into the atmosphere and is
highly mobile in ground water.
The top of an unconfmed (unpressurized) aquifer, below which the void spaces resulting
from the granular texture or fractures of earthen material are saturated with water.
Materials such as concrete, piping, metal, and stone that line and support a well and
prevent it from collapsing.
National Pesticide Survey: Phase I Report
-------
Glossary and List of Acronyms 97
List of Acronyms
ADQ
CATI
CCS
CWS
DCS
DQO
DRASTIC
EDL
EMSL-CI
EPA
ES&E
FIFRA
FRDS
GC
GC/MS
HAL
HPLC
IOC
LOQ
MCL
Mg/L
MQL
MRL
ND
NPS
NPSIS
Audits of Data Quality
Computer Assisted Telephone Interviews
Communications Control System
Community Water System
Document Control System
Data Quality Objective
DepthyRecharge/Aquifer/Soil/Topography/Impact/Conductiviry
Estimated Detection Limit
Environmental Monitoring and Systems Laboratory - Cincinnati
Environmental Protection Agency
Environmental Science and Engineering
Federal Insecticide, Fungicide, and Rodenticide Act
Federal Reporting Data System
Gas chromatography
Gas chromatography/mass spectrometry
Health Advisory Level
High pressure liquid chromatography
Inorganic compound
Level of quantification
Maximum contaminant level
Milligrams per Liter
Micrograms per Liter
Minimum quantification limit
Minimum reporting limit
Non-detected
National Pesticide Survey
National Pesticide Survey Information System
National Pesticide Survey: Phase ( Report
-------
98 Glossary and List of Acronyms
NPTN National Pesticide Survey Telecommunications Network
NTIS National Technical Information Service
ODW Office of Drinking Water
OPP Office of Pesticide Programs
PE Performance Evaluation
PIC Public Information Center
PPB Parts per billion
PPM Parts per million
PWSIN Public Water Supply Identification Number
QA Quality Assurance
QAC Quality Assurance Coordinator
QAMS Quality Assurance Management Staff
QAO Quality Assurance Officer
QAPjP Quality Assurance Project Plan
QAPP Quality Assurance Program Plan
QA/QC Quality Assurance/Quality Control
QC Quality Control
RDD Random Digit Dialing
SAP Scientific Advisory Panel
SCS Sampling Control System
SDWA Safe Drinking Water Act
SOC Synthetic Organic Compound
SOP Standard operating procedure
TSA Technical System Audit
TSD Technical Support Division
USDA United States Department of Agriculture
USGS United States Geologic Survey
VOC Volatile Organic Compound
National Pesticide Survey: Phase I Report
-------
National Pesticide Survey
Appendix A: Background
-------
Appendix A: Background
A1 Introduction
The National Survey of Pesticides in Drinking Water Wells originated in 1984 and is one of the most
complex surveys ever conducted by EPA This appendix reviews the background, context, and objectives of
the Survey; discusses the management and planning of the Survey up to the end of 1987, when a pilot study
was successfully completed and its lessons incorporated into the Survey design; and records some of the major
substantive issues and decisions encountered in the course of planning the Survey through the completion of
the pilot survey. Appendix C discusses management and implementation of the Survey from the completion
of the pilot study to completion of the full scale Survey.
Appreciation of the fragility and importance of ground water is a recent phenomenon. Ground water
once was thought to be protected from pesticide contamination by impervious layers of subsoil, rock, and clay,
and by chemical degradation processes in and on the soil. One of the first steps in dismantling this long-held
belief was the discovery of the pesticide dibromochloropropane (DBCP) in about 2,500 wells in California.
DBCP was soon found in ground water in four other States as well. Another pesticide, aldicarb, was found
in 1979 in wells on Long Island, New York, and subsequently in Wisconsin and 11 other States. The
discoveries of ethylene dibromide (EDB) in wells in California, Georgia, and other States in 1982-1983 raised
concerns to a new level.
By the mid-1980s, there were indications of a larger problem of pesticides in ground water. By 1987,
at least 17 pesticides had been found in ground water in 23 States.1 Studies of pesticides in ground water
were undertaken by the States of California, Florida, Maryland, Minnesota, New York, Washington, and
Wisconsin, among others.
Most of these studies, however, were limited to a small number of pesticides and specific geographic
areas and did not attempt a comprehensive nationwide assessment. Furthermore, not all States were active
in monitoring for pesticides. Finally, in selecting wells for sampling, most State surveys did not use a statistical
method that would allow the results to be extrapolated to a larger population.
EPA undertook the National Pesticide Survey (NPS) in response to the need for nationwide
information on pesticides in drinking water. The Survey was predicated on the concern that agricultural
chemicals, both pesticides and fertilizers, might be playing a significant role in ground-water contamination
and that there was cause for concern for the quality of ground water as an irreplaceable drinking water source.
A particularly disturbing realization was that one of the major sources of the presence of agricultural chemicals
in ground water might be the normal, approved use of these chemicals. From a health perspective, the
pervasiveness, toxicity, and persistence of many pesticides in the environment were of concern. A number of
pesticides are known or suspected to cause a variety of adverse health effects, ranging from eye and skin
irritation to cancer. With growing public concern about the health and environmental implications of
chemicals in ground water, EPA issued a ground-water protection strategy in 1984, and began the development
of a longer-term effort, referred to as the Pesticides in Ground Water Strategy.
1 Coben, S.Z., C. Eiden, and M.N. Lorber, "Monitoring Ground Water for Pesticides in the U.S.," in Evaluation of
Pesticides in Ground Water. ACS Symposium Series No. 315, ed. by W.Y. Garner, R.C Honeycutt, and H.N. Nigg, American
Chemical Society, Washington, D.C., 1986. See also: Pesticides in Ground Water Background Document. U.S. EPA, Office
of Ground Water Protection, May 1986.
National Pesticide Survey: Phase I Report
-------
A-2 Appendix A: Background
The National Pesticide Survey is a major component of the Agency's overall effort to understand and
characterize the problem of agricultural chemicals in ground water. The National Pesticide Survey was
undertaken by EPA's Office of Pesticide Programs (OPP) and Office of Drinking Water (ODW) with two
principal objectives in mind: (1) to determine the frequency and concentration of pesticides, pesticide
degradates, and nitrate in the drinking water wells of the nation; and (2) to improve our understanding of how
the presence of these substances is associated with patterns of agricultural pesticide usage and the vulnerability
of ground water to pollution. Survey information was collected to help support the development, evaluation
and enforcement of regulations and guidance for drinking water and pesticides under the Safe Drinking Water
Act (SDWA) (Sections 1442 and 1445) and the Federal Insecticide, Fungicide, and Rodenticide Act (FIFRA)
(Section 20C). Exhibit A-l provides a list of the uses of NPS data that were expected at the time.
With adequate Survey information on the presence of different pesticides in wells around the country,
EPA can better design its regulatory programs to target pesticides of concern. Pesticides shown to pose
potential hazards by their ability to leach into ground water could be subject to a range of further regulatory
actions under FIFRA, including changes in label directions, use restrictions, or suspension or cancellation of
a pesticide's registration. The Agency also intended to use information from the Survey to implement SDWA
requirements. New maximum contaminant levels and monitoring requirements could be proposed for
pesticides shown to pose a hazard in public drinking water.
A2 Management and Planning of NPS
A2.1 Early Proposals: 1979-1984
A study of pesticides in ground water was first discussed at EPA in a September 1979 OPP
memorandum that outlined criteria for selecting pesticides and soils.2 The original focal points of interest
were whether pesticides are leaching from the soil surface to aquifers, and whether there are any predictive
tools that could warn of such contamination problems. By December 1979, a list of 26 candidate pesticides
had been developed for the proposed study, based on their physical properties, soil mobility, persistence, usage
on crops, and human and animal toxicity.
The study was not funded at the time, but the proposal resurfaced when EPA suspended use of a soil
fumigant, ethylene dibromide (EDB) in September 1983. During 1984, OPP and ODW agreed on a joint
study, officially laid out in a memorandum of agreement between the two offices. A major issue was whether
to limit the study only to ground water (favored by OPP and the option ultimately taken) or also to include
surface water (favored by ODW).
Full-scale planning of a ground-water study was authorized in August 1984 in a joint memorandum
from the Directors of ODW and OPP. One representative each from ODW and OPP were designated as
coordinators of the management work plan. Thirteen task groups were set up, including an analyte selection
workgroup, an analytical methods workgroup, a health advisory workgroup, a hydrogeologic workgroup, a
pesticide-use workgroup, a quality assurance workgroup, a State liaison workgroup, a sampling workgroup, and
a data management workgroup. EPA contracted with Research Triangle Institute to do the statistical design
of the Survey.
2 More information on the early years of the survey and an extensive bibliography may be obtained from the report,
National Survey of Pesticides in Drinlcing Water Wells: A Review of the Planning Process and The Data Quality Objectives.
by Monica Nees and Cynthia Salmons, Contract No. 68-01-7350, Research Triangle Institute, Research Triangle Park, NC,
July 6, 1987.
National Pesticide Survey: Phase I Report
-------
Appendix A: Background A-3
Exhibit A-1
Anticipated Uses of National Pesticide Survey Data
Enhance EPA's ability to meet the growing demands for accurate data on the
presence of pesticides in drinking water wells.
Identify pesticide compounds which occur at levels of health concern or with
sufficient frequency to be considered high priority candidates for regulation
through the maximum contaminant level (MCL) process under SDWA.
Identify the range of pesticide concentrations found in community water
systems, so that the feasibility of drinking water treatment can be assessed.
Identify and examine the relationship of pesticide contamination to
ground-water vulnerability and pesticide usage in order to help direct
compliance monitoring requirements for community water systems.
Facilitate regulatory impact analyses by building an adequate data base on
the distribution of pesticide levels in drinking water.
Develop working relationships with the States to facilitate implementation of
drinking water programs and to build State programs relating to pesticides.
Promote coordinated ongoing monitoring programs with States and USGS,
building on the NPS data base.
Determine the scope of the national problem of pesticides in drinking water
wells to enable planning and priority-setting for pesticide regulations across
types of pesticides and individual compounds.
Determine the frequency and range in concentrations, and examine the
relationships among pesticide contamination, ground-water vulnerability, and
pesticide usage to help focus registrant monitoring requirements (both
where' to monitor and "what" to monitor).
Enable targeted retrospective field studies by registrants.
Ascertain whether contamination exists at levels of health concern with
sufficient frequency, in association with vulnerability and usage conditions,
to move directly, if necessary, to labeling and registration decisions.
Share Survey results with States to help focus State monitoring and
regulatory efforts, and State training and certification programs.
Determine statistically reliable national estimates of nitrate in ground water
for use in implementing the Agricultural Chemicals in Ground Water Strategy.
National Pesticide Survey: Phase I Report
-------
A-4 Appendix A: Background
In November 1984, two key options for the design of the Survey were examined:
Limiting sampling to areas of high pesticide usage and high vulnerability for ground water
contamination; or
Designing the Survey to characterize the distribution of contamination by randomly sampling
areas with different levels of pesticide usage and different geophysical characteristics.
EPA chose to design the Survey to accomplish both objectives within specified cost limits, accepting
the need for limitations in the certainty of the Survey's results.
A2.2 Planning Years: 1984-86
From late 1984 through 1985, the Survey was managed by a Steering Committee of representatives
from ODW and OPP. By 1985, estimates of the cost of the Survey were put at $5-6 million.
In September, 1985, the Survey design was submitted for peer review to a subpanel of EPA's FIFRA
Scientific Advisory Panel (SAP). A subpanel public meeting on the design review was held on October 3,
1985.3 The Survey design was also reviewed by the National Agricultural Chemicals Association (NACA).
The SAP subpanel recommended that the Survey design emphasize private wells, exclude wells not at risk and
where explanatory variables are not present, include rural non-farm private wells, and collect considerably
more local information to be used in selection of wells. In addition, the SAP subpanel and NACA cautioned
that the use of the DRASTIC model might be insufficient (see discussion in the next section) and
recommended conducting a pilot survey.
A full-time manager of the Survey was appointed in April 1986. Ongoing coordination continued with
other Federal Government agencies, such as the Department of Agriculture, industry associations, health and
environmental groups, and State and county officials, including the Association of State Drinking Water
Administrators (ASDWA), the State FIFRA Issues Research and Evaluation Group (SFIREG), and the
National Association of County Health Officials (NACHO). NPS staff attended meetings on a monthly basis
with U.S. Geological Survey staff.
With changes made to incorporate many of the SAP comments into the Survey design, the addition
of a pilot study, and an expansion in plans for collecting local information, estimates of the cost of the Survey
rose to $9-10 million. In light of these cost projections, the NPS underwent a series of intensive reviews by
EPA management with respect to cost containment measures. In September 1986, the Assistant
Administrators of the Office of Water and the Office of Pesticides and Toxic Substances made several cost-
containment decisions, including the decision to seek voluntary State sampling of community water systems
and to eliminate most data collection around community water systems.
In addition, a National Pesticide Survey Regional/State Workgroup was convened prior to
implementation of the pilot study and subsequently at key phases of the Survey. The aim of the workgroup
was to ensure adequate State and Regional input throughout the process. The workgroup had representatives
from several States and four EPA Regional Offices, as well as a representative from SFIREG and key EPA
personnel. The group reviewed and evaluated all aspects and phases of the Survey and served as a forum for
State input and exchange of information.
50 Federal Register 37904, September 18, 1985.
National Pesticide Survey: Phase I Report
-------
Appendix A: Background A-5
Also during this time period, EPA's Office of Drinking Water was engaged in the development of
health advisory levels and documentation for numerous pesticides and other potential drinking water
contaminants. The first set of 15 draft health advisories were presented to EPA's Scientific Advisory Panel
in June 1986. The panel's critique included the recommendations that the health advisories include multiple
calculations of risk, reflect differences among studies upon which the calculations are based, and provide upper
and lower confidence limits and maximum likely estimates where appropriate.
Using these recommendations as a guide in revising the process for developing health advisories, in
1987 and 1988 EPA issued 99 health advisories, covering 25 organic chemicals, 7 inorganic chemicals, one
microbial agent, and 66 pesticides (including 62 pesticides then of interest to the Survey).
In conjunction with this effort, the Survey staff recognized a need for developing and distributing non-
technical summaries of the health advisory documents to the public (and especially to well owners and
operators and householders participating in the Survey who would be concerned if pesticides were found in
their well water). A draft set of health advisory summaries was reviewed by EPA's Regional Offices, the
Conservation Foundation, Concern, Inc., the League of Women Voters Education Fund, NACA, and several
focus groups. Commenters stressed the need to clarify the discussion of health effects, the need to explain
technical terms sensitively and simply, the need to provide information on practical treatment options for
contaminated drinking water, and the need to eliminate risk comparisons when the risks are not comparable
on important dimensions (such as the voluntary or involuntary nature of the risk). The health advisory
summaries were revised and 59 summaries were issued in January 1989.
A2.3 Pilot Study: 1987
In March 1987, EPA launched a pilot study to field test the major components of the Survey and to
provide an opportunity for any necessary revisions or modifications before implementing the full Survey.4
Between March and August, the pilot study for the NPS was conducted in three States: California, Minnesota,
and Mississippi. The States were selected to provide geographic diversity and because of their high level of
interest and cooperation. Two of the States, Minnesota and California, had considerable previous experience
in State pesticide monitoring programs. Mississippi presented an example of a State with strong interest but
little prior experience in this area.
Water sampling was conducted at 48 wells in the pilot study, including both domestic rural wells and
community water system wells. Domestic wells were sampled by EPA's contractor, the Research Triangle
Institute (RTI), whose staff also conducted interviews with householders regarding the usage and construction
of their wells. In addition, data were collected on conditions within the immediate vicinity of the well and
within a half mile radius of the well. CWS wells in each pilot State were sampled by State health department
officials, after training by RTI. Questionnaires on the construction and characteristics of the wells were also
administered to CWS operators.
Because of the concern for data validity, heavy emphasis was placed on quality assurance and quality
control procedures in the pilot study. This resulted in an average of 50 bottles of water being taken at each
well sampled. The water samples were analyzed in the pilot study by EPA's contractor laboratories (Battelle
Columbus and Southwest Research Institute) and by four EPA quality control laboratories.
4 For more information on the pilot study, see National Pesticide Survey. Pilot Study Evaluation Summary Report. U.S.
EPA, September 1987, or National Pesticide Survey. Pilot Evaluation Technical Report. Contract No. 68-01-7350, Research
Triangle Institute, Research Triangle Park, NC. January 29, 1988.
National Pesticide Survey: Phase I Report
-------
A-6 Appendix A: Background
Overall, the pilot study confirmed that major components of the Survey design were in good working
order. Specifically:
Sampling was conducted satisfactorily, both technically and logistically, at selected wells;
The nine analytic methods were tested. Modifications to Method 3 were accomplished in mid-
study,
The statistical methods developed for the Survey to select domestic wells functioned
appropriately,
Agricultural extension agents were cooperative and informative in providing cropping data;
Interview questionnaires were straightforward to administer;
No difficulties were encountered collecting observational data around the wells;
Survey staff met a high degree of cooperation and interest on the part of householders, with
a participation rate of over 90 percent;
Informed consent procedures were implemented successfully, while maintaining satisfactory
participation rates; and
States were enthusiastic and cooperative, assuming a major role in sampling CWSs.
The pilot study was considered particularly valuable in providing a learning period for trying
alternative approaches and for improving procedures and techniques. As a result of the pilot study,
recommendations were made for changes in the interview questionnaires, training manuals, and sampling and
communications procedures, in anticipation of the full Survey. In addition, the pilot study raised specific issues
related to the Survey design, which were analyzed and presented to the Scientific Advisory Panel in September
1987. (See section A3.3 below for a discussion of these issues.)
The SAP gave strong support to the design and approach of the Survey and selected among the
options presented on specific design issues. The panel recommended dropping VOCs entirely from the Survey
due to problems of obtaining non-aerated water samples, but did not support making any other cost-saving
cuts and recommended collecting additional data to enhance the Survey's relational analyses. EPA dropped
Analytic Method Number 8, for VOC's, but added three chemicals from Method 8 to Method 7.
In the summer of 1987, protocols for the six new analytic methods developed specifically for the
National Pesticide Survey underwent peer review under the auspices of EPA's Environmental Monitoring and
Support Laboratory (Cincinnati).
The fall of 1987 saw renewed attention to cost-cutting measures and further management reviews.
Measures considered included eliminating one or more analytic methods, sampling fewer wells, eliminating
either the domestic or the community well side of the Survey, eliminating well-specific questionnaire and data
collection efforts, and extending the project schedule to 1991. It was believed that each of these measures
would have seriously compromised the Survey's results or usefulness; none of the measures was selected.
Prior to launching the full Survey, during October and November 1987, NPS staff held a series of
meetings with State and Regional staff in every EPA Regional Office around the country. Attendees at the
meetings included officials from State departments of agriculture and drinking water supply agencies, EPA
Regional Office staff, district representatives of the U.S. Geological Survey, the State Cooperative Extension
National Pesticide Survey: Phase I Report
-------
Appendix A: Background A-7
Services, and representatives of county organizations. The meetings provided EPA with the opportunity to
explain the design of the Survey and its implementation, to review the experience of the pilot study, and to
bear and discuss State and Regional concerns and suggestions.
In December 1987, the contract for implementing the full Survey was awarded to ICF Incorporated
with Westat Incorporated as a subcontractor. Certain changes were subsequently made in the Survey design,
most notably in the use of Random Digit (telephone) Dialing rather than face-to-race interviewing in the third
stage selection of the domestic well survey design, a redefinition of the subcounty "cropped and vulnerable"
strata in the domestic well portion of the Survey, and changes recommended by the SAP review in light of the
pilot study experience. (See Appendix B for a discussion of the statistical design used in the full Survey.) In
addition, changes were made in the methods and materials for implementing the Survey. The questionnaires
were revised; new training manuals and activities were developed; and sampling, communications, and logistics
procedures were refined. (See Appendix C for a discussion of Survey implementation.)
A3 Issues and Decisions
The planning, design, and implementation of the National Pesticide Survey were analyzed thoroughly
during the course of the project. This section reviews some of the major issues encountered during the period
from 1984 through 1987, grouped under five categories:
1. Purpose and Scope of Survey
2. Early Design Decisions
3. Pilot Study/SAP Issues
4. Communications and Role of States
5. Confidentiality.
A3.1 Purpose and Scope of Survey
At various points in the course of planning the National Pesticide Survey, EPA considered the merits
of other related approaches. Early options for the Survey included as objectives the assessment of pesticide
contamination of ground water or an assessment of population exposure to pesticides in drinking water. The
1985 SAP review of the Survey design noted that these two objectives were not compatible. The Survey being
contemplated could not serve simultaneously as a ground-water study and a human exposure study, although
some types of inferences about ground-water vulnerability and the general magnitude of human exposure to
pesticides through drinking water were still considered possible with the data that the Survey would generate.
Another alternative considered was to include only domestic wells in the Survey, and exclude
community water system wells. The theory was that domestic wells, generally shallower than CWS wells, could
function as a sort of early warning indication of a pesticide contamination problem. This alternative was
rejected because it would not provide a nationwide assessment of the problem of pesticides in drinking water
wells, since community well systems are the major source of drinking water for the majority of the population.
Eliminating domestic wells from the Survey, on the other hand, would mean forgoing most of the research and
regulatory benefits of the Survey for the Office of Pesticide Programs.
Still another alternative was to focus only on areas where the ground water is especially susceptible
to pesticide contamination or on areas where pesticide usage is known to be high. This so-called "hot spots"
approach was rejected, however, because it could not provide results that could be generalized to all wells.
In other words, there would still be no national estimates for the country as a whole. Moreover, there was
insufficient information to help pick the wells to investigate. Similar reasons led to rejection of both a "case
studies" approach that would investigate only a few areas in great detail and the option of conducting statistical
surveys in only a few States.
National Pesticide Survey: Phase I Report
-------
A-8 Appendix A: Background
Other measures limiting the scope of the Survey were examined for purposes of containing costs.
These included:
Reducing the number of analytes and analytic methods;
Changing the precision requirements for the results and/or sampling fewer wells;
Changing the requirements for confirmation of positive detections;
Using fewer strata in the design of the Survey (e.g., collapsing the four usage and three
ground-water vulnerability strata to a two-by-two matrix);
Selecting CWSs as planned but then surveying private wells only in the vicinity of the selected
CWSs;
Eliminating questionnaire interviews and the collection of observational data entirely, or
eliminating them for CWS wells;
Relying only on first stage stratification to select wells randomly both for domestic and CWS
wells; and
Eliminating random sampling at the third stage and instead selecting wells on a quota basis
within clusters.
Most of these approaches would have narrowed the focus and usefulness of the Survey's results, in
some cases drastically reducing the credibility of the results for use in regulatory programs, in other cases
running counter to repeated SAP recommendations specifically supporting the need for the data in question.
A3.2 Early Design Decisions5
Agricultural Pesticide Use
Originally, all types of pesticide usage were to be included in the Survey: agricultural, home and
garden, industrial, rights-of-way, commercial, and government. However, because no detailed information
existed even on the national level in 1984 for the last four categories, the decision was made to focus the first
stage data collection effort on agricultural pesticide use. The pesticide usage information task group formed
in 1984 examined 12 options for developing county-level agricultural use data for the pesticides of greatest
concern.
Two promising resources were the Census of Agriculture for crop acreage and Doane Marketing
Research, Inc. for regional pesticide usage for 28 major and minor crops. A small study of five pesticides used
on four crops in Wisconsin was conducted using Census of Agriculture data to determine if county-level
estimates of the agricultural use of specific pesticides were possible. The Doane data were also compared to
a database compiled by Resources for the Future, which was viewed on balance as no better suited to the
purposes of the Survey.
5 For more information on this section, see National Survey of Pesticides in DrinldDg Water Wells: A Review of the
Planning Process and The Data Quality Objectives, by Monica Nees and Cynthia Salmons, Contract No. 68-01-7350, Research
Triangle Institute, Research Triangle Park, NC, July 6, 1987.
National Pesticide Survey: Phase I Report
-------
Appendix A: Background A-9
The 1985 SAP reviewers approved the use of the Doane Marketing Research data for the first stage
of the Survey but stressed the need to supplement the information with "as much local information as possible
on the historical pattern of pesticide use in the immediate location of the well, the climatological, soil and
hydrogeological conditions, and the characteristics of the wells themselves."6 The subpanel steered EPA
towards obtaining local information from pesticide registrants and local pesticide distribution centers, State
departments of agriculture, and local offices of USGS, SCS and the cooperative extension service.
Subsequent debates centered on the type of localized crop and pesticide usage data that should or
could be collected in the field in the second or third stages of the Survey. By January 1987, consensus had
emerged that some information on either crops or pesticide usage was necessary for the relational analyses
to be performed, that there was a need for consistency in the resources and effort devoted to the different data
elements in the Survey, and that the pilot study would be useful in testing different data collection methods.
A number of options were considered at the time (some were tested in the pilot study), including:
Conducting a "drive-around" or "walk-around" informal look at cropping activity within a mile
or half-mile radius of each domestic well;
Using a farm questionnaire to collect more detailed information for the estimated 100-125
domestic wells that would be located on farms (this was dropped because it would yield good
information for too small a sample);
Using more sophisticated methods (e.g., a combination of satellite data and fly-overs) to
obtain detailed crop information for a specified radius around each domestic well (this was
dropped for reasons of cost); and
Using data from the National Agricultural Statistical Service (NASS). (NASS's area sampling
units were investigated as possible alternatives to the use of Census Bureau enumeration
districts and block groups as the basic sampling units of the Survey. The NASS sampling
units also carry agriculturally related land-use information that would have been useful for
the second stage stratification. However, it was determined that the NASS sampling units
would not have been compatible with the Survey design, which at the time relied on
enumeration districts and block groups.)
Ground-Water Vulnerability
The use of DRASTIC to characterize vulnerability of ground water was discussed extensively in early
stages of the Survey design. The National Water Well Association was asked to review EPA's use of
DRASTIC. They gave a positive review but cautioned that DRASTIC had not been designed to yield a single
numeric vulnerability score for an entire county and that this problem would be magnified in larger counties
that contained varied hydrogeological features.
The 1985 SAP subpanel urged EPA to use local deterministic data to supplement DRASTIC. In their
view, DRASTIC was capable only of a rough categorization of counties, it pertained only to shallow water
vulnerability, and some of its parameters overlapped or were not strictly related to actual vulnerability. The
SAP subpanel questioned why EPA did not use its own PRZM (Pesticide Root Zone Model) model which
6 The results of the SAP Subpanel were reported on October 9, 1985: Federal Insecticide, Fungicide, and Rodenticide
Act Scientific Advisory Panel Subpanel Review, Technical Review of the Preliminary Design of the National Survey of
Pesticides in Ground Water - Report of SAP Subpanel Recommendations, p. 4.
National Pesticide Survey: Phase I Report
-------
A-10 Appendix A: Background
"emphasizes climate and the upper layers of soil that are the primary determinants of pesticide persistence and
mobility."7
However, PRZM was considered by Survey designers to be a site-specific model that would have
required extensive modification to be applicable to county-scale parameters and to include relevant
characteristics that influence ground-water vulnerability. Other site-specific approaches were also available;
in general, however, there were very few large area approaches available and none with the consensus of
DRASTIC. Moreover, DRASTIC was being proposed for use as an average, relative ranking index, not as a
quantitative predictor. Its utility was primarily as an aid to stratification rather than as an explanatory variable
in its own right. Finally, to validate DRASTIC scoring, 220 counties were rescored independently by one or
two hydrogeologists. The results demonstrated that any variability in DRASTIC scoring was not large enough
to obscure actual differences in county-wide vulnerability categories and that the stratification needs for the
first stage of the Survey could be achieved with DRASTIC.8
The SAP reviewers also noted that the lack of well construction information and knowledge of well
casing integrity (particularly for domestic wells) could introduce even more uncertainty into the Survey's results
than DRASTIC. As part of the pilot study, EPA undertook a well depth validation study to determine the
significance of this source of uncertainty and concluded, as described in Section A3.3 of this appendix, that
reliable information on well depth was desirable but difficult to obtain.
Precision Requirements
During the summer of 1986, a number of options for defining the precision requirements of the Survey
were analyzed in detail. The options examined the size of the sample required to produce results of a given
power or relative standard error under varying assumptions about the "true" rate of contamination of drinking
water wells (e.g., 0.5 percent, 1 percent, 2 percent, and 5 percent). If reasonable estimates had been available
of the proportion of contaminated wells or systems, an efficient allocation of samples across strata would have
been possible. Since they were not available, the samples were allocated to ensure that frequencies of
occurrence as low as 0.5 percent could be detected at the national level.
Thus, the CWS sampling scheme devised was intended to provide a 90 percent probability of detecting
the presence of pesticides in the sample if 0.5 percent of all community water systems in the country contained
pesticides. This is fairly conservative; the precision of the results would of course be much better if more than
0.5 percent of all CWSs in fact contained pesticides.
The domestic well statistical design was intended to yield a range of probabilities of detecting
pesticides in different domains. The strategy was to employ a sampling design to obtain the highest power
(higher power being defined as higher probability of detecting occurrence) in the vulnerable/high use counties
and in the cropped and vulnerable subcounty areas of those counties, and to accept a slightly lower power in
the national level estimates. Areas of particular concern for pesticide detection were oversampled in order
to yield better, more precise estimates. Thus, the likelihood of the Survey detecting pesticides in domestic
wells was 63 percent at the national level (assuming a 1 percent rate nationally), but the Survey had a 97
percent chance of detecting pesticides in cropped and vulnerable areas (assuming a 0.25 percent actual rate
of pesticide presence in wells in these areas).
7 October 9, 1985, Report of SAP Subpanel Recommendations, p. 5.
8 Alexander, WJ. et al. Ground-Water Vulnerability Assessment in Support of the First Stage of the National Pesticide
Survey. EPA Contract No. 68-01-6646, Research Triangle Institute, Research Triangle Park, NC, 1986.
National Pesticide Survey: Phase I Report
-------
Appendix A: Background A-11
Analytical Methods Development
Among the issues in preparing for the Survey was the question of the reporting limits that could be
achieved, whether they were sufficiently low and whether lowering them still further could be done at a
reasonable cost. Issues of data quality, cost, and comprehensiveness were closely interrelated. The sensitivity
of the analytic methods declined slightly as the number of chemicals analyzed per method incfeased. In
addition, ODW and OPP had somewhat different concerns in the reporting limits issue. ODW was primarily
interested in determining whether pesticides existed at levels of health concern. OPP was also interested in
characterizing the nature and extent of pesticide presence in well water, and hence favored low analytical
limits. Conceptually, it was agreed that in setting reporting limits, EPA should strive for levels that are
commensurate with the levels found by States in other surveys and that the public is familiar with; that a
reporting level should be well below the health advisory level for the pesticide; and that reporting levels should
be capable of giving an adequate picture of the extent of pesticide presence in wells.
A3.3 Pilot Study/SAP Issues
Four major issues raised by the pilot study were presented to a subpanel of EPA's Scientific Advisory
Panel in September 1987.9 The first was the ability of the Survey to meet the objectives specified by EPA,
particularly with respect to the Survey data analyses. Additional issues for consideration included the well
selection method for community water systems, temporal variation, and the instability of certain analytes.
Meeting Survey Objectives
Number of Wells Sampled. A key objective of the NPS was to conduct "relational analyses"
comparing the results of the well sampling with cropping or pesticide use patterns and hydrogeological data.
There was concern that the anticipated sample size of approximately 750 domestic wells would be minimal for
conducting the relational analyses, especially for pesticides with highly limited or specialized geographical uses.
A sample size of 1,500-2,000 domestic wells would be preferable for the relational analyses.
Options considered to remedy the situation included adding more wells in the domestic well
component of the Survey (infeasible due to high costs per well) and collecting "second stage" cropping and
hydrogeological data for the community water system wells (which would not necessarily be as helpful as the
same data for the domestic wells). The SAP responded that the number of wells sampled could not be
increased by a significant enough number to affect the confidence of the interpretations. Based on this
response, EPA maintained the sample size.
Well Depth Data. The "well depth validation study" conducted during the pilot study identified and
examined well construction records available from householders, State and municipal archives, well drillers,
and the WATSTORE data base maintained by USGS. The purpose of the study was to determine whether
interviewees had sufficiently accurate knowledge or recall of well information; the extent to which records with
information on well depth and construction were available to corroborate or supplement householders' recall;
the location of these records; and the costs of locating them.
The conclusion of the study was that it would be extremely difficult to find reliable, direct information
on well depth for all sample wells. However, the States tended to have information on well construction in
the areas of the sample wells that could be used to infer well depth and the type of aquifer tapped for many
9 52 Federal Register 33991, September 9, 1987. The results of the SAP Subpanel review were reported on October
9, 1987: Federal Insecticide, Fungicide, and Rodenticide Act Scientific Advisory Panel Subpanel; A Set of Scientific Issues
Being Considered by the Agency in Connection With the National Pesticide Survey Pilot Study - Report of SAP Subpanel
Recommendations.
National Pesticide Survey: Phase I Report
-------
A-12 Appendix A: Background
of the sample wells. The SAP recommended continuing with the same "search" procedures in the full Survey,
contacting State officials and well-drillers, and when necessary, using inferential data from neighboring wells.
During second stage data collection for domestic wells, EPA implemented the panel's recommendations. For
CWSs, EPA determined that system operators would, in general, have better information on well depth than
other sources.
Second Stage Data. The Survey design called for collecting second stage data to help direct the
stratification of the sample. The second stage cropping data and DRASTIC scores were consistently available;
however, the second stage data were expensive to develop, and they covered large subcounty areas (e.g.,
hydrogeologic settings), which might not accurately reflect conditions around the sample well. Conversely, data
on well characteristics collected through Survey questionnaires and records searches were more accurately
targeted to the sample wells than the second stage information; however, there was concern that the data
simply might not be available for a large number of wells.
In an effort to be sure that the gains achieved by the second stage stratification were worth the
expense of the second stage data collection effort, EPA came up with a number of alternatives to the original
plans for using DRASTIC:
Use of type settings developed by the National Water Well Association to direct the second
stage stratification. This approach would reduce the hydrogeologic data gathering effort,
although at the cost of some accuracy.
A simplified DRASTIC scoring could be undertaken, using a smaller set of hydrogeological
characteristics (such as topsoil, slope, and depth to water) to categorize high, moderate, and
low vulnerability areas within each county. Again, this approach would reduce the
hydrogeologic data gathering effort, reducing accuracy as well.
Use of a geographic information system (GIS) to save time and money in averaging the
DRASTIC scores and transferring them to subcounty areas. The overall cost savings to the
project of using this approach were not clear.
The SAP recommended the following:
Continued stratification at the second stage but using a simpler approach than DRASTIC
scoring (e.g., type settings) and a refinement in collection of cropping data at the second
stage.
Refining and augmenting well-specific information collection, by asking well owners more
information, expanding the local area data collection effort, and asking for State help in
searching for well depth data.
If enough positives appeared after three months, pursuing added well-specific data collection:
DRASTIC scoring of half-mile radius around well; Survey of agricultural chemical dealers;
farm property Survey.
After careful consideration of the alternatives and SAP's recommendations, EPA decided to use a
simplified version of DRASTIC and a GIS program to conserve time and money. In addition, EPA expanded
the data collection effort at the second stage to include more information on crops and pesticide use. As part
of this data expansion, questions on well-specific characteristics and fanning practices were added to the
questionnaires.
National Pesticide Survey: Phase I Report
-------
Appendix A: Background A-13
Well Selection for Community Water Systems
The study design called for community water systems to be selected for the Survey from the Federal
Data Reporting System (FRDS) data base. The pilot study uncovered numerous problems with FRDS,
necessitating changes in the statistical design for the community water system component of the Survey.
The most serious problem encountered was with the FRDS data on number of wells per system.
According to the FRDS data base for 1984-85, 83 percent of CWSs had one single well; 7 percent had two
wells; and only 10 percent had three or more wells. Based on that information, the Survey was designed to
estimate the number of community systems in the country with at least one well containing pesticides. The
implication was that virtually every well in every selected system would be sampled and analyzed.
In the course of screening the systems for the pilot study, however, it was found that the selected
CWSs had many more wells than were shown on FRDS ~ on average 5.75 wells per FRDS record rather than
the expected 1.5 wells per system. Although one could not generalize from the pilot study systems, any
number substantially higher than 1.5 wells per system would have made sampling each well in each selected
system, as originally planned, extremely costly. (It would also have focused CWS sampling on fewer systems,
thereby limiting observations to fewer geographic areas of the country and reducing the precision requirements
of the Survey.)
To solve this problem, EPA proposed to develop a list of CWS wells (rather than systems), from
which a sample of CWS wells could be selected, using a three step design. In step one, a sample of systems
would be selected with equal probability from FRDS without replacement from within strata. The sample
would be screened by telephone to determine the number of wells operated by each sample system. This
information would be used at step two to select a subsample with probability proportional to size (number
of wells). At the final step three, a single well would be selected from each of the systems in the subsample.
The SAP found this approach generally acceptable and suggested additional options: cleaning the
FRDS data base prior to using it (considered too costly); continued oversampling at stage one; exploring the
"false negatives" problem (considered too costly); exploring the use of counting rules to deal with multiplicities;
and oversampling higher risk wells as well as systems. EPA followed the recommendation of the panel and
implemented the three step approach.
Temporal Variation
A statistical design issue affecting the appropriate interpretation of the Survey results for both the
community water system and the domestic well components of the Survey is the temporal or seasonal variation
of pesticide detection. The Survey was expected to be conducted over a two year period, but it was not
originally designed to provide a random sample over time, and the Survey design did not "control for"
seasonality. Because of laboratory capacity constraints, it did not appear feasible to eliminate all effects of
seasonality by condensing all the sampling into a small time period (i.e., three months).
Among the options considered by EPA were: (a) to develop a reasonable-cost method of distributing
the sampling for any given region of the country over the course of a year in order to incorporate temporal
variation into the sampling schedule; (b) to consider temporal variation in the Survey data analyses, but not
in the sampling schedule; and (c) not to pursue this issue further because of scientific uncertainties and the
nature of the Survey design.
The SAP advised randomizing the sampling over time using practical approaches, or, if randomizing
at the well level proved too costly, to randomize at the county, system, or State level. The SAP also advised
EPA not to overinterpret the results of the NFS in terms of temporal variation because the Survey data base
National Pesticide Survey: Phase I Report
-------
A-14 Appendix A: Background
was not designed to incorporate the wide variability in the factors that influence temporal changes. Based on
the panel's recommendations, EPA decided to randomize sampling of community wells at the well level and
domestic wells at the county level.
Analyte Instability
Limited time storage studies showed that some pesticide analytes appeared to be unstable during the
period between collection and analysis of the sample. EPA asked the SAP's opinion on the following
questions:
If there is a stability problem, is it appropriate to address the issue in the context of the
Survey? Or should the issue be addressed more appropriately as a separate research project?
If there is a stability problem, should unstable analytes be deleted from the Survey analyte
list? What criteria should be used to make such decisions? Would it be advisable to delete
the "problem analytes" from the list of Survey analytes, but have the laboratories continue to
test for them and report them if they are found and confirmed?
How should EPA report results to domestic well owners and community water system
operators, if the Agency suspects that certain analytes are unstable?
The SAP responded that this issue should be addressed in the context of the Survey, beginning with
a review of literature and EPA data bases and proceeding to time storage studies. The panel further advised
not deleting any analytes until the nature of the instability was better understood. If a real instability problem
was found that could not be adjusted for, the panel advised not to analyze or report data for those analytes.
A3.4 Communications and Role of States
Communications represent an integral part of any large-scale effort dealing with the public. The
National Pesticide Survey involved the cooperation of all 50 States, the active involvement of all ten EPA
Regional Offices, the services of eight different laboratories, the assistance of the U.S. Department of
Agriculture and the U.S. Geological Survey, and the good will of numerous individuals across the country.
A great deal of attention was therefore given to establishing communications mechanisms throughout
the course of the Survey, with four primary goals: (1) establishing and maintaining a mechanism of ongoing
information exchange; (2) ensuring the identification and involvement of all potential project participants and
interested parties; (3) ensuring adequate project information dissemination; and (4) ensuring adequate
collection of project-related information and the consistency and quality of the Survey results.
Role of the States. State agencies played a key role in implementing the Survey. State drinking
water supply agencies were asked to sample the selected community wells and to notify both domestic well
owners and community water system operators of the results of the NPS sampling. Both State drinking water
supply agencies and departments of agriculture were asked to serve as points of contact for communications
with interested parties and to provide follow-up and technical assistance.
This important role of the States was decided upon for several reasons. First, State agreement to take
samples would greatly reduce the costs (particularly transportation costs) to EPA of sending out sampling
teams across the country. (Otherwise, since the CWSs were to be sampled on a randomly-based schedule,
samplers would have to return to the same State a number of times, raising the costs of the Survey.) A second
and important reason was the existing working relationship between community water system and State
drinking water supply agencies. It was anticipated that having State agency staff act as the samplers would
National Pesticide Survey: Phase I Report
-------
Appendix A: Background A-15
result in better cooperation on the part of the CWSs and therefore provide more assurance that the samples
would be taken. State officials were also concerned about possible interference with their existing relationships
with CWS personnel.
Several disadvantages to this extensive State role were raised in planning the Survey. First, there was
the potential for inconsistency in sampling practices across States. Even if State officials were experienced
samplers, their methods would likely vary widely across States. To address this issue, EPA's contractors carried
out 54 intensive training courses in 45 States for EPA and State samplers. Second, there was the fear that
for reasons of insufficient resources or too many demands on their schedule, States would not be able to meet
the NPS schedule, thereby throwing off the efforts to spread sampling across the year as much as possible.
This was largely overcome through careful attention to communications (in terms of going over the schedule
with the States in advance and then monitoring the sampling dates) and by offering States a two-week period
in which to conduct the sampling.
States also played a prominent role in notifying well owners, operators, and householders of the results
of the well sampling. Householders and CWS owner/operators were likely to contact the State anyway when
they obtained their results, as were media and community officials. It was therefore decided that everyone
would be best served if the States were provided with the Survey results as soon as they became available, on
an ongoing, routine basis. State personnel also coordinated participation by county officials, who also played
a large role in responding to public inquiries about the Survey and in providing information about
hydrogeologic characteristics and agricultural practices. (The issue of confidentiality as related to the States'
involvement in notification is discussed in the section below.)
Numerous other communications issues were addressed during the course of planning the Survey,
particularly with regard to notification procedures. For example:
Notification by Mail or Telephone. Both mail and telephone have advantages and
disadvantages as a means of communication - telephone calls can be unnecessarily alarming
or intrusive but may communicate better; letters may not be read or understood but are more
permanent records of the sampling results. It was decided that in the event of a positive
rinding above health advisory levels, the Survey Director would call the State and Region to
alert them the same day as the laboratory reported the confirmed results to EPA This was
known as the Rapid Reporting System. The States would then be asked to call well owners
and operators to inform them of the results, followed up with a notification letter. Within
three days of receiving written confirmation of the results by the laboratory, EPA would
notify the State in writing of the findings. State contacts in the Departments of Health were
encouraged to send notification letters and supplementary information to well owners and
operators within one week of receipt of the results and to make follow-up calls if necessary.
Notifying Well Owners vs. Householders. Both for health reasons and potential legal
concerns, it was determined that both well owners and the tenants (where applicable) needed
to give permission for sampling and to be notified individually of the sampling results.
Media Attention. Attention was given to notifying the proper State and local officials in
each State prior to sampling in order to ensure that each level of government was aware of
EPA's presence before the Survey was reported in the media. Another concern, raised by the
Office of Management and Budget during its review of EPA's information collection plans,
related to the pilot study results. There was concern that EPA not publicize the pilot study
results, because they would not be representative of the national situation and could easily
be misinterpreted by the media or the public. (For this reason, the pilot study results were
not included in the pilot study evaluation reports.)
National Pesticide Survey: Phase I Report
-------
A-16 Appendix A: Background
Resampling. Concerns were raised in planning the Survey about the need for, cost, and
implications of resampling wells that were found to contain pesticide residues. Issues of
protocols and responsibilities for resampling were raised by the SAP as far back as 1985 -
for example, when should resampling be done and by whom, how should the results of
resampling be presented to well owners particularly in light of seasonal variations,
confidentiality concerns; and whether positive findings should be reported in the absence of
resampling. Several States raised concerns about taking any type of action with respect to
a drinking water supply based solely on one sample. It was agreed that States would conduct
the resampling in the event of a positive finding and that EPA's laboratories would conduct
the sample analyses. Because of the possibility that the results of the resampling would bias
results, however, the resamples were not included in the Survey analysis.
A3.5 Confidentiality
EPA's policy on the release of well sampling results of the National Pesticide Survey favored full
disclosure of information. However, issues of confidentiality of the results were raised early in the design of
the Survey and discussed extensively. The key issue was the degree to which EPA should protect participants'
privacy and encourage them to participate in the Survey.
A general policy on confidentiality was also needed to guide the handling of requests for information
under the Freedom of Information Act (FOIA), although each FOIA request must be considered on a case-by-
case basis. Under FOIA, EPA is required to release information unless EPA determines that one of nine
FOIA exemptions applies to the information. EPA cannot, for example, temporarily delay release of
information to allow a State time to respond to the situation. EPA also cannot delay release of information
until the end of the Survey when the results are considered "final." The exemption that was most clearly
applicable to NPS data is Exemption 6, which protects personal privacy.
The concern about confidentiality related only to domestic wells (none of the CWS data could be
considered to raise a personal privacy issue). The focus of the concern related to the names and addresses
of the Survey participants, as well as the Domestic Well Questionnaire data they provided and the results of
the sampling (to the extent these data could be identified with the specific participants).
The competing issues involved included the following:
Maintaining the States as Partners In the Survey. The partnership approach of the
National Pesticide Survey with the States was unique among Agency surveys. It helped
develop constructive working relationships between EPA and the States and fostered closer
relations among ground water, drinking water, and pesticide agencies. It also comported with
the States' general responsibilities for the water supplies of their residents. Many States
made it clear, however, that they would participate as partners only if they received full
information from EPA on the Survey, including the results of the domestic and community
well sampling, linked to the names and addresses of Survey participants.
Some States, however, also indicated that they could not keep the sampling results and
participant identities confidential. California, for example, is required by law to make all such
information available to the appropriate local health departments, who have varying policies
about releasing the information to the public. Other States were not constrained by law, but
by policy, or by data management systems that were not geared towards protection of
personal information.
National Pesticide Survey: Phase I Report
-------
Appendix A: Background A-17
The variations among States in confidentiality policies created several difficulties. First, to
the extent that information was in the public domain (or potentially so) in States that could
not or would not maintain confidentiality, the Agency's case for withholding the information
under FOIA request was greatly weakened. EPA would likely not be able to withhold in the
States in question and might not even be able to withhold in all States under FOIA.
A second issue was what EPA should tell participants about confidentiality, given that State
policies varied. If EPA could not make a strong assurance of confidentiality, it was not clear
whether EPA should give participants a written statement on confidentiality, or instead, not
raise the issue at all but instruct contractor personnel on how to answer any questions that
might arise.
Protecting the Privacy of Survey Participants. Some of the information asked on the
questionnaire could conceivably be considered sensitive or embarrassing or more generally,
the kind of information that people customarily maintain in confidence. EPA was also
concerned that participants be protected from intrusion and potential harassment by
commercial vendors and the media. It was thought that potential NFS participants might
require confidentiality out of fear that public disclosure of adverse information about their
property would lower their property values and make their homes more difficult to sell. Also,
it was thought some participants might be concerned about their personal liability for
contaminated water supplies - they might worry that their neighbors would find out about
pesticide detection and sue them or that the State would take action against them.
Ensuring the Statistical Integrity of the Survey. Assurances of confidentiality are
generally believed necessary to convince people to participate in a survey. Because of the
concern that people with a higher likelihood of having well water containing pesticides would
disproportionately refuse to participate in the Survey without a promise of confidentiality,
there was also concern that this could affect the statistical integrity of the Survey. Failure to
obtain an accurate national picture of the presence of pesticides would then run counter to
the larger interests of the public. Empirically, the results of the pilot study seemed to
indicate that confidentiality would not be a major concern on the part of individual
householders and that they would not mind sharing the analytic results with the States.
NPS staff also reviewed the experience of other EPA Surveys. The Radon Survey conducted
by the Office of Radiation Programs planned to release data only to homeowners, with none
of the data to be released to the State; individual household results were expected to be
withheld under FOIA The Asbestos in Schools Survey conducted by EPA's Office of Toxic
Substances inspected for asbestos in school buildings and kept all the information
confidential. Results were not even reported to the school boards themselves. After the data
base was complete and "clean," the Survey destroyed the names and locations of the schools.
Protecting the Public Health. From a public health perspective, it was anticipated that
communities where NPS sampling was taking place might have a strong interest in the
sampling results. For example, if only one aquifer was used for drinking water in the area
near an NPS well, and the sampling results gave reason to suspect wider contamination, the
health of the residents of the surrounding area might be at risk as well. This would argue in
favor of a more open NPS disclosure policy.
On the other hand, it was not clear that the public interest extended beyond the sampling
results obtained in the community, to the names and addresses of the Survey participants.
To strengthen the public health protection of the communities involved with the Survey, EPA
National Pesticide Survey: Phase I Report
-------
A-18 Appendix A: Background
could, for example, explicitly recommend to the States that they inform communities of
suspected problems with particular aquifers or particular types of wells but without releasing
the participant identities.
EPA considered a variety of options in an effort to meet these competing interests and goals. They
included the following:
1. That EPA defer to State law and policy in each State, in order to be consistent with the
State's standard procedures. However, this policy would not have been legally supportable
under FOIA. EPA is not permitted to withhold information merely because State law
protects the information.
2. That each well owner be allowed to decide whether or not to have the State receive the
sampling results. This policy would have been a logistical nightmare for data management
and notification; it would also have severely limited the State role in the NPS and required
EPA to offer follow-up assistance.
3. That all information (identities, sampling results, and questionnaire data) be maintained by
the contractors at their offices rather than at EPA, because from an operational point of
view, there is no necessity for EPA to maintain these records. Under this approach, the
records would not be subject to FOIA. This option was rejected as impractical for all except
the questionnaire data, given the desired involvement of the States and Regional Offices in
the Survey notification process.
4. That although EPA would not publicize participants' identities on its own initiative, the
Agency would determine in advance that participant identities are generally not withholdable
under Exemption 6 and consequently that EPA would generally make the information
available upon request unless there was reason to believe that the disclosure could be harmful
to the participants.
5. That EPA would seek to withhold participant identities except in response to written FOIA
requests where the Agency found that the public interest in disclosure of the data was
relatively stronger than the privacy interest.
6. That EPA would share the sampling results only with States that could withhold personal
information. States that could not or did not intend to protect personal information would
not receive the participant identities, and would receive the sampling results only in coded
form. In those States, EPA would notify participants of their results.
The option selected (#5) was incorporated into the confidentiality policy adopted for the National
Pesticide Survey (shown in Exhibit A-2). Appendix C describes the operation of the confidentiality policy and
reporting system during implementation of the full Survey.
National Pesticide Survey: Phase I Report
-------
Appendix A: Background A-19
Exhibit A-2
NFS Confidentiality Policy
EPA's policy on the release of well sampling results of the National Pesticide Survey
favors full disclosure of information. Results from the sampling of community water
system wells will be made public as soon as the State and CWS have been notified.
However, where the sampling results relate to the wells of private individuals, the
policy will be modified to protect individual privacy, while still permitting disclosure
of sampling results. EPA will also seek to obtain agreement from the States to
protect the confidence of personal information on Survey participants.
Efforts to maintain confidentiality include the following:
Owners of private wells will be asked to read and sign the Water Sample
Consent and Data Release Agreement Form which acknowledges that their
well sample results and questionnaire data will be shared with the State.
Spot checks of these forms will be conducted to ensure that they are being
completed.
Each State may be asked to provide a written statement of its ability to
maintain confidentiality. In those States where reasonable assurances
cannot be given, a revised release form may be used, giving the well owner
the option of having the sampling results go no further than EPA and the
contractor.
EPA will ensure that Survey records are not accessible by the name or
address of any individual. Files will be organized instead by an anonymous
internal code to be assigned to each well in the Survey (e.g., "Well #23, Well
#37'). The 'cross-reference' file linking the internal well code with the name
and address of the well owner will be accessible only through the well code.
Access to the cross-reference file will be restricted to those with specific
project responsibilities.
The contract laboratories will be legally bound to observe confidentiality
procedures. The contract laboratories will not have access to the names and
addresses of the well owners; samples will be linked to the well from which
they originated by the internal project code number only. Spot checks of
data management procedures will be conducted during the regular audits
of the laboratories.
Results of domestic well sampling and other information will be released in
reports prepared for EPA only in appropriately aggregated form (e.g., "X
percent of all wells...') or by anonymous well code. Personal information (i.e.,
names and addresses and other identifying details) will not be included in
EPA reports.
National Pesticide Survey: Phase I Report
-------
National Pesticide Survey
Appendix B: Statistical Design and Analysis
-------
Appendix B: Statistical Design and Analysis
B1 Introduction
This appendix describes the key elements of the statistical design and analysis for the National Pesticide
Survey. With respect to the statistical design, this appendix presents the following information:
the final precision requirements that the designers of the Survey established as the basic level
of confidence for the results of the Survey;
the mathematical procedures that were used initially to determine the numbers of community
water system wells and rural domestic wells that would constitute the sample of wells;
changes in statistical design based on the results of the pilot study and other reviews of the initial
Survey planning;
the final procedures used to screen wells and identify those wells from which data would be
obtained during the field sampling phase of the NFS; and
the actual response rates for the Survey (i.e., the numbers of wells from which the Survey
obtained data).
In addition to describing the design of the Survey, this appendix also explains the key statistical concepts
and implementation procedures that were used to produce the results presented in Chapter 6 of the Phase I
Report. In particular, this appendix discusses the following analytic topics:
scheduling of well sampling, and the analysis of whether temporal variation affected Survey
results;
procedures used to handle missing data and questionable data, including imputation procedures;
weighting procedures through which the wells actually sampled represent wells in the population;
procedures for calculating national population estimates based on well sampling results,
weighting, and other statistical procedures; and
a comparison of the results presented in Chapter 6 with the initial precision requirements of the
Survey.
As the first national survey of drinking water wells, the NFS involved a number of design assumptions
and innovations. In addition, at certain points the design was partially revised to reflect changes in goals or
procedures. The purpose of this appendix is to describe and state the design assumptions, identify the data
sources used in the Survey design, explain statistical concepts and procedures and provide the most important
mathematical equations underlying the Survey design and analysis.
Section Bl provides an overview of the Survey design goals and the basic sample design for the CWS
and rural domestic well surveys. Section B2 describes the first stage stratification that is common to both the
CWS and rural domestic well surveys. Section B3 provides further details of the sample design for the CWS
well survey and Section B4 similarly provides further detail of the sample design for the rural domestic well
survey. Section B5 discusses the temporal allocation of sampled wells in both surveys. Section B6 provides
details of the statistical procedures required to produce the results presented in Chapter 6 of this report and
discusses the actual precision achieved by the survey in comparison to the precision requirements specified
for the design.
-------
B-2 Appendix B: Statistical Design and Analysis
61.1 Key Elements of the Survey Design
The Survey was designed to collect samples from drinking water wells across the country. The Survey
also collected information about well construction, physical surroundings, and land use for investigation of
possible relationships among analyte detections and ground-water vulnerability and pesticide use. The Survey
was not designed to assess ground water or drinking water. Well water samples were collected before
treatment.
Originally, the Survey was designed to provide estimates of characteristics of community water systems,
and rural domestic wells. Before determining the number of community water systems, and rural domestic
wells to include in the Survey, the respective populations needed to be defined, and the population sizes
estimated.
Community water systems are defined under the Safe Drinking Water Act and its regulations as systems
providing piped drinking water with at least 15 connections, or serving a population of at least 25 persons who
are permanent residents of the service area. For the purposes of the Survey, a CWS was further defined as
a system with at least one working well that is used to obtain drinking water. The estimate of the number of
community water systems was based on the Federal Reporting Data System (FRDS) listing for the period from
July 1984 through June 1985. FRDS is a database that is maintained by the EPA Office of Drinking Water
that contains information on public water supply systems. The number of community water systems was
initially estimated to be 50,771.
Rural domestic wells were defined for the Survey as drinking water wells that supply occupied housing
units located in rural areas of the United States, except for wells located on government reservations. The
United States Bureau of the Census (Census Bureau) defines rural areas as areas outside of incorporated or
unincorporated places with a population of 2,500 or more, or as areas outside urban fringe areas (for more
information on the definition of urban and rural areas see Appendix C). The number of private rural domestic
wells was initially estimated as 13,102,000 based on the 1980 Census.
The Survey uses a stratified sampling design. In general, stratified designs are used to improve the
precision of parameter estimates by controlling the distribution of the sample. First-stage stratification was
common to both CWS and rural domestic well surveys. Each of the 3,137 counties or county equivalents in
the United States was categorized by pesticide use and ground-water vulnerability. This ensured that samples
were taken from wells located in areas with different levels of pesticide use and ground-water vulnerability,
and provided a mechanism for oversampling wells in areas that had greater a priori expected detection rates.
The first stage stratification was the same for both the community water systems and rural domestic wells, and
is explained in section B2. Further details of the sampling designs for CWS and rural domestic wells are
provided in Sections B3 and B4 respectively.
The Survey design also identified particular subgroups of the populations of drinking water wells that
were of special interest. These subgroups, termed "domains of interest," constituted categories of wells about
which it was particularly important to obtain information. These domains were based on the various strata
introduced into the Survey design. The seven domains of interest were:
All CWS wells in the United States;
CWS wells in counties with high ground-water vulnerability;
All rural domestic wells in the United States;
Rural domestic wells in counties with high pesticide use;
Rural domestic wells in counties with high ground-water vulnerability,
Rural domestic wells in "cropped and vulnerable" county subregions; and
Rural domestic wells in counties with high pesticide use and high ground-water vulnerability.
National Pesticide Survey: Phase I Report
-------
Appendix B: Statistical Design and Analysis B-3
Once the populations had been established, the number of samples necessary to provide specified levels
of precision were calculated. A detection rate was assumed for each of the seven domains, and the precision
requirements (in terms of variances, relative standard errors, detection probabilities, and confidence intervals)
and estimated costs for each stage were specified (the costs were specified in terms of dollars per well). The
sample allocation was calculated to achieve the required precision for the least overall cost. The mathematical
procedure used to perform this constrained optimization problem is based on Kuhn-Tucker theory. The
mathematical procedure for the CWS and rural domestic well Surveys is explained in Sections B3 and B4
respectively.1
The Survey was designed to estimate detection rates for the presence of any of the pesticides or
pesticide degradates (101 pesticides and 25 pesticide degradates). The precision requirements for the Survey
were specified for the expected detection of at least one of these pesticides or pesticide degradates, rather than
for the detection of individual analytes. Survey results for the presence of at least one pesticide are in
accordance with the pre-specified precision requirements. The Survey was not designed to provide estimates
for individual pesticides or pesticide degradates, and in general, a much larger sample would be needed to
achieve the specified precision for individual pesticides given the assumed detection rates. Furthermore, if the
Survey had been designed to determine the presence of any one pesticide then a different stratification would
have been used (for instance, highlighting use areas for that pesticide specifically). For these reasons, results
presented in Chapter 6 for individual analytes generally contain larger relative standard errors than results
presented for all pesticides together. If actual detection rates in the Survey are different from the assumed
rates used for the survey design, then the precision of survey estimates will be different than original precision
specifications.
B1.2 Pilot Study and Initial Design Review
EPA conducted a pilot study from March to August, 1987, to test the major components of the Survey
and to provide an opportunity for any necessary revisions or modifications before the full Survey was
implemented.2 The pilot study was conducted in three States: California, Minnesota, and Mississippi. The
pilot study led to improved procedures and techniques in several areas. In October, 1987, the FIFRA
Scientific Advisory Panel (SAP) subpanel commented on several topics, including the proposed method of
selecting CWS wells for water sampling, and the issue of temporal variation (seasonal patterns of pesticide
application, precipitation, and other similar phenomena) as a factor influencing when, and in what
concentration, pesticides and nitrate impact drinking water wells.
In response to the SAP subpanel's recommendations on methods of selecting CWS wells, the CWS
survey design was altered from a survey of community water systems to a survey of CWS wells. Further details
are given in Section B3.
The NFS pilot survey is described in Appendix A There were three parts of the pilot survey that
directly influenced modifications made in the statistical design of the full Survey: adequacy of the CWS frame,
use of an area probability design for sampling within first-stage units, and selection of wells to be sampled
within some of the counties.
The frame for the CWS survey was the 1984 Federal Reporting Data System (FRDS) file. This multi-
purpose data base includes all community water systems reported to the Federal Government by State
1 Parts of the sample allocation procedure presented in this Appendix were originally presented in the Research Triangle
Institute (RTT) report entitled "National Survey of Pesticide Results in Community System and Rural Domestic Wells -
Sample Allocation Report," April 6, 1988, under EPA contract number 68-01-7350.
2 National Pesticide Survey Pilot Evaluation Report, Office of Drinking Water and Office of Pesticide Programs, U.S.
EPA, September 1987.
-------
B-4 Appendix B: Statistical Design and Analysis
agencies. At the time of the pilot survey it was believed that the variable "number of water sources" indicated
the number of wells used by those systems that stated they used ground water. Since this number averaged
under two per system, the CWS was originally designed as a sample of systems, where every well in a selected
system would be sampled. The results of the pilot survey indicated that many systems interpreted this variable
to mean ground water and surface water as two water sources, regardless of how many wells or streams were
being used. The average number of wells for the systems included in the pilot sample was five-and-a-half.
Multiple wells from the same system are likely to much closer together than wells from different systems and
therefore more correlated with regard to pesticide contamination. Thus it was decided to switch to a survey
of wells used by community water systems.
The CWS pilot survey also discovered that in at least one state, California, the FRDS file contained
numerous multiplicities. Some of the "systems" on FRDS were actually water companies that operated many
smaller community water systems that could be located many miles from each other and administered
separately. It was also possible for a system to be listed twice, once by itself and once as part of a larger water
company. Finally, some systems had merged with other systems over time.
Finally, the pilot survey found ineligible systems on the FRDS frame. Ineligible systems were caused
by numerous factors. A system might no longer be operating; it might never have qualified as a community
water system; or it might have been impossible to locate the system.
As a result of these findings it was decided that not all wells could be economically sampled from every
sampled system; thus the CWS would be a survey of wells used by CWSs, not a survey of systems. The need
to know how many wells a system used, combined with the need to sort out multiplicities, led to the use of
a two-stage sample of CWSs for the main survey, with a screener survey conducted of the systems selected in
the first-stage which tried to discover multiplicities, confirmed eligibility for the survey, and determined the
number of wells used by the system.
In response to the SAP subpanel's concerns about temporal variability, the Survey design randomly
assigned available time periods to the wells selected for sampling. This randomization was performed within
the design strata and across the CWS well and rural domestic well surveys. Further details are provided in
Section B5.
B2 First-Stage Stratification
The first-stage stratification was the same for the CWS and rural domestic well Surveys. Each of the
3,137 counties or county equivalents in the United States was categorized by pesticide use and ground-water
vulnerability.
EPA developed pesticide use estimates for each county based on information about crop acreages from
the 1982 Census of Agriculture and on private marketing data on pesticides (provided by Doanes Marketing
Research, Inc.). Total acreage using 52 specific chemicals was calculated for 28 crops.3 The list of chemicals
used is provided in Exhibit B-l. Exhibit B-2 provides a list of the crops. The county-level use estimates were
derived by a simple proportioning scheme. For each crop, data on acreage for that crop in each county in the
United States (except Alaska) was extracted from the 1982 Census of Agriculture. The ratio of the county
acreage for a crop to the corresponding regional acreage for that crop gave a proportion that was multiplied
by the regional estimates of individual pesticides used on that same crop. Not all counties grow all of the 28
crops for which regional nesticide use data was obtained. Furthermore, not all pesticides are used on all crops
nor is a pesticide used on a crop always used in all regions growing that crop. The estimation process assigned
zeros to many of the estimates. A series of use estimates were prepared for each county and for each
pesticide/crop pairing for which regional use data were available.
3 Derivation of County-Level Pesticide Usage Estimates For Design of the Groundwater Pesticide Survey, Economic
Analysis Branch, Benefits and Use Division, Office of Pesticide Programs, September 1985.
National Pesticide Survey: Phase I Report
-------
Appendix B: Statistical Design and Analysis B-5
Exhibit B-1
Chemicals Used for County-Level Characterization of Pesticide Use
Acifluorfen
Alachlor
Aldicarb
Ametryn
Atrazine
Bentazon
Bromacil
Butylate
Carbaryl
Carbofuran
Carboxin
Chloramben
Chlordane
Chlorothalonil
Cyanazine
Cycloate
Dalapon
DBCP
Dacthal
Diazinon
Dicamba
2,4-D
Dieldrin
Dinoseb
Diphenamid
Disuffoton
Diuron
EDB
Fenamiphos
Fluometuron
Fonofos
Hexazinone
Mateic Hydrazide
MCPA
Methomyl
Methyl Parathion
Metolachlor
Metribuzin
Oxamyl
Paraquat
PCNB
Picloram
Prometon
Pronamide
Propazine
Propham
Simazine
TrifJuralin
Trialiate
2,4,5-T
Tebuthiuron
Terbacil
Exhibit B-2
Crops Used for County-Level Characterization of Pesticide Use
Alfalfa Tomatoes Apples
Corn Tobacco Pears
Cotton Sugar Cane Peaches
Sorghum Sugarbeets Cherries
Soybeans Sunflowers Plums and Prunes
Wheat Rice Grapefruit and Oranges
Pasture Dry Beans and Peas Lemons
Peanuts Walnuts Grapes
Potatoes Pecans
Sweet Corn Almonds
The next step in the process was to do a simple summation of the county-level use estimates across all
crops for each of the pesticides. This step provided county-level use by pesticide. The basic underlying
assumption of this approach was that among growers of a crop in a region there is a homogenity of pesticide
use practices. Summing the estimates and dividing by the total area of the county provided an index that could
be used to compare the relative intensity of agricultural pesticude use among counties. Agricultural pesticide
use was described as high, moderate, low, or uncommon, based on the following percent cut-offs.
High Pesticide Use: Pesticides applied to 75 percent or more of the land area of the county.
Moderate Pesticide Use: Pesticides applied to at least 30 percent and less than 75 percent of
the land area.
Low Pesticide Use: Pesticides applied to at least 5 percent and less than 30 percent of the land
area.
Uncommon Pesticide Use: Pesticides applied to less than 5 percent of the land area.
-------
B-6 Appendix B: Statistical Design and Analysis
Within each of these levels of pesticide use, counties were further stratified according to their ground-
water vulnerability.4 Ground-water vulnerability was estimated using the Agricultural DRASTIC index. As
described in Chapter 2, the DRASTIC Index is a numerical scoring system devised by the National Water Well
Association (NWWA) based on the weights, ranges, and rating of seven factors. High scores were selected
to represent areas where physical characteristics afforded little protection from ground-water pollution. An
Agricultural DRASTIC Index was designed by NWWA for which the specific concern was the application of
pesticides to the land surface. The Agricultural DRASTIC Index differs from the DRASTIC Index only in the
assignment of different weights for the seven factors. The Agricultural DRASTIC Index places a higher
weighting on topography (percent slope) and soil media than does the DRASTIC Index. Topography is
important when considering the application of pesticides to the land surface because slope controls the
likelihood that a pollutant will run off or remain on the land surface in one area long enough to infiltrate.
The attenuating processes (filtration, biodegradation, sorption, or volatilization) that may be available in the
upper few feet of the soil media can also influence the migration of pesticides.
Although the DRASTIC system was designed by NWWA as a screening tool for county-size areas down
to a resolution of approximately 100 acres, an application of a modified system was considered suitable for
the county-level assessment needed to meet the first stage of stratification of the National Pesticide Survey.
Recognizing the physical variation that exists in areas as large as counties, a county data collection form was
designed that allowed each factor to be coded on a percentage basis.
Coders performed DRASTIC scoring using two slightly different approaches that were reviewed and
coordinated in quality control procedures. In one, the seven hydrogeologic factors for each county were
divided among several coders so that a few coders ranked the factors dealing with geologic criteria (depth to
water, aquifer media, impact of the vadose zone media, and hydraulic conductivity), others ranked soils, and
others ranked the topography and net recharge. The composite numerical score for each county coded using
this system may have, therefore, resulted from the combined efforts of at least three individuals. This division
of labor was intended to promote consistency in the coding of specific hydrogeologic factors among counties.
Hydrogeologic factors such as depth to water, aquifer media, soil media, and vadose zone media were ranked
based on literature review and conversations with State and regional representatives (e.g., State Geological
Surveys, U.S. Geological Survey, and Soil Conservation Service). Net recharge was ranked based on a formula
to incorporate information on surface water baseflow, published recharge rates, precipitation, topography, and
soil media. Topographic slope was determined using USGS 2° Topographic Maps (1:250,000) and a
standardized slope estimation technique. Hydraulic conductivity estimates for each county were based on
literature review, the contractor's knowledge of the aquifer media, and general ranges of hydraulic conductivity.
The second system assigned entire States to individual coders. Each coder was therefore responsible
for ranking all seven hydrogeologic factors for every county within a State. Such an approach was used to
promote overall consistency among the DRASTIC scores of all the counties within a State. Six of the seven
DRASTIC factors (all but topography) were ranked in this system based on literature review, discussions with
State and regional representatives, and the applicable field experience of the coder. Topographic slope was
estimated based on information from USGS 2° Topographic Maps and from individual State general soils
maps.
The quality control process randomly selected 7% of the counties after coding and assigned them for
receding to coders who followed the other approach. A total variable measurement error estimate then was
computed. In addition, a randomly chosen set of this subsample was also recoded by the prime contractor.
Finally, 6 counties were coded by several nationally recognized hydrogeologists as a qualitative check.
4 Ground-Water Vulnerability Assessment In Support Of the First Stage of the National Pesticide Survey, Research
Triangle Institute, February 14, 1986.
National Pesticide Survey: Phase I Report
-------
Appendix B: Statistical Design and Analysis B-7
Originally-coded county data were edited for completeness, correctness, and legibility. The majority of
the fields on the forms were found to be complete but occasional omissions were not uncommon for a given
DRASTIC factor. The forms were examined for correctness with respect to the percentage allotted for a given
factor.
Following the editing process, the forms were screened for reasonableness. Total numerical scores were
compared with those of adjacent counties or States, and the individual DRASTIC factors were compared for
reasonable agreement on a given form. General agreement between aquifer media, the impact of the vadose
zone, and the hydraulic conductivity was found in many instances. Completed forms were occasionally
reviewed for reasonableness with respect to known regional hydrogeologic conditions by independent reviewers.
Upon completion of the editing and preliminary review process, all fields on the forms were keyed and
the forms were verified by the key operators. The keyed data were copied to EPA's National Computer Center
(NCC) at Research Triangle Park. Data were read from tape into Statistical Analysis System (SAS) data sets.
Three numerical scores were calculated from the data set:
Total Score. An unweighted score corresponding to the summation of the highest percentage
for the seven DRASTIC factors.
Weighted Score. A weighted score corresponding to the summation of all percentage ratings
for the seven DRASTIC factors. The weighted score accounts for the variability that occurs in
areas as large as counties. In counties with little variability, little difference exists between the
total score and weighted score.
VARSCORE. Represents the weighted score including (+ or -) the index of variability, where
used. The VARSCORE or the adjusted VARSCORE is considered to be the most appropriate
score for the purposes of the NPS in that it best accounts for intracounty variability.
The index of variability was used because the Agricultural DRASTIC Score that was determined is based
on typical values that do not account for heterogeneities such as degree of fracturing in the aquifer media or
the vadose zone. The term referred to as "index of variability" can be used to indicate the hydrogeologist's
opinion on the variability that might be associated with the previously determined score. Aquifer media and
the impact of the vadose zone are the primary DRASTIC factors where index of variability may apply.
Ground-water vulnerability was classified as high, moderate, or low, separately within each level of
pesticide use. The proportions of high, moderate, and low ground-water vulnerability could then be set equally
for each level of pesticide use. The high ground-water vulnerability stratum contained approximately 25
percent of all households with wells, as did the low ground-water vulnerability stratum. The middle group is
about twice as large as either of the other two. (These numbers are not exact as the DRASTIC index was
scored for whole counties and, for instance, counties were included in the high ground-water vulnerability
stratum when the number of corresponding households with wells was nearest to 25%.) Three groups of
counties are identified by ground-water vulnerability within each pesticide stratum. The stratum definitions
are summarized in Exhibit B-3. The actual DRASTIC index values that define the ground-water vulnerability
strata are also included in this exhibit.
Klntlnnal P*»e*irM» ftiiruau* Dhaea I Dan/ir*
-------
B-8 Appendix B: Statistical Design and Analysis
Exhibit B-3
Strata for the NPS First-Stage Survey Design
Agricultural
Pesticide Use
High
Moderate
Low
Uncommon
Ground-Water
Vulnerability
High
Moderate
Low
High
Moderate
Low
High
Moderate
Low
High
Moderate
Low
Defining
DRASTIC Scores
148 and over
116 to 147
115 and under
163 and over
113 to 162
112 and under
159 and over
132 to 158
131 and under
152 and over
121 to 151
120 and under
From this point, the CWS and rural domestic well surveys follow separate, though sometimes similar, paths.
The CWS well survey design is further detailed in the next section, while further details of the rural domestic
well survey are provided in Section B4.
B3 CWS Well Survey Design
The sampling design for the CWS survey was originally a stratified, two stage design for community
water systems. As the Survey developed, recommendations were made for changing the CWS survey from a
survey of community water systems to a survey of CWS wells. As a survey of wells, the CWS survey is
regarded as a three stage design where the third stage of the design selects a well from the systems sampled
at the second stage. This section presents elements of the original design and the design changes that were
made as a result of the change in target populations. The first stage stratification for the Survey was presented
in the previous section. The stratum sizes in terms of the number of community water systems in the
population are presented in Exhibit B-4.
Section B3.1 outlines the precision requirements in terms of variance constraints imposed on the two
CWS domains of interest Sections B3.2 and B3.3 outline the variance and cost components of the
optimization procedure used to determine the desired sample sizes. The optimization procedure is explained
in Section B3.4. Section B3.5 explains non-response adjustments to the sample allocation. To this point, the
discussion concerns the sample allocation according to the design requirements. Section B3.6 and B3.7
describe some of the implementation problems and the actual survey design that was used.
National Pesticide Survey: Phase I Report
-------
Appendix B: Statistical Design and Analysis B-9
Exhibit B-4
Community Water System First Stage Stratum Sizes
Pesticide Use
Stratum
High
Moderate
Low
Uncommon
Total
High
Moderate
Low
Uncommon
Total
Vulnerability
Stratum
High
Moderate
Low
High
Moderate
Low
High
Moderate
Low
High
Moderate
Low
High
Moderate
Low
Total
Total
Total
Total
Total
Number of FRDS
Records*
1,310
2,669
1,432
2,183
3,352
2,559
3,578
7,199
5,545
3,472
9,511
7,961
10,543
22,731
17,497
5,411
8,094
16,322
20,944
50,771
* Entries are the numbers of FRDS records that are identified as
active community water systems with at least one well or ground-water
source operated by the system. Seven records for which county
identification was missing are excluded.
Source: Mason, R.E., C.H. Benrud, and V.G. lannacchione.
Stratification Proposed for the National Pesticide Survey, Research
Triangle Institute, Report No. RTI/3030/03-06F, 1986.
National Pesticide Survey: Phase I Reoort
-------
B-10 Appendix B: Statistical Design and Analysis
B3.1 Precision Requirements
The sampling design for the CWS Survey was initially developed to provide estimates for drinking
water systems rather than the individual wells. Original expectations were that the number of wells per system
would average about 1.5. However, the NFS pilot survey indicated an average of about 6 wells per system.
Accordingly, individual CWS wells replaced active community water systems as the basis for the design.
The precision requirements were stated in terms of the maximum values of the sampling variances to
be associated with estimates of system level proportions of a specified size.5 As a result of the SAP subpanel
recommendations, the precision requirements were changed to represent constraints on well level
proportions.6 It is these constraints that are presented in Exhibit B-5. The relative domain size is an
assumed percentage of all CWS wells in the population that contain at least one Survey analyte. For example,
0.5% of all CWS wells were assumed to contain detectable levels of at elast one pesticide. The precision
requirements are specified in terms of the variances to be associated with sample estimates for the specified
relative domain sizes. Separate variance constraints were imposed on the two CWS domains of interest, as
Exhibit B-5 indicates.
Exhibit B-5
Precision Requirements for the Community Water System
Well Survey
Domain Description
Item
Value
1. CWS wells nationally
Relative domain size
Relative standard error
Detection probability
Confidence interval
0.005
0.658
0.90
0.0-0.012
2. CWS Wells in counties with high ground-
water vulnerability
Relative domain size
Relative standard error
Detection probability
Confidence interval
0.001
1.040
0.60
0.0 - 0.003
They can be interpreted as follows, for example:
The relative standard error for the detection of pesticides in CWS wells nationally in 0.658;
Requiring a 90% probability of detecting the presence of at least one pesticide in the sample
of CWS wells nationally if 0.5% of such wells contain pesticides;
The 95% upper confidence bound for the proportion of CWS wells nationally that contain
pesticides is 1.2% if 0.5% contain pesticides; and7
5 Mason, R.E., et al., 1988, National Pesticide Survey Pilot Evaluation Technical Report, Research Triangle Institute,
report number RTV7801/06-02F, 405 p.
* Mason, R.E., and Lucas, R.M., 1988, National Survey of Pesticide Residues in Community System and Rural Domestic
Wells, Sample Allocation Report, Research Triangle Institute, report number RTV7801/04-04F. Terms used in this report
have been retained for purposes of consistency and documentation.
7 Confidence intervals were not design requirements and were calculated using a normal distribution approximation.
National Pesticide Survey: Phase I Report
-------
Appendix B: Statistical Design and Analysis B-11
Requiring a 60% probability of detecting the presence of at least one pesticide in the sample
of CWS wells in counties with high vulnerability if 0.1% of CWS wells nationally are in
counties with high vulnerability and contain pesticides.
Other quantities derived from the variances were used to evaluate alternative survey designs and are presented
in Exhibit B-5. They are:
1. Relative Standard Errors. The proportion that the square root of the variance (the
standard error) is of the relative domain size (also referred to as the coefficient of variation):
RSE(Pd) = [VAR (Pd)f I Pd
2. Detection Probabilities. The proportion of repeated samples of size 8 (using sample
random sampling) in which at least one well in that domain containing pesticides is observed:
DP (Pd) = 1 - [l - P
where
e =
VAR (Pd)
3. Confidence Intervals. The proportion of repeated samples for which the computed interval
contains the relative domain size (in this case, approximate 95% confidence intervals). The
lower bound of the confidence interval is the value, P^, that most nearly satisfies the equality
in the constraint:
0.025
and the upper bound of the confidence interval is the value, Pu, that most nearly satisfies the
equality in the constraint:
p.'(i-p.)fl--,o.o25
The estimates, Pd, used in the above equations are the assumed relative domain sizes (indexed by d for the
domain). B is a surrogate for the sample sizes required to satisfy the precision constraints, and a is the sample
estimate of the number of wells containing at least one pesticide.
The formulas are presented assuming that the underlying distribution for the relative domain size is
binomial. In fact, the populations are finite so the underlying distribution is hypergeometric, but the
population sizes are large enough that use of the binomial distribution approximation to the hypergeometric
distribution is appropriate. The confidence intervals presented in Exhibit B-5 used the normal distribution
approximation to the binomial distribution. When the proportions are extremely small this approximation
is not justified and the binomial distribution should be used directly. Relative standard errors and detection
probabilities presented in Exhibit B-5 are a more accurate reflection of the precision requirements for the
Survey.
National Pesticide Survev Phaae
-------
B-12 Appendix B: Statistical Design and Analysis
The CWS survey design can be regarded as a three stage design, where the first and second stage
sampling units are community water systems and the third stage sampling units are CWS wells. Initially a
large first stage sample was selected with conditionally equal probabilities, given the design stratum. This
sample is termed the "screener" sample. The number of operable wells contained in each of these systems was
determined by talking with the system operators. Given this information, a second stage subsample of systems
was selected with probability proportional to number of operable wells in the system. This sample is termed
the "main" sample. At the third stage, a single sample well was selected at random from each system in the
main (second stage) sample. (Note that if a system was selected twice at the second stage then two wells were
selected at the third stage, etc.)
The sample allocation model takes the three design stages into account. That is, appropriate sample
sizes were calculated for each stage based on the precision requirements (the sample sizes are the same at the
second and third stages). The procedure for calculating the sample sizes necessary to satisfy the precision
requirements for the CWS well survey required two components, a variance model that accounts for the
precision requirements and the three design stages, and a cost model. Section B3.2 describes the variance
model; section B3.3 describes the cost model. Output from these two models, the variance constraints and
the cost estimates, were used in a nonlinear programming procedure based on Kuhn-Tucker constrained
optimization theory to derive the sample allocation that satisfies the precision requirements for the minimum
cost. Application of this procedure is described in Section B3.4.
B3.2 Variance Model for the CWS Well Survey Design
The estimator of the mean proportion of CWS wells with detectable levels of at least one pesticide
that was used to design the Survey was modeled within each domain as:
12 N(f>
Pd = E E S(i,j) Pd(i,j) I N
where the subscript d denotes the domain of interest (either CWS wells nationally or CWS wells in counties
with high ground-water vulnerability); i indexes the twelve first stage strata (see Exhibit B-3); ; indexes the
community water systems within strata; N represents the total number of community water systems (assumed
N=50,71l); N(i) represents the number of community water systems contained in the i first stage stratum
(see Exhibit B-4); S(i,j) represents the number of wells contained in they system in the i strata; and Pd(i,j)
represents the estimate of the proportion of wells in domain d with detectable levels of at least one pesticide
in the (i,j)tb stratum.
National Pesticide Survey: Phase I Report
-------
Appendix B: Statistical Design and Analysis B-13
The variance (VAR(P^) can be expressed as the sum of the variance components associated with the
two stages of sampling as follows:
where:
12
VAR
j (< J) / #(<) =
estimated proportion of all CWS wells that are in first stage
stratum z of domain d, and have detectable levels of at least one
analyte.
(OP m
= E S(ij) Pfij) I N(f) = proportion of all CWS wells that are in first stage stratum z of domain
'"l d, and have detectable levels of at least one pesticide.
p = intracluster correlation among wells within systems.
f(k), g(k) are functions of the distribution of the number of wells per system assumed for the CWS well
survey design (see Exhibit B-6 below).
n^(i) denotes the number of community water systems selected into the first stage (screener) sample.
n2(i) denotes the number of community water systems selected into the second stage (main) subsample.
The first variance component (Fj) represents the variance associated with the first stage of sampling
(i.e., sampling systems from the population).
The second variance component (P^) represents the variance associated with the second stage of
sampling (i.e., sampling systems from the first stage subsample).
The third variance component (F3) represents the variance associated with the sampling of only one
well from each of the second stage subsamples.
National Pesticide Survey: Phase I Report
-------
B-14 Appendix B: Statistical Design and Analysis
Exhibit B-6 shows the assumed distribution of the number of wells per system used in the community
system design. The factors f(k), g(k) are defined with respect to this exhibit as:
/,(*)
*(*)
1
*(*)
and
*(*)
1-1
m
Exhibit B-6
Assumed Distribution of the Number of Wells Per System
Used in the Community System Design
Group Index
Value (k)
1
2
3
4
5
6
Total:
Mean (m)
Mean Number
of Wells in
Group K (M(k))
1
2
3
4
5
10.5
-
3.56
Percent Relative
Frequency of Number
of Systems (R(k))
22.0%
31.7%
17.1 %
7.3 %
7.3%
14.6 %
100.0%
-
The estimated values ofM(k), R(k), and m are based on the results of the pilot study, which was the
only reliable information available. For instance, 22.0% of community water systems have 1 well, and 14.6%
of systems have more than 5 wells. The average number of wells per system is 3.56, and the average number
of wells per system for systems that have at least 6 wells is 10.5. The value of p was assigned to be 0.1 for
all strata and both domains. This assumed intracluster correlation was the most arbitrary part of the sample
allocation process. If a smaller value of p had been specified then the variance constraints could be achieved
with a smaller sample size. If a larger value had been specified, then a larger sample size would be required.
National Pesticide Survey: Phase I Report
-------
Appendix B: Statistical Design and Analysis B-15
B3.3 Cost Model for the CWS Well Survey
Costs were estimated to calculate a cost efficient sample size that would achieve the precision
requirements specified by EPA Actual costs of the Survey were different than these initial estimates. The
cost model for the CWS well survey was specified in the form:
12
C(B) = £ [n,(OC,(0
M
where n^i) and n2(i) are the sample sizes for the first stage (screener) sample and the second stage (main)
sample, C(n) denotes the total cost of the survey (the vector n represents the sample sizes for the first two
design stages and all twelve first stage strata), and
Crfi) = the first stage per system cost ($13.21) of sampling frame construction and
stratification, sample selection, data collection, and data processing and analysis
averaged over all systems in the i stratum;
C2(i) = the second stage cost per system ($508.19); and
C3(i) = the average third stage cost per well ($3,013.84) of sampling and collecting,
preserving, transporting, and chemically analyzing the water samples.
B3.4 Sample Allocation for the CWS Well Survey
Given the variance constraints and the cost component, a nonlinear programming procedure based
on Kuhn-Tucker constrained optimization theory was used to derive the sample allocation that achieves the
precision requirements for the minimum cost.
This procedure can be expressed as minimizing the following function:
F(a) = C(a) + £ A, [Var (?) - Vd]
where:
Ad = a LaGrange multiplier for the d domain corresponding to the d precision
constraint.
Fd = variance constraint imposed for the d domain (as given in Exhibit B-5).
National Pesticide Survey: Phase I Report
-------
B-16 Appendix B: Statistical Design and Analysis
The sample size solutions are found by taking partial derivatives with respect to the sample sizes n-^ft) and
n2(i), equating to zero and solving. The solutions take the following form:
n,(0 =
2
E
-------
Appendix B: Statistical Design and Analysis B-17
Any failure to obtain complete information, including eligibility information, water samples, or
questionnaire data, from a selected system results in a non-response. Examples of non-response include failure
to locate a system, or lack of cooperation of system operators. Initial estimates of non-response rates were
revised subsequent to results of the pilot study, and incorporation of the screener survey into the survey design.
Exhibit B-8 presents the number of wells that were included in the CWS well survey after a non-
response factor has been taken into account. For instance, 7,086 systems were selected for the screener survey,
of which 5,872 were expected to be eligible. Similarly, 599 wells were in the main survey. Note, these numbers
reflect design parameters. As explained in the following section the actual implementation used slightly
different values.
Exhibit B-8
Sample Sizes for the CWS Well Survey
Stratum
1
2
3
4
5
6
7
8
9
10
11
12
Total
Pesticide
Use
High
Moderate
Low
Uncommon
Ground-Water
Vulnerability
High
Moderate
Low
High
Moderate
Low
High
Moderate
Low
High
Moderate
Low
First Stage
Sample Size*
Required
264
249
134
439
313
239
720
671
517
698
887
742
5,872
Adjusted
293
277
148
556
355
265
867
772
647
776
1,069
1,061
7,086
Second Stage
Sample Size
Required
25
24
13
42
30
23
69
65
50
67
85
71
564
Adjusted
27
25
14
45
32
24
73
69
53
71
90
76
599
Sample sizes required to achieve survey precision requirements and adjusted sample sizes that account for
expected non-response.
National Pesticide Survey: Phase I Report
-------
B-18 Appendix B: Statistical Design and Analysis
B3.6 Screening Procedures for Community Water System Wells
The sampling frame for the community water system survey was the 1984 Federal Reporting Data
System (FRDS) database. FRDS includes the name, address, and selected information on all CWSs reported
by State. The 1984 file was the most recently available version of FRDS at the time of the selection. Federal
money is available for CWSs listed on FRDS, thus providing an incentive for complete reporting. There is
no comparable incentive for eliminating inoperative systems. Therefore it was assumed that FRDS contained
a complete list of CWSs (except those created since 1984) but also could contain systems that were not eligible
for the survey. Within each stratum the CWSs on the FRDS frame were sorted by EPA Region, then State,
and finally by their public water system identification number (also referred to as their FRDS ID or PWS ID).
The frame for the CWS survey was known to contain records of non-CWSs (e.g., systems that no longer
produced drinking water) mixed among the CWSs. The pilot survey also found that there were no reliable
data on the number of wells in a system on the frame. There was also the possibility of multiplicities on the
frame, both one system being reported twice and one "system" reporting for multiple actual systems. It was
therefore necessary to conduct a screener survey to verify and clean up the information on the frame. This
was carried out on a large sample of CWSs.
The sample that was screened included the oversampling of systems in the most vulnerable counties.
From those that were found to be eligible, a systematic sample of systems was selected with replacement with
probabilities proportional to the number of wells within each system. One well was then selected from within
each system. This combination of probabilities resulted in a sample that was close to having all wells having
the same probabilities of selection (except for the oversampling in vulnerable counties). A systematic random
sample was then selected of the size specified in Exhibit B-8.
Using FRDS it was possible to identify 50,771 systems that reported using ground water as a source of
drinking water. The first step was to assign each CWS to a county so as to place each system into one of the
twelve first-stage strata. This was done as part of the pilot survey. When interpreting the results of the CWS
survey for the "most vulnerable counties" domain, it is important to remember that the county is assigned
based upon the mailing address for the system, usually where the administrative office is located. In general
this office is located near the CWS well(s). It is possible, however, especially in large systems, for individual
wells to be located in a different county from the mailing address of the administrative office. For example,
one system with a mailing address in Reno, Nevada had wells in the Lake Tahoe region.
Lists of the selected community water systems were sent to the appropriate EPA Regional offices, and
forwarded by them to the States for review before screening telephone calls commenced. The purpose of the
State review was two-fold: to permit the States to notify the selected systems of the imminent telephone calls,
if they desired; and to have the States identify any errors in the names, addresses and telephone numbers of
the sampled systems. The States were not asked to comment on the appropriateness of including the system
in the sample.
While many States submitted minor changes to the screener sample, it was discovered that there were
major problems with the FRDS data for California. Specifically, a number of the systems on the frame for
California turned out not to be community water systems. In addition, there were no small community water
systems on the frame, only large systems (200 or more connections). The nature and extent of these frame
errors, concentrated in one State, required that the screening survey sample be corrected.
California officials had discovered this problem with their 1984 FRDS data and had submitted a revised
data set, but this revision had not been incorporated into FRDS when the original design of specific sample
sizes for the screener and main survey samples was prepared. This problem became known during the pilot
survey, but could not be rectified before the beginning of the main survey. To correct the screener sample,
a stratified sample of systems was selected from the revised California FRDS data set using the same sampling
rates, and adjusting for non-response, as in the earlier data base. This resulted in a final screener sample of
National Pesticide Survey: Phase I Report
-------
Appendix B: Statistical Design and Analysis B-19
6,814 systems (rather than 7,086) after adjusting for non-response. Similarly, the adjustments to the screener
survey size affected by the relative proportions of CWS from California led to a new desired eligible sample
size of 5,648 systems.
The screener survey found 5,660 of the 6,814 systems surveyed were eligible community water systems
willing to participate in the main survey. The screener achieved a 96.9 percent response rate, with almost all
non-response caused by non-locatable cases even after extensive tracing (i.e., 96.9 percent of the 6,814 systems
surveyed were respondents). Four percent of the screener sample was found not to qualify as a CWS. A large
percent (over 10 percent) of the ineligible cases contained out-of-date or incorrect contact information.
Without extensive tracking procedures (see Appendix C for more details) these would all probably have been
deemed out of business. The tracking procedures determined that one percent had definitely gone out of
business and another three percent could not be traced or located and, therefore, were assumed to be out of
business. For more details on the procedures used to contact these systems see Appendix C.
B3.7 CWS Main Survey Sample Selection
A minimum sample of 564 completed CWS well samples was necessary to meet the original design
accuracy requirements given the assumed rates of detection. Because all CWSs are required by law to allow
their water to be tested it was assumed that there would be only a 5 percent non-response and ineligibility rate
with this sample. The 5% was applied separately within each of the 12 first stage strata. This corresponded
to an initial sample size of 599 CWS wells.
The 5,660 eligible community water systems were sorted as for the screener survey, by EPA Region and
State. The systematic selection of wells for the main survey sample used probabilities proportional to size,
where the number of drinking water wells reported in the screener survey was used as the measure of size.
Community water systems that had refused to cooperate on the screener, or for which language problems had
prevented the gathering of information, were assumed to be eligible and given a measure of size equal to one
well. This subset of the non-respondents corresponded to 36 of the 6,814 systems.
Community water systems with a large number of wells were selected for the main survey with certainty.
The certainty cut-off was determined for each stratum by taking the total number of wells divided by the
number of sampled wells to be selected. All systems exceeding these cut-offs were placed together in the
systematic sorting procedure and given a chance of being selected multiple times. The number of times a
system was selected determined the number of wells to be tested at the system. For example, assume the
certainty cut-off in a given stratum is 30.0 and that one community water system has 72 wells. Since
72/30=2.4, a systematic sample with a random start has .4 probability of selecting the system 3 times and .6
probability of selecting the system 2 times. If it was selected 3 times, then 3 of the 72 wells were sampled; if
it was selected twice, then 2 of the 72 wells were sampled.
EPA used a random well selection procedure for each selected CWS system to determine the one well
to be tested. A sequence of six random numbers was selected for each system between 1 and the number of
wells reported on the screening survey. The interviewers obtained a listing of the wells in the system and then
selected the well corresponding to the first random number on the list. If that well was not able to be sampled
(e.g. no sampling port available pre-treatment), the interviewer was to select the well corresponding to the
second random number. This procedure continued until all six numbers had been attempted. If none of these
wells could be sampled, then the interviewer was to proceed through the list one well at a time (beginning with
the sixth random number), until a well was finally selected that could be sampled.
When a system was to be sampled multiple times, the above procedure only applied to the first visit
The CWS Field Director randomly selected the well to be sampled for all successive visits, telephoning the
CWS to confirm its ability to be sampled.
National Pesticide Survey: Phase I Report
-------
B-20 Appendix B: Statistical Design and Analysis
During the CWS survey a number of questions arose over the eligibility of certain wells for inclusion
in the NPS. Some well owners used chlorine to purify their well water, and useful water samples could not
be taken from such wells because the laboratory analyses could not be accurately performed on water
containing chlorine. Wells for which no sampling port was available prior to chemical treatment could not
have samples drawn from them. In some systems, the water from several wells was co-mingled in common
pipes (tubular well systems). Because it was impossible in such cases to allocate the attributes of the water
to the individual wells, these systems could not be sampled as part of the survey, but were still considered part
of the target population.
During the field period it became apparent that the target sample size of 564 wells would not be
achieved. More wells than anticipated were ineligible or impossible to sample in the field. There were a
number of reasons for this. In some cases the number of wells reported on the screener far exceeded those
found at the time of sampling. There were numerous wells with chlorine problems or which were pan of
tubular systems. For some wells, a port was not available from which to draw a water sample prior to the
point of chemical treatment. Some well owners refused to participate in the survey; these amounted to about
one percent of the sample. Other data were lost to quality control failures associated with taking water
samples. It was estimated that the original sample of 599 wells would only provide approximately 512
completed samples. The shortfall was most pronounced in stratum 12; with a more moderate impact in strata
8, 9, and 10; and only a minor impact on the other eight strata. These three strata groupings were treated
separately for selection of the supplemental sample. A supplemental sample was selected that was large
enough to increase the final number of expected completed samplings to the minimum sample size of 564
wells. The estimated shortfall was determined for each stratum, and inflated by the appropriate percent for
each sample to account for ineligibles and wells impossible to sample. This resulted in a supplemental sample
of 67 wells.
In order to minimize the response burden on any CWS a procedure was used that minimized the
probability of selecting systems for the supplemental sample that were already part of the original sample.
This was achieved in two steps. First, rather than selecting an independent random (RS) start for the
supplemental sample, the random start for the original sample plus (or minus) one-half of the original interval
(I) was used. The supplemental skip interval was set equal to a multiple of the original interval. In formulas
(with primes for the supplemental sample):
RS + 112 ifRSs 1/2
KS - 7/2 ifRS > 112
and
/' = */
where
13 if stratum - 12
7 if stratum = 8, 9, or 10
21 for all other strata
where RS represents the random start number; I represents the sampling interval (i.e., one system is selected
out of every / systems), and it is a stratum dependent constant that determines how many more systems are
selected for each stratum.
A system could be selected in both samples only if its number of wells exceeded 1/2 and happened to
be the 1, Jfc+1, 2k+l, etc. element of the original sample.
At the time the supplemental sample was selected there were only 8 weeks left in the data collection
period. The supplemental sample in each stratum was then allocated evenly among the remaining four two-
week sampling periods (see Section 5 for further details on the revised temporal allocation).
National Pesticide Survey: Phase I Report
-------
Appendix B: Statistical Design and Analysis B-21
A total of 592 eligible wells were selected as pan of either the original or supplemental samples. Water
samples were successfully analyzed from 540 of these wells, giving an overall response rate of 91.2 percent.
Stratum level response rates varied from a high of 97.4 percent in stratum 4 to a low of 85.7 percent in
stratum 3 (the smallest stratum with only 14 eligible sampled wells).
B4 Rural Domestic Well Survey Design
The original survey design for the rural domestic well survey was a three stage design with stratification
imposed on the first and second stages. The first stage stratification was described in Section B2. Unlike the
CWS survey, where the first stage sampling units were systems, the first stage sampling units in the rural
domestic well survey were counties (or county equivalents). Second stage sampling units for the rural domestic
well survey were county subregions (enumeration districts and block gropus) constructed within the sample
counties. The subregions comprised mutually exclusive and exhaustive partitions of the total rural areas
defined for the Survey. The subregions defined two levels: more and less heavily "cropped and vulnerable11
areas. The third stage sampling units were operable rural domestic wells.
The first stage stratum population sizes are presented in Exhibit B-9. The total number of rural
domestic wells was assumed to be approximately 13,102,00 for the purposes of calculating sample sizes
necessary to achieve Survey precision requirements. This number was obtained from the 1980 Census.
Section B4.1 outlines the precision requirements in terms of variance constraints imposed on the five
rural domestic well survey domains of interest Sections B4.2 and B4.3 outline the variance and cost
components of the optimization procedure used to determine the desired sample sizes. The optimization
procedure is explained in Section B4.4. Section B4.5 explains non-response adjustments to the sample
allocation. To this point, the discussion concerns the sample allocation according to the design requirements.
Section B4.6 discusses the final survey design. The final design includes several changes in well selection
procedures, such as the incorporation of random digit dialing at the second stage, and its effect on sample
allocation.
National Pesticide Survey: Phase I Report
-------
B-22 Appendix B: Statistical Design and Analysis
Exhibit B-9
Rural Domestic Well Survey First Stage Stratum Sizes
Pesticide Use
High
Moderate
Low
Uncommon
Total
High
Moderate
Low
Uncommon
Total
Ground-Water
Vulnerability
High
Moderate
Low
High
Moderate
Low
High
Moderate
Low
High
Moderate
Low
High
Moderate
Low
Total
Total
Total
Total
Total
Number
of Counties
106
234
129
110
204
267
193
375
404
186
513
416
595
1,326
1,216
469
581
972
1,115
3,137
Number of Rural
Domestic Wells
(thousands)
455
916
440
684
1,417
671
1,154
2,270
1,170
1,034
1,894
997
3,327
6,496
3,279
1,810
2,772
4,594
3,926
13,102
Number of Rural
Domestic Wells In
Cropped and
Vulnerable Areas'
(thousands)
114
229
110
171
354
168
289
568
293
261
474
249
835
1,625
820
453
693
1,150
984
3,280
B4.1
Precision Requirements for the Rural Domestic Well Survey
The precision requirements for the rural domestic well survey were stated in terms of specified
acceptable variances for assumed proportions of rural domestic wells containing detectable levels of pesticides.
Five domains of interest were specified for the rural domestic well survey. The corresponding precision
requirements are given in Exhibit B-10. The interpretation of these requirements is similar to the
interpretation described in Section B3.2 for the CWS well survey.
National Pesticide Survey: Phase I Report
-------
Appendix B: Statistical Design and Analysis B-23
Exhibit B-10
Precision Requirements for the Rural Domestic Well Survey
Domain Description
Item
Value
1. All rural domestic wells in the United States
Relative domain size
Relative standard error
Detection probability
Confidence interval
0.01
1.0
0.63
0.0 - 0.30
2. Wells in counties with highest average pesticide
use
Relative domain size
Relative standard error
Detection probability
Confidence interval
0.0014
1.85
0.75
0.0 - 0.004
3. Welfs in counties with highest average ground-
water vulnerability
Relative domain size
Relative standard error
Detection probability
Confidence interval
0.0025
0.85
0.75
0.0 - 0.007
4. Wells in the 'cropped and vulnerable1 county
subregions
Relative domain size
Relative standard error
Detection probability
Confidence interval
0.0025
1.525
0.97
0.0 - 0.005
5. Wells in counties with highest average pesticide
use and high ground-water vulnerability
Relative domain size
Relative standard error
Detection probability
Confidence interval
0.0003
1.25
0.47
0.0-0.11
The rural domestic well component of the Survey was designed to yield a range of detection
probabilities (relative standard errors, confidence intervals) for the different domains. The strategy was to
employ a sampling design that provided greater chance of detection in domains of greater interest. The
precision requirements yielded estimates of sample sizes that corresponded to comparative oversampling for
these domains. For instance, the probability of detecting the presence of at least one pesticide in at least one
well in the "cropped and vulnerable" domain was specified to be 97%. The corresponding probability of
detection for the national domain was specified to be only 63%. To achieve the greater precision, the
"cropped and vulnerable" domain had to be comparatively oversampled. The relative domain sizes presented
in Exhibit B-10 correspond to the proportion of rural domestic wells in the nation that contain detectable
levels of at least one pesticide and that are in the listed domain of interest. The listed variance constraints
are calculated assuming a 0.01 proportion within each domain. The relative domain sizes are calculated as
0.01 multiplied by the proportion of wells in that domain. For further details of the relationships between
variances, relative standard errors, detection probabilities, and confidence intervals see Section B3.2.
B4.2 Variance Model for the Rural Domestic Well Survey Design
The procedure for computing the sample sizes necessary to satisfy the precision requirements was
generally the same as that for the CWS well survey. A variance model and a cost model were constructed and
the optimal sample sizes were calculated that achieve the precision requirements (through the variance model)
for the least cost.
National Pesticide Survey: Phase I Report
-------
B-24 Appendix B: Statistical Design and Analysis
The estimator of the mean proportion of rural domestic wells with detectable levels of at least one
pesticide used for designing the Survey was modeled within each domain as:
E £
i-l ;'«1
N
where
N is the (assumed known) number of rural domestic wells nationally, N = 13,102,000.
N(ij) is the number of rural domestic wells in the i first stage and; second stage strata.
Pd(i,j) is the estimated proportion of rural domestic wells nationally that are in domain d
and have a detectable level of at least one pesticide.
The variance of the sample estimate can be expressed as the sum of variance components associated with the
three stages (Vv F2, K3) of sampling as follows:
12
-E
where
i-l
",(0
is the estimated proportion of rural domestic wells nationally that are in domain d
of the i stratum and have a detectable level of at least one pesticide.
AKO?
N \
o2(») p,
J
J
°2(«V) d - P!) d - Pa)
o2ft)
Pi
P2
Pd(W V -
is the proportion of rural domestic wells nationally that are in the i stratum of
domain d and have a detectable level of at least one pesticide.
is the intracluster correlation of detections of at least one pesticide within counties
(intracounty correlation due to selecting clusters within counties).
is the intracluster correlation of detections of at least one pesticide within second-
stage subcounty regions (intracluster correlation due to selecting wells within
clusters).
represents the sample sizes to be selected at the three stages of sampling
(k f {1, 2, 3}).
National Pesticide Survey: Phase I Report
-------
Appendix B: Statistical Design and Analysis B-25
The variances o2^) and o2(i,j), are computed using the specified relative domain sizes presented in
Exhibit B-9 and the stratum sizes presented in Exhibit B-8.
The correlation parameters, pl and p2 were assigned values 0.01 and 0.1, respectively, for all domains
and strata.
B4.3 Cost Model for the Rural Domestic Well Survey
Costs were estimated to calculate a cost efficient sample size that would achieve the precision
requirements specified by EPA Actual costs of the survey were different than these initial estimates. The
cost model used for designing the rural domestic well survey was specified as:
C(a) =£
2
/I
where n^ft), n2(i,j), and n3(y) are the sample sizes for the first, second, and third stages of the survey; C(n-$
denotes the total cost of the survey (the vector n represents the sample sizes at all stages of the Survey, at the
first stage this is the number of counties; at the second stage, the number of subcounty regions (clusters); and
at the third stage, the number of rural domestic wells).
Crfi) is the per county cost ($3,947) of sampling frame construction and stratification,
sample selection, data collection and data processing averaged over all counties
contained in the ith stratum;
C2(i,j) is the per subcounty region (cluster) cost ($678) similarly; and
C3fyj is the per well cost ($2,832) of sampling and collecting, preserving, transporting, and
chemically analyzing the water samples.
B4.4 Sample Allocation for the Rural Domestic Well Survey
Given the variance constraints and the cost component a nonlinear programming procedure, similar
to that used for the CWS well survey, was used to calculate the sample allocation that achieves the precision
requirements at minimum cost. The method is described in Section B3.4. The sample size solutions take the
following form:
5
£
4*1
c,(o
C,(0
d-l
National Pesticide Survey: Phase I Report
-------
B-26 Appendix B: Statistical Design and Analysis
-------
Appendix B: Statistical Design and Analysis B-27
I
O
S
0)
o
.05
i£
Wc
o
a
a
CO
.S 5-
S
55
SS
O C
8 1
fff
8 ?8 § 8
ll
o a: o
?
o> « jr o>
-
w < o) w
lIlIU
O C O (T O C
to at <
S82J8
JS « .2
£ ^J J3
IfHIf
«
alo°l6°l
CO J
» 45 S
z I -
-------
B-28 Appendix B: Statistical Design and Analysis
1
CO
J
o o
= 1
ii
CD O
II
iD o
o
Q.
(0
CO
i
I
c
1
-
S8228S
5*
n n
885S8*
» s
55
0 o
«- CM -
I S I
itirif
3 O 3 O 3 O
> O > O > O
2 i 2 I 2
O cr o c o
iped and vul
ainder of co
8
8
0» >
I |
I
t
111
Klnttnnal Paetlr.Mo <%tirw<»w Phftfte I Renort
-------
Appendix B: Statistical Design and Analysis B-29
of three to one relative to the remainder of the county. This rate was a compromise between the desire to
maximize the number of samples drawn from wells likely to be contaminated and the absence of knowledge
of the location of such wells.
The pilot survey used an area probability sample to select domestic wells within each sampled county.
This involved selecting a sample of enumeration districts/block groups (ED/BGs) based on 1980 Census data,
sending interviewers into the field to list all housing units in those ED/BGs, selecting a sample of housing
units for screening, then returning to the field to screen those housing units for use of well water and finally
to conduct the interview and sample the well. ED/BGs in the most cropped and vulnerable parts of each
county were oversampled to increase the probability of including contaminated wells in the sample.
The pilot study process was very time consuming. It also relied heavily on local interviewers in each
county, since they had to visit the county twice, once to list the housing units and the second time to screen
and conduct the interviews.
Therefore, the sample selection for the main rural domestic well survey was revised, and the sample of
housing units was selected through use of random digit dialing (RDD). This procedure requires less time in
the field in each county, better controls the sample size, allows for the use of traveling interviewers (taking
advantage of improved quality from experienced NFS staff), and spreads the sample more completely
throughout the sampled county. The only drawback to using RDD was that households using well water that
did not have a telephone were excluded from being selected for the sample. Given the small proportion of
households with wells that are without telephones (10 percent). Even with small differences in contamination
rates between wells used by households with and without telephones this modification to the target population
was not thought to introduce a significant source of bias.
Counties were selected with probabilities proportional to size, where size is defined as the minimum
of the number of rural households or number of wells in the county as reported in the 1980 Census. This
measure of size is a surrogate for the number of rural wells in the county, which is not provided by the Census.
An examination of the first stage weights (the inverse of this probability of selection) for the sampled counties
indicates that two of the counties were not selected from the same stratum to which they were assigned.
Clarendon County, South Carolina is in stratum 10, but its probability of selection corresponds to the counties
selected in stratum 4. Similarly Oakland County, Michigan is in stratum 5, but its probability of selection
corresponds to stratum 8. In both cases such a reassignment to a new strata could be the result of revised
pesticide use information becoming available after the sample of counties was selected. For the analysis of
Survey data, both counties were considered to be in their reassigned strata while retaining their original
probability of selection.
The cropped and vulnerable areas of sampled counties were oversampled at a rate of 3:1 relative to the
remainder of the county. This rate was a compromise between the desire to maximize the number of samples
drawn from wells likely to be contaminated and the lack of knowledge over the location of such wells. For
design purposes it was assumed that approximately one percent of all wells were contaminated by some
analyte. If no oversampling were conducted one percent of all wells were contaminated by some analyte. If
no oversampling were conducted one would expect a sample of 732 wells to yield only 7 or 8 contaminated
wells. This would obviously be too few for use in any relational analyses, one of the two main goals of the
NFS. It was therefore important to try and oversample areas of likely contamination. Unfortunately it was
not known how strongly being in the "cropped and vulnerable" part of a county would increase the probability
that a well contained pesticides. For example, a well without a sanitary seal at the top might become
contaminated by water entering the top of the well, regardless of how vulnerable the underlying aquifer was
to contamination. Thus it was quite possible that wells not in the cropped and vulnerable parts of a county
might be contaminated. If the oversampling rate was too high the resulting weights of contaminated wells
could vary significantly. This would greatly reduce the accuracy of estimates regarding all contaminated wells.
Klatlnnal D*>o»li~lrt« Cuo/ai/- DKaea I Dftnftrt
-------
B-30 Appendix B: Statistical Design and Analysis
Because the survey is of rural domestic wells, any well in an area defined by the 1980 Census as urban
or urbanized was ineligible. The boundaries of such areas did not always correspond with boundaries that were
easy to describe. It was sometimes necessary to extend the urban area to the next closest boundary. Thus a
few wells that were indeed rural, but located in these "extended urban" areas, had to be excluded from the
survey.
A similar situation arose because of boundaries for the oversampled areas. There were actually four
categories in which a well could be placed. Actual target areas, extended target, non-target, and undefinable
target (too small to describe without including a much larger non-target area). The first two were oversampled
while the last two were not. Determination of whether or not to include a well in the "cropped and
vulnerable" area for analysis was based on the actual observed location recorded by the field interviewers, not
the telephone center's location.
The process of selecting clusters of telephone numbers for the rural areas of sampled counties began
with a list of telephone number prefixes and five-digit zip codes. The list indicates which prefixes intersect
which zip codes. Before sample selection, the list was examined to see if any zip codes lay entirely within
urban areas of the county and if some prefix areas were entirely within these zip codes. Such prefixes were
deleted from the set to be sampled.
The maps showing the areas to be oversampled were superimposed on zip code maps to identify the
zip codes that intersected the oversampled areas and those that were completely outside the oversampled areas.
The latter were classified as nonoversampled prefix areas. The former were mostly mixed prefix areas, that
is, areas in which some households with wells were oversampled and others not oversampled.
The sample design called for 734 wells to be sampled from the 90 counties. To account for potential
non-response at sample time the RDD screening survey was designed with a goal of 843 completed screening
interviews on the assumption of an approximate 15% non-response rate. That is, the size of the screening
sample needed to yield (in expectation) 843 eligible wells for sampling.
The RDD process identified clusters of 100 telephone numbers sharing the same six digit telephone
number prefix (i.e., the area code and the next three digits). For counties in which all telephone prefixes cover
both oversampled ("cropped and vulnerable") and other areas of the county, the size of the screening sample
was calculated as:
S = n l[rp d + (i-r)/ 3}]
where
n is the sample size required for the main survey.
r is the response rate; and
p is the proportion of households that qualify as rural domestic households and have wells
calculated from 1980 decennial census reports;
t is the proportion of wells in the oversampled part of the county;
National Pesticide Survey: Phase I Report
-------
Appendix B: Statistical Design and Analysis B-31
For counties in which some telephone prefixes cover only oversampled areas of the county, the size of
the screening sample was calculated as:
S = n I [rp {/+v+(l-f-v)/ 3}]
where v is the proportion of telephone numbers in prefix areas that cover only non-oversampled areas.
In order to minimize the variance increase associated with RDD clustering, the number of RDD clusters
was selected so that, on average, each RDD cluster would contain one rural domestic well. This is achieved
by setting the number of RDD clusters equal to the number of wells to be sampled:
c = ips(l-2v / 3)
However, it was anticipated that there could be distributional problems when the proportion of rural
households with wells,/*, was small. "Cropped and vulnerable" areas are typically less urban and more likely
to have rural domestic wells. To guard against this problem, the number of RDD clusters was increased by
50% in counties with/7 < 0.3. Hence:
\rps(l -2v / 3) when p * .3
C ~ |l.5 rps(l ~2v I 3) when p < .3
Typically, 20 to 25 percent of all 10,000 numbers with the same telephone number prefix (the first six
digits, i.e., the area code and the next three digits) are used by residential households. However, residential
household numbers tend to be clustered together. Consequently, the RDD process was carried out by
identifying clusters of 100 numbers containing some residential numbers. Using this method, the proportion
of residential households called can increase to 60% in those clusters.
A cluster of telephone numbers consists of 100 telephone numbers with the same first eight digits
(where a telephone number has ten digits in total). The number of telephone numbers needed to identify c
RDD clusters of active household numbers was calculated as c divided by the proportion of telephone numbers
for residential households in each county. This proportion is 20 to 25 percent nationwide, but varies
considerably among counties.
Clusters were then identified by the following procedure. The prefix numbers in each county were
arrayed in numerical order. A systematic random sample of clusters, 50 percent greater than the number
needed (in order to provide a surplus), was selected. A randomly selected seed number was then selected from
each cluster. Each sampled seed telephone number was called and asked if it was a household and was located
in the designated county. Clusters whose seed numbers passed both criteria were retained; others were
rejected. This procedure continued until c clusters were identified.
A modified version of the Waksberg method was used for sample selection of households within
clusters. A constant number of telephone numbers (within the county) per cluster was selected instead of a
constant number of households. This produced a sample that was not quite self-weighting, but weights were
available and the brief time interval available for the RDD work in a county made the modified method
preferable.
The screening interview first asked if the household was in the designated county. Out-of-county
households were ineligible. The next question asked if the household was in an urban part of the county.
Urban areas were also ineligible. The third geographic question asked if the household was within the
boundaries of the oversampled part of the county. If so, the household was retained and became part of the
National Pesticide Siirv«w Phase I ftf>nnrt
-------
B-32 Appendix B: Statistical Design and Analysis
sample. If not, a one-third subsampling rate was applied to select the households to be retained in the final
sample.
For some counties it was possible to identify certain telephone prefixes that were only located in zip
codes that were entirely in areas which were not cropped and vulnerable. Households with a well in these
prefixes were to be subsampled at a rate of 1 in 3. To save time and money the above procedures were slightly
modified for such prefixes. Rather than subsampling eligible households, only one-third as many clusters were
called from these prefixes. Any eligible respondent in these prefixes was then included in the sample. This
retained the same probabilities of selection but required fewer phone calls.
The screening interview also asked whether the household water was obtained from a well. In a few
cases households would report that either more than one house shared the same well (but not enough to
qualify as a CWS) or that the household used multiple wells for drinking, bathing, or washing. Wells used
by multiple households were subsampled in order to achieve equal selection probabilities at the well level.
This was done by determining the number of households using the well and then retaining the household with
probability of 1 in that number. If a household used more than one well, the well most frequently used was
the well included in the survey.
It is also possible for a household to have two (or more) residential phone numbers. In such cases the
household was retained with a 1 in 2 probability, which retained an equal probability of selection at the well
level.
B4.7 Rural Domestic Well Survey Response Rates
During implementation of the main survey it became clear that the target sample size (indicated in
Exhibit B-10) was not going to be met due to a greater non-response rate than anticipated. At the point when
50 counties had been sampled it was estimated that an additional 41 completed samples would be needed to
satisfy the design specifications. To account for this 15% non-response rate, an additional sample of 48 wells
was added. The increase in sample sizes was two per county in the high pesticide use strata and one per
county in other strata.
Because this modification was made relatively early in the Survey implementation, its impact on the final
sample weights was minimal. In addition, because of the temporal allocation, the impact of this modification
was spread across all strata. There is, however, a small increased (minimal) potential for non-response bias
resulting from the greater than expected non-response rate.
The ROD screener for each county lasted three weeks and was allocated temporally in accordance with
the field schedule. It started eight weeks prior to the beginning of the two-week sampling period. Three
weeks was felt to be the minimum time period necessary for adequate refusal conversion efforts. It was
necessary to leave five weeks from when the sample was selected until the field work would commence to allow
for time to send out survey materials to selected households, confirm scheduled interviews, and finalize travel
schedules. Leaving eight weeks between being selected for the NFS and actually being interviewed/sampled
was a potential source of increased nonresponse to the survey.
RDD screener response rates in the telephone center averaged 86.8 percent (nonrespondents being
those known households which refused, were contacted the maximum number of times without completing the
screener, had language problems, or had other similar problems). This rate was consistently higher in truly
rural counties than in those containing large urban populations. There were two reasons for this. First was
a difference between urban and rural populations. The second was the fact that when an unexpectedly large
number of urban, and therefore ineligible, households were found it was necessary to introduce more
telephone numbers into the telephone screening process during the three week period. These numbers
therefore did not have available the full three weeks for refusal conversion. They also could not have their
National Pesticide Survey: Phase I Report
-------
Appendix B: Statistical Design and Analysis B-33
attempted contacts spread out over as long a time period. Thus these numbers introduced late in the
screening period were less likely to be completed, further reducing the response rate.
For rural domestic wells, a total of 864 eligible wells were selected as part of either the Pilot Survey
(for the first six counties) or the telephone screening (for the other 84 counties). Completed water sampling
and interviewing was conducted for 752 (87 percent) of these wells. Exhibit B-12 indicates the response rates
achieved in each of the five design domains of interest for the rural domestic well survey.
Exhibit B-12
Rural Domestic Well Survey Response Rates
Domain
All Rural Domestic Wells
High Ground-water Vulnerability Counties
High Pesticide Use Counties
High Pesticide Use and Ground-water
Vulnerability
Cropped and Vulnerable" County Subregions
Eliglbles
864
248
298
99
413
Respondents
752
217
264
83
343
Percent
87.0
87.5
88.6
83.8
83.1
B5 Temporal Allocation
The pilot study led to improved procedures and techniques in several areas of the Survey. Issues raised
by the pilot study were analyzed and presented to the EPA Science Advisory Panel (SAP) in September 1987.
One of these issues concerned temporal variation as a factor influencing when and in what concentrations
pesticide and nitrate impact drinking water wells, and how to address temporal variation in water sampling.
When the Science Advisory Panel subpanel reviewed the statistical design proposed in the pilot survey
they expressed concern that the design should minimize the impact of any temporal (seasonal) variability in
pesticide contamination levels. There are many possible reasons for such variability. Among these are: date
of pesticide application, variation in ground water levels, rainy seasons versus dry seasons. The impact of all
of these is likely to produce complicated patterns of variation in contamination levels. These patterns will also
vary from one part of the country to another and from one soil type to another. The SAP therefore urged
that rather than trying to explain the temporal variability the NPS should minimize the impact on the results.
The best way to minimize such variability is to spread the samples throughout the year. A second
advantage of spreading the sample throughout the year was that this made it possible for the water analyses
to be done in a small number of contract laboratories which had limited capacity for analyzing the samples
within the 14 days of sampling required by NPS quality assurance protocols. Each well was assigned to a two
week time period during which the water samples were to be drawn.
The laboratories' capacities required a ceiling of approximately 40 well samples every two weeks. To
satisfy this constraint the samples were allocated across the year-and-a-half data collection period. Since the
design domains of interest corresponded to individual and groups of first-stage strata, the most efficient
procedure was to allocate the samples across time separately for each stratum.
Klatinnal P*»ctirlHo Snruau- Dhoea I Qarwtrt
-------
B-34 Appendix B: Statistical Design and Analysis
In response to the SAP subpanel's recommendations, EPA selected a method of allocating sampling
dates to wells across the data collection period randomly within each stratum to minimize the possible impact
of temporal bias. The capacities of the analytic laboratories established the ceiling for the number of wells
that could be sampled in a particular period. Within this constraint, EPA allocated counties for rural domestic
well sampling to two-week time periods spread over an 18-month period, from September 1988 to February
1990. The survey was conducted from March 1988 to August 1988 for the six counties used in the pilot study.
Sampling of community water systems was similarly spread out in two-week periods from August 1988 to
December 1989.
The initial temporal design was characterized by 28 available two-week time periods (time-slots) and
analytic laboratory limitations of 40 samples per time-slot. The initial CWS well sample design specified the
28 time slots from June 1988 to August 6,1989. The initial rural domestic well sample design specified time-
slots beginning August 22,1988 and ending December 3,1989.
The distribution of the 599 CWS wells selected for sampling provided just over 20 samples per time slot
for the analytic laboratories. Each county selected for sampling in the rural domestic well survey was expected
to contribute approximately 9 samples for the analytic laboratories. Thus, in time-slots where CWS well
sampling occurred, two counties could be sampled in the same time-slot for the rural domestic well survey,
and in other time-slots four counties could be sampled.
The initial temporal allocation of CWS wells is presented in Exhibit B-13. The available time-slots were
allocated randomly within strata to the 599 wells selected for sampling. For instance, in the first time-slot one
CWS well was sampled from strata 1, 3, 5, and 6; 2 each were sampled from strata 4, 9, and 10; and 3 each
were sampled from strata 7, 8, 11, and 12; no wells were sampled in the 2nd stratum in the first time-slot.
The CWS well survey and rural domestic well main survey (i.e., not including the pilot survey') were
initially expected to coincide from August 22,1988 through August 6,1989, corresponding to CWS time-slots
4 through 28. This corresponds to 25 time-slots for which 2 counties were sampled in the rural domestic well
survey. The remaining 34 counties could be sampled at twice the rate, corresponding to S1A more time-slots
(that is, the 34th time-slot would only have approximately 20 samples for the analytic laboratories). The
allocation of pairs of counties to the strata is presented in Exhibit B-14. The first 25 pairs were assigned to
time-slots coinciding with the CWS temporal allocation. The remaining 17 pairs were sampled at twice the
rate (two pairs per time-slot).
The temporal allocation was designed to be random within strata in the sense that, for the CWS well
survey, time-slots were randomly allocated within strata to the CWS wells selected for sampling, and for the
rural domestic well survey, the 42 "pairs" of counties were randomly allocated within strata to the selected
counties. These 42 "pairs" correspond to the 34'/2 time-slots as indicated above.
The Survey implementation did not manage to adhere exactly to the temporal design, for a variety of
reasons. Frequently (114 out of 540 cases) the CWS interviewing/sampling team was not able to comply with
the selected sampling period. There were many reasons for this, but most frequently cited was the limitation
of travel funds and time available for the NPS, given their regular schedule of activities. The team would
frequently request that the sampling period be switched to accommodate its schedule. It was therefore
necessary to try and balance the desire to retain the cooperation of State interviewing/sampling teams while
TniniTni7.ing the impact on the temporal sampling design.
National Pesticide Survey: Phase I Report
-------
Appendix B: Statistical Design and Analysis B-35
I
(0
o*.
<
"5
1
J
i*
R
8
5
8
8
K
*
f*
ID
*
w
~
«
*
r*
«
rt
«
-
I
K
W
-
-
-
-
-
-
-
-
-
-
-
-
-
-
o
-
-
<
-
'
8
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
o
M
Jf
O
-
O
o
-
o
o
o
0
e
-
o
'
e
-
e
-
o
-
n
9
N
-
M
M
-
n
n
n
M
n
"
ft
-
en
M
n
-
ft
n
«»
8
-
-
-
-
-
-
-
-
-
-
N
-
-
"
-
-
-
-
-
«
8
-
-
-
-
-
-
-
-
-
-
0
-
-
-
-
-
-
-
-
ff
n
n
n
M
n
m
n
m
n
N
n
n
«9
M
W
*»
n
«
n
K
«
«f
M
-------
B-36 Appendix B: Statistical Design and Analysis
fl
ft
1
o
1
0>
o
2
DC
- o
T «
m£
i§
x O
111 «-
o
£
a
"o
o
3
0
o
3
^
75
c
5*
o
4
o
ii
s
8
N
«
0
«
n
»
N
^
e
N
»
*
r>
M
-
I
n
X
X
X
-
»
X
X
X
X
N
M
X
X
n
«
X
X
<
^
X
X
X
X
0
n
X
X
X
a
n
X
X
X
K
K
X
X
X
X
X
X
X
»
X
X
X
X
«
X
X
o
K
X
X
X
X
X
X
X
-
-
X
N
9
National Pesticide Survey: Phase I Report
-------
Appendix B: Statistical Design and Analysis B-37
(D
s
oo
I
o
I
o
Q
0> 3
StE
£ O
S**
o S
*7 o
mo
tf «
"o
c
75
5*
O
11
*
I
9
*
s
ft
ft
$
X
8
*
3
a
n
ft
8
8
R
X
8
X
8
8
I
*
X
X
X
X
-
n
X
X
X
N
n
X
X
n
n
X
X
X
»
10
X
X
X
X
X
a
o
«
F»
X
X
X
X
N
N
X
X
X
X
X
X
X
«
N
X
X
*
X
X
X
X
o
-
X
X
X
X
'
K>
X
X
X
X
«
s
National Pesticide Survey: Phase I Report
-------
B-38 Appendix B: Statistical Design and Analysis
Two types of time switches were immediately accepted. Any well could be switched to an adjoining two-
week period without objection. It was also allowable to switch a sampling date between the same two-week
period of 1988 and 1989. Fifty-seven wells moved to adjacent time periods and 14 switched year (5 of these
moved a year plus one sampling period).
Many teams requested time changes of more than two weeks (but not one year). The first response was
to explain the importance of the temporal design and try to convince the team to retain the scheduled date.
When this failed EPA tried to arrange to trade sampling times between two wells in the same stratum and
EPA region, thus minimizing the impact on the temporal design. For example, both Alaska wells were
assigned sampling periods when the well would be inaccessible due to the cold weather. Their sampling
periods were then switched with two wells in the state of Washington which were in the same first-stage
stratum. A total of 48 wells were moved more than one two-week period from that originally proposed by
the temporal design.
A second temporal issue addressed by the sample design was how to sample seasonal wells. The
determination of what procedure to follow depends upon whether one wants to make estimates (e.g. percent
of wells contaminated) for all wells ever used for drinking, bathing, or showering, or for all wells used on a
typical day. This can be demonstrated with the following extreme example: assume that there are 730 wells
in the country, 365 used year round and 365 each used one day a year. Further assume that the 365 used for
only one day are all contaminated while none of the year round wells are contaminated. The survey could be
designed to report that half of all wells in the country ever used (365 out of 730) are contaminated; or that
less than one percent of all wells in use on a typical day (1 out of 366) are contaminated. It was decided that
the less than one percent number would be of more use for policy purposes.
Procedures were developed and implemented to ensure that the probability that a well was included in
the survey would be proportional to its frequency of use, and thus provide national estimates for the
percentage of wells typically in use. These procedures were different for community system wells and rural
domestic wells. For CWS wells they included the following measures: If a system was shut down during the
time period the screener was conducted with no one available to answer the phone, the system would have
been determined to not be in existence and declared ineligible. If someone was available to answer the phone
and could explain the seasonal nature of the system, it was considered eligible for the survey. Actual cases
of seasonal wells identified during the survey were extremely rare.
If a sampled community water system was not in use at the selected field sampling time period, it was
rescheduled for when it was in use. This is because the population of systems had already been sampled to
provide an estimate of all wells in use at a point in time. For domestic wells there was no national screening
at a particular point in time. While different counties were assigned sampling intervals throughout the year,
we only sampled a given county during one time interval. If a household used well water but the well was not
used during the entire two week sampling period (e.g., the well was dry) we did not reschedule and considered
it to be an ineligible well. Indeed, if a household was on vacation throughout the sampling period the
telephone center would obviously be unable to reach them and they would therefore not be able to be included
in the survey. Thus, as with community water system wells, by including rural domestic wells in proportion
to their frequency of use, we are able to provide estimates for all wells used at any particular time. The actual
times of sampling are indicated in Exhibit B-15 (CWS wells), and B-16 (rural domestic wells).
National Pesticide Survey: Phase I Report
-------
Appendix B: Statistical Design and Analysis B-39
«
I
a
|
i-
JO
I
in
T
00
fi
UJ
Q,
I
I
i
a
2
ff
8
at
00
r*
>fi
^~
n
CM
~
o
a>
00
i>-
V
f>
f
CM
-
V
~
O
"-
+
o
t*»
>r»
«
*»
»-
»-
^
**,.
o
o
I
»
o
r-
^*
-
o
-
<~
*~
-
0
-
-
o
o
o
-
1 Stratum 2
r~
O
O
»
o
o
o
o
0
~
o
<^
«
»"
o
0
o
«
o
CM
^
o
-
o
^
-
-
n
CM
CM
CM
CM
'-
0
o
| Stratum 4
-
"
o
0
»
0
*>
O-
O
N.
cv
CM
CM
«
«-
O
CM
CM
CM
CM
»
-
*
«
CM
-
O
«
*
o
CM
»
«»
O
-
"
o
**
e*
e»
»
n
to-
t>
0
01
«
CM
CM
CM
«
O
-
w
CM
-
CM
W
«
CM
CV
o
o
0
«
cu
10
*
«
o
«
w
«
CM
«9
N
*
v*
W
O
o
*
»*
*
CM
-
CM
CM
O
^
^r
w
CM
-
^
«r
n
^
o
e
CM
8
»
»
K
s
V-
W
^T*
*
8
w
SH
S
2
S
«
«
T
«
| Total In Period
-------
B-40 Appendix B: Statistical Design and Analysis
0>
^ TJ
D C
0) «
i E
CD -o
_ j,
5 a
x a
UJ (A
£
T>
O
e
XI
3
E
B
I
8
£
8
8
S
3
%
n
8
CM
IS
&
8
£
3
S3
8
CM
{5
»
-
O
o
0
o
-
o
o
0
o
T-
'-
*-
01
<-
o
i
s
o
o
o
0
o
co
*-
'-
-
-
o
o
o
V
-
o
'-
| Stratum 2
CM
o
o
o
0
o
o
o
0
o
-
o
*-
'-
o
o
0
o
n
i
S
CO
"
o
CM
O
o
CM
0
o
o
o
o
0
-
*
»-
CM
»
CO
fc
o
'-
o
»-
"
0
«-
o
o
o
*»
*-
f
V"
»-
"
V-
to
s
o
o
o
o
CM
-
0
'-
-
o
o
o
o
-
V-
-
*
1 Stratum 6
£
CM
-
O
n
CM
*
eg
0
o
M
M
»-
*-
n
co
CM
CM
K
«
55
s
o
"
o
CM
CM
»-
CM
"
-
-
*-
--
CM
CM
-
-
n
Stratum 8
9
o
o
-
M
w
CM
"
CM
CM
^
T~
*"
o
o
CO
CM
M
1 Stratum 9
s
o
CO
'-
CM
CM
CM
CO
V-
"
-
'-
"
-
CM
CM
CO
CO
II Stratum 10
2
0
«M
T"
w
P5
W
10
e
CM
o
*~
0
«-
to
«
M>
^«
v»
§
S
O
O
*
CM
rs.
*
CM
'-
CM
^
CM
CM
CM
O
-
CM
*-
CM
co
i
40
O
f
%
%
J.
s
^-
o
-^
«
0
n
a
-------
Appendix B: Statistical Design and Analysis B-41
£
I
O)
£
a
? I
UJ
I
cr
!
o
O
z
s
a
N
8
O>
CO
^
N.
CO
K
in
»
n
-
O
o
o
o
o
»-
o
o
o
o
o
o
o
o
o
e>
o
o
o
o
o
«o
o
o
o
o
-
o
o
o
o
o
o
-
0
0
o
o
o
o
"
o
o
o
o
o
| Stratum 6
0
o
o
o
o
0
o
-
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o
V*
K
o
-
o
-
0
o
o
o
0
-
o
o
o
o
0
o
o
o
o
o
o
o
o
o
I Stratum 8
*
o
o
«
«
0
»
o
o
o
e
o
o
o
o
e
*« :
O
0
o
o
o
o
e
«
o
o
*"
o
o
o
o
e
»
o
o
o
o
0
o
o
0
o
0
o
o
o
o
*-
0
I
o
6
o
e
V-
0
O
49
e
o
o
o
»
o
o
o
0
o
o
e
o
o
o
o
il
o
0
o
o
0
o
o
-
0
0
0
o
o
o
0
o
»
0
o
e
o
o
o
o
w
<4
«v
«i
w
«M
0
«tf
CD
w
-
1*
-
«
o
o
o
«M
O
-
«
o
o
o
w
£
I
-------
B-42 Appendix B: Statistical Design and Analysis
_o
1
JC
o
o
eg
E
JO
c W
o
<*,
(O
i
i i
j= o
X Q
s
1
o
I
K
5
5?
3
$
*
5
$
*
S
8
8
£
8
8
3
3
S
w
%
8
8
ft
8
16
*~
o
o
-
o
o
*
0
0
-------
Appendix B: Statistical Design and Analysis B-43
B6 Statistical Analysis Procedures
The survey design described in this appendix was implemented during the field sampling and data
collection phase of the Survey. Thousands of water samples and large amounts of questionnaire data were
collected, analyzed, and sorted into the Survey data base. The following sections deal with the preparation
of the Survey data base for analysis, and with the statistical methods used to produce the results contained in
Chapter 6 of this report. The statistical topics that are addressed include:
missing data and imputation;
weighting; and
calculation of population estimates and confidence intervals.
This section concludes with a brief comparison of the Survey precision requirements with results of the
preliminary statistical analysis of proportions of wells containing detectable levels of at least one pesticide or
pesticide degradate.
B6.1 Missing Data and Imputation
The Survey experienced two basic types of missing data: unit non-response (i.e., no data for a sampled
well could be recorded); and item non-response (i.e., the data record for a sampled well is missing a few
responses, which can pertain either to analytes or questionnaires). Unit non-response is accounted for by a
non-response adjustment in the weighting procedures, which are discussed in Section B6.2. Item non-response
is accounted for either by data retrieval or imputation, which are discussed in this section.
B6.1.1 Data Retrieval
The initial methods for addressing missing data or inconsistencies in the data include logic checks and
data retrieval. A number of logic checks were performed on the Survey data base (see Appendix C for further
details). For instance, data on the depth of a well screen could be receded to be less than or equal to the
depth of the well. In some instances, general rules were implemented to make the data more consistent. Data
retrieval was attempted for a number of questionnaire items. When these methods for handling missing data
could not be used, imputation was performed (for item non-response).
B6.1.2 Hot-Deck Imputation Procedures
Two methods of imputation were used to prepare the Survey data data base for Phase I data analysis:
using another item from the same data record; and
using the same item from a different record.
An example of the first method is the following. If the question "In 1987, were crops farmed within
one-half mile of the well?" is unanswered, then the answer to the question "In 1988, were crops fanned within
one-half mile of the well?" can be used to infer the missing item. That is, it seems likely that if there was
farming this year, then there was probably farming last year.
The second imputation method, using the same item from another record, is applied using hot-deck
procedures.8 Hot-deck procedures were first developed by the U.S. Bureau of the Census.
8 Little, R. and Rubin, D. Statistical Analysis with Missing Data. (New York: John Wiley & Sons, 1987)
-------
B-44 Appendix B: Statistical Design and Analysis
Hot-deck imputation involves substituting for a missing item of the "recipient" record with the value
of the same item of a "donor" record. The donor record and recipient record should be similar. That is, the
response to other items should be equal or nearly equal on both the donor and recipient records. For example,
suppose an item listing pesticides used on crops is missing. Then the list of pesticides from a donor record
is copied to the corresponding missing data item on the recipient record so long as responses to other items
are similar. For instance, the effects of these pesticides on well water for the donor and recipient record may
be the same only if the first-stage stratum (defining ground water vulnerability and pesticide use) is identical
for both the donor and recipient records. Other items should also be similar.
The methodology of hot-deck imputation involves the stratification of the records into cells. This
imputation stratification is performed using variables of interest or importance (such as first-stage
stratification). For instance, imputation stratification using only the first stage strata would result in 12
stratification cells, where the records within any one cell all have a common first stage stratum. Records
within a cell are then sorted according to another item (or items) for which similar values are desired, such
as geographic region. In this manner, records that are most similar are adjacent in the list of records within
the cell. The record immediately preceding the recipient record is normally selected to be the donor record.
It is not selected only if (1) the record immediately preceding the recipient record is also a recipient record;
then the next preceding record is considered; (2) no preceding record is available as the donor record; then
the first available succeeding record is the donor record; (3) no record is permitted to be a donor record more
than twice (thus preventing any single record from dominating results, though domination is unlikely to occur
unless a record is a donor record many more times than twice).
Each missing data item could be imputed using both a different imputation stratification into cells and
a different sort within cells. Considerable economies can be made, however, if different items have identical
imputation strata and sorting variables. The following sections describe the stratification and sorting variables
used for items on the analytic results and questionnaires for both the CWS well and rural domestic well
surveys.
Imputation for Analyte Data
Many analytes were not detected in any water samples. Hot-decking imputed a value of not detected
for such missing items. These analytes are imputed without having to perform stratification and sorting.
For analytes with at least one detection, the imputation strata were, in order:
1. First-stage strata;
2. Responses to the farmed/not farmed indicator (for agricultural chemicals and nitrogen
only);
3. "Cropped and vulnerable" area indicator (for the rural domestic well Survey only);
4. An indicator for the use of that analyte within Vi mile of the well; and
5. Crops on which this analyte is used (for agricultural chemicals only).
Within cells the records were sorted by geographic region. For the CWS well survey, the geographic
sorting items were EPA Region and State. For the rural domestic well survey, the geographic sorting item
was the county.
Often, especially with 5 stratification items for imputation, cells contained very few records and
imputation of the records within cells was infeasible. In this case cells were collapsed by dropping out the
least important stratification item (i.e., the last stratification item above). If imputation remained infeasible,
then the next to last stratification item was dropped, and so on, until all records with missing items were
imputed.
National Pesticide Survey: Phase I Report
-------
Appendix B: Statistical Design and Analysis B-45
Imputation for Questionnaire Data
For missing questionnaire items from the Local Area Questionnaire and the Main Questionnaires for
both CWS wells and rural domestic wells, the imputation strata were, in order:
1. First-stage strata;
2. Responses to the farmed/not farmed indicator (Local Area Questionnaires only);
3. The size of the CWS in terms of the number of operating wells (CWS Main
Questionnaire, sections A and B only);
4. Crops used within Vi mile of the well (Local Area Questionnaires, and Main
Questionnaire, section C only); and
5. Pesticides used within !/a mile of the well (Local Area Questionnaires, and Main
Questionnaires, section C only).
Within cells the records were sorted by geographic regions, analogous to those used for the analyte
imputation procedure, and collapsing across cells occurred if there were too few items in a cell. Rules for
imputing missing CWS data using another item are shown in Exhibits B-17 and B-18. Similar rules apply for
the rural domestic well survey.
B6.2 Weighting Procedures
This section describes the weighting procedures that were followed to develop national estimates based
on the results from the wells that were sampled by the Survey.
Each well in the CWS and rural domestic well surveys was assigned a weight that corresponds to the
number of wells in the population that the well represents. The base weight for a well is the inverse of its
probability of selection. The base weights are adjusted for non-response. For the rural domestic well Survey,
trimming and post-stratification were used to make further adjustments to the weights.
B6.2.1 Weighting Procedures for the CWS Well Survey
The weighting procedure for the CWS wells consisted of a two-step process. First, a base weight was
assigned to each CWS well. Then an adjustment was made to the weights to account for non-response.
CWS Base Weight
The base weight assigned to a well is the inverse of the probability of selection. That probability is the
product of three factors associated with each stage of the sample. The CWS well survey was a three-stage'
design. The first stage consisted of the selection of 6,814 CWSs from the FRDS frame of 50,771 CWSs. The
6,814 CWSs were screened to determine eligibility for sampling (5,660 wells were determined to be eligible)
and to determine an estimate of the number of wells in each CWS. The second stage sample consisted of the
selection of 666 CWSs from the 5,660 eligible CWSs. The third stage consisted of the selection of one well
from each of the 666 CWSs (this is a slight oversimplification of the procedure, since some CWSs were
selected more than once at the second stage). The final sample consisted of 592 eligible wells. Analyte and
questionnaire data were collected at 540 of these wells.
The first stage weighting factor for a well is the inverse probability of selecting a community water
system:
-------
B-46 Appendix B: Statistical Design and Analysis
Exhibit B-17
CWS Main Questionnaire Items Imputed Using Respondent Information
Question No.
Abbreviated Question
Imputation
A.12
What is the distance from the ground
surface to the bottom of the casing?
Impute the value recorded in A.5 minus
some final value (i.e., 10 feet) since the
well is not cased to the total depth.
A.17
What is the total distance from the ground
surface to the bottom of the deepest
screen?
Impute the value recorded in A.5 (i.e., well
depth) if A.11 = 1. If A.11 = 2, then use
A.12. If A.11 = 98, then impute 98, 'Don't
Know."
B.4(a)
In 1988, what non-farm pesticides were
stored for more than one month on the
property where the well is located?
Impute the series of pesticides for which a
positive response was either recorded in
question B.4(b) or B.4(c).
B.4(b)
In 1987, what non-farm pesticides were
stored for more than one month on the
property where the well is located?
Impute the series of pesticides for which a
positive response was either recorded in
question B.4(a) or B.4(c).
B.4(c)
In 1986, what non-farm pesticides were
stored for more than one month on the
property where the well is located?
Impute the series of pesticides for which a
positive response was either recorded in
question B.4(a) or B.4(b).
C.1
Has anyone farmed the property since
January 1, 1984?
Impute code 01 "Yes," if a positive
response was recorded in either questions
C.2 or C.3; otherwise use hot-deck.
C.5(a)
In 1988, what farm pesticides were stored,
for more than one month, on the property
where the well is located?
Impute the series of pesticides for which a
positive response was either recorded in
question C.5(b), C.5(c), C.5(d), or C.5(e).
C.5(b)
In 1987, what farm pesticides were stored,
for more than one month, on the property
where the well is located?
Impute the series of pesticides for which a
positive response was either recorded in
qeustion C.5(a), C.5(c), C.5(d), or C.5(e).
C.5(c)
In 1986, what farm pesticides were stored,
for more than one month, on the property
where the well is located?
Impute the series of pesticides for which a
positive response was either recorded in
question C.5(a), C.5(b), C.5(d), or C.5(e).
C.5(d)
In 1985, what farm pesticides were stored,
tor more than one mont, on the property
where the well is located?
Impute the series of pesticides for which a
positive response was either recorded in
question C.5(a), C.5(b), C.5(c), or C.5(e).
C.5(e)
In 1984, what farm pesticides were stored,
for more than one month, on the property
where the well is located?
Impute the series of pesticides for which a
positive response was either recorded in
question C.5(a), C.5(b), C.5(c), or C.5(d).
National Pesticide Survey: Phase I Report
-------
Appendix B: Statistical Design and Analysis B-47
Exhibit B-18
Local Area CWS Questionnaire Items Imputed Using Respondent Information
Question No.
Abbreviated Question
Imputation
2a.
In 1988, what crops were farmed and
what pesticides were used within one-
half mile of the well?
Impute the series of crops and
pesticides for which a positive
response was first recorded in
question 2b or 2c.
2b.
In 1987, what crops were farmed and
what pesticides were used within one-
half mile of the well?
Impute the series of pesticides for
which a positive response was first
recorded in question 2a or 2c. Note:
Select the responses from only one of
the previous years.
2c.
In 1987, what crops were farmed and
what pesticides were used within one-
half mile of the well?
Impute the series of pesticides for
which a positive response was first
recorded in question 2a or 2b. Note;
Select the responses from only one of
the previous years.
4a.
In 1988, what pesticides were used on
the pasture?
Impute the series of pesticides for
which a positive response was first
recorded in question 4b or 4c. Note:
Select the responses from only one of
the previous years.
4b.
In 1987, what pesticides were used on
the pasture?
Impute the series of pesticides for
which a positive response was first
recorded in question 4a or 4c. Note:
Select the responses from only one of
the previous years.
4c.
In 1986, what pesticides were used on
the pasture?
Impute the series of pesticides for
which a positive response was first
recorded in question 4a or 4b. Note:
Select the responses from only one of
the previous years.
-------
B-48 Appendix B: Statistical Design and Analysis
where Sj is the first stage sampling interval within the strata (i.e. on average 1 in every S: community water
systems was selected). The second stage factor for a well in a community water system is defined as:
W. = JL _L +_L
2 if I p
S2'
where S2 is the second stage sampling interval within the strata for the original sample, and S^ is the
corresponding sampling interval for the supplemental sample. By using a supplemental sample random start
equal to the original random start plus half of the original sampling interval, the chance of a system being
selected in both samples was minimized. Afe is the estimated measure of size (estimate of the number of wells
in the CWS from the screener survey). If this factor (W-fi is greater than 1 then it is the inverse of the
conditional probability of selecting the system (conditional on the first stage CWSs). If this factor is less than
1 then it is the inverse of the conditional expectation of the number of times the CWS is selected in the
second stage sample.
For each CWS selected in the second stage the actual number of wells belonging to that system was
ascertained at the time of sampling. In eight CWSs, the actual number of wells was found to be much less
than the expected number of wells. Instead of selecting multiple wells from these CWSs (as would be required
by the second stage sample), only one well was selected. The second stage weight factor for these 8 wells was
redefined as W2 = 1.
The third stage factor for a CWS well is the inverse probability of selecting one well from the system,
W3-Ma
where Ma is the actual number of wells in the system (determined at the third stage, i.e., at the time of water
sampling). The base weight of a well is then,
Wb = Wl x W2 x W3.
CWS Non-response Adjustment
There were 592 wells in the actual CWS well and analyte sample, and questionnaire data was collected
at 540 of these wells. The remaining 52 wells experienced unit non-response. An adjustment to the weights
was required to account for those 52 wells. This adjustment was performed by redistributing the unit non-
respondent wells among the 540 respondent wells.
The wells are partitioned into cells by stratum and State, and a separate non-response adjustment factor
was computed for each cell in this partition. The non-response adjustment factor is,
f
''
National Pesticide Survey: Phase I Report
-------
Appendix B: Statistical Design and Analysis B-49
where ^ * is the sum of the weights of the respondent wells in the cell and ^ * is the sum of the weights
of the nonrespondem wells in the cell. For the few cells in which there were no respondent wells this non-
response adjustment factor is defined to be one. A second non-response adjustment factor was then required,
since the weights of the nonrespondents in those few cells were not redistributed. In that case the wells were
partitioned by stratum and EPA Region. The non-response adjustment factor for cells for this partition is,
EO^*/,) +E <***/«)
* *
where the summation is over the stratum by EPA Region. There are no cells in this partition without at least
one respondent well. The non-response adjustment factor for a well in these cells is the product of the two
factors (with different values for wells in different cells),
For wells in non-collapsed cells, the non-response adjustment factor is /=/j
Final CWS Weight
The final weight of a well in the CWS well survey was the product of its base weight and its non-
response adjustment factor,
W = Wb x /
B6.2.2 Rural Domestic Well Weighting Procedures
The weighting procedure for rural domestic wells consisted of a multi-step process. First, a base weight
was calculated for each rural domestic well. Calculation of this base weight varied, however, depending on
whether the rural domestic well was sampled in one of six pilot study counties or in one of the other 84
counties included in the Survey.
The second stage weights also required an adjustment due to the RDD procedure. A non-response
adjustment factor was also required. Finally, a post-stratification adjustment factor was applied due to updated
information from the 1987 American Housing Survey administered by the Census Bureau.
Rural Domestic Well Base Weights
The rural domestic well sample is a two stage design. The first stage is a sample of 90 counties in the
U.S. The second stage is a sample of rural domestic wells within each county. For the 6 pilot counties, the
second stage sample was selected through field interviewing during the pilot study. For the other 84 counties,
the second stage sample was selected using random digit dialing (RDD) procedures. The 3,137 counties and
county equivalents in the U.S. were stratified into twelve strata (see Exhibit B-2). The 90 first stage counties
were then selected within each stratum with probability proportional to size. The measure of size is the
minimum of the number of rural households and the number of wells9. These numbers are obtained from
9 For more details on the first-stage sample selection see RTI's "National Pesticide Survey Pilot Evaluation Technical
Report," January 29, 1988.
-------
B-50 Appendix B: Statistical Design and Analysis
the 1980 census. The first stage weight for a county is the inverse of the probability of selecting the county,
denoted W^.
The sampling rate within each strata was determined using common sampling rates within certain
groups of strata; (1) high ground-water vulnerability, (2) high pesticide use, (3) both high vulnerability and
high pesticide use, and (4) other strata. After the selection of counties was complete, two of the 90 counties
were re-stratified (the 2 counties had been mis-classified with regard to pesticide use). The weights for these
2 counties are unaffected by this re-stratification.
Second Stage Weights for Pilot Counties
The wells selected within the 6 pilot counties were assumed to be a simple random sample. The
number of rural wells in the county was assumed to be the minimum of the 1980 Census counts of the number
of rural households and the number of wells, denoted N. The second stage weight for a well in the county
is the inverse probability of selection specified as:
***
where n is the total number of wells selected for sampling in the county.
Second Stage Weights for Other Counties
The wells selected within the 84 other counties were selected using RDD. The RDD survey was
implemented using a modification of what is frequently referred to as the Waksberg method.10 The
Waksberg method of RDD draws a sample of households where each household has the same probability of
selection. The modified Waksberg method of RDD draws a sample where households may have varying
probabilities of selection, although the relative probabilities, and hence relative weights, are known. The
purpose of using the modified Waksberg method is that a sample can be obtained more quickly. The Survey
identified rural households within a county and assigned weight, denoted ^RDD» to eac* household. The
weight for the selection of a household in the county is the inverse probability of selection specified as:
j^
WH = ^ X WKDD
L "HDD
where R is the number of rural households in the county (obtained from the 1980 Census), and SHj^o is the
total of the weights of the households sampled in the county.
The wells were selected from the households through a series of steps. First, only rural households with
wells were retained. Second, all rural households with wells in the "cropped and vulnerable" area of the county
were retained, while 1 in 3 of the wells in the "not cropped and vulnerable" area were retained. Third, a
sample of wells was drawn from the remaining sample of households with wells (since a well may serve
multiple households, a well was retained with probability equal to the inverse of the number of households
served by the well). Fourth, a subsample of wells was drawn, since a predetermined number of wells in each
county was required for the sample. A weight, denoted Ws, associated with each well in the sample, accounts
for this series of steps. The second stage weight for a well in the county is the product of the two weights,
JF2 = WH x Ws
10 Waksberg, J. (1978). " Sampling Methods for Random Digit Dialing", Journal of the American Statistical Association,
Volume 73, No 361, pp 40-46.
National Pesticide Survey: Phase I Report
-------
Appendix B: Statistical Design and Analysis B-51
Base Weight Computation
The base weight for a well in the sample is then the product of the first and second-stage weights,
W^-W1x W2
where the second stage weight depends on whether the county is included in the pilot survey or the main
survey.
Non-response Adjustment
Each rural domestic well was classed as a respondent or nonrespondent. An adjustment to the weights
was required to account for the nonrespondents. The weights of the nonrespondents were redistributed among
the weights of the respondents.
The wells were partitioned by stratum and target. The variable target indicates whether the well is in
a "cropped and vulnerable" area of the county or not. A non-response adjustment factor was computed for
each cell in the partition. The non-response factor is,
f =
IX
K
v^ w v^ w
where ^ is the sum of the weights of the respondents wells in the cell and ^ is the sum of the weights
of the nonrespondent wells in the cell. This is the same adjustment factor as the first factor used in the CWS
weight calculations. The adjusted weight of a well is the product of its base weight and non-response
adjustment factor,
W0 = Wtxf
B6.2.3 Post-Stratification
The 1987 American Housing Survey (AHS) administered by the Census Bureau was used for post-
stratification. The 1987 AHS supplies estimates of the number of wells in rural areas by census region and
indicates whether or not the wells are on farms. These data update the 1980 Census information and supply
more appropriate control totals. There are 4 census regions and farming is classified yes or no according to
crop sales. The rural domestic well sample estimates corresponding to these strata (census region and
farming) are adjusted to agree with the 1987 AHS. The post-stratification adjustment was computed for each
cell of the stratification and is specified as:
/,=
IX
where A is 1987 AHS estimate of the number of rural wells in the stratification cell, and %W is the weighted
total of the sample wells in the stratification cell.
-------
B-52 Appendix B: Statistical Design and Analysis
Final Rural Domestic Well Weight
The final weight of a well is the product of the adjusted weight and post-stratification adjustment,
Wf-W.xf,
Weight Trimming
After completing post-stratification the distribution of the individual rural domestic well weights was
examined to determine if the variability in the weights might produce estimates with exceedingly large mean
square errors (MSE). When weights are highly variable it is often possible to reduce the MSE by trimming
the largest weights.11 Trimming will reduce the variability of the weights, but may introduce a source of bias.
If this bias is relatively small the net effect of trimming the weights is to reduce the MSE, thereby improving
the accuracy of the estimates.
The variability in weights for the rural domestic well survey primarily comes from four sources:
sampling error, cluster weights (the probability of selecting a particular cluster of telephone numbers for
inclusion in the RDD survey is inversely proportional to the number of households contacted in the cluster),
requiring a fixed sample size per county within each stratum (rather than a fixed sampling rate), and relative
changes in the number of rural households in a county from the 1980 Census (post-stratification is only able
to control the total for a region of the country, not each county). Trimming weights will only introduce bias
into the estimates if the reason for the large weights was highly correlated with the variables of interest. Given
the above sources of variation in weights, it is doubtful that trimming introduced any significant bias.
The average weight of the 752 rural domestic well respondents was 13,975; with a largest weight of
108,473. The nine wells (1% of the sample) with the largest weights represented 831,000 wells (9% of the
population). It was therefore decided to trim these 9 weights that exceeded 60,000 by replacing their weights
with 60,000, then re-post-stratifying to retain the national total number of wells. This final component for the
rural domestic well weights is incorporated as:
W = aan(W, 60,000) x
, 60,000)
This component reallocated a total weight of approximately 270,000 among the wells in the same post-
stratification cells as those that were trimmed. Since this represents less than 3 percent of the national total,
its impact on any biases is likely to be minimal.
B6.3 Calculation of Population Estimates and Confidence Intervals
The population estimates and confidence intervals presented in Chapter 6 are the result of applying the
sample design through the sample weights to appropriate sample statistics. The following sections describe
the methods used for calculating the population estimates (in terms of numbers and proportions), and the
confidence intervals for those estimates.
11 For example, Potter, R, "Survey of Procedures to Control Extreme Sampling Weights," Proceedings of the Section
on Survey Research Methods, the American Statistical Association, 1988, pp. 453-458.
National Pesticide Survey: Phase I Report
-------
Appendix B: Statistical Design and Analysis B-53
B6.3.1 Population Estimates
A population estimate is simply the weighted sum of the sample statistics. Thus, the calculation of
population estimates accounts for the complex sample design through the weights. A population estimate,
Xd, (for example, the number of wells in domain d that contain a Survey analyte) is calculated as:
where Wi is the weight associated with the i& well in the d domain, «d represents the number of sampled
wells in the d domain, and X{ is an indicator for the presence or absence of the analyte.12 That is, Xi =
0 or 1 where the value of 1 represents presence of the characteristic. The estimate of the proportion, PA is
then calculated as
X.
where vVd is the estimated number of wells in the domain population. That is:
hence
^
Percent estimates are then calculated as 100 x Pd. (Note that for most domains, Nd is also a random variable,
and this affects the calculation of variance estimates as described below.)
B6.3.2 Confidence Intervals
Two methods were used for calculating confidence intervals when the sample estimate is greater than
zero in the Survey. Initially, confidence intervals were produced for all characteristics of interest using the
WESVAR13 procedure. The confidence intervals produced in this manner assume an underlying normal
distribution for the estimates. Use of the normal distribution is appropriate in the Survey when the binomial
parameter of interest is not extreme (i.e., not near zero or one), and the sample size is sufficiently large.
Binomial confidence intervals were also calculated for many Survey characteristics when the binomial pro-
portion is very small. Finally, in the most extreme cases, i.e., when the binomial proportion is zero, upper
confidence bounds were calculated as described in this section. The three methods are discussed separately
below.
12 The underlying distribution for these populations is Hypergeometric because the populations are finite, but the
populations are large enough that a Binomial distribution assumption (a well either does or does not contain a Survey analyte)
is justifiable.
13 WESVAR is a SAS procedure written and maintained by Westat, Inc.
-------
B-54 Appendix B: Statistical Design and Analysis
Confidence Intervals Based on the Normal Distribution
The Survey used a complex design with unequal probabilities of selection for both the CWS and rural
domestic well Surveys. This requires the use of complex computation procedures to calculate design-based
variance estimates, and hence standard errors and confidence intervals.
Confidence intervals for the Survey were initially calculated using the WESVAR procedure. Briefly,
the method used takes a number of subsamples of the actual sample and computes the parameter of interest
for each subsample. A variance estimate is then computed as the variance of the values of the parameter
estimates from each subsample. These subsamples are usually referred to as replicates. The method is
referred to as the Jackknife, or Jackknife replicates, method.14
Because of differences in the designs, different approaches to the formation of replicates for the variance
estimation were used for the CWS and rural domestic well Surveys. The CWS sample was a systematic sample
within each of the 12 first-stage strata with systems as primary sampling units. The variance estimation
procedure must reflect this design. Within each of the strata (ground-water vulnerability by pesticide use) the
selection order of the wells is known. The wells within a stratum are further partitioned into a number of
pairs of variance strata. For instance, the 28 wells in first stage stratum one (high ground-water vulnerability
and high pesticide use) are partitioned into 4 variance strata;
variance stratum 1A: 1,3,5,7,9,11,13
variance stratum IB: 2,4,6,8,10,12,14
variance stratum 2A: 15,17,19,21,23,25,27
variance stratum 2B: 16,18,20,22,24,26,28
where the integers 1 to 28 correspond to the selection order of the wells in the first-stage stratum. The
variance strata 1A and IB form a pair, and the variance strata 2A and 2B forms another pair. Similarly, each
of the other 11 first stage strata are partitioned into a number of pairs of variance strata. In this manner, a
total of 38 pairs of variance strata (i.e., 76 variance strata) are created. Exhibit B-19 shows the number of
variance strata in each of the twelve strata. Variance estimation then employs 38 replicates, where a replicate
is formed by selecting one of the 38 pairs of variance strata, deleting the wells from one of the variance strata
in that pair, and increasing the weights of the wells in the other variance stratum of the pair.
14 Wolter, M. (1985). Introduction to Variance Estimation. Springer-Verlag, New York.
National Pesticide Survey: Phase I Report
-------
Appendix B: Statistical Design and Analysis B-55
Exhibit B-19
Partition of First Stage Strata Into Variance Strata
First Stage Strata
1
2
3
4
5
6
7
8
9
10
11
12
Total
Number of Wells
28
26
14
39
32
24
71
70
52
75
89
72
592
Number of Variance
Strata
4
2
2
4
4
4
10
10
6
10
10
10
76
Number of Pairs of
Variance Strata
2
1
1
2
2
2
5
5
3
5
5
5
38
The DWS sample design is a clustered sample within first stage stratum, with 90 counties as the primary
sampling units in the first stage sample (see Section B4.4). Thus, 90 replicates are formed, where a replicate
is formed by deleting the wells from a county and increasing the weights of the wells in the remaining counties
from the same stratum. The jackknife variance estimate thus accounts for the clustered sample within first-
stage stratum design.
The variance estimates do not assume an underlying distribution; they are simply variances of a data
set. Confidence intervals can be constructed from these variance estimates by making a distributional
assumption, such as normality. The form of a 95% confidence interval for an estimate, P, then is:
where P is the estimate of the parameter of interest and the value 1.96 corresponds to the 95% confidence
interval for the Normal distribution.
Using this method, confidence intervals can be calculated for all parameters of interest in the Survey
by assuming an underlying Normal distribution for the parameters, i.e., estimates of population numbers, &
and population proportions, P. Recall:
and
x =
N
In most domains the population total, N, is also a random variable. (N was assumed fixed in the "rural
domestic wells nationally" domain. A value of 10,509,373 was used based on the 1987 American Housing
Survey administered by the U.S. Bureau of the Census.) However, variance estimation is performed from the
sample data directly. The calculation of variances of ratio estimators is, in some sense, avoided using the
-------
B-56 Appendix B: Statistical Design and Analysis
Jackknife replicates method. One effect of the ratio of random variables, however, is that the variance of the
proportion does not "match" the variance of the corresponding population number. That is, if N is a constant
(as for the "rural domestic wells nationally" domain) then:
VAR (P) = VAR(X)
and the variance of the proportions are a constant multiple of the variance of the numbers. But, when N is
a random variable (as is the case for all other domains):
VAR(P) = [VAR(X) + VAR(N) - 2 COV(X,N)}
(Nf
and the respective variances of P and % are not constant multiples of one another. The jackknife replicates
method accounts for this difference implicitly rather than explicitly.
Confidence Intervals Based on the Binomial Distribution
As discussed previously, the underlying distribution of proportions of Survey characteristics is binomial.
When the estimated proportion is extreme and the sample size is not large enough to overcome this extremity,
the normal distribution approximation to the binomical distribution is not justifiable. For these cases,
confidence intervals were based on the Binomial distribution.
The upper bound for the 95% confidence interval for the binomial distribution is calculated by solving
for Pu in the equation:
i-O
n] f'll- P.)""' * 0.025
where n is the sample size and a is the sample estimate of the number of wells for which the characteristic
of interest is present. Solving for Pu corresponds to computing Pu such that for this value the probability of
obtaining a or less of the characteristic is less than 0.025.
A similar formulation is necessary for the lower bound estimate />L:
£ tn\? 1(1-^)* 0.025
Pratt and Gibbons15 (1981) note that the binomial confidence bounds are quantiles of the Beta distribution.
The values presented in Chapter 6 were taken from appropriate Beta distribution charts.
In the earlier discussion of confidence intervals based on the normal distribution, the confidence
intervals for population estimates did not match those for estimates of proportions (except in the case of rural
domestic wells nationally). Unlike proportions, (which follow a binomial distribution), the underlying
distribution for population estimates is not clear. Confidence intervals for population estimates corresponding
to extreme proportions were calculated by simply adjusting the binomial confidence intervals for proportions
to the population size. This makes an implicit assumption that the total number of rural domestic wells (N)
15 Pratt J., and Gibbons J. (1981) Concepts of Nonparametric Theory. Springer-Verlag, New York.
National Pesticide Survey: Phase I Report
-------
Appendix B: Statistical Design and Analysis B-57
is a constant, and not a random variable. This method was also used when the sample estimate is zero. Other
methods may be used for the Phase II report, which may produce different values.
Calculation of confidence intervals reflected the complex sample design, which must be taken into
account when computing error bounds. If the Survey had used a simple random sample, then the binomial
confidence interval would depend on the sample estimate of the proportion of wells with a particular
characteristic of interest (for example, containing detectable levels of an analyte), and the sample size.
However, the Survey used a more complex design. The effective sample size is calculated for each domain as:
where nd is the actual sample size in domain d and 5d is the design effect.16 The design effect depends on
the weights associated with each well in the domain of interest:
V
»,£
«,-*
If the weights are all equal, as is the case with simple random sampling, then the design effect is one. Exhibit
B-20 shows the design effects computed for the seven domains of interest in the Survey.
Kish, L (1965) Survey Sampling. Wiley, New York.
-------
B-58 Appendix B: Statistical Design and Analysis
Exhibit B-20
Design Effects for the Domains of Interest
Domain
Community water system wells
nationally
Community water system wells in
counties with high vulnerability
Rural domestic wells nationally
Rural domestic wells in counties with
high pesticide use
Rural domestic wells in counties with
high vulnerability
Rural domestic wells in 'cropped and
vulnerable" regions
Rural domestic wells in counties with
high pesticide use and high
vulnerability
Design Effect
(6)
1.27
1.16
1.84
1.77
1.91
2.06
1.44
Sample Size
(n)
540
197
752
217
264
343
83
Effective
Sample Size
(n")
425
170
409
123
138
167
58
In the rural domestic well survey, the design effect also depends on a cluster effect (an effect due to
.selecting wells in only 90 counties). It is not possible to estimate directly the intra-county correlation for
analytes without any detections. Considering analytes that were detected, the intra-county correlation is
estimated to be small. The additional factor in the design effect due to the county clustering is therefore
negligible, and the design effect in the rural domestic well survey is estimated using the identical formula used
for the CWS well survey.
Confidence Bounds When the Sample Estimate is Zero
When the sample estimate,/), of the proportion of a population characteristic of interest (for example,
presence of pesticides above the HAL/MCL in CWS wells) is zero, this does not necessarily imply that the
number in the population is also zero. Some percentage of the population may have the characteristic, but
the Survey found none by random chance. In these cases an upper confidence bound (95%) is calculated.
This confidence bound corresponds to the probability that the Survey does not find the characteristic of
interest given the proportion in the population that do have that characteristic. In the case of detectable levels
of pesticides above the HAL/MCL in the CWS Survey this can be presented as the probability of no detections
in the Survey, given the proportion in the population, Pr (no detections \ p). If the Survey used a simple
random sample, then each selected well would have a detectable level of the analyte (above HAL/MCL) with
probability/), and hence would not have a detectable level above the HAL/MCL with probability 1 -p. Using
the binomial distribution assumption, a sample of n wells would then have no detections (above HAL/MCL)
National Pesticide Survey: Phase I Report
-------
Appendix B: Statistical Design and Analysis B-59
if each well was found not to have such a detection. The actual Survey used a complex design, so the weights
are accounted for by the design effect as indicated above. Hence:
Pr (no detections in domain d \ />) = (1 - />)"'
A 95% upper confidence bound for the proportion in the population, Pd, is the solution to:
(1 - £,)"' = 0.05
or equivalently,
pd = 1 - 0.051"1'
The calculations for the 95% upper confidence bounds also account for non-response. These upper bound
estimates were initially required for analytes that were not detected in the Survey. For example, pebulate was
not detected in the Survey, but, for the reasons outlined above, this does not mean that pebulate is not present
in any CWS wells or rural domestic wells nationwide.
The design effects were calculated based on the complete sample within domains. However, the initial
sample size used for these binomial upper confidence bounds corresponds to the number of respondent wells.
This value differs for each analyte. The most conservative upper bounds correspond to the analyte that has
greatest non-response. That is, a different upper bound exists for each analyte. The one presented here
corresponds to the analyte with the greatest non-response and hence the smallest initial sample size. Exhibit
B-21 presents the actual sample sizes, effective sample sizes, and 95% upper confidence bounds for the seven
domains of interest. The exhibit also includes the name of the analyte that had greatest non-response in each
domain.
National Pesticide Snruau- OH*
-------
B-60 Appendix B: Statistical Design and Analysis
Exhibit B-21
Confidence Bounds for Population Estimates of Zero
Domain
Community water system wells
nationally
Community water system wells in
counties with high vulnerability
Rural domestic wells nationally
Rural domestic wells in counties
with high pesticide use
Rural domestic wells in counties
with high vulnerability
Rural domestic wells in 'cropped
and vulnerable' regions
Rural domestic wells in counties
with high pesticide use and high
vulnerability
Sample Size
479
168
694
192
236
305
66
Effective
Sample Size
377
145
377
108
124
148
46
95% Upper
Confidence
Bound
.008
.020
.008
.027
.024
.020
.063
Analyte
Used
Pebulate
Pebulate
Butachlor
Ametryn
Butachlor
Butachlor
Butachlor
The estimates presented in Exhibit B-21 were presented in Chapter 6 whenever a population estimate
was zero (or, the sample estimate was zero). A separate calculation should be made in each case to account
for non-response for each particular characteristic of interest. The estimates used are conservative for all
analytes as described above. Calculation of more accurate upper bounds for each individual case would have
been time consuming and would not have resulted in substantially different estimates.
Geometric Means and Confidence Intervals
Geometric means, with confidence intervals, are provided in Chapter 6 for nitrate, atrazine, and DCPA
acid metabolites, separately for both CWS wells and rural domestic wells. This section explains how they are
calculated.
The geometric mean, GM, of a set of positive numbers (xa, x2,... x,,) is defined as:
GM = exp{-
National Pesticide Survey: Phase I Report
-------
Appendix B: Statistical Design and Analysis B-61
This formulation makes no distributional assumptions and may be regarded as a summary statistic. As the
data in the NFS is weighted, a weighted geometric mean, GM^ is a more appropriate summary statistic and
is the one presented in Chapter 6:
where N is the population size, or more generally, A , and J^ represents the weight of a particular
observation. ~ M '
To obtain confidence intervals for the geometric mean requires distributional assumptions. The most
usual assumptions (for unweighted data) are that the natural logarithm transformation for the data follows
a normal distribution, so that the data follows a lognormal distribution. A sample estimate of the mean of
Yis:
" - - T Y - -
n n
The sample geometric mean can then be calculated as e^. Confidence intervals at 95% for the mean of log
transformed data can also be calculated as
A ± 1.966 /
where a is the usual estimator of the standard deviation for normal data V "~
Confidence intervals for the geometric mean of sample data can then be calculated by taking exponents of the
limits of the confidence interval for /x. This procedure provides confidence intervals for the geometric mean
in the sense that the NFS data is weighted to account for the Survey design, and a weighted geometric mean
is appropriate. Accordingly, confidence intervals for (j. were calculated using the WESVAR procedure.
If the geometric mean and confidence bounds are denoted ' " > e l> cau* e " then the geometric standard
deviation (GSD) can be calculated as follows:
\LL = n - 1.96 o /
-------
B-62 Appendix B: Statistical Design and Analysis
hence
1.96
The geometric standard deviation is the exponential of this result, ea. The sample size used in this calculation
should account for the design through the design effect (see Exhibit B-20). The geometric standard deviation
is not the standard deviation of the lognonnal distribution.
The NFS concentration data is truncated at the lower end by the MRL associated with each chemical.
Consequently, the geometric mean, which is an estimate for the NFS data only, and the confidence intervals,
which relied on an assumption of normality for the log transformed data, should be regarded as preliminary.
Concentration distributions will be estimated in Phase II.
B6.4 Comparison of Phase I Results With Precision Requirements
The precision requirements for the Survey are presented in Exhibits B-5 (for the CWS well survey) and
B-10 (for the rural domestic well survey). Exhibit B-22 provides a comparison of the precision requirements
in terms of relative standard errors and detection probabilities, and the corresponding values achieved by the
Survey for the proportions of wells with detectable levels of at least one pesticide for each of the seven
domains of interest. Necessary formulas are provided in Section B3.1.
For comparison purposes, relative domain sizes are provided though the relative standard errors and
detection probabilities are calculated based on assumed (for the precision requirements) and estimated (from
the Survey data) detection rates within each domain. That is, the assumed P for the precision requirements
is 0.5% for both CWS domains, and 1% for all rural domestic well domains, and the estimated £ for the values
achieved by the Survey are (in order) 10.41%, 9.30%, 4.24%, 3.55%, 2.77%, 5.48%, and 1.41% (as shown in
Chapter 6 of the Phase I Report). The difference between these values and the relative domain sizes is
accounted for by the assumed and estimated domain population sizes. For example, of the estimated 94,600
CWS wells nationally, 20,800 were in counties with high ground-water vulnerability.
As can be seen (Exhibit B-22), the Survey surpassed the design specified precision requirements. The
actual relative standard errors are all lower, and, consequently, the actual detection probabilities are all higher.
The reasons for this are twofold. The specified precision specifications were purposefully conservative and
the proportions of wells that contain detectable levels of at least one pesticide was greater than expected,
especially for the CWS well survey.
National Pesticide Survey: Phase I Report
-------
Appendix B: Statistical Design and Analysis B-63
Exhibit B-22
Comparison of Survey Design Precision Requirements
With Survey Results
Community water system
wells nationally
Community water system
wells in counties with high
vulnerability
Rural domestic wells
nationally
Rural domestic wells in
counties with high pesticide
use
Rural domestic wells in
counties with high
vulnerability
Rural domestic wells in
cropped and vulnerable1
regions
Rural domestic wells in
counties with high pesticide
use and high vulnerability
II
Design Specifications j| Survey Results
Relative
Domain Size
0.5%
0.1%
1.0%
0.14%
0.25%
0.25%
0.03%
RSE
0.658
1.040
1.0
0.85
0.85
0.525
1.25
Detection II Domain
Probability || Size
0.90
0.60
0.63
0.75
0.75
0.97
0.47
10.81%
2.16%
4.24%
0.39%
0.70%
1.93%
0.05%
RSE
0.14
0.34
0.23
0.79
0.57
0.35
2.31
Detection
Probability
1.00
1.00
1.00
0.99
0.98
1.00
0.56
National Pecflrlrte
-------
National Pesticide Survey
Appendix C: Implementation of the Survey
-------
Appendix C: implementation of the Survey
C1 Introduction
This Appendix describes the activities that occurred between January 1988 and February 1990, in order
to carry out well sampling and data collection for the National Pesticide Survey. The activities that formed
the core of NFS implementation include training of field sampling and interviewing teams, selecting eligible
wells, field operations and data collection, and database development. They are described below in roughly
the chronological order in which they were carried out.
C2 Training of Field Sampling/Interviewing Teams
The National Pesticide Survey (NPS) Well Sampling Training Program provided NFS samplers with
technical training in sampling and interviewing procedures to ensure the consistency and integrity of the data
collected during the Survey. All EPA Regional participants and State personnel conducting well sampling or
administering questionnaires received the NPS CWS Training Program. For the rural domestic well survey,
separate training sessions were held for samplers and interviewers. Participants received formal personal
instruction from NPS staff for specific procedures and were given written training materials, including the CWS
or Rural Domestic Well Training Manuals.1 NPS training of EPA Regional and State sampling teams was
presented in one-day or one-and-a-half day workshops (depending on schedules and travel budgets of State
or Regional personnel) by CWS trainers. The NPS prime contractor, ICF Incorporated, conducted rural
domestic well sampling training at its headquarters and regional offices and observed at scheduled sampling
sites to provide hands-on field sampling training to staff. For the rural domestic well survey, interviewers from
the NPS survey subcontractor, Westat Incorporated, received a minimum of four hours of general interviewer
training and nine hours of NPS-specific training, covering both interviewing and assisting with water sampling.
These training sessions were conducted at Westat's headquarters. All well sampling teams received training
manuals providing guidance on field team organization, well area data collection, sampling of wells, and
administration of questionnaires. The Field Logbook provided detailed guidelines, checklists, and forms for
NPS team members to conduct their work. NPS sampling team trainers conducted training for contractor,
State and Regional staff from May 1988 to June 1989.
C2.1 Training Topics and Materials
The NPS CWS training covered the following topics:
An overview of the NPS;
Field team organization;
Well area data collection;
General instructions for sampling CWS wells;
Field well site water sampling procedures;
Instructions for administering CWS well questionnaires;
Chlorine test procedures; and
Communications.
NPS rural domestic well training was split into two parts. Water samplers were trained on the
sampling procedures; interviewers were trained in interviewing and water sampling assistance.
1 Water sample collection training procedures for Community Water System (CWS) wells and rural domestic wells were
similar. When the training varied, the differences are described.
National Pesticide Survey: Phase I Report
-------
C-2 Appendix C: Implementation of the Survey
The NPS Well Sampling and Interviewing Training Overview. The NFS well
sampling and interviewing training began with an overview of the goals and objectives of the NPS and an
explanation of the background, history, and need for the Survey. The trainer also outlined the key elements
of the four major components of the Survey:
1. Selecting a representative set of drinking water wells (NPS statistical sampling design);
2. Collecting and analyzing well water samples to measure the types and amounts of pesticides
present;
3. Issuing Health Advisories to well owners/residents2 with sampled wells containing levels of
pesticide or nitrate concentrations that may pose a health risk; and
4. Collecting key information through questionnaires and forms to obtain data on additional
factors affecting the presence of pesticides or nitrate in drinking water wells.
Training In Field Team Organization and Responsibilities. This training section
emphasized the organization and assignment of field duties, logistical arrangements, and the practice of field
procedures. Field sampling teams learned the duties of the three individuals on each sampling team: the field
team leader, well sample collector, and questionnaire interviewer. (For many CWS and rural domestic wells,
these three roles were handled by two people.) The field team leader received instructions for overall
coordination of sampling (and data collection for CWS wells) activities such as preparation, mobilization,
sampling, interviewing (for CWS wells), shipment of sample container kits to the laboratories, and return of
the field equipment after completion of all well assignments. The field team leader also learned to monitor
all activities (sampling only for rural domestic wells) to uphold quality assurance protocols. The well sample
collector learned all appropriate procedures to ensure proper sampling techniques in the field. The designated
questionnaire interviewer learned to schedule interviews with local officials, review and prepare questionnaires
prior to field visits, conduct interviews, edit each questionnaire upon completion, and return all completed
questionnaires and other interviewing materials to the appropriate NPS staff (the field team leader for CWS
and the field supervisor for rural domestic well sampling).
Training In Well Area Data Collection. The complexity of prefield preparation and
sampling of well sites required extensive hands-on training and supporting written documents to ensure
consistency during the Survey. NPS staff responsible for collecting data on the well area were trained to use
the NPS Field Logbook, the main instrument for recordkeeping during well sampling. The Field Logbook
contained forms, checklists, and instructions for sampling (and data collection for CWS wells). Exhibit C-l
provides a list of all training materials provided in the Field Logbook as well as other supporting materials.
Training for Sampling Wells. The field sampling teams received instructions for collecting
well water samples. Participants learned about health and safety aspects of sampling, quality assurance/quality
control procedures, and communications protocols with well owners, county officials, and the news media.
Health and Safety. This training emphasized the safe handling of sample preservatives such as
mercuric chloride, sulfuric acid, and pH buffer solution. The sampling teams received a copy of the
Materials Safety Data Sheets for each preservative.
2 The term "owners/residents" is used to encompass the public or private use of both CWS and domestic wells. CWS
wells belong to a public water system that can involve several "residents." Domestic wells are privately owned wells used by
the "owner" or by "residents" who rent the property from the owner.
National Pesticide Survey: Phase I Report
-------
Appendix C: Implementation of the Survey C-3
Exhibit C-1
Training Materials for the
Well Sampling Training Program3
Field Logbook Materials
Well Sampling Information Sheet;
Instructions for use of the Field Logbook;
Supply Kit Equipment Checklist;
CWS/Rural Domestic Well Purging Parameters Record;
Final CWS/Rural Domestic Well Checklist;
Shipping Confirmation Sheet;
Team Leader Introduction and Well Observation Record Booklet;
Field Well Site Water Sampling Manual; and
Interviewing Package including a CWS Questionnaire and a Local Area CWS
Questionnaire.
Other Supporting Materials
Instructions for the Conductivity Tester;
Instructions for the pH Tester;
Instructions for the Temperature Tester;
Materials Safety Data Sheets; and
Federal Express Airbill.
3 These materials are contained in the NFS Field Logbook and the appendices to the CWS Training Manual and the
Domestic WeJI Manual.
National Pesticide Survey: Phase I Report
-------
C-4 Appendix C: Implementation of the Survey
Quality Assurance/Quality Control. This training stressed the need to crosscheck other team
members' work (for rural domestic wells such crosschecking was limited to water sampling activities).
Completion of collection forms, bottle labeling, and packaging were covered; sampling team members
were taught to use the NPS hotline for consultation about sampling problems. Sampling team
members were trained to review and edit questionnaires in the field, to ensure that they were
complete and legible. CWS sampling team members were also asked to contact the CWS field
director, via the NPS hotline, with questions about completing the questionnaires. Rural domestic
well interviewers were trained to contact the rural domestic well field director directly.
Communications. This training taught field team members how to answer technical questions about
the NPS, and to inform the questioner about other sources of information, especially the toll-free NPS
hotline.
Training In Field Well Site Sampling Procedures. Each field team member received
training in conducting well sampling that included prefield preparation, CWS/rural domestic well sampling
procedures, and procedures to follow after completion of sample collection. The training also discussed the
rationale for technical procedures during sampling, such as the use of different sized sample bottles to
accommodate various analytic procedures at the contract laboratories. (Section C4.3 provides a description
of sampling procedures.)
Prefield Preparation. Field team members learned the detailed steps in the Well Sampling
Information Sheet and the Team Leader Introduction and Well Observation Record contained in the
Field Logbook.
CWS/Rural Domestic Well Sampling Procedures. The training for sampling procedures included
explanations of how to:
Randomly select one well site among all wells comprising the water system (CWS
only);
Locate a sampling port or tap as close to the well head as possible;
Measure the air temperature around the well tap;
Open the well sampling port or tap valve and purge the well (measure pH,
temperature, and conductivity);
Collect a sample of water;
Properly handle and label sample bottles;
Review and complete relevant paperwork on the Sample Tracking Form to check
that all sample bottles were collected in the sample container kit; and
Conduct the final pH, conductivity, and temperature measurements, and
appropriately mark the readings on the CWS/Rural Domestic Well Purging
Parameters Record.
After Sample Collection. Field team members learned that shipping the samples according to
procedures was essential to preserve the integrity of the samples. NPS field sampling teams learned
how to pack sample container kits in ice, properly complete the Federal Express Airbill, complete the
Well Observation Record, and ship the samples to laboratories.
National Pesticide Survey: Phase I Report
-------
Appendix C: Implementation of the Survey C-5
Training for Administering CWS/Rural Domestic Well Questionnaires. Designated
interviewers learned how to administer Main and Local Area questionnaires. The topics covered included:
General Interviewing Techniques. Participants learned how to obtain the cooperation of the
respondent, ask questions in an unbiased manner, and record the data so that it could be easily
transferred to a computerized data base.
Technical Information. The trainer reviewed each question in the questionnaires, giving the intent
of the question, providing the interviewer with written definitions of technical terms to be read to the
respondent in answer to his or her questions, and teaching the interviewer how to record the response.
Rural domestic well training included role playing to practice handling different potential situations.
Answers to Respondents' Questions. Participants were provided with a list of answers to
questions commonly raised by respondents during interviews concerning the objectives of the Survey
and how they would be notified about the results of their water samples.
Training In Chlorine Test Procedures. During the Survey, NPS contract laboratories
discovered free chlorine in some well water samples that had been collected before treatment of the water
supply. Therefore, sampling teams received additional instructions through letters and phone calls on how
to resolve this problem and received new test procedures in the chlorine test kit enclosed in the equipment
supply kit. There was a slight variation between chlorine testing for CWS and rural domestic wells, but both
procedures still assured accuracy and consistency of sample results. CWS sampling teams learned how to
conduct a simpler qualitative chlorine test procedure; rural domestic well sampling teams received training to
implement a more complex quantitative chlorine determination procedure.
Training In Communications and Confidentiality Issues (rural domestic well
training only). Rural domestic well sampling teams received special training to address confidentiality
issues for the owners/residents of sampled rural domestic wells. EPA established a confidentiality policy to
protect the privacy of rural domestic well participants because of the potentially sensitive nature of the
sampling and data information. The policy does not apply to community water systems since these systems
serve the public and therefore any CWS data is considered public information. Field sampling teams learned
to communicate EPA's confidentiality policy to rural domestic well participants and assure them that sample
results would only be released to the State and EPA. No questionnaire information would be made available
to anyone unless stripped of all identifiers. Where States had a more stringent confidentiality policy or a
conflicting policy, this information was provided to the field sampling teams in their communications packets
and explained to participants when inquiries were made.
C2.2 Scheduling CWS and Rural Domestic Well Sampling Training
Dozens of CWS and rural domestic well sampling teams required training in specific NPS sampling
and data collection procedures. The State and Regional CWS sampling teams were located in 49 States and
the NPS training group manager had to ensure that each team was prepared to conduct well sampling and data
collection activities prior to their first scheduled NPS assignment. It was also desirable to schedule each CWS
training workshop as close as practical to the date of the first sampling assignment for those being trained to
maximize the value of instruction.
Scheduling CWS Training
EPA Regional and State NPS sampling teams were selected by EPA Regional and State contacts.
Once all CWS sampling dates were scheduled and sampling teams were identified in a given State, the NPS
National Pesticide Survey: Phase I Report
-------
C-6 Appendix C: Implementation of the Survey
communications manager notified the State CWS Sampling Contact to select a date and location for NFS CWS
training. The NFS CWS training program was presented in a one-and-a-half day workshop; however, if
participants' schedules or travel budgets could not accommodate an overnight stay, the workshop was
condensed into a one-day session. To the extent feasible, training programs were scheduled close to the
sampling team's first scheduled sampling assignment. If possible, the training was scheduled to allow the
trainers to accompany the State or Regional sampling team on their first sampling assignment to provide them
with hands-on field training. In large States, multiple CWS training programs were scheduled to accommodate
sampling teams located in different regions within the State.
The NFS communications coordinator arranged logistics for the training session, including locating
a training facility, obtaining directions to the training site, and providing the State Sampling Contact, in
writing, with this information. Training manuals were shipped to the training location prior to the scheduled
training date to provide participants an opportunity to review the materials.
The NFS training group manager assigned a CWS well sampling trainer for each training program.
The NPS provided instructors for interviewer training. The interviewing instructors underwent a training
program themselves which included a trainer's manual and monitoring by another trainer during their first
training session. This was done to ensure consistent quality for all training sessions. The NPS
communications coordinator prepared a monthly training schedule for all CWS trainers and provided them
with logistical information for each program.
Scheduling Rural Domestic Well Training
NFS contractor staff were independently trained in rural domestic wellwater sampling and interviewing
procedures. The interviewing trainers instructed rural domestic well interviewers on survey administration
protocols and procedures for assisting with water sampling prior to the implementation of NPS rural domestic
wellwater field sampling. The NPS training group manager selected dates for NPS rural domestic well
sampling training for field office sampling teams. To the extent possible, water sampling training programs
were scheduled close to the sampling team's first scheduled sampling assignment. If possible, the training was
scheduled to allow the trainers to accompany the well sampling team on their first scheduled sampling
assignment to provide them with hands-on field sampling training.
The rural domestic wellwater sampling teams were trained in the prime contractor's regional offices,
where sampling team members were located. The NPS training group manager coordinated with the NPS field
office manager to schedule the training and make necessary logistical arrangements. The NPS field office
manager was responsible for informing the field office sampling team of the training dates and location and
verifying with the NPS training group manager the number of participants expected to attend the training
program in that region. Training manuals were shipped to the training locations prior to the scheduled
training dates. Before the rural domestic well sampling training was conducted in a region, each NPS field
office manager assigned the field sampling team to specific county assignments and informed the NPS well
sampling group manager of these assignments.
The NPS training group manager assigned a rural domestic well trainer to each training program. The
NPS training group manager prepared a monthly training schedule for all rural domestic well sampling trainers
and provided them with logistical information for each program.
The rural domestic well survey field interviewers were trained at the subcontractor's headquarters in
Rockville, Maryland. Three rural domestic well survey field interviewing training sessions were conducted
during the course of the Survey. Training sessions were scheduled as new interviewers were required to meet
the Survey schedule.
National Pesticide Survey: Phase I Report
-------
Appendix C: Implementation of the Survey C-7
Scheduling CWS and Rural Domestic Well Refresher Training
NFS sampling training workshops were held from May 1988 until June 1989. If an extended period
of time lapsed between the first sampling assignment and subsequent sampling assignments, the NPS training
group manager contacted the sampling team by phone to review sampling protocols prior to the next sampling
assignment. The NFS training group manager provided refresher training instruction to CWS and rural
domestic well sampling teams until January 1990. (After training was completed in Minnesota, the Minnesota
sampling teams requested that District Engineers be allowed to conduct the interviews with CWS
owners/operators. The District Engineers were trained to complete the questionnaires by the CWS field
director over the telephone.)
In addition, steps were taken to identify and retrain team interviewers. These steps are described in
detail in Section C4.4 under field supervision.
C3 Selecting Eligible Wells
C3.1 CWS
Procedures for Scheduling Community Well Field Sampling
From July 1988 through December 1989, EPA sampled 566 Community Water System (CWS) wells
in 49 states. EPA had to identify a representative sample of CWS that met specific NPS eligibility criteria
prior to collecting drinking water samples. EPA developed a master CWS sampling schedule and conveyed
this schedule to State and Regional CWS Contacts and CWS owners/operators. The Sampling Team contacted
the CWS owner/operator to arrange a convenient field sampling date. The NPS implementation support
subcontractor screened the frame, selected the sample, and provided the first draft of the project schedule.
The NPS prime implementation support contractor facilitated the final scheduling activities.
Screening to Identify Eligible Community Wells
EPA initiated screening to identify eligible community wells with the 1984 Federal Reporting Data
System (FRDS). A community water system was defined in the NPS study as a water distribution system that:
Provides piped drinking water;
Has 15 or more connections; and/or
Provides water to 25 or more people who are permanent residents of the service area.
The Federal Reporting Data System (FRDS) was selected as an initial sample frame to identify the
population of community water systems. Using the definitions listed above, over 51,000 entries in the FRDS
were identified as possible eligible CWS. Of these 51,000 entities, 7,083 were randomly selected for continued
consideration for eligibility.
Each EPA Region was sent a copy of the list of CWS in its Region that were selected to be in the
sample of 7,083 systems. The Regions then sent this information to each individual State for review.
Comments and corrections were returned by States and Regions and integrated into the FRDS sub-sample.
The experience with the State of California illustrates the value of this methodology. Upon reviewing the
sample selected for California, the State determined that the FRDS file used as the sampling frame was
missing all small community water systems in the State. (California had provided the submission to FRDS,
National Pesticide Survey: Phase I Report
-------
C-8 Appendix C: Implementation of the Survey
but only after the NPS sampling frame had been extracted.) California provided EPA with an updated frame
for the State and a revised CWS sample was selected in California.
The resulting sample of systems with corrections provided by the States and Regions, was forwarded
to the next eligibility determination phase, the telephone administration of the "CWS Screener" questionnaire.
The CWS Screener contained twenty questions and took approximately five to ten minutes to administer. The
CWS Screener was administered by telephone interviewers to a representative of each CWS. The screener
helped to:
Verify that each entry was a CWS with at least one working drinking water well;
Determine the number of working drinking water wells at each CWS;
Identify the entries in the sample that represented multiple CWS; and
Identify a contact person at each CWS to aid in the collection of additional questionnaire
data and water samples.
Based on the responses to the screening questionnaire, it was possible to determine whether the CWS
was eligible for inclusion in the follow-up Survey. Based on probability sampling procedures (described in
Appendix B), 599 CWS wells were selected from eligible systems for administration of the CWS
Questionnaires and well water sampling.4
The sample of Community Water Systems obtained from the FRDS presented considerable location
problems. It was found that 13.4 percent of all cases would require extensive tracing. (Tracing refers to the
steps and procedures followed to locate the well owner/operator and the information being collected.)
Specifically, the telephone interviewers called one or more of the following places to locate a CWS:
City Hall and Town Clerk
Local or County Health Department
Local or County Environmental Resources
Post Office
Rural Mail Carriers
Libraries
Local Water and Utilities Department
Public Works
Tax Assessor
Planning Commission
Chamber of Commerce
Housing Development Authority
Real Estate Agents
Police Departments
Fire Stations
Small Community Liquor Stores
Community Centers
Gas Stations
4 The CWS sample comprised all States except Kentucky. At the request of the State of Kentucky, EPA sampled
Kentucky CWS wells but excluded these well sample results from the Survey findings so as to not compromise the sample
design.
National Pesticide Survey: Phase I Report
-------
Appendix C: Implementation of the Survey C-9
The tracing effort resulted in successfully locating 85.7 percent of all initial tracing cases. The final
screening success rate was 96.9 percent, with only 2.6 percent of the total sample not located.
Once the CWS Screeners were completed, the CWS site identification number and the total number
of wells per site were entered into a computer data base file. After all completed cases were entered in the
file, random sampling procedures were initiated that resulted in 599 CWS being selected for the sample of
wells to be tested and surveyed. The contact information (well owner's name, address, phone number) was
then entered for the 599 CWS in the file.
Each eligible system selected was allocated a two-week sampling window. This effort resulted in the
master CWS sampling schedule.
Communicating With CWS Contacts and CWS Owner/Operator
EPA distributed the master CWS sampling schedule to EPA Regional and State contacts in June 1988,
requesting that the contacts indicate a preferred sampling week within the two-week window identified. State
contacts then sent a letter to each CWS owner/operator indicating that their system was selected for sampling
as pan of the National Pesticide Survey. The letter stated they would be contacted by a CWS Sampling Team
member to confirm an exact sampling appointment.
Scheduling Field Sampling Dates
The NFS State or Regional sampler responsible for each CWS was contacted to finalize the actual
sampling date and ensure that the necessary sampling equipment was sent on time. The sampler was
responsible for telephoning the CWS owner/operator to arrange a convenient date and relaying this
information to the scheduler. In addition to scheduling a sampling and interviewing visit, the CWS sampler
also had to arrange an interview with the local area expert (usually the county agricultural extension agent)
to discuss local fanning practices, and pesticide use.
IB some instances during the scheduling procedure, the CWS sampler discovered that the system no
longer met NPS criteria. For example, the system might have closed down since the initial screening call was
placed, or the CWS design did not permit sampling prior to treatment. This information was always confirmed
and then entered into the NPS Information System (NPSIS). NPSIS is described in detail in Section C4.7.
Finalizing Field Sampling Dates
The CWS Field Sampling Logbook was sent to the sampler approximately eight weeks prior to the
preferred sampling week. (The Logbook contained the name, address, and phone number of the CWS
owner/operator and a script for telephoning the individual to schedule a sampling appointment.) After the
samplers contacted the owner/operator, and confirmed an exact sampling date, the sampling date was entered
into NPSIS. When CWS confirmation dates fell outside of the two week sampling period, statisticians
determined whether the deviation was permissible based on the temporal sampling guidelines (see
Appendix B). Any unallowable deviations were resolved by contacting the samplers to reschedule a more
acceptable date. All deviations were recorded in the Tracking System to be used during the analysis phase.
CWS with unconfirmed dates were coded in the NPSIS communications data base to track outstanding
sampling dates, and a report was generated for daily review. A copy of the NPSIS summary report of
scheduled sampling dates for one State is provided in Exhibit C-2. Four weeks prior to the preferred sampling
date, the contractor conducted a "gentle reminder call" to samplers included in the unconfirmed CWS schedule
report Two weeks prior to the sampling period, sampling kits and supplies were shipped to the sampler. All
well schedules were usually confirmed by this time.
National Pesticide Survey: Phase I Report
-------
C-10 Appendix C: Implementation of the Survey
Exhibit C-2
Scheduled Sampling Dates
Sampling Time Frame
Number of
Community Wells In
Well Systems Stratum System
cws
cws
cws
cws
cws
cws
cws
cws
cws
cws
cws
cws
cws
cws
cws
* Note:
1
2
3*
4
5
6
7
8
9
10
11
12
3*
13
14
Two
4
11
12
8
7
4
11
12
11
7
4
10
12
11
8
wells from
2
3
40
2
2
3
1
2
12
20
1
2
40
2
1
CWS 3 sampled
Week for Sampling
Confirmed
With CWS
10/17/88 -
12/12/88 -
01/09/89 -
01/09/89 -
02/20/89-
03/06/89 -
03/20/89-
04/17/89 -
05/01/89 -
05/01/89 -
05/15/89 -
06/12/89 -
08/07/89 -
09/04/89 -
12/04/89 -
in different
10/21/88
12/16/88
01/13/89
01/13/89
02/24/89
03/10/89
03/24/89
04/21/89
05/05/89
05/05/89
05/19/89
06/16/89
08/11/89
09/08/89
12/08/89
sampling
Potential
Sampling Week 1
10/17/88
12/12/88
01/09/89
01/09/89
02/20/89
06/06/89
03/20/89
04/17/89
05/01/89
05/01/89
05/15/89
06/12/89
08/07/89
09/04/89
12/04/89
periods.
- 10/21/88
- 12/16/88
- 01/13/89
- 01/13/89
-02/24/89
- 03/10/89
- 03/24/89
- 04/21/89
- 05/05/89
- 05/05/89
- 05/19/89
- 06/16/89
-08/11/89
- 09/08/89
- 12/08/89
Potential
Sampling Week 2
10/24/88
01/02/89
01/16/89
01/16/89
02/27/89
03/13/89
03/27/89
04/24/89
05/08/89
05/08/89
05/22/89
06/19/89
08/14/89
09/11/89
12/11/89
- 10/28/88
- 01/06/89
- 01/20/89
- 01/20/89
- 03/03/89
- 03/17/89
- 03/31/89
-04/28/89
- 05/12/89
- 05/12/89
- 05/26/89
- 06/23/89
- 08/18/89
- 09/15/89
- 12/15/89
C3.2 Rural Domestic Wells
From April 1988 through February 1990, EPA sampled 783 rural domestic wells in 90 counties in 38
States. Prior to actually collecting drinking water samples from these wells, EPA had to identify a
representative sample of wells in each county that met specific NPS eligibility criteria.5 Once these eligible
wells were identified, EPA had to solicit voluntary participation in the Survey from rural domestic well owners
and schedule sampling and data collection during designated two-week sampling periods. This scheduling
process was cooperatively administered by the NPS data collection and processing staff and the NPS
communications staff.
Identification of Geographic Areas to be Sampled
The rural domestic well portion of NPS was designed to obtain information on rural domestic well
water use and to collect water samples from approximately 750 rural domestic wells located in 90 counties
across the country. Of the 90 counties, six counties had wells selected as part of the pilot study of NPS
procedures. During the full field survey, 84 counties remained in which to select wells to be sampled. EPA
was particularly interested in sampling water from wells located in areas believed to be more likely to contain
5 See Appendix B for further explanation of eligibility requirements, numbers of wells in specific counties, and counties
selected for sampling.
National Pesticide Survey: Phase I Report
-------
Appendix C: Implementation of the Survey C-11
pesticide contamination. Thus, the selection of the initial 90 counties oversampled both counties with high
pesticide use and those with highly vulnerable groundwater.
Starting in March 1988, second stage interviews with county agricultural extension agents were
conducted in the 84 counties. The purpose of the interviews was to collect information on cropping patterns
and pesticide use for second-stage stratification of rural domestic well sampling. The county agents were asked
questions about crop types, pesticide usage, soil textures, crop management practices, irrigation methods, and
conservation measures in the counties they served.
Information from the questionnaires was used to make determinations about cropping patterns for
each 7.5 minute USGS quadrangle map comprising the county. The maps were then assigned a score to
represent the amount of agricultural activity in the quadrangle: greater than 50 percent cropped (high), 26 to
50 percent cropped (moderate), 25 percent or less cropped (low). These cropped areas were overlaid with
ground-water vulnerability information (DRASTIC) to identify "cropped and vulnerable" areas for
oversampling.
Additional characteristics of the counties were also identified. Some characteristics made an area
ineligible (e.g., military bases were excluded from the eligible areas). Other characteristics affected the
probability of sampling wells in that area (e.g., all selected wells in vulnerable areas were scheduled for
sampling whereas only one well out of three selected were sampled in non-vulnerable areas). Typically, there
was an overlap between the land areas with these characteristics. Once the areas were identified on the maps,
the ZIP codes for the areas were determined. After identifying different characteristics of areas, the mapping
staff defined the boundaries of areas using standard and well-known place designators. These designators (e.g.,
street names, river names, county boundaries) would be used at a later point in the rural domestic well survey
to help respondents determine if their wells were in an eligible, oversampled area.
The steps followed in the mapping process were:
Locate and obtain a detailed county map which included highways and other landmarks;
Compare a computerized map of ZIP codes in the county with the 1987 Rand McNally ZIP-
Code Map and resolve discrepancies by calling local post offices within the county;
Indicate target area on each map;
Determine if the county contained any urbanized areas or was considered rural;
Draw the urbanized area(s) on the target area map using 1980 Census information; and
Develop a list of cities for each county which are to be considered ineligible land areas and
excluded from the study.
Identification of Telephone Exchange Clusters for
Random Digit Dialing (RDD) of Homes
The rural domestic well survey was designed to reach respondents (well owners/residents) and identify
private wells by randomly dialing homes in the sampled geographic areas. In order to efficiently identify phone
numbers to call, the contractor utilized a modified form of the Waksberg sampling method for random digit
dialing. First, a recent list of area codes and telephone exchanges was obtained for the 84 counties selected
in the initial sampling. For each area code/exchange combination (six-digits) in each county, all possible values
of the next two digits were joined to the six-digit area code and exchange, forming a series of eight-digit
National Pesticide Survey: Phase I Report
-------
C-12 Appendix C: Implementation of the Survey
numbers. The sampling frame for a county was the set of all possible eight-digit telephone clusters identified
for a county.
To select the sample, a random selection was made of eight-digit clusters. A random number in the
range of 0 to 99 was also generated and appended to the selected cluster to form a ten-digit telephone number.
The number was dialed. If it turned out to be a residential number in the sampled county, the entire cluster
of numbers was retained for use in the random digit dialing effort. If the number was not residential, the
cluster of telephone numbers was rejected. Estimates were made of the number of wells that would be
sampled as a result of calling within a cluster and a determination was made concerning the number of clusters
to identify.
Development of the Rural Domestic Well Screening Questionnaire to
Determine if a Respondent's Well was Within the Eligible Geographic Area
County residents were called as a result of the random digit dialing effort described above. After
verifying that the phone was in a residence and not a business, the respondents were asked a series of
questions to establish the location of the residence vis-i-vis the eligibility areas of the county. Using
geographic information on the counties developed earlier, the rural domestic well survey research staff drafted
a unique questionnaire for each of the 84 counties. The questions identified the precise location of the well.
The interviewers posed questions to:
Identify ZIP codes which included target areas;
Exclude cities (e.g., Do you live within the city limits of Richmond?);
Determine the boundaries of the urbanized area when city limit questions did not cover the
entire urbanized area; and
Identify vulnerable (target) areas.
When writing boundary questions for urbanized areas, if the boundaries did not coincide with easily
identified landmarks, the boundaries of the urbanized area were enlarged. It was most probable that urbanized
areas had grown since the last Census was conducted and the given urban boundaries defined. Similarly,
mappers erred to the side of overestimating the boundaries of vulnerable areas (referred to as target area) in
order to coincide with well-known and definable boundaries.
The purpose of the activity was to describe the boundaries in such a way that the respondent was able
to recognize the area by the description given over the phone. For example, the questions could not include
directional references like "north," "south," "east" and Vest." (Many people do not correctly interpret such
references.) Instead, questions were phrased to describe the areas between two known landmarks, e.g., "Do
you live between the Red River and Highway 71?" Estimates of distances were also used, e.g., "Are you within
five miles of Gopher Gulch Road?"
Administration of the Rural Domestic Well Screening Questionnaire and
Scheduling of Field Team Appointments
The location questions for each of the 84 counties were drafted, typed, and reviewed. The mapping
specialists reviewed the questions for each county and compared the named landmarks to the county maps to
verify that the correct areas were being singled out. The exercise was repeated with the mapping supervisor.
The supervisor was often called upon to make the final decisions about enlarging urban and target area
boundaries and other issues that may have affected the sampling probabilities.
National Pesticide Survey: Phase I Report
-------
Appendix C: Implementation of the Survey C-13
Rural domestic well screening interviewers used Computer Assisted Telephone Interviewing (CATI)
to administer the Rural Domestic Well Screening Questionnaire. A series of questions on location of the well
were integrated into the standard questionnaire administered to the respondent reached through ROD. The
RDD calling and administration of Rural Domestic Well Screening questionnaires took place over a period
of 17 months. Calls were made to approximately five to ten counties each month with calls to each individual
county lasting approximately three weeks.
The Rural Domestic Well Screening questionnaire had the following flow. The interview:
Determined if the home or well site was eligible for the Survey:
~ Private residence;
Located in the sampled county, and
A permanent residence for household members.
Determined if the residence was in an urban area or other area determined to be ineligible
for sampling;
Determined if the residence used water from a private well for domestic purposes (i.e.,
drinking, bathing or washing);
If respondent and/or well were deemed to be ineligible based on any of the above criteria, the
interview was immediately concluded;
Determined if the well was in a target area (oversampled);
If the well was not located in a target area, the CATI program applied a sampling algorithm
which included the well with a probability of one in three. The interview was immediately
concluded if the respondent's well was not selected. All wells in the target area were selected;
For sampled cases, the questionnaire continued with questions concerning characteristics of
the well; and
If the well provided water for domestic uses to more than one household, the CATI program
applied a second sampling algorithm, and included or excluded the well with a probability
based on the total number of households serviced by the well. Similarly, if a household had
more than one residential telephone number, it was retained with a probability of one in two.
The interview was immediately concluded if the respondent's well was not selected.
Once it was determined that a respondent's well was eligible to be included in the NPS, the screening
interviewer asked the respondent to participate in the field Survey. The screening interviewer explained to
the eligible respondent that EPA would send a letter and a brochure explaining the purpose of the Survey and
the requirements for participation. The screening interviewer then scheduled a sampling appointment within
the two-week sampling period for that county, unless the respondent would not schedule a particular time.
The screening interviewer also determined whether the respondent was the owner of the well or a resident of
the well site who did not own the well. If the respondent was a resident of the site, but did not own the well,
the interviewer asked the resident to identify the well owner. In addition, the screening interviewer attempted
to identify fanners of the well site property, if applicable. The screening interviewer attempted to get the
names, addresses and telephone numbers for farmers of well properties so that interviews with farmers could
be arranged.
National Pesticide Survey: Phase I Report
-------
C-14 Appendix C: Implementation of the Survey
If the well owner was different from the respondent who agreed to participate, the screening
interviewer called the well owner and obtained his or her permission for the field sampling and scheduled a
tentative appointment for an interview. If the property resident was different from the respondent who agreed
to participate, the screening interviewer called the property resident and got their cooperation for the field
Survey and attempted to schedule a tentative appointment.
Confirming Sampling Appointments and Soliciting Participation of Farmers
Once the representative sample was selected for a given county and all screening questionnaires were
checked and reviewed, the telephone survey Operations Manager forwarded respondent information sheets
logging rural domestic well owners', residents', and farmers' names, addresses, telephone numbers, and
sampling appointments to the NFS Communications Manager. (See Exhibit C-3 for a sample respondent
information sheet.) The NFS communications staff entered this information into the NPSIS communications
data base and forwarded a letter and brochure to each eligible participant (including well owners and farmers)
explaining the purpose of the NFS. The well owner was asked to be present during sampling to confirm his
or her permission to sample the well, and participate in an interview; a resident who did not own the well was
asked to be present during sampling and to participate in an interview; or the fanner was asked to participate
in an interview. (See Exhibit C4 for a sample NFS request for participation letter.) After mailing a letter
and brochure to each eligible rural domestic well participant and allowing a reasonable amount of time for
mail delivery, NFS communications staff called each eligible well owner in a given county to discuss the NFS.
During this call, the interviewer verified each participant's address and phone number and obtained directions
to the well site. If the resident had not scheduled an exact time for the appointment during the screener, the
appointment was set up during this phone call.
Sampling dates and interview times were recorded in the NPSIS communications data base and a
sampling schedule was produced for the sampling team assigned to the county. If the well owner retracted
permission to sample, the NFS Communications Manager recontacted the eligible owner within 72 hours and
attempted to convert the refusal to a willingness to participate.
One week prior to field sampling, NFS communications staff recontacted rural domestic well
participants to confirm the date and time of sampling and interviews. Once in the field, the NFS Sampling
Team Leader contacted the participants before the scheduled sampling date to confirm all scheduled
appointments. Occasionally, non-resident well owners or well property fanners were "discovered" during the
interviews conducted at the sampling appointments. The NFS interviewer contacted these additional owners
and farmers and scheduled appointments for interviews. The interviewer also confirmed these appointments
one day prior to the interview. These follow-up measures helped to maximize participation in the Survey.
Scheduling Local Area Expert Interviews
In addition to scheduling well sampling and participant interviews, the NFS interviewer also scheduled
interviews with County Agricultural Extension Agents or the designated local area expert. The Westat Field
Supervisor contacted the county agent by telephone, asked him or her to participate in an interview to provide
NFS staff with relevant cropping and pesticide use information, and scheduled a convenient interview time
during the two-week sampling period for that county. Interviewers administered the Local Area Expert
Questionnaire. This questionnaire collected data on land use in a one-half mile area for each sampled well
within a county.
National Pesticide Survey: Phase I Report
-------
Appendix C: Implementation of the Survey C-15
Exhibit C-3
Sample Rural Domestic Well Selection Respondent Information Sheet
Respondent Information Sheet
Well ID # XX-XX-XXXX
Date: 06/05/89
Mr.
Address:
Phone #:
Box1
Sampled Household
John Doe
Star Route A Box 100
123-456-7777
Box 5
Person who lives on property where well is located.
Mr. John Doe
Address: Star Route A Box 100
Phone #: 123-456-7777
Box 2
Appointment with Sampled Household:
Date: 07/10/89
Time: 09 : 00 a.m.
Mr.
Address:
Phone #:
Box 6
Person who owns the well.
Richard Roe
City, State
123-999-9999
Box 3
Best time to call:
Usually in the evenings except Wednesday
Box 7
Person who farms property where well is located.
Mr. Harry Hoe
Address: Star Route C Box 2000
Phone #: 123-888-8888
Box 4
Sampled Household
Lives on property where the well is located. Yes
Owns the well. No
Farms the property where the well is located. No
Box 8
Total number of respondents = 3
Comments:
Respondent could not schedule appointment between June 27 and July 6.
National Pesticide Survey: Phase I Report
-------
C-16 Appendix C: Implementation of the Survey
Exhibit C-4
Sample Letter Requesting Participation in Survey
Dear [Resident]:
Recently a member of your household was contacted by telephone about participating in
the national survey of pesticides in drinking water wells, which is currently being conducted by
the U.S. Environmental Protection Agency.
The purpose of the National Pesticide Survey is to provide EPA with information on the
levels of pesticides found in drinking water wells nationwide, and it is the first study of its kind to
be conducted on a national scale. It will provide information critical to effective protection of our
drinking water resources.
The selection of your well was made through a random statistical procedure; we do not
know of or suspect any problems with your water.
EPA has hired ICF Incorporated to conduct the Survey. ICF staff will be contacting you
by, telephone to arrange a convenient time to visjt_vour. home during [the last week of
September]. During the visit, you will be asked to complete a 20-minute interview regarding the
use and construction of your well and to allow water samples to be taken from an outdoor tap.
The water samples will be analyzed in our laboratories, free of charge, and the results will be sent
to you.
Your participation is voluntary, and you may refuse to answer any question asked.
However, your participation is extremely important to the success of this research and cannot be
replaced by someone else. We greatly appreciate your help and cooperation.
Yours sincerely,
Acting Director
National Pesticide Survey
National Pesticide Survey: Phase I Report
-------
Appendix C: Implementation of the Survey C-17
C4 Field Operations and Data Collection
C4.1 Communications
Because of the large number of persons involved in the NPS and the widespread geographical scope
of NPS activities, an effective communications system was necessary to coordinate activities. NFS
communications activities served to:
Ensure smooth, timely, and regular communications among project participants;
Maintain confidentiality regarding rural domestic well owner participation and well sampling
results;
Provide accurate and useful information to well owners/operators and to the numerous parties
interested in the Survey;
Maintain high participation rates among CWS owners/operators and rural domestic well
owners/residents; and
Ensure timely notification of sampling results to EPA and State personnel, CWS
owners/operators, and rural domestic well owners/residents.
Communications played a key role during the water sampling phase of the Survey. NPS
communications staff supported water sampling and data collection activities by training CWS sampling teams
on NPS communications protocols, conducting scheduling of CWS field sampling, providing information to
well owners and sampling team members, and notifying State officials and well owners of well sampling results.
NPS Communications Resources
In addition to the NPSIS communications control system used to manage all scheduling activities (see
Section C3.7), other major NPS communications resources included the following:
NPS Hotline. EPA established a toll-free NPS telephone information line in March 1988.
The hotline was used extensively by State and county officials, CWS and rural domestic well
sampling teams, CWS owners/operators, rural domestic well ownersAesidents, and the general
public. The NPS communications staff operated the hotline and responded to inquiries on
Survey activities, including scheduling, field sampling, and the status of well sampling results.
NPS Mailing Ust. EPA maintained an extensive NPS mailing list and periodically mailed
the "NPS Project Update" to all interested parties to keep them informed of the Survey's
progress. In addition to Survey participants and members of Congress, industry
representatives, public interest groups, and trade associations were interested in all aspects
of the Survey, from the statistical design of NPS to data analysis and policy development.
NPS Project Updates. EPA published periodic newsletters to inform project participants
and interested parties of the progress of the National Pesticide Survey implementation efforts.
Health Advisories. EPA developed Health Advisories for 59 priority pesticides and nitrate
using information collected on physicochemical properties, uses, chemical fate, health effects,
treatment, and existing criteria and guidelines. These Health Advisories are technical,
scientific guidance documents developed to assist in evaluating the results of well sampling
and to determine whether the contamination levels found warrant further action. As part of
National Pesticide Survey: Phase I Report
-------
C-18 Appendix C: Implementation of the Survey
EPA's effort to improve risk communication, the Agency also prepared non-technical Health
Advisory Summaries and distributed these summaries, explaining the health effects of
exposure to pesticides and nitrate in simple language, to the owners, residents, and operators
of sampled wells where contamination was found.
Press Advisories. The EPA Headquarters Press Office issued press advisories and news
releases on NPS activities, and EPA Regional contacts responded to local media inquiries to
keep the public informed.
Rural Domestic Well Communications. Rural Domestic Well Field Team Leaders were
provided a communications packet for each county sampling assignment. This packet
contained the following materials:
Sampling schedule;
EPA's NPS confidentiality policy;
State's confidentiality policy, if applicable;
List of EPA and State NPS contacts;
Rural domestic well question-and-answer brochure;
Non-technical answers to commonly asked questions;
Copy of letter to private well owners/residents inviting them to participate in the
Survey;
Door notice to be used when well owner/resident did not keep a scheduled
appointment;
Additional copies of materials that could be distributed to local officials, the media,
and NPS participants; and
NPS interviewers were provided with definitions for terms used in the questionnaire,
to be read to the respondent when necessary interviewers were also given answers to
questions likely to be asked by the respondent during the interview.
C4.2 Sampling Operations Management
NPS sampling operations were coordinated from a sample preparation room, set up in April 1988.
In this preparation room, sample containers were prepared for use; portable meters for measuring
temperature, pH, and electrical conductivity at the sampling site were maintained and calibrated; sampling kits
were assembled and shipped to the assigned sampling team; and sampling kits and samples were tracked.
Sample Preparation Room Operations
Sampling kit preparation took place at the preparation room at the headquarters of the NPS
implementation support contractor in Fairfax, Virginia. The preparation room was a secured area containing
facilities for storage of supplies, kit preparation, terminals for the NPS Information System (NPSIS1), and
telephones for the NPS Hotline. All sampling operations management thus occurred under secured
conditions. The preparation room manager was assigned responsibility for all preparation room activities.
National Pesticide Survey: Phase I Report
-------
Appendix C: Implementation of the Survey C-19
Sample Container Supply, Cleaning, and Preservative Preparation Requirements. An
independent vendor provided boxes of cleaned water sample bottles containing the appropriate amounts of
sample preservative to the preparation room, where they were stored. Bottle cleaning and the addition of
sample preservatives were carried out by the vendor according to a Standard Operating Procedure.6 After
the sample containers were cleaned, the appropriate type and volume of preservative was added to each sample
container, depending on the assigned analytic method. Exhibit C-5 summarizes the bottle and cap
requirements and preservative requirements for each of the NFS analytic methods. After the bottles were
cleaned, filled with the proper amounts of preservative, and capped, they were not reopened until needed at
the sampling site.
Preparation of Sampling Kits. Preparation room personnel received shipments of sample bottles,
as well as other materials used to prepare sampling kits such as ice chests (coolers), styrofoam containers,
styrofoam inserts, plastic liners, and cardboard boxes, and stored all materials in the preparation room and
at a readily accessible off-site warehouse. Preparation room personnel then prepared sampling kits, as
indicated by the sampling schedule, by carrying out the following steps:
Printing the appropriate labels and forms;
Placing the correct bottles and inserts into styrofoam containers, according to the sampling
scenario for the site and the laboratory, based on the sampling schedule, and placing the
completed kits into cardboard boxes;
Preparing a sampling supply kit, containing tools and equipment, and placing the kits into
a cooler for shipment;
Shipping the sampling kits and coolers to the proper sampling team; and
Updating the computer record to show the airbill number, kit type, and date shipped for each
kit prepared.
Each of these steps is described in detail in this section.
The overall goals for number of wells to be sampled, shipping blanks, spiked samples, and time storage
samples were all entered into NPSISl and were used to generate the schedule of sites to be sampled for two-
week sampling periods.
6 The cleaning procedure for glass sample containers generally involved a hot water/detergent wash, deionized water rinse,
and drying at high temperature (400-450°C). Caps for the glass sample containers were washed with detergent and rinsed
with deionized water, followed by a solvent rinse. Plastic sample containers and caps were washed with hot water and
detergent, followed by a tap and deionized water rinse.
National Pesticide Survey: Phase I Report
-------
C-20 Appendix C: Implementation of the Survey
Exhibit C-5
Sample Container Preservation Requirements
NPS METHOD BOTTLE. CAP. AND PRESERVATIVE*
NPS-1 Clear l-liter borosilicate glass bottles with teflon-lined caps
with 10 ml of mercuric chloride.
NPS-2 Clear l-liter borosilicate glass bottles with teflon-lined caps
with 10 ml of mercuric chloride.
NPS-3 Clear l-liter borosilicate glass bottles with teflon-lined caps
with 10 ml of mercuric chloride.
NPS-4 Clear l-liter borosilicate glass bottles with teflon-lined caps
with 10 ml of mercuric chloride.
NPS-5 250 ml amber screw-cap glass bottles with teflon
faced septa with 7.5 ml of pH 3 buffer.
NPS-6 60 ml screw-cap glass bottles with teflon-faced septa with
0.6 ml of mercuric chloride.
NPS-7 60 ml screw-cap glass bottles with teflon-faced septa with
0.6 ml of mercuric chloride.
NPS-9 125 ml polyethylene bottles with 0.25 ml of sulfuric acid.
Volume of preservative should be <. 1% of the final sample volume for all methods.
Concentration of the preservative should be 10 mg/L for all methods.
National Pesticide Survey: Phase I Report
-------
Appendix C: Implementation of the Survey C-21
The current schedule of sites to be sampled indicated the "sampling scenario" to be employed for each
laboratory analyzing samples from that site. The sampling scenario, in turn, determined the number of each
bottle type to be included in the styrofoam container (the "kit").7 A sampling scenario for one two-week
period for rural domestic wells is shown in Exhibit C-6.
The preparation room staff entered into NPSIS1 the information specific to each laboratory analyzing
samples from the site, including the well ID number, the sampling scenario, and the lab three-letter code. In
addition, if a spiked sample was assigned to the lab for a particular well, that information was also entered.
NPSIS1 then printed the correct number of labels for each type of bottle to be included in each kit for each
laboratory. As many as three kits might be prepared for a single laboratory. NPSIS1 also printed the Sample
Tracking Form, which listed all sample bottles contained in each kit The Sample Tracking Form served as
the packing list and the means of tracking the sample bottles from the preparation room to the field and from
the field to the lab. Exhibit C-7 presents a copy of the Sample Tracking Form. NPSIS1 also printed the Well
Sampling Information Sheet, with all pertinent information about the sampling site: the sampling location
and address; sampling date, well owner/resident name and phone number, name and address of the person
responsible for sampling, and kit numbers. Exhibit C-8 provides a copy of an example Well Sampling
Information Sheet.
Sample kits were then prepared using styrofoam containers, styrofoam inserts, and boxes of the correct
size and type for the particular kit. Preparation room staff verified that all sample bottles contained
preservative prior to affixing sample bottle labels and verified that the correct labels were attached to the
correct bottle types. Samples designated as "shipping blanks" were filled with deionized water, and an orange
sticker was placed on the top of the cap of the bottle to alert the field team that this sample bottle should not
be opened.
After labeling and verifying the correct labeling of bottles, the preparation room staff placed the
bottles within the styrofoam container, placed the container in a plastic bag, and placed the container and bag
into a cardboard box. Preprinted airbills for shipping the kit from the field to the lab, with the kit number
written on the airbill, and the lab and field copies of the Sample Tracking Form were then placed in a plastic
bag taped to the top of the styrofoam container. After verifying the address on the airbill corresponded to
the address of the lab to which the kit was to be sent from the field, the preparation room staff sealed the
cardboard shipping box.
The preparation room staff then placed all necessary field equipment and tools (wrenches, pens, tape)
in an equipment bag and placed the bag inside a cooler. Questionnaires were mailed to CWS field teams with
the Field Logbook one month in advance of well sampling. Airbills were then placed on the cooler and on
all cardboard shipping boxes containing sampling kits. NPSIS1 was updated to indicate the well identification
numbers for the kits being shipped and the bottle identification numbers for the bottles in the kits. This
shipping data updated the Well Sampling Data base.
7 One regular sample and one backup sample were taken for eacb analytical method at each site. In addition, one
shipping blank was always prepared for analytic method 7 for each site. Samples for lab spikes were collected at 30 percent
of the sites for Method 1; 20 percent of the sites for Methods 2, 3, and 4; and 10 percent of the sites for Methods 5, 6, 7,
and 9 (Method 8 was excluded from the Survey). Samples and duplicate samples for time storage studies were collected at
15 percent of the sites for Method 1; 10 percent of the sites for Methods 2, 3, and 4; and 5 percent of the sites for Methods
5, 6, 7, and 9. Finally, one additional sample for each analytical method was taken at 1,50 sites for the EPA referee
laboratories. The sampling scenario specified for a particular site which of these alternatives the sampling kit should be
prepared. When confirmation or follow-up contacts with the domestic well owner or operator suggested that sampling would
not be allowed at a well, the NPS preparation room staff were notified by NFS communications staff immediately. If the
sampling scenario indicated that time storage samples, referee laboratory samples, or lab spikes should be obtained at that
well, the sampling scenario for another well was adjusted to include such samples, to ensure that the necessary number of
time storage samples, referee laboratory samples, and lab spikes would be obtained.
National Pesticide Survey: Phase I Report
-------
C-22 Appendix C: Implementation of the Survey
Exhibit C-6
NPS Rural Domestic Well Sampling Scenario and Lab Spike Assignments
for Two-Week Period (for codes see next page)
No.
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
Well
ID
Start
Date
08/21/89
08/21/89
08/21/89
08/21/89
County
County A
County B
County C
County D
Lab
*
ESE**
ATI
JMM
BCL
JMM
JMM
ATI
RAD
ATI
JMM
JMM
JMM
RAD
ESE
BCL
ATI
Scenario
No.
1
2 '
1
2
1
1
4
1
8
2
3
2
1
2
1
1
2
1
1
1
1
1
1
2
4
3
7
2
2
1
1
1
2
2
1
1
Type
FS
LS
FS
LS
FS
FS
LS
FS
LS
LS
LS
LS
FS
LS
FS
FS
LS
FS
FS
FS
FS
FS
FS
LS
LS
LS
TS
LS
LS
FS
FS
FS
LS
LS
FS
FS
Analytical Method
1
B1
BO
AO
CO
B1
A2
3
BO
AO
B1
9
A2
AO
2
B1
AO
A1
BO
4
AO
A1
5
AO
A1
7
AO
A1
6
AO
A1
Referee Lab
BSL
1
1
1
1
TSD
1
1
1
1
When no laboratory is indicated, samples were analyzed by all contract laboratories.
When a particular laboratory is indicated, additional samples were analyzed by that laboratory.
National Pesticide Survey: Phase I Report
-------
Appendix C: Implementation of the Survey C-23
Exhibit C-6 (continued)
Description of Sample Code Numbers for Sampling
Scenario and Lab Spike Assignments
PD-0001-1-1-01
Well Type
ID
Number Lab Name
Method
Number
Sample Type
PC = Community Well 0001
PD = Domestic Well
PR = Resampled Well .
PB = PE Sample
1500
1 = JMM
2 = ATI
3 = RAD
4 = ESE
5 = BCL
6 = BSL
7 = TSD
1 01 = Field Sample
2 02 = Shipping Blank
3 03 = Backup sample
4 04 = Lab spike (mix A, level 1)=AO
5 05 = Lab spike (mix A, level 2)=A1
6 06 = Lab spike or time storage - Day 0 (mix A, level 3)=A2
7 07 = Lab spike (mix B, level 1)=BO
9 08 = Lab spike (mix B, level 2)=B1
09 = Lab spike or time storage - Day 0 (mix B, level 3)=B2
10 = Lab spike (mix C, level 1)=CO
11 = Lab spike (mix C, level 2)=C1
12 = Lab spike or time storage - Day 0 (mix C, level 3)=C2
13 = Time storage (Day 0 duplicate)
14 = Time storage - Day fourteen
15 = Time storage - Day fourteen duplicate
Lab performing the analyses for the NPS:
1 = JMM (Montgomery Laboratories)
2 = ATI (Alliance Technologies Inc./Clean Harbors Analytical Services)
3 = RAD (Radian, Inc.)
4 = ESE (ES&E)
5 = BCL (Battelle, Columbus Division)
6 = BSL (Bay St. Louis (EPA/Environmental Chemistry Lab))
7 = TSD (EPA/Technical Support Division Lab)
FS = Field Sample
LS = Lab Spike
TS = Time Storage
National Pesticide Survey: Phase I Report
-------
C-24 Appendix C: Implementation of the Survey
Exhibit C-7
Sample Tracking Form
SAMPLE TRACKING FORM
EPA NATIONAL PESTICIDE SURVEY
Well I.D. No.: 2201
FRDS I.D. No. (CWS Well Only): ID1280098
Sample Collection Date: / /
Tracking Form Completed by:
Lab: JMM
Scenario: 4
Kit No.: PC-2201-141
Box 1 of 2
TO BE COMPLETED BY:
CF
Sample Number BotHe Sample
Size Description
PC-2201-1-1-01 1000 Field Sample
PC-2201-1-1-03 1000 Backup Sample
PC- 2201-1 -3-01 1000 Field Sample
FIELD THAW
Sampler
(Initial)
Time
Sampled
Comments (1)
LAB
Received
Y N
_Y N_
Y N
Comment* (2)
Chlorine Test:
(No Color Change, Light Pink, Dark Pink)
SHIPPED BY:
Date:
Time'
Sent to
Jane Doe
Example State Agency Building
111 State Agency Road
Capital City, EX 99990
Lab Address:
James M. Montgomery Laboratories
555 East Walnut SL
Pasadena, CA 81101
RECEIVED AT LAB BY:
Date:
Time:
Condition (3):
(1) For Example Bottle broken, bottle missing, overfilled bottle, cap was dropped
(2) For Example- Bottle broken, bottle missing, bottle contaminated, temperature criteria not met
(3) For Example: Ice melted, box leaking
(Lab comments should concur with NPSB1 Sample Receipt)
National Pesticide Survey: Phase I Report
-------
Appendix C: Implementation of the Survey C-25
Exhibit C-8
Example Well Sampling Information Sheet
WELL SAMPLING INFORMATION SHEET
EPA NATIONAL PESTICIDE SURVEY
NFS Well ID No.: 1000 Sampling Date: 01/01/90
Well Type: C
(C=CWS, D=Domestic, R=Resample)
Well Contact & Address: Well Sampling Assigned to:
John Doe Jane Doe
Example Well System Corp. Example State Agency Bldg.
100 Example Ave. 111 State Agency Road
Example County Capital City, EX 99990
Exampletown, EX 99999
Tel. (999) 999-9999 Extn: 999
Closest Federal Express Office to Sampling Site:
222 FEDERAL EXPRESS ROAD. METROPOLIS. EX 99980
Federal Express Toil-Free Number: (800) 238-5355
List of Sampling Kits For This Well:
KIT NO. LAB
1. PC-1000-111 JMM
2. PC-1000-421 ESE
TOTAL NUMBER OF SAMPLE CONTAINER KITS SENT TO SAMPLING TEAM: 2
ICF's Toll-Free Hotline Number: (800) 451-7896
County Agent (for administration of Local Area Expert Questionnaire):
Name: CWS Training Materials
National Pesticide Survey: Phase I Report
-------
C-26 Appendix C: Implementation of the Survey
As the final step of the initial kit preparation and shipment, the cooler and sampling kits were cleared
for shipment to the scheduled sampler. All materials were sent about two weeks in advance of the sampling
date for CWS and one week in advance for rural domestic well sampling. The Sampling Team Leader was
instructed to call the NFS Hotline when the kit was received. Upon receipt of the call, the Sample Kit
Tracking System was updated to confirm that the kit was in the possession of the Sampling Team. If the
preparation room staff did not receive confirmation from the Sampling Team within three business days of
the expected date of receipt for CWS and one business day for rural domestic well sampling, the preparation
room staff called the Sampling Team to verify receipt of the kit. If the kit had not been received, or was
received with damaged bottles, a Kit Damage Report and a Missing or Damaged Bottle Report were printed,
the Sample Bottle Tracking System was updated, and replacement bottles were prepared and shipped.
Finally, after sampling was completed and the kits sent from the field to the labs, the sampling team
called the NPS hotline to confirm shipment. Preparation room personnel updated the Sample Kit Tracking
System to indicate that the kits had been shipped from the field to the laboratories.
The National Pesticide Survey Information System (NPSIS1) generated management reports designed
to track sampling kits, record sampling activities and sampling problems, and provide information about
completed wells. Exhibit C-9 lists the reports the Preparation Room staff used in managing the Preparation
Room operations.
C4.3 Sampling and Data Collection Techniques and Procedures
NPS sampling and data collection techniques and procedures for both CWS and rural domestic wells
were complex, detailed, and precise, and required careful quality assurance to guarantee consistency throughout
the Survey. NPS designers divided these techniques and procedures into six major steps, which included
prefield sampling preparation, sample collection procedures, after sample collection activities, shipment of
samples to laboratories, collection of additional information from local experts, and return of completed
questionnaires and other paperwork. Sampling and data collection techniques were discussed in detail in the
Field Well Site Water Sampling Manual that was distributed to sampling teams. This section will discuss the
major techniques and procedures for collecting water samples; sections C4.4 and C4.5 discuss administration
of CWS and rural domestic well questionnaires. Communication procedures were discussed in C4.1.
Prefield Sampling Preparation
Approximately two weeks prior to collecting well water samples, the field team leader received the
necessary equipment for sample collection activities from the NPS preparation room staff. This equipment
included five to thirteen sample container kits with bottles for each site and a cooler with a pullout tote bag
containing sampling equipment supplies. The Field Logbook containing the Well Sampling Information Sheet
was provided to each sampling team approximately four weeks prior to sampling. A list of all sample
container kits was attached to the Well Sampling Information Sheet for a preliminary inventory check of the
kits. Upon the receipt of the sample container kit, the sample team conducted the following activities:
For each well, removed the Well Sampling Information Sheet from the Field Logbook and
conducted an inventory check of the sample container kit numbers listed on the Well
Sampling Information Sheet for that well;
Conducted an inventory of the equipment provided in the tote bag against the Supply Kit
Equipment Checklist;
For each well to be sampled, checked that the well location, contact person, and sampling
date were correct on the Well Sampling Information Sheet;
National Pesticide Survey: Phase I Report
-------
Appendix C: Implementation of the Survey C-27
Exhibit C-9
Reports Used in Managing Preparation Room Operations
Well Schedule Report used to monitor the sampling schedule of wells on a biweekly basis;
Unconfirmed Well Schedule Report used to monitor selected unconfirmed sampling dates on a biweekly basis;
Lab Schedule Report used to monitor the kits the laboratories received for analysis on a weekly basis;
Lab Spike and Time Storage Mix/Level Distribution Report used to monitor the distribution of lab spikes and
time storage samples on a weekly basis;
Kit Tracking Report used to monitor the kits received by the field teams and NFS laboratories on a daily basis.
All outstanding kits were monitored for possible delays or other problems throughout the cycle;
Sample Bottle Tracking Report used to monitor sample kit bottles on a daily basis;
Misplaced Bottle Tracking Report used to monitor any misplaced bottles on a daily basis. This report was
used only when damages were indicated on the Sample Kit Tracking Report;
Damaged Kit Report used to monitor sample kit damage. This report was used only when damages were
indicated on the Sample Kit Tracking Report;
Missing or Damaged Bottle Report used to monitor sample kit bottles on a daily basis. This report was used
only when damages were indicated on the Sample Kit Tracking Report;
Well Purging Parameters Report used to monitor well purging parameters and comments that the field teams
recorded in field logbooks on a monthly basis;
Incomplete Well Purging Data Report used to monitor wells with incomplete well purging data on a monthly
basis;
Equipment Tracking Report used to monitor field equipment shipments and damages on a weekly basis;
Method 5 pH Value Report used to monitor pH values for water samples on a monthly basis;
NIPS Completed Wells Report used to tally the completion of wells sampled and the samples received by the
NPS laboratories on a weekly basis;
Process Completion Report used to summarize the preparation and shipment of sampling kits to sampling
teams, on a weekly basis;
Inventory Information Report used to monitor the Preparation Room inventory and the NPS off-site storage
of equipment and supplies; and
Electronic Mail System (EMAIL) served as a computerized communications network between the NPS
Preparation Room and the laboratories. EMAIL was used to transmit critical messages from the NPS
Preparation Room to the NPS labs about schedule changes, kit damages, and other logistical problems.
When a problem was encountered, the Preparation Room staff routinely called the NPS laboratories to convey
the necessary information and sent a follow-up message via EMAIL
National Pesticide Survey: Phase I Report
-------
C-28 Appendix C: Implementation of the Survey
Called Federal Express to determine if the office location printed on the Well Sampling
Information Sheet was the nearest office to the sampling site;
Called the NFS toll-free Hotline (1-800-451-7896) to report receipt of field equipment and
reported if kits and supplies had arrived in acceptable condition, or if there were any
problems with the kits;
Removed old Federal Express bar code labels and airbills from the sample container kits and
cooler, and crossed out the Federal Express three-character Airport code to avoid confusion
when preparing shipments to the laboratories;
Reconfirmed the sampling appointment and meeting time with the well owner or contact
person as needed, and obtained directions on how to get to the well site;
Assigned specific responsibilities for completion of the Field Logbook and questionnaires,
Sample Tracking Forms, and sample collection; and
Reviewed NFS sampling procedures, forms, and questionnaires and practiced the use of the
temperature meter, pH meter, and conductivity meter.
Sample Collection Procedures
For CWS wells, on the day of sampling, the Team Leader arranged to meet the local contact and
introduced him/her to the other team members. For rural domestic wells, the introduction to the resident
took place upon arrival at the well sites. The team leader explained procedures such as the need to sample
from a port or tap closest to the wellhead before any treatment has occurred and that the sampling point
should be where water was freshly pumped from the well. The Team Leader also noted that the well pump
would need to operate continuously for one to two hours for purging and sampling tasks and that the well port
or tap would be kept open until the collection of the last sample. (For CWS wells operating on an automatic
mechanism, sampling was done during the automatic pumping cycle or during a manual override of the
automatic system).
Before sampling began, the well owner (or resident) answered the questions in the CWS (or Rural
Domestic Well) Team Leader Introduction Questionnaire. Upon completion of this brief interview, the team
members followed these sampling procedures:
Located a sampling port or tap as close to the well head as possible and placed a bucket
under the tap to collect excess water as needed;
Measured the air temperature around the well tap using the temperature meter provided in
the supply kit and recorded this in the CWS/Rural Domestic Well Purging Parameters
Record;
Opened the well sampling port or tap valve and collected a sample of water in the 1 liter
plastic beaker provided in the cooler and immersed the temperature, pH, and conductivity
meters into the beaker of water. This initial sample of water from the pipe provided a
baseline against which to measure the parameters of the ground-water samples;
Allowed the water to run five minutes and repeated the three water readings again in the
1 liter plastic beaker. The purpose of this sampling procedure was to observe changes in the
parameters from the baseline;
National Pesticide Survey: Phase I Report
-------
Appendix C: Implementation of the Survey C-29
Again, allowed the water to run for five minutes and retook the three water readings. The
purpose of this sampling procedure was to record the parameters as ground water was
reached and the levels stabilized. Repeated checks were made at five minute intervals until
stability was reached or a maximum of 30 minutes had elapsed. If stability was not reached,
the well was not sampled.
Selected a sample container, removed the enclosed Sample Tracking Form, and checked that
there were no broken bottles, that bottles were labeled, and bottle label numbers
corresponded to the sample numbers shown on the Sample Tracking Form;
If the kit contained a shipping blank sample bottle, the sampling team initialed and dated the
bottle label with the provided waterproof pen; and
Using the opened container kit, sampling began with the largest bottle size of 1 L, followed
by the 250 milliliter (mL) glass amber bottle, the 125 mL plastic bottle, and the 125 mL glass
amber bottle. (Detailed descriptions for the collection of water in each of these bottles is
provided in the Field Well Site Water Sampling Manual).
After Sample Collection
Field Team members conducted the following activities upon completion of sample collection:
Replaced or added ice as needed in each of the sample container kits;
Removed the Federal Express airbill form inside the ziplock bag outside of the sample
container kit lid and resealed the ziplock bag to protect the enclosed Sample Tracking Form;
Carefully pulled and taped together the ends of the plastic bag that were between the kit and
cardboard box;
Taped the box shut, being careful not to cover the color-coded labels located on two opposite
sides of the box and reinforced the bottoms and corners of the boxes with extra tape if signs
of wear were apparent;
Placed the pre-addressed Federal Express airbill in the clear adhesive pouch and affixed it on
the appropriate box;
Completed the CWS/Rural Domestic Well Final Checklist that was contained in the Field
Logbook; and
Searched the well site to ensure that all trash and extra sampling materials had been picked
up.
Shipment of Samples to Laboratories
Field team members completed the following procedures to ship samples for laboratory testing:
Called Federal Express and arranged a pick-up or took the shipment to the nearest Federal
Express office;
Relinquished the sample container kits only to the custody of a Federal Express employee;
National Pesticide Survey: Phase I Report
-------
C-30 Appendix C: Implementation of the Survey
Collected a copy of each of the airbills for the sample container kits from the Federal Express
agent;
Returned all unused sampling and supply kit materials, communication materials, forms,
labels, and airbills by placing these materials back in the supply kit cooler and taping the
cooler shut;
Returned all unused communications materials to the NFS Communications manager so that
the materials could be reused;
Shipped the supply kit cooler to the NFS preparation room staff by Federal Express
"Standard Air" two-day shipment; and
Called the NFS Hotline toll-free number and requested the "Sample Tracking System" to
confirm that sample container kits had been sent to the laboratories.
Collection of Additional Information from Local Area Expert
One member of the sampling team was responsible for completing the CWS/Rural Domestic Well
Local Area Questionnaire by interviewing the local area expert (e.g., county agricultural extension agent). The
interviewer made the following arrangements:
For CWS wells, meet with the local area expert as soon as possible after the completion of
well sampling; and
For rural domestic wells, the Westat Field Supervisor arranged an appointment for the field
interviewer to meet the local area expert near the end of the two-week data collection period.
At that time, the local expert answered questions regarding the areas surrounding every sampled well
in the county.
Return of Completed Questionnaires and Other Paperwork
The CWS sampling team conducted the following checklist after completion of the questionnaires and
all other paperwork:
Conducted an inventory of the shipment to make sure all the following paperwork was
included in the package:
A completed CWS Purging Parameters Record form;
A completed CWS Main Questionnaire;
A completed CWS Team Leader Introduction/Well Observation Record;
A completed CWS/Rural Domestic Well Local Area Questionnaire;
The Sample Tracking Form (pink copies). The originals from this
form were sent with the samples to the appropriate laboratories;
and
- Completed airbills (sender's copies).
Returned the completed paper work along with the Field Logbook to the EPA Survey
contractor in the Federal Express "Courier-Pak" envelope provided.
National Pesticide Survey: Phase I Report
-------
Appendix C: Implementation of the Survey C-31
The Rural Domestic Well Questionnaires were returned separately by the interviewer upon completion
of the entire county.
C4.4 Preparation of CWS Field Survey Materials
Survey contractor staff members were responsible for the preparation of all materials needed in the
field to conduct CWS interviews. The required materials included:
Three different questionnaires;
Bar code labels;
Random number lists;
Mailing labels; and
Custody receipt form.
The schedule for the production and mailout of these materials was directed by a computer program.
Contractor staff prepared packages of materials to send to the NPS preparation room, from which the
packages were forwarded to the field teams.
The packages included one copy of each of the three questionnaires used in the survey:
1. Team Leader Introduction/Well Observation Record. This questionnaire was
administered to the CWS owner/operator by a staff member of the State's environmental
protection agency or EPA Regional office or State Health Department or, on occasion, by
a contractor interviewer at the time the well water samples were drawn. This questionnaire
collected information concerning the eligibility, number, plumbing, and location of wells at
the CWS.
2. CWS Main Questionnaire. This instrument was administered to the CWS owner/operators
by a staff member of the State environmental protection agency at the time the water samples
were collected. This form collected details concerning the use of the water from the wells
and pesticide usage on the property on which the well is located.
3. Local Area Questionnaire. This questionnaire was administered to the county's agricultural
extension agent for the area. It collected information on pesticide and other chemical uses
within one-half mile of each sampled well to help explain potential sources of well
contamination. (The CWS and rural domestic versions of this questionnaire were identical.)
The questions covered four general areas: (1) pesticides applied within one-half mile of each
sampled well; (2) accidental spills of pesticides and hazardous chemicals in the vicinity of each
sampled well; (3) other possible influences on ground water within one-half mile of each well
site such as golf courses, waste treatment facilities, chemical plants, etc.; and (4) irrigation
used within one-half mile of each well.
Included in the packages were continuation sheets to capture lengthy responses. Affixed to each of
the questionnaires were bar code labels depicting the CWS ID number. The bar code labels were essential
to the subsequent tracking of the Survey instruments throughout the Survey and data processing phases of the
study.
National Pesticide Survey: Phase I Report
-------
C-32 Appendix C: Implementation of the Survey
The Survey design called for the sampling of one well in each CWS. In the instances where there were
multiple wells at a CWS, the field team needed to pick one well at random. To assist the field team, the
tracking system generated a sufficient number of random numbers for each CWS and printed these numbers
on a label. This label was affixed to the Team Leader Introduction. In case a random selection of a well was
needed, the interviewer had only to refer to the label and instructions in the questionnaire for guidance.
The creation of the survey materials package and its shipment to the NFS preparation room
automatically marked the starting point for the CWS questionnaire tracking system. After the bar code labels
were affixed to blank questionnaires, the bar codes were electronically read, and an entry (date of shipment)
made into the tracking system. The tracking system produced a Custody Receipt form to accompany the
shipment of packages to the NFS preparation room. The form listed each CWS well ID included in the
shipment. Upon receipt, the NFS preparation room staff verified the shipment against the Custody Receipt
form, signed it, and returned it.
Custody Receipt. Based on the mailout schedule, all questionnaires were logged into a CWS
Questionnaire Tracking System computer program that recorded each one with the current date. The program
also simultaneously printed out a Custody Receipt Form showing the identification number of all cases being
mailed out. This sheet was checked and the shipment's completeness was verified. All forms were kept in the
Custody Receipt Logbook located in the NFS data preparation area.
Used Field Logbooks were shipped back to the Survey preparation room, but not before all
questionnaire-related information was removed. The CWS Tracking System generated a Custody Receipt
Form showing the ID number of all cases received. The form was then checked against the logbooks on hand
to ensure that none were misplaced. The logbooks were then shipped to the Survey preparation room along
with the Custody Receipt Form, and receipt of the shipment was verified.
Repeat Visit Cases. If a CWS had a large number of wells, the Survey design included sampling
more than one well from that system. These cases were labeled "Repeat Visit Cases' and required material
preparation in addition to that described above. Specifically, after the first visit to a CWS had been completed
and the questionnaires were returned, the Westat CWS Field Director reviewed the list of wells and randomly
selected three of the unsampled wells. The Field Director called the owner/operator of the designated CWS
to verify the operability of the wells. To simplify the field operations, the Director incorporated this
information in the Team Leader Introduction questionnaire for each sampled well. This questionnaire thus
indicated which well to sample and precisely which questions needed to be asked. The tracking system was
updated to report the preparation and mailout of the "Repeat Visit Cases." Once all these steps were
completed, the case was mailed to the NFS preparation room for shipment to the field team. Though lengthy,
this method proved effective in minimizing confusion on the field team's part in selecting wells for water
sampling purposes.
Monitoring CWS Progress Through the CWS Tracking System
The CWS questionnaire tracking system ensured the smooth functioning of the survey materials
preparation processes. For example, based on scheduling information obtained from ICF and the field, the
"Mailout Report" told Westat when to begin survey material preparation and for which specific wells.
Requests to the tracking system also prompted the production of the custody receipt forms. In addition, the
tracking system anticipated the return of questionnaires. Any case that had not been received within three
weeks of the interview/water sampling date was identified overdue on the Overdue Case Report. These
overdue notices prompted a telephone call to the field team for an explanation.
National Pesticide Survey: Phase I Report
-------
Appendix C: Implementation of the Survey C-33
The receipt of the questionnaires and their entry into the tracking system as received precluded the
Overdue notice. The tracking system monitored cases throughout the subsequent processing steps. A Status
Report could be called for at any time which detailed the number of questionnaires in each of the stages. A
menu option provided the user with the location of any case, including if it was being held back pending some
data clarification on the Tracking System.
Review of Instruments Prior to Data Processing
All materials returned by the field teams were reviewed by editors to identify any problems that
needed to be resolved while the team was still in the field. This initial edit included a check for completeness,
correct usage of skip patterns, and proper identification of the sampled well. Any problems uncovered at this
stage were forwarded to the Field Director for review. The Field Director would determine whether a call
needed to be made to the responsible field team members. When materials were incomplete or missing, the
Field Director would initiate Data Retrieval, which involved calling the interviewer, and sometimes the
respondent, to provide the missing items.
In many of the cases referred to the Field Director, it was apparent that there was a procedural
problem, e.g., the field team failed to observe correct procedure in a specific situation. If the Field Director
felt it was appropriate, a call was made to the field team to retrain the individual. At the approximate half-
way point of the Survey, the Field Director prepared a letter for all field team members noting what aspects
of field procedures had proved most problematic, and specifying the correct procedures to be used in those
circumstances. This letter was sent out with every subsequent shipment of questionnaires to the field.
C4.5 Preparation of Rural Domestic Well Field Survey Materials
Survey Questionnaires
Three questionnaires were designed for field use for the rural domestic well portion of NFS:
1: Rural Domestic Well Team Leader Introduction/Well Observation Record. The Rural
Domestic Well Team Leader Introduction collected information concerning eligibility of the well,
household water usage during the 24 hours before the water samples were taken from the well, the
well piping system, and whether septic systems were located on the well property.
2. Rural Domestic Well Questionnaire. The Rural Domestic Well Questionnaire was designed to
collect information on water usage, farming and non-farming activities, construction of the well,
and pesticide application on the property where the well is located.
3. Local Area Rural Domestic Well Questionnaire. The Local Area Rural Domestic Well
Questionnaire (LAWQ) was designed to collect information about the area of land within one-half
mile of each sampled well to help explain potential sources of well contamination. (The CWS and
rural domestic well versions of this questionnaire were identical.) The questions gathered
information regarding: (1) pesticides applied within one-half mile of each sampled well; (2)
accidental spills of pesticides and hazardous chemicals in the vicinity of each sampled well; (3)
other possible influences on ground water within one-half mile of each well site such as golf
courses, waste treatment facilities, chemical plants, etc.; and (4) irrigation used within one-half
mile of each well. The respondent for this instrument was the local county agriculture extension
agent.
National Pesticide Survey: Phase I Report
-------
C-34 Appendix C: Implementation of the Survey
Two modified versions of the Rural Domestic Well Questionnaire were used for special circumstances:
Team Leader Introduction for Interviews Without Water Sampling; and
Rural Domestic Well Questionnaire Section C for Multiple Fanners.
Preparation of Survey Materials and Field Arrangements
The field schedule produced as a result of the CATI screening was reviewed and modified to address any
logistical problems. Materials for the field assignments were then prepared by the Survey implementation
subcontractor. This preparation included:
Arranging for a field team to administer the Survey materials;
Making all travel and interviewer hotel arrangements for field teams;
Making appointments for county agent interviews;
Printing name/address labels for each Survey package and for the individual questionnaires;
Selecting, labeling, and bundling the appropriate questionnaires;
Forwarding Survey packet to field teams with the interview schedule; and
Initializing monitoring system to track progress of Survey materials throughout processes.
C4.6 Rural Domestic Well Field Operations and Monitoring of Survey Task
Progress
The field team received the materials and schedule at the beginning of the two-week field period in each
county. Twenty-four hours in advance of each scheduled interview, the field team called the home to confirm the
home visit on the following day. Frequently at this point, but sometimes at the door of the house, respondents
requested changes in the appointment times. Based on their schedule of other appointments, the field team
rescheduled as best it could. Changes were called into the Field Director and the NPS hotline.
The field team visited each respondent's home at the designated time. After the Team Leader introduced
the team, the interviewer would attempt to identify the correct respondent for the instrument. The object was to
speak with the person who was the most knowledgeable about the well and surrounding land activities. It was
preferred that one person answer all questions about the well. On occasion it became necessary to interview more
than one person. A well owner may lease the land surrounding the well to a former. In such a case, the well owner
would be asked questions about water usage, the construction of the well, and the non-farm activities (Sections A,
B, and D of Rural Domestic Well Questionnaire), but the lease-holder who actually farms the land would be asked
about farm use of pesticides (Section C).
More complex situations occurred when, for example, a household used well water from a well on someone
else's property. In this situation the household who used the well water was asked water usage questions
(Section A), and the person who owned the well was asked about well construction and non-farm activities around
the well (Sections B and D).
National Pesticide Survey: Phase I Report
-------
Appendix C: Implementation of the Survey C-35
The questions in Section C of the Rural Domestic Well Questionnaire ask the fanner of the well property
to recall farming activities for the five years prior to the interview. On the occasions when property was Canned
by more than one farmer over the past five years, the interviewer administered the Rural Domestic Well
Questionnaire Section C, Multiple Fanners Questionnaire, to each fanner.
On rare occasions, the appropriate respondent for the Team Leader Introduction and Section A for the
Rural Domestic Well Questionnaire was not present when water samples were collected from the well. The
interviewer returned to the household later and administered the appropriate sections of the Rural Domestic Well
Questionnaire and the Team Leader Introduction for Interviews Without Water Sampling to the correct respondent.
The third questionnaire, the Local Area Rural Domestic Well Questionnaire, was administered to the local
county agriculture extension agent. If the Field Director had been unable to schedule the appointment, the field
interviewer attempted to schedule an appointment with the pertinent extension agent and administer the
questionnaire.
C4.7 Implementation Data Management
The NPSISI data bases and information system were developed to automate, organize, and centrally locate
sampling information, (e.g., scheduling and kit-packing) vital to the day-to-day functions of the NFS. EPA
determined that a cost-efficient and effective management strategy would include the use of computers and a
customized software system to manage sampling data. The resulting system, named NPSIS (NPS Information
System) in its earlier form, enabled the preparation room staff, NPS managers, and communications team to enter
and use computerized information. The system used a computer network and communications software that
allowed remote access to NPSIS at EPA, the contract laboratories, EPA laboratories, and at NPS contractor offices.
It provided an effective means of tracking kits and bottles through the sampling process.
NPSIS I Systems Requirements
The hardware configuration of the system included the following:8
IBM Personal Computers (2) ~ not including personal computers used for remote access by
laboratories or other EPA or contractor personnel;
IBM Token Ring Network;
IBM Token Ring Network Adapter Cards and Cabling(2) - not including network adapter cards
needed for access by other contractor personnel;
Hayes (or compatible) 2400 Modem, Modem Cabling, and dedicated phone line capable of data
transmission;
Printer - Epson FX286e Dot Matrix Printer to impact text on polyurethane bottle labels;
dBase III Plus software package (two);
Nantucket Clipper language and compiler,
Carbon Copy communications software (eight) - one copy for the communications PC and one
copy for each remote user;
Reference to particular products or trade names does not imply EPA endorsement.
National Pesticide Survey: Phase I Report
-------
C-36 Appendix C: Implementation of the Survey
DOS software, version 3.3 or above;
Polyurethane adhesive bottle labels and triplicate printer forms; and
The NPSIS I computer software.
NPSIS I Systems Configuration
The NPSIS I system configuration consisted of placing two IBM personal computers in the preparation
room. The preparation room was equipped with network plugs and cabling for three personal computers and a
phone jack connected to a dedicated (data transmission) phone line.
One PC was programmed to automatically come up as a network server when turned on or rebooted and
selected disk directories and their file contents were placed in a network share mode. These shared directories and
files house the NPSIS I data bases. The NPSIS I customized software was copied onto the computer.
Another IBM personal computer was installed with an internal modem and programmed to come up on
the network as a redirector and to share the NPSIS I data bases with the other computer. Carbon Copy was
installed on this PC as the host copy and activated in "call waiting" mode so that the PC would answer any incoming
phone calls from the laboratories to the data transmission phone line. The NPSIS I customized software and the
Sample Receipts Program, a specialized module of NPSIS I, were copied onto the computer.
A backup system was implemented to ensure against loss of data. Two copies of the data base system and
software were saved each night by the preparation room staff using a program written to automate the backup
process. Copies of the data and software were stored in the preparation room and an office of an NPS data
management team member.
A batch file was copied onto computers elsewhere in the NPS implementation support contractor's
building, from which NPSIS I would be accessed. This file performed the necessary commands to access the server
computer and start the NPSIS software.
All remote users were sent the Carbon Copy software, a systems requirements document, and a users
manual describing how to install Carbon Copy and access NPSIS I.
Implementation Data Management Components
There were five components of the NPSIS I data management system used in implementing the Survey:
the Sampling Control System (SCS), the Sample Receipts Program (SRP), the Communications Control System
(CCS), the Document Control System (DCS), and the NPS Problem File. Each of these implementation data
management systems is described below.
Sampling Control System (SCS)
The SCS was developed to automate information collection, storage, and retrieval for various sampling
tasks. The SCS combines data entry, storage, dissemination, and management through a series of user-friendly
screens. The SCS was used to facilitate Survey implementation in the following ways:
Scheduling wells for sampling;
Assigning the analysis method and kit configurations for each well;
Printing sample tracking forms, bottle labels, and kit labels;
National Pesticide Survey: Phase I Report
-------
Appendix C: Implementation of the Survey C-37
TrackiDg samples through various stages of sampling, from kit preparation through receipt of
sample kits at the laboratories, by collecting airbill numbers and providing the capability for the
laboratories to report sample receipt and condition;
Tracking equipment and inventory supplies;
Providing a key entry system for recording well observation log purging parameters; and
Generating management reports which describe all phases of sampling.
The NFS Sample Kit preparation staff entered all necessary information, according to standard operating
procedures, into the SCS during daily operations. To the fullest extent possible, sampling implementation
information was automatically generated by the SCS computer programs. Examples of automatically generated
information are: 1) sample bottle information such as identification numbers, bottle size, and sample type, and 2)
assignment of bottles and kits to particular laboratories for analysis based on the analysis method used and kit
configurations.
Sample Receipts Program (SRP)
The SRP was developed to enable the contract and EPA laboratories to log on to NPSISI to report sample
receipts, report problems with the condition of the kits and samples, check the sampling schedule to determine
future sampling flows, and send electronic mail, to other laboratories or the NPS implementation support
contractor.
The SRP includes a user-friendly set of menus and screens, complete with instructions and help suggestions.
The communications software allows remote users to completely control the NPSIS I computer as if they were
sitting at the keyboard instead of at their remote terminals. The SRP performs automated QA/QC by rejecting
certain types of invalid information, such as invalid ranges or characters for numeric fields. SRP eliminates the
need for file transfer, does not include the automated QA/QC features, and does not allow interactive data input
or retrieval.
Security features were incorporated into the design of the SRP. Passwords were required for remote access
and access through the NPS implementation support contractor's network.
Communications Control System (CCS)
A separate computer system was developed and integrated with the SCS to assist the NPS Communications
Group to track rural domestic well sampling schedules, communications information needed to secure voluntary
participation from rural domestic well owners, and participation rates.
NPS communications staff entered scheduling and communications information into the computer, and then
used CCS to track and manage rural domestic well sampling communications tasks and produce rural domestic well
sampling schedules for contractor field sampling teams. This information was also accessible to users of the SCS.
Document Control System (DCS)
A computerized document control system (DCS) was created to provide an efficient way to store and
retrieve project records. Documents filed in DCS were given a unique number and identity based on the following
items:
Date;
Task Number;
National Pesticide Survey: Phase I Report
-------
C-38 Appendix C: Implementation of the Survey
Status (i.e., deliverable, draft, final);
Author;
Person the document was submitted to;
Title;
Type of document (i.e., briefing, correspondence, publication, report, other); and
Activity code (i.e. analytical methods, well selection, data management, training, etc.).
This information was key entered into DCS. Documents are located in filing cabinets by searching the
document control data base using DCS's menus. The user can identify a particular document or a group of
documents based on the identifying items listed above.
The NPS Problem File
A computerized system was developed to document problems associated with sampling that could affect
the validity of the analytic results. The system was designed to maintain data on all problems and to retrieve
information based on key identifiers, such as well identification numbers and the type of problem encountered.
NPS staff were responsible for key entry of information into the problem file. Information stored in this system
could be accessed by all NPS staff or disseminated by report.
C4.8 Notification of Results
EPA established strict procedures to ensure the proper dissemination of well sampling results. The NPS
communication staff notified well owners and State officials of well sampling results by instituting these procedures
to ensure that the privacy of rural domestic well residents and/or owners was protected and to comply with Freedom
of Information Act regulations.
Once an individual test result was completed, the contract laboratory sent the results to the NPS Project
Director and the NPS implementation contractor. If no pesticides were detected or the concentration of pesticides
was below the rapid reporting level, notification of the results was made through the mail; if the concentration of
pesticides was found to be above the rapid reporting level, notification was made through an expedited process.
Rapid reporting levels for each of the pesticides represented pesticide concentrations that warranted immediate
action, and were based on EPA's Lifetime Health Advisory Levels (HALS). Each of these processes is explained
in detail below. Except where noted, the notification process was identical for community well system and rural
domestic drinking well owners/operators.
Regular Notification
Once the test results were completed, the contract laboratory sent the results to the NPS Project Director
and the NPS implementation contractor. If the sample results were below the rapid reporting level or no pesticides
or nitrate were detected, the NPS implementation contractor prepared notification packets for the EPA NPS
Director that included a transmittal memo, sampling results, sample notification letters, Health Advisory Summaries
for any pesticides found in a well, a list of pesticides included in the Survey, a list of pesticides not analyzed, and
full analytical results for each method.
The NPS Director reviewed the notification packet, signed the transmittal memo, and forwarded the packet
to the Regional and State contacts. The States were urged to notify each system owner/operator in writing within
one week of receipt of the results; it was also suggested that States follow up the notification letters with calls to
owner/operators whose wells were found to contain pesticides. When notifying rural domestic well
owners/operators, the contacts were reminded that the summary results and full laboratory results were confidential.
When notifying community well system owners/operators, States were reminded that all sampling information
pertaining to community water systems is releasable to the public. However, the States were urged to inform
National Pesticide Survey: Phase I Report
-------
Appendix C: Implementation of the Survey C-39
owners/operators of test results in advance of public release of the information. Exhibit C-10 contains a copy of
the letter sent to owners/operators of CWS wells that had pesticide or nitrate concentrations below the rapid
reporting level.
Once notification was complete, the State's primary responsibility was to answer any questions raised by
owners/operators, especially questions related to potential health effects. Each State had its own procedures for
responding to questions.
Rapid Notification
In the event of results higher than the rapid reporting level, the NFS Director telephoned the State and
Regional contacts to inform them of the results within 24 hours of a laboratory confirmation. Notification packets
were express mailed to States within three days of a laboratory confirmation of the results. The State contact was
required to call within two days to inform the owner/operator of the results.
In addition to being notified immediately by phone, the owners/operators were sent written notification
of the results. The NFS implementation contractor prepared a notification packet containing a transmittal memo,
a summary of results, a personalized well owner and/or household resident notification letter, a list of pesticides
included in the Survey, and a Health Advisory Summary. The NFS Director reviewed the packet, signed the
transmittal memo, and forwarded the packet to the Regional and State contacts.
Upon receipt of the packet, the State photocopied the personal notification letter onto its letterhead, added
the date, appropriate State contact information, signature, and sent the packet to the owner and/or household
resident within three days of receipt of the packet. Exhibit C-ll contains a copy of the letter sent to
owners/operators of rural domestic wells that had pesticide concentrations above the rapid reporting level. If the
notification concerned rural domestic wells, States were reminded that the test results and the notification letter
were confidential. In cases of community well system notification, the States were reminded that the results were
releasable to the public. In addition, the States were urged to include instructions to the system owner/operator
on actions required in light of the test results.
During the time that the results were distributed, the State agencies responded to questions regarding the
results of the Survey, potential health effects due to pesticides in wells, and any necessary corrective actions. Each
State had its own procedure for responding to questions.
C5 Data Base Development
C5.1 NPSISII
As sampling tasks came to completion, a larger computerized data base and information system, named
NPSIS II, was designed to manage the entire body of NFS data. This system includes the NPSIS I data bases as
well as the final sample analytic results, questionnaire data, sampling weights, and ground-water related data. The
NPSIS II data base system is installed on EPA's mainframe computer system.
Systems Requirements and Configuration
NPSIS II resides on EPA's Cincinnati logical mainframe and the EPA National Computer Center IBM
mainframe in a partitioned dataset containing Statistical Analysis System (SAS) files, documentation, and programs
to assist users in using and interpreting the data base system. To access NPSIS II, an EPA computer account and
computer funding are required.
National Pesticide Survey: Phase I Report
-------
C-40 Appendix C: Implementation of the Survey
Exhibit C-10
CWS Standard Test Results Letter
(Positive Contamination Below Health Advisory Levels)
Date:
Well Code #
Name of Owner/Operator
Name of CWS
Address
City, State, Zip
Dear CWS Owner/Operator:
Several months ago, water samples were taken from the [name of system! community water
system [(FTDS # )] in [City/County, State] as part of the National Pesticide Survey being
conducted by the U.S. Environmental Protection Agency (EPA). Testing has been completed for over
100 pesticides and nitrate/nitrite. (See the enclosed List of Pesticides Included in the National
Pesticide Survey.)
[Earlier you were informed of our initial findings of pesticides or nitrate/nitrite in your water
samples.] Your [remaining] test results are shown below. The levels at which contaminants were
found are not considered to pose a health risk. An EPA Health Advisory Summary is enclosed for
each contaminant found.
Stare's instructions to the CWS owner/operator on actions to take, if any, should be inserted here.
For more information, please contact the State office listed below:
Contact:
Office:
Tel:
Yours sincerely,
Enclosures
National Pesticide Survey: Phase I Report
-------
Appendix C: Implementation of the Survey C-41
Exhibit C-11
Rural Domestic Well Owner Standard Test Results Letter
(Positive Contamination Above Health Advisory Level)
Date:
Well Code #
Name of Owner
Address
City, State, Zip
Dear Well Owner:
Recently, water samples were taken from your well as part of the National Pesticide Survey
being conducted by the U.S. Environmental Protection Agency (EPA). The samples are being tested
for over 100 pesticides and nitrate/nitrite. (See the enclosed list.)
In testing the samples, to date we have discovered the following: Your water sample
contained milligrams per liter* of . This is above EPA's Guidance Level of
milligrams per liter of .
An EPA Health Advisory Summary is enclosed for the contaminant found. EPA is continuing
further tests on your water samples for the presence of additional pesticides. We will inform you of
those results as soon as they are available.
For more information, please contact the State office listed below:
Contact:
Office:
Tel:
Yours sincerely,
* Micrograms per liter is the unit of measurement for contaminants in water, equivalent to parts per
billion. Milligrams per liter is a common unit of measurement for nitrate/nitrite in water, equivalent to
parts per million.
National Pesticide Survey: Phase I Report
-------
C-42 Appendix C: Implementation of the Survey
Analytic Results Data Management
The analytic results were entered by contract and EPA laboratory staffs. Analytic results were processed
at the EPA laboratory in Cincinnati, where QA/QC checks were performed. The analytic results were retrieved
from the EPA mainframe computer system and linked, or merged, with the information stored in SCS. The merged
data were then processed to generate the following reports:
Community water system owner/operator notification of results reports;
Rural domestic well sampling homeowner (and resident, if homeowner does not reside at well site)
notification of results reports;
Summary results reports; and
Freedom of Information Act reports.
Questionnaire Data
Survey questionnaire data were collected during sampling to augment the analytic results data. The Survey
instruments and methods used to collect this data are:
For rural domestic well sampling: Rural Domestic Well Survey Main Questionnaire, Local Area
Questionnaire, Team Leader Introduction and Well Observation Record; and
For community water system sampling: Community Water System Main Questionnaire, Local
Area Questionnaire, Team Leader Introduction and Well Observation Record.
Questionnaire data were key entered and QA/QC checks were performed according to standard operating
procedures. The data were then integrated into the final data analysis system.
C5.2 Questionnaire Data Base Development
This section details the steps and procedures that were observed in preparing the data collected through
the questionnaires for computerized processing and the creation of national estimates.
Overall, both CWS and rural domestic well questionnaire materials were processed in accordance with the
following standard procedures, and modified, when necessary, to apply to the unique aspects of the CWS and rural
domestic well survey. The general steps followed in preparing questionnaire data are:
Receipt of materials from field staff;
Entry of receipt into Tracking System;
Scan edit for serious problems (CWS) or thorough edit by the Field Supevisor (DWS);
Data retrieval of vital information, if necessary;
Coding batch formation via the Tracking System;
Batches coded for clarity and assignment of codes to open-ended questions;
Key entry batch formation via the Tracking System;
National Pesticide Survey: Phase I Report
-------
Appendix C: Implementation of the Survey C-43
Batches forwarded to key entry;
Batches key entered and key verified;
Questionnaires returned to coders/editors;
Computerized range and logic checks conducted, errors corrected;
Advanced/additional computerized logic checks conducted, errors corrected;
Data retrieval calls made to appropriate respondents if necessary to resolve discrepancy;
Corrections entered into data;
"Checking and correcting" reexecuted until data base is considered "clean"; and
Data base further developed (imputation performed and weighting applied) so that national
estimates can be made for the data items.
Details concerning imputations and weighting of the data base for national estimates are found in
Appendix B.
Review of Completed Survey Instruments Prior to Data Processing
The field team returned the Survey package materials at the conclusion of the interviewing in a county
(rural domestic wells only). The packages were opened and receipt notations were entered in the manual receipt
control system. Bar code labels on each Survey instrument indicating the well ID were scanned into the
computerized receipt control system. These bar codes were pivotal to the tracking of each instrument throughout
the data processing phase.
The Field Director reviewed each return initially to spot major breaches in established procedure. When
necessary, calls were made to the field team to clarify points or to send the team back to the home or well for data
clarification. Materials were then batched and forwarded to the coder/editor teams.
Advanced Preparation
In advance of the actual return of Survey materials, code books were established for the CWS and rural
domestic well survey forms. The code books specified characteristics of each variable (e.g., number of positions,
type of format), the record structure and the placement of fields on the records, unambiguous and clear code
systems (l=Yes, 2=No), and consistencies that should be observed in the data (e.g., logical limits on values of
variables, relationships between variables.) These range limits and logical relationships were built into the
computerized data checking software (COED) and applied to the incoming data. After the first several batches
of actual materials were processed, the criteria were reviewed and some changes were made to criteria in light of
actual field findings.
Receipt of Field Materials and Coding
Upon receipt, all field materials were logged into the computerized document tracking system. The CWS
questionnaires were submitted to an initial scan edit to identify any immediate and serious problems that needed
to be resolved by the field team. Questionnaires for rural domestic wells were reviewed thoroughly by the Field
Supervisor. After this initial edit, the questionnaires were sorted into batches of approximately 20 using the
Tracking System. Logs were printed that listed each questionnaire in the batch.
National Pesticide Survey: Phase I Report
-------
C-44 Appendix C: Implementation of the Survey
The questionnaires were passed to coders who reviewed all data fields and made corrections and
clarifications in accordance with the code book and data coding guides. A special feature of the coding instructions
was a notation that coders could turn on if they felt that a particular question needed evaluation by a specialist with
a hydrogeological background. A special program was written that singled out cases with such flags for the most
important variables, which were then given extended and substantive review.
Data Entry
After coding, the batches of 20 questionnaires were combined into batches of approximately 40 via the
Tracking System and were forwarded to the data entry staff for key entry. Logs were printed that listed each
questionnaire in that batch.
The data entry staff, in advance, had prepared customized key entry programs to facilitate the speedy and
accurate entry of CWS and rural domestic well information into the system. The materials received by data entry
were keyed and 100 percent rekeyed by a different operator for verification. Any discrepancies were flagged and
evaluated. Simple errors were corrected; ambiguities were referred to the Coding Supervisor for clarification and
resolution. After each correction was made, the record was 100 percent rekeyed and key verified to ensure that
errors were not introduced in the course of making corrections.
Expert Review
The expert review of confusing or difficult field findings was conducted by the Field Director and a
contractor staff member familiar with the Survey. This specialist reviewed the comments provided by the Field
Team and the coder and made a judgment as to the correct code to be applied to the data item(s) in question in
accordance with coding procedures.
Data Retrieval
Missing or inconsistent responses in NPS data were corrected through a process known as data retrieval.
Data retrieval began with the identification of problems by the computerized edit checking programs conducted on
the data files. The problems were summarized in writing and submitted to the data retrieval staff, who obtained
the appropriate original Survey questionnaire and made a telephone call to the appropriate respondent. The
problem was explained to the respondent and the resolution made. The data retrieval specialist wrote up the
correction and indicated the correction to be imposed on the raw data base. This correction was submitted to key
entry and the data base was updated. The computerized logic checks were rerun to verify that the problem no
longer existed and that new problems were not introduced.
Data retrieval efforts were limited to correcting key variables. In order to be subject to data retrieval, a
variable had to be designated as "key* by EPA or it had to have a logical relationship with a key variable.
The instructions for data retrieval were very specific about who could provide answers for the data retrieval
questions. Acceptable respondents were detailed on a question-by-question basis. This constraint was imposed to
ensure that accurate data were obtained whenever possible.
Passage of Data Files to Analytic Staff
When all data preparation steps were completed, the data files were moved to a separate computer accoun
for use by the analytic staff for the purpose of imputing missing data fields and applying weights that would produc
national estimates.
National Pesticide Survey: Phase I Report
-------
National Pesticide Survey
Appendix D: Questionnaire Data and Questionnaires
-------
Appendix D: Questionnaire Data and Questionnaires
D1 Introduction
One of the purposes of the NFS is to investigate the relationship between the presence of pesticides
and nitrate in drinking water wells and a number of factors potentially affecting that presence. In order to
gather data on well construction and maintenance, farming activities conducted in the vicinity of wells, and
pesticide use, several detailed questionnaires were administered to the owners or operators of wells,
householders using rural domestic wells, farmers, and local area experts such as county agricultural extension
agents. Questionnaires used to compile the data include the CWS and Domestic Well Team Leader
Introduction (TLI) and Well Observation Record (WOR), the CWS and Domestic Well Main Questionnaires
(CWS Main and DWS Main), and the CWS and Domestic Well Local Area Questionnaires (LAQ).
This appendix consists of seven exhibits, one for each domain of interest, that present estimates and
confidence intervals for well characteristics. In addition this appendix contains copies of the questionnaires
used to obtain the raw data.
The seven exhibits present national estimates based on questionnaire data for:
CWS wells in the United States (Exhibit D-l);
CWS wells in counties with high vulnerability (Exhibit D-2);
Rural domestic wells in the United States (Exhibit D-3);
Rural domestic wells in counties with high pesticide use (Exhibit D-4);
Rural domestic wells in counties with high ground-water vulnerability (Exhibit D-5);
Rural domestic wells in "cropped and vulnerable" county subregions (Exhibit D-6); and
Rural domestic wells in counties with high pesticide use and high ground-water vulnerability
(Exhibit D-7).
Caution should be exercised in making comparisons between domains for similar characteristics
because the confidence intervals (indicated in the exhibits as 95% C.I.) do not account for the relationship
between domains or between characteristics (i.e., each confidence interval is calculated separately or
independently). All confidence intervals presented in this appendix are based on the normal distribution
approximation to the binomial distribution.
A parallel numbering system is used in the exhibits, but that system does not correspond to the
numbering used in Survey questionnaires. The exhibits also provide the identity of the questionnaire(s) and
question(s) from which the data are derived (e.g., CWS TLI #14 indicates that the data source is question #14
of the CWS Team Leader Introduction).
Some questionnaire items contain skip patterns (i.e., instructions to skip particular questions
depending upon previous answers). A questionnaire item is presented as a single item including sub-items
resulting from skip patterns, which are separated from the leading item with a blank line.
Number and percent estimates for responses to particular questions sum exactly to the domain
population size and one hundred percent, respectively, when the possible Survey responses are mutually
exclusive and exhaustive. Rounding can cause the summations to be inexact (number estimates are generally
rounded to 3 significant digits, and percent estimates are rounded to 1 decimal place). In addition,
summations may not equal the population size or one hundred percent for a few questionnaire items that do
not contain a mutually exclusive or exhaustive list of possible responses.
For particular responses to questionnaire items when the percent is extreme (i.e., near 0.0% or
100.0%) and the sample size is relatively small, the 95% confidence intervals should be treated with caution.
When the percent is equal to 0.0% or 100.0%, traditional 95% confidence intervals based solely on sampling
error are not applicable. For further details see Section B6.4 of Appendix B.
-------
D-2 Appendix D: Questionnaire Data and Questionnaires
Exhibit D-1
Estimated Number and Percent of
Community Water System Wells In the United States
Item
1.1
cws
TLI
#8
I.2
CWS
TLI
#14
I.3
CWS
WOR
#1
I.4
CWS
WOR
#4
I.5
CWS
WOR
#3
Questionnaire Item
Number of wells that belong to
systems with:
One working well
Two working wells
3-5 working wells
6-1 0 working wells
> 1 0 working wells
Septic system within 300 feet of
well:
Septic tank
Septic field
Cesspool
Topographic setting at well:
Hilltop
Hillside
Flat valley
Depression
Plateau
Other
Don't know
Description of land area within
300 feet of well:
Roofed
Paved
Graveled
Exposed or vegetated soil
Other
Well protection at surface:
Well house or shed
Concrete pad
Sanitary or grouted seal
Covered pit
Other
National Estimates for CWS Wells
Number (95% C.I.)
15,900 (±3,080)
26,200 (±3,810)
27,000 (± 4,390)
7,850 (± 2,220)
17,800 (±2,710)
19,100 (±2,780)
17,700 (±2,890)
2,220 (± 1,650)
11,000 (±2,750)
18,900 (±3,040)
30,700 (±4,180)
3,380 (±1,380)
9,510 (±2,620)
20,600 (± 3,700)
570 (± 600)*
4,060 (±1,600)
3,520 (±1,380)
4,320 (± 1,480)
68,500 (± 4,300)
14,200 (± 4,070)
53,000 (±5,170)
73,800 (±4,160)
65,500 (±5,150)
7,500 (±2,120)
15,600 (±3,370)
Percent (95% C.I.)
16.8 (± 3.4)
27.7 (± 4.0)
28.5 (± 4.2)
8.3 (± 2.4)
18.8 (± 2.7)
20.2 (± 2.8)
18.8 (± 2.9)
2.4 (± 1.7)
11.6 (± 2.9)
20.0 (± 3.4)
32.5 (± 4.1)
3.6 (± 1.5)
10.1 (± 2.9)
21.7 (± 3.5)
0.6 (± 0.6)*
4.3 (± 1.7)
3.7 (± 1.5)
4.6 (± 1.6)
72.4 (± 4.3)
15.0 (± 4.1)
56.0 (± 4.3)
78.0 (± 3.1)
69.2 (± 4.3)
7.9 (± 2.3)
16.5 (± 3.4)
National Pesticide Survey: Phase I Report
-------
Appendix 0: Questionnaire Data and Questionnaires D-3
Exhibit D-1 (cont'd)
Estimated Number and Percent of
Community Water System Wells In the United States
Item
I.6
CWS
Main
A.1
I.7
CWS
Main
A.2
I.8
CWS
Main
A.5
i.g
CWS
Main
A.7
CWS
Main
A.11
CWS
Main
A.13
CWS
Main
A.18
Questionnaire Item
Age of well:
< 5 years
5 to 10 years
11 to 20 years
> 20 years
Don't know
Well redrilled:
Yes
No
Don't know
Well depth:
< 20 feet
20 to 50 feet
51 to 100 feet
101 to 200 feet
201 to 500 feet
> 500 feet
Don't know
Well casing information:
Well cased
Well not cased
Don't know casing
Cased to total depth
Cased < total depth
Don't know casing depth
Screens in casing
No screens in casing
Don't know screening
Well grouted
Well not grouted
Don't know grouting
National Estimates for CWS Wells
Number (95% C.I.)
5,900 (±1,550)
16,300 (±2,700)
26,700 (± 4,290)
33,600 (± 5,750)
12,000 (± 3,360)
4,530 (±1,570)
84,700 (± 4,370)
5,330 (± 2,220)
260 (± 510)*
4,410 (±2,470)
14,900 (±3,180)
16,600 (±3,300)
33,400 (± 4,450)
19,800 (±3,530)
5,150 (± 2,270)
93,600 (± 3,720)
220 (± 300)*
720 (± 670)
43,300 (±4,410)
40,300 (± 4,260)
10,100 (±3,100)
49,000 (±3,130)
32,300 (±4,110)
12,400 (± 2,770)
65,200 (± 4,700)
10,700 (± 2,660)
17,700 (±3,400)
Percent (95% C.I.)
6.2 (± 1.6)
17.2 (± 3.1)
28.2 (± 4.5)
35.6 (± 5.4)
12.7 (± 3.6)
4.8 (± 1.6)
89.6 (± 2.7)
5.6 (± 2.4)
0.3 (± 0.6)*
4.7 (± 2.6)
15.8 (± 3.2)
17.6 (± 3.4)
35.3 (± 4.4)
20.9 (± 3.8)
5.5 (± 2.4)
99.0 (± 0.8)
0.2 (± 0.3)*
0.8 (± 0.7)
46.2 (± 4.6)
43.0 (± 4.2)
10.8 (± 3.2)
52.3 (± 3.2)
34.5 (± 4.0)
13.3 (± 2.9)
69.6 (± 4.1)
11.5 (± 2.8)
18.9 (± 3.6)
-------
D-4 Appendix D: Questionnaire Data and Questionnaires
Exhibit D-1 (cont'd)
Estimated Number and Percent of
Community Water System Wells In the United States
Kern
1.10
cws
Main
A.21
1.11
CWS
Main
A.23
1.12
CWS
Main
A.25
1.13
CWS
Main
A.37
1.14
CWS
Main
B.1
CWS
Main
B.3
Questionnaire Item
Well capped at ground surface:
Well capped
Well not capped
Don't know if capped
Distance from the ground surface
to the water surface in the well:
< 20 feet
20 to 50 feet
51 to 100 feet
101 to 200 feet
201 to 500 feet
> 500 feet
Don't know
Aquifer:
Confined
Unconfined
Don't know
Abandoned or non-operating
wells within 500 feet of the well:
Yes
No
Don't know
Non-farm pesticides used in the
last 3 years:
Non-farm pesticides used
Not Used
Don't know whether used
Used < 1 00 feet from well
Used >. 100 feet from well
Don't know distance
National Estimates for CWS Wells
Number (95% C.I.)
79,300 (± 5,440)
15,100 (±3,200)
110 (± 190)*
13,700 (± 3,350)
20,500 (± 3,430)
14,100 (±3,620)
12,800 (± 3,020)
9,520 (±1,990)
690 (± 740)*
23,200 (± 4,320)
41,100 (±5,080)
36,400 (± 4,400)
18,930 (±3,800)
20,800 (±4,160)
69,000 (± 4,960)
4,740 (±1,930)
24,100 (±3,420)
67,300 (± 4,670)
3,190 (± 1,520)
17,700 (±3,500)
3,400 (±1,750)
2,990 (±1,120)
Percent (95% C.I.)
83.9 (± 3.6)
16.0 (± 3.6)
0.1 (± 0.2)*
14.5 (± 3.4)
21.7 (± 3.4)
14.9 (± 3.7)
13.6 (± 3.2)
10.1 (± 2.1)
0.7 (± 0.8)*
24.6 (± 4.8)
43.5 (± 5.0)
38.5 (± 4.3)
20.0 (± 4.0)
22.0 (± 4.2)
73.0 (± 4.8)
5.0 (± 2.0)
25.5 (± 3.5)
71.2 (± 3.7)
3.4 (± 1.6)
73,5 (± 9.1)
14.1 (± 7.0)
12.4 (± 4.8)
National Pesticide Survey: Phase I Report
-------
Appendix D: Questionnaire Data and Questionnaires D-5
Exhibit D-1 (cont'd)
Estimated Number and Percent of
Community Water System Wells In the United States
Item
1.15
CWS
Main
B.4
CWS
Main
B.5
1.16
CWS
Main
B.6
1.17
CWS
Main
B.8
1.18
CWS
Main
C.1
CWS
Main
C.2
1.19
CWS
Main
C.5
Questionnaire Item
Non-farm pesticides stored for
more than 1 month in the last 3
years:
Non-farm pesticides stored
Not stored
Don't know whether stored
Closest storage < 100 feet
Closest storage > 100 feet
Don't know distance
Non-farm pesticide container
disposal in the last 3 years:
Containers disposed
No disposal
Don't know
Accidental spill of non-farm
pesticide in the last 3 years:
Non-farm pesticides spilled
No spill
Don't know
Property farmed in the last 5
years:
Yes
No
Don't know
Pesticides used
Not used
Don't know whether used
Farm pesticides stored for more
than 1 month in the last 5 years:
Farm pesticides stored
Not stored
Don't know whether stored
National Estimates for CWS Wells
Number (95% C.I.)
5,040 (±2,150)
88,500 (± 3,950)
1,050 (± 1,170)*
3,150 (± 1,860)
1,890 (±1,460)
0 N/A
190 (± 380)*
93,500 (± 3,660)
900 (± 680)
0 N/A
93,300 (± 3,780)
1,320 (± 890)
7,040 (±3,510)
87,100 (± 5,030)
430 (± 270)
2,420 (± 1,800)
2,020 (± 1,480)
2,600 (± 1,760)
640 (± 820)*
93,100 (±3,630)
890 (± 820)
Percent (95% C.l.)
5.3 (± 2.2)
93.6 (± 2.5)
1.1 (± 1.3)*
62.4 (±24.7)
37.6 (±24.7)
0.00 N/A
0.2 (± 0.4)*
98.9 (± 0.8)
1.0 (± 0.7)
0.0 N/A
98.6 (± 0.9)
1.4 (± 0.9)
7.4 (± 3.7)
92.1 (± 3.7)
0.5 (± 0.3)
34.4 (±18.0)
28.7 (±15.1)
36.9 (±21.7)
0.7 (± 0.9)*
98.4 (± 1.3)
0.9 (± 0.9)
-------
D-6 Appendix D: Questionnaire Data and Questionnaires
Exhibit D-1 (cont'd)
Estimated Number and Percent of
Community Water System Wells In the United States
Item
I.20
CWS
Main
C.7
1.21
CWS
Main
C.9
I.22
CWS
LAQ
#1
I.23
CWS
LAQ
#3
CWS
LAQ
#4
I.24
CWS
LAQ
#5
Questionnaire Item
Farm pesticide containers
disposed on the property in the
last 5 years:
Containers disposed
No disposal
Don't know
Accidental spill of farm pesticide
on the property in the last 5
years:
Farm pesticides spilled
No spill
Don't know whether spill
Crops farmed within 1/2 mile of
well in the last 3 years:
Yes
No
Don't know
Land used for pasture within 1/2
mile of well in the last 3 years:
Yes
No
Don't know
Pesticides Used
Not used
Don't know
Major non-agricultural pesticide
application within 1/2 mile of well
in the last 3 years:
Yes
No
Don't know
National Estimates for CWS Wells
Number (95% C.I.)
190 (± 380)*
92,900 (± 3,830)
1,490 (± 1,120)
150 (± 300)*
91,400 (±3,740)
3,030 (± 1,280)
38,200 (± 3,790)
48,800 (±4,610)
7,620 (±3,180)
38,800 (±4,310)
45,400 (± 4,500)
10,500 (±2,700)
21,700 (±3,120)
16,100 (±2,860)
200 (± 390)*
18,700 (± 2,860)
54,400 (± 4,670)
21,500 (±4,300)
Percent (95% C.I.)
0.2 (± 0.4)*
98.2 (± 1.3)
1.6 (± 1.2)
0.2 (± 0.3)*
96.6 (± 1.4)
3.2 (± 1.3)
40.3 (± 3.7)
51.6 (± 4.4)
8.1 (± 3.4)
41.0 (± 3.9)
48.0 (± 4.6)
11.1 (± 2.9)
57.2 (± 5.6)
42.3 (± 5.5)
1.5 (± 1.0)*
19.8 (± 2.9)
57.5 (± 4.4)
22.8 (± 4.5)
National Pesticide Survey: Phase I Report
-------
Appendix D: Questionnaire Data and Questionnaires D-7
Exhibit D-1 (cont'd)
Estimated Number and Percent of
Community Water System Wells In the United States
Item
(.25
CWS
LAQ
#7
I.26
CWS
LAQ
#9
127
CWS
LAQ
#11j
128
CWS
LAQ
#2
1.29
CWS
LAQ
#13
Questionnaire Item
Accidental spilt of any pesticide
within 1/2 mile of well in the last 3
years:
Yes
No
Don't know
Accidental spill of any hazardous
chemical within 1/2 mile of well in
the last 3 years:
Yes
No
Don't know
Golf course within 1/2 mile of well:
Yes
No
Don't know
River, canal, bay, spring, pond,
etc. within 1/2 mile of well:
Yes
No
Don't know
Irrigation within 1/2 mile of well:
Yes
No
Don't know
National Estimates for CWS Wells
Number (95% C.I.)
95 (± 190)*
68,400 (± 4,630)
26,100 (± 4,680)
1,340 (± 730)
64,600 (± 4,750)
28,700 (± 4,320)
8,990 (±2,210)
84,400 (± 3,790)
1,220 (± 870)
78,800 (± 4,390)
9,490 (±1,980)
6,290 (± 2,620)
23,700 (± 3,330)
63,600 (± 4,840)
7,340 (± 2,740)
Percent (95% C.I.)
0.1 (± 0.2)*
72.3 (± 4.6)
27.6 (± 4.6)
1.4 (± 0.8)
68.3 (± 4.5)
30.3 (± 4.4)
9.5 (± 2.3)
89.2 (± 2.6)
1.3 (± 0.9)
83.3 (± 3.2)
10.0 (± 2.2)
6.7 (± 2.7)
25.0 (± 3.6)
67.2 (± 4.2)
7.8 (± 2.8)
For particular responses to questionnaire items where the percent is extreme (i.e., near 0.0% or 100.0%) and the sample
size is relatively small, the 95% confidence intervals should be treated with caution. For further details see Section B6.4
of Appendix B.
N/A For particular responses to questionnaire items where the percent is equal to 0.0% or 100.0%, traditional 95% confidence
intervals based solely on sampling error are not applicable. For further details see Section 86.4 of Appendix 6.
-------
D-8 Appendix D: Questionnaire Data and Questionnaires
Exhibit D-2
Estimated Number and Percent of Community Water System Wells
Belonging to Counties with High Vulnerability
Item
11.1
cws
TLI
#8
II.2
CWS
TLI
#14
H.3
CWS
WOR
#1
II.4
CWS
WOR
#4
II.5
CWS
WOR
#3
Questionnaire Item
Number of wells that belong to
systems with:
One working well
Two working wells
3-5 working wells
6-1 0 working wells
> 1 0 working wells
Septic system within 300 feet of
well:
Septic tank
Septic field
Cesspool
Topographic setting at well:
Hilltop
Hillside
Flat valley
Depression
Plateau
Other
Don't know
Description of land area within
300 feet of well:
Roofed
Paved
Graveled
Exposed or vegetated soil
Other
Well protection at surface:
Well house or shed
Concrete pad
Sanitary or grouted seal
Covered pit
Other
Estimates for CWS Wells In
Counties with High Vulnerability
Number (95% C.I.)
3,160 (±1,130)
5,690 (± 970)
5,300 (±1,940)
2,470 (±1,000)
4,180 (± 890)
5,230 (±1,310)
4,350 (±1,040)
190 (± 370)*
1,540 (± 640)
2,110 (±1,000)
6,130 (± 1,010)
700 (± 570)
2,010 (±1,050)
8,210 (±1,770)
100 (± 230)*
1,590 (± 630)
1,240 (± 380)
650 (± 700)*
14,900 (± 1,220)
2,440 (± 890)
12,100 (± 1,370)
17,300 (±1,080)
14,700 (±1,830)
1,900 (± 540)
3,080 (±1,110)
Percent (95% C.I.)
15.2 (± 5.7)
27.3 (± 4.0)
25.5 (± 9.4)
11.9 (± 4.7)
20.1 (± 3.8)
25.1 (± 6.4)
20.9 (± 5.1)
0.9 (± 1.8)*
7.4 (± 3.1)
10.2 (± 4.8)
29.5 (± 5.4)
3.4 (± 2.7)
9.7 (± 5.1)
39.5 (± 7.5)
0.5 (± 1.1)*
7.7 (± 3.0)
6.0 (± 1.9)
3.1 (± 3.4)*
71.5 (± 4.8)
11.8 (± 4.2)
58.2 (± 6.2)
83.3 (± 3.6)
71.0 (± 7.4)
9.2 (± 2.8)
14.8 (± 5.1)
National Pesticide Survey: Phase I Report
-------
Appendix D: Questionnaire Data and Questionnaires D-9
Exhibit D-2 (cont'd)
Estimated Number and Percent of Community Water System Wells
Belonging to Counties with High Vulnerability
Kern
11.6
CWS
Main
A.1
II.7
CWS
Main
A.2
II.8
CWS
Main
A. 5
II.9
CWS
Main
A. 7
CWS
Main
A.11
CWS
Main
A.13
CWS
Main
A.18
Questionnaire Item
Age of well:
< 5 years
5 to 10 years
11 to 20 years
> 20 years
Don't know
Well redrilled:
Yes
No
Don't know
Well depth:
< 20 feet
20 to 50 feet
51 to 100 feet
101 to 200 feet
201 to 500 feet
> 500 feet
Don't know
Well casing information:
Well cased
Well not cased
Don't know casing
Cased to total depth
Cased < total depth
Don't know casing depth
Screens in casing
No screens in casing
Don't know screening
Well grouted
Well not grouted
Don't know grouting
Estimates for CWS Wells in
High Vulnerability Counties
Number (95% C.I.)
1,150 (± 410)
3,460 (± 1,440)
7,370 (± 1,800)
6,020 (±1,110)
2,790 (± 1,260)
870 (± 630)
19,000 (± 1,410)
910 (± 950)*
260 (± 510)*
540 (± 710)*
4,270 (± 1,330)
4,530 (± 1,340)
6,200 (± 780)
3,870 (± 840)
1,130 (± 580)
20,600 (±1,120)*
110 (± 210)*
100 (± 200)*
8,070 (±1,460)
9,550 (±1,700)
2,970 (± 1,310)
10,500 (± 1,160)
6,730 (± 1,840)
3,380 (± 1,380)
14,700 (±2,060)
1,950 (± 750)
3,940 (± 1,070)
Percent (95% C.I.)
5.5 (± 2.1)
16.6 (± 6.7)
35.5 (± 8.1)
29.0 (± 4.9)
13.4 (± 6.3)
4.2 (± 3.0)
91.4 (± 5.1)
4.4 (± 4.6)*
1.2 (± 2.5)*
2.6 (± 3.4)*
20.5 (± 6.2)
21.8 (± 6.7)
29.8 (± 3.6)
18.6 (± 3.8)
5.4 (± 2.9)
99.0 (± 1.4)*
0.5 (± 1.0)*
0.5 (± 1.0)*
39.2 (± 7.5)
46.4 (± 6.7)
14.4 (± 6.5)
50.9 (± 4.5)
32.7 (± 7.9)
16.4 (± 7.3)
71.7 (± 7.7)
9.5 (± 3.8)
19.2 (± 5.5)
-------
D-10 Appendix D: Questionnaire Data and Questionnaires
Exhibit D-2 (cont'd)
Estimated Number and Percent of Community Water System Wells
Belonging to Counties with High Vulnerability
Item
11.10
cws
MAin
A.21
11.11
CWS
Main
A.23
11.12
CWS
Main
A.25
11.13
CWS
Main
A.37
11.14
CWS
Main
B.1
CWS
Main
B.3
Questionnaire Item
Well capped at ground surface:
Well capped
Well not capped
Don't know if capped
Distance from the ground surface
to the water surface in the well:
< 20 feet
20 to 50 feet
51 to 100 feet
101 to 200 feet
201 to 500 feet
> 500 feet
Don't know
Aquifer:
Confined
Unconfined
Don't know
Abandoned or non-operating
wells within 500 feet of the well:
Yes
No
Don't know
Non-farm pesticides used in the
last 3 years:
Non-farm pesticides used
Not used
Don't know whether used
Used < 100 feet from well
Used > 100 feet from well
Don't know distance
Estimates for CWS Wells In
High Vulnerability Counties
Number (95% C.I.)
16,600 (± 1,170)
4,050 (±1,050)
110 (± 190)*
4,900 (±1,620)
5,230 (± 1,230)
2,650 (± 960)
1,380 (± 660)
370 (± 430)*
0 N/A
6,280 (± 1 ,470)
9,170 (±1,650)
6,860 (±1,860)
5,280 (±1,590)
5,330 (± 1,630)
14,300 (±1,940)
1,210 (± 670)
5,450 (±1,450)
14,600 (± 1,190)
740 (± 590)
3,710 (± 1,330)
1,560 (± 890)
180 (± 100)
Percent (95% C.I.)
80.0 (± 4.9)
19.5 (± 4.7)
0.5 (± 0.9)*
23.5 (± 7.6)
25.2 (± 5.6)
12.8 (± 4.3)
6.6 (± 3.1)
1.8 (± 2.0)*
0.0 N/A
30.2 (± 7.9)
44.1 (± 7.8)
33.0 (± 7.8)
25.4 (± 8.2)
25.6 (± 7.9)
65.8 (± 8.1)
5.8 (± 3.2)
26.2 (± 6.2)
70.3 (± 6.2)
3.5 (± 2.9)
68.1 (±14.6)
28.6 (±14.8)
3.3 (± 1.9)
National Pesticide Survey: Phase I Report
-------
Appendix D: Questionnaire Data and Questionnaires D-11
Exhibit D-2 (cont'd)
Estimated Number and Percent of Community Water System Wells
Belonging to Counties with High Vulnerability
Item
11.15
CWS
Main
B.4
CWS
Main
B.5
11.16
CWS
Main
B.6
11.17
CWS
Main
B.8
11.18
CWS
Main
C.1
CWS
Main
C.2
11.19
CWS
Main
C.5
Questionnaire Item
Non-farm pesticides stored for
more than 1 month in the last 3
years:
Non-farm pesticides stored
Not stored
Don't know whether stored
Closest Storage < 1 00 feet
Closest Storage > 100 feet
Don't know distance
Non-farm pesticide container
disposal in the last 3 years:
Containers disposed
No disposal
Don't know
Accidental spill of non-farm
pesticide in the last 3 years:
Non-farm pesticides spilled
No spill
Don't know
Property farmed in the last 5
years:
Yes
No
Don't know
Pesticides used
Not used
Don't know whether used
Farm pesticides stored for more
than 1 month in the last 5 years:
Farm pesticides stored
Not stored
Don't know whether stored
Estimates for CWS Wells in
High Vulnerability Counties
Number (95% C.I.)
1,380 (± 1,370)
19,000 (± 1,570)
420 (± 390)
400 (± 580)*
980 (±1,170)*
0 N/A
0 N/A
20,300 (± 1,240)
520 (± 440)
0 N/A
20,600 (± 970)*
240 (± 340)*
690 (± 490)
19,800 (± 970)
360 (± 230)
390 (± 520)*
200 (± 250)*
100 (± 200)*
260 (± 370)*
20,100 (± 1,240)*
410 (± 560)*
Percent (95% C.I.)
6.6 (± 6.5)
91.4 (± 6.3)
2.0 (± 1.9)
29.1 (±34.1)*
70.9 (±34.1)*
0.0 N/A
0.0 N/A
97.5 (± 2.2)
2.5 (± 2.2)
0.0 N/A
98.9 (± 1.6)*
1.2 (± 1.6)*
3.3 (± 2.3)
95.0 (± 2.6)
1.7 (± 1.1)
56.4 (±33.2)*
28.9 (±39.9)*
14.7 (±30.0)*
1.3 (± 1.8)*
96.8 (± 4.0)*
2.0 (± 2.7)*
National
-------
D-12 Appendix D: Questionnaire Data and Questionnaires
Exhibit D-2 (cont'd)
Estimated Number and Percent of Community Water System Wells
Belonging to Counties with High Vulnerability
Hem
II.20
CWS
Main
C.7
11.21
CWS
Main
C.9
II. 22
CWS
LAG
#1
H.23
CWS
LAG
#3
CWS
LAQ
#4
II.24
CWS
LAQ
#5
Questionnaire Item
Farm pesticide container disposal
on the property in the last 5
years:
Containers disposed
No disposal
Don't know
Accidental spill of farm pesticides
on the property in the last 5
years:
Farm pesticides spilled
No spill
Don't know whether spill
Crops farmed within 1/2 mile of
well in the last 3 years:
Yes
No
Don't know
Land used for pasture within 1/2
mile of well in the last 3 years:
Yes
No
Don't know
Pesticides Used
Not used
Don't know
Major non-agricultural pesticide
application within 1/2 mile of well
in the last 3 years:
Yes
No
Don't know
Estimates for CWS Wells In
High Vulnerability Counties
Number (95% C.I.)
0 N/A
20,200 (±1,120)
610 (± 560)
150 (± 300)*
19,900 (±1,220)
710 (± 590)
8,250 (± 980)
11,300 (± 740)
1,220 (± 690)
4,090 (± 600)
14,700 (± 1,180)
2,010 (± 940)
2,410 (± 820)
1,680 (± 860)
0 N/A
5,870 (±1,170)
8,770 (±1,500)
6,170 (± 1,890)
Percent (95% C.I.)
0.0 N/A
97.1 (± 2.7)
2.9 (± 2.7)
0.7 (± 1.4)*
95.9 (± 3.2)
3.4 (± 2.9)
39.6 (± 3.6)
54.5 (± 4.5)
5.9 (± 3.2)
19.7 (± 2.4)
70.7 (± 5.4)
9.7 (± 4.5)
58.9 (±19.3)
41.1 (±19.3)
0.0 N/A
28.2 (± 5.6)
42.2 (± 7.4)
29.6 (± 8.4)
National Pesticide Survey: Phase I Report
-------
Appendix D: Questionnaire Data and Questionnaires D-13
Exhibit D-2 (cont'd)
Estimated Number and Percent of Community Water System Wells
Belonging to Counties with High Vulnerability
Item
II.25
CWS
LAQ
#7
II.26
CWS
LAQ
#9
II.27
CWS
LAQ
#11j
II. 28
CWS
LAQ
#2
II.29
CWS
LAQ
#13
Questionnaire Item
Accidental spill of any pesticide
within 1/2 mile of well in the last 3
years:
Yes
No
Don't know
Accidental spill of any hazardous
chemical within 1/2 mile of well in
the last 3 years:
Yes
No
Don't know
Golf course within 1/2 mile of well:
Yes
No
Don't know
River, canal, bay, spring, pond,
etc. within 1/2 mile of well:
Yes
No
Don't know
Irrigation within 1/2 mile of well:
Yes
No
Don't know
Estimates for CWS Wells In
High Vulnerability Counties
Number (95% C.I.)
95 (± 190)*
14,100 (± 1,060)
6,560 (± 1,390)
600 (± 400)
12,700 (± 1,960)
7,540 (± 1,800)
2,190 (± 840)
18,200 (± 1,020)
390 (± 290)
17,500 (± 1,050)
1,920 (± 630)
1,370 (± 420)
7,560 (± 1,580)
11,500 (±1,690)
1,760 (± 690)
Percent (95% C.I.)
0.5 (± 0.9)*
68.0 (± 5.2)
31.5 (± 5.9)
2.9 (± 1.9)
60.9 (± 9.0)
36.3 (± 8.6)
10.5 (± 3.9)
87.6 (± 4.3)
1.9 (± 1.4)
84.2 (± 4.7)
9.3 (± 2.9)
6.6 (± 2.0)
36.4 (± 7.0)
55.2 (± 7.9)
8.4 (± 3.4)
For particular responses to questionnaire items where the percent is extreme (i.e., near 0.0% or 100.0% and the sample
size is relatively small), the 95% confidence intervals should be treated with caution. For further details see Section B6.4
of Appendix B.
N/A For particular responses to questionnaire Hems where the percent is equal to 0.0% or 100.0%, traditional 95% confidence
intervals based solely on sampling error are not applicable. For further details see Section B6.4 of Appendix B.
-------
D-14 Appendix D: Questionnaire Data and Questionnaires
Exhibit D-3
Estimated Number and Percent of Rural Domestic Wells
In the United States
Item
1.1
DWS
TLI
#6
I.2
DWS
WOR
#1
I.3
DWS
WOR
#4
I.4
DWS
WOR
#3
I.5
DWS
Main
D.3
Questionnaire Item
Septic system on the property:
Septic tank
Septic field
Cesspool
Topographic setting at well:
Hilltop
Hillside
Flat valley
Depression
Plateau
Other
Don't know
Description of land area within
300 feet of well:
Roofed
Paved
Graveled
Exposed or vegetated soil
Other
Well protection at surface:
Well house or shed
Concrete pad
Covered pit
Other
Age of well:
< 5 years
5 to 1 0 years
11 to 20 years
> 20 years
Don't know
National Estimates for
Rural Domestic Wells
Number (95% C.I.)
9,730,000 (± 355,000)
9,050,000 (± 470,000)
853,000 (± 347,000)
1,750,000 (± 468,000)
3,520,000 (± 611,000)
3,640,000 (± 793,000)
99,300 (± 108,000)*
298,000 (± 206,000)
1,190,000 (± 493,000)
0 N/A
222,000 (± 202,000)
0 N/A
69,700 (± 94,000)*
9,570,000 (± 523,000)
646,000 (± 475,000)
3,240,000 (± 550,000)
2,170,000 (± 539,000)
1,350,000 (± 522,000)
5,570,000 (± 672,000)
1,640,000 (± 389,000)
1,710,000 (± 420,000)
3,070,000 (± 450,000)
2,520,000 (± 501,000)
1,570,000 (± 398,000)
Percent (95% C.I.)
92.6 (± 3.4)
86.1 (± 4.5)
8.1 (± 3.3)
16.7 (± 4.5)
33.5 (± 5.8)
34.6 (± 7.6)
1.0 (± 1.0)*
2.8 (± 2.0)
11.4 (± 4.7)
0.0 N/A
2.1 (± 1.9)
0.0 N/A
0.7 (± 0.9)*
91.1 (± 5.0)
6.2 (± 4.5)
30.9 (± 5.2)
20.6 (± 5.1)
12.9 (± 5.0)
53.0 (± 6.4)
15.6 (± 3.7)
16.3 (± 4.0)
29.3 (± 4.3)
24.0 (± 4.8)
14.9 (± 3.8)
National Pesticide Survey: Phase I Report
-------
Appendix D: Questionnaire Data and Questionnaires D-15
Exhibit D-3 (cont'd)
Estimated Number and Percent of Rural Domestic Weiis
In the United States
Item
I.6
DWS
Main
D.4
I.7
DWS
Main
D.6
I.8
DWS
Main
D.8
DWS
Main
D.10
I.9
DWS
WOR
#2
1.10
DWS
Main
D.17
Questionnaire Item
Well redrilled:
Yes
No
Don't know
Well depth:
< 20 feet
20 to 50 feet
51 to 100 feet
101 to 200 feet
201 to 500 feet
> 500 feet
Don't know
Well casing information:
Well cased
Not cased
Don't know casing
Cased to total depth
Cased < total depth
Don't know casing depth
Well closed at ground surface:
Yes
No
Don't know
Aquifer:
Single
Double
Don't know
National Estimates for
Rural Domestic Wells
Number (95% C.I.)
534,000 (± 201,000)
9,210,000 (± 403,000)
764,000 (± 336,000)
385,000 (± 226,000)
1,160,000 (±306,000)
2,170,000 (± 335,000)
2,860,000 (± 475,000)
1,930,000 (± 398,000)
424,000 (± 340,000)
1,580,000 (± 361,000)
9,010,000 (± 390,000)
491,000 (± 246,000)
1,000,000 (± 316,000)
4,280,000 (± 604,000)
2,990,000 (± 490,000)
1,740,000 (± 344,000)
8,380,000 (± 544,000)
1,190,000 (± 486,000)
940,000 (± 385,000)
2,750,000 (± 474,000)
4,390,000 (± 563,000)
3,380,000 (± 552,000)
Percent (95% C.I.)
5.1 (± 1.9)
87.7 (± 3.8)
7.3 (± 3.2)
3.7 (± 2.2)
11.0 (± 2.9)
20.7 (± 3.2)
27.2 (± 4.5)
18.4 (± 3.8)
4.0 (± 3.2)
15.1 (± 3.4)
85.8 (± 3.7)
4.7 (± 2.3)
9.6 (± 3.0)
47.5 (± 6.4)
33.2 (± 5.1)
19.3 (± 3.8)
79.8 (± 5.2)
11.3 (± 4.6)
8.9 (± 3.7)
26.1 (± 4.5)
41.7 (± 5.4)
32.1 (± 5.3)
-------
D-16 Appendix D: Questionnaire Data and Questionnaires
Exhibit D-3 (cont'd)
Estimated Number and Percent of Rural Domestic Wells
In the United States
Item
1.11
DWS
Main
B.14
1.12
DWS
Main
B.1
1.13
DWS
Main
B.2
1.14
DWS
Main
B.3
1.15
DWS
Main
B.4
DWS
Main
B.5
Questionnaire Item
Non-operating wells within 500
feet of the well:
Yes
No
Don't know
Non-farm pesticides used in the
last 3 years:
Used inside house
Not used inside house
Don't know
Non-farm pesticides used in last 3
years:
Used on lawn
Not used on lawn
Don't know
Non-farm pesticides used in last 3
years:
Used in garden
Not used in garden
Don't know
Non-farm pesticides stored for
more than 1 month in the last 3
years:
Non-farm pesticides stored
Not stored
Don't know whether stored
Stored < 100 feet from well
Stored > 100 feet from well
Don't know where stored
National Estimates for
Rural Domestic Wells
Number (95% C.I.)
2,950,000 (± 490,000)
7,000,000 (± 517,000)
559,000 (± 259,000)
7,950,000 (± 539,000)
2,540,000 (± 537,000)
18,200 (± 37,100)*
4,200,000 (± 675,000)
6,270,000 (± 662,000)
35,700 (± 57,500)*
4,400,000 (± 461,000)
6,060,000 (± 471 ,000)
44,800 (± 67,800)*
7,120,000 (± 613,000)
3,380,000 (± 614,000)
9,110 (± 18,500)*
5,120,000 (± 556,000)
1,890,000 (± 352,000)
103,000 (± 103,000)*
Percent (95% C.I.)
28.1 (± 4.7)
66.6 (± 4.9)
5.3 (± 2.5)
75.7 (± 5.1)
24.1 (± 5.1)
0.2 (± 0.4)*
40.0 (± 6.4)
59.7 (± 6.3)
0.3 (± 0.6)*
41.9 (± 4.4)
57.7 (± 4.5)
0.4 (± 0.6)*
67.7 (± 5.8)
32.2 (± 5.8)
0.1 (± 0.2)*
72.0 (± 4.5)
26.6 (± 4.4)
1.4 (± 1.5)*
National Pesticide Survey: Phase I Report
-------
Appendix 0: Questionnaire Data and Questionnaires D-17
Exhibit D-3 (cont'd)
Estimated Number and Percent of Rural Domestic Weils
In the United States
Item
1.16
DWS
Main
B.6
DWS
Main
B.7
1.17
DWS
Main
B.17
and
B.18
DWS
Main
C.6
1.18
DWS
Main
C.11
and
B.20
Questionnaire Item
Non-farm pesticide container
disposal in the last 3 years:
Containers disposed
No disposal
Don't know
< 100 feet from well
> 1 00 feet from well
Don't know how far
Property farmed:!
Yes
No
Pesticides used for farming in the
fast 5 years:
Pesticides used
Not used
Don't know whether used
Farm pesticides stored for more
than 1 month in the last 5 years:
Farm pesticides stored
Not stored
Don't know whether stored
National Estimates for
Rural Domestic Weils
Number (95% C.I.)
642,000 (± 247,000)
9,860,000 (± 251,000)
9,110 (± 18,500)*
285,000 (± 144,000)
357,000 (± 190,000)
0 N/A
1,230,000
9,280,000
1,020,000 (± 120,000)
200,000 (± 105,000)
0 N/A
473,000 (± 153,000)
10,000,000 (± 172,000)
34,900 (± 40,800)*
Percent (95% C.I.)
6.1 (± 2.4)
93.8 (± 2.4)
0.1 (± 0.2)*
44.4 (±18.5)
55.6 (±18.5)
0.0 N/A
11.7
88.3
83.6 (± 8.7)
16.4 (± 8.7)
0.0 N/A
4.5 (± 1.4)
95.2 (± 1.6)
0.3 (± 0.4)*
-------
D-18 Appendix D: Questionnaire Data and Questionnaires
Exhibit D-3 (cont'd)
Estimated Number and Percent of Rural Domestic Wells
In the United States
Hem
1.19
DWS
Main
C.14
and
B.22
DWS
Main
C.16
and
B.23
I.20
DWS
Main
C.17
and
B.24
1.21
DWS
LAG
#1
I.22
DWS
LAQ
#3
Questionnaire Item
Farm pesticide container disposal
in the last 5 years:
Containers disposed
No disposal
Don't know
< 1 00 feet from well
> 100 feet from well
Don't know how far
Accidental spill of farm pesticide
in the last 5 years:
Farm pesticides spilled
No spill
Don't know whether spilled
Crops farmed within 1/2 mile of
well in the last 3 years:
Yes
No
Don't know
Pasture land within 1/2 mile of
well in the last 3 years:
Yes
No
Don't know
National Estimates for
Rural Domestic Wells
Number (95% C.I.)
337,000 (± 138,000)
1,070,000 (± 167,000)
0 N/A
22,500 (± 26,100)
311,000 (± 141,000)
3,650 (± 5,520)*
33,700 (± 46,300)*
1,370,000 (± 139,000)
0 N/A
6,950,000 (± 800,000)
3,300,000 (± 753,000)
258,000 (± 213,000)
6,870,000 (± 788,000)
3,430,000 (± 778,000)
203,000 (± 274,000)*
Percent (95% C.I.)
24.0 (± 9.6)
76.0 (± 9.6)
0.0 N/A
6.7 (± 8.3)
92.2 (± 8.6)
1.1 (± 1.7)*
2.4 (± 3.3)*
97.6 (± 3.3)
0.0 N/A
66.2 (± 7.6)
31.4 (± 7.2)
2.5 (± 2.0)
65.4 (± 7.5)
32.7 (± 7.4)
1.9 (± 2.6)*
National Pesticide Survey: Phase I Report
-------
Appendix D: Questionnaire Data and Questionnaires D-19
Exhibit D-3 (cont'd)
Estimated Number and Percent of Rural Domestic Wells
In the United States
Item
I.23
DWS
LAQ
#4
(I.22 tub-
tern)
I.24
DWS
LAQ
#5
I.25
DWS
LAQ
#7
(.26
DWS
LAQ
#9
I.27
DWS
LAQ
#11]
Questionnaire Item
Pesticides used on pasture land
within 1/2 mile of well in the last 3
years:
Yes
No
Don't know
Major non-agricultural pesticide
application within 1/2 mile of well
in the last 3 years:
Yes
No
Don't know
Accidental spill of any pesticide
wrthin 1/2 mile of well in the last 3
years:
Yes
No
Don't know
Accidental spill of any hazardous
chemical within 1/2 mile of well in
the last 3 years:
Yes
No
Don't know
Golf course within 1/2 mile of well:
Yes
No
Don't know
National Estimates for
Rural Domestic Wells
Number (95% C.I.)
2,820,000 (± 736,000)
4,050,000 (± 799,000)
0 N/A
1,460,000 (± 500,000)
8,580,000 (± 559,000)
474,000 (± 322,000)
10,100 (± 14,500)*
1,010,000 (±383,000)*
362,000 (± 382,000)*
17,200 (± 28,200)*
10,100,000 (± 377,000)
431,000 (± 378,000)
523,000 (± 259,000)
9,820,000 (± 327,000)
165,000 (± 230,000)*
Percent (95% C.I.)
41.1 (± 9.6)
58.9 (± 9.6)
0.0 N/A
13.9 (± 4.8)
81.6 (± 5.3)
4.5 (± 3.1)
0.1 (± 0.1)*
96.5 (± 3.6)*
3.5 (± 3.6)*
0.2 (± 0.3)*
95.7 (± 3.6)
4.1 (± 3.6)
5.0 (± 2.5)
93.5 (± 3.1)
1.6 (± 2.2)*
National Pesticide Survev: Phase I Renort
-------
D-20 Appendix D: Questionnaire Data and Questionnaires
Exhibit D-3 (cont'd)
Estimated Number and Percent of Rural Domestic Wells
In the United States
item
1.28
DWS
UQ
#12
I.29
DWS
IAQ
#13
Questionnaire Item
River, canal, bay, spring, pond,
etc. within 1/2 mile of well:
Yes
No
Don't know
Irrigation within 1/2 mile of well:
Yes
No
Don't know
National Estimates for
Rural Domestic Wells
Number (95% C.I.)
9,860,000 (± 363,000)
567,000 (± 350,000)
82,000 (± 98,000)*
1,910,000 (± 564,000)
8,550,000 (± 582,000)
49,500 (± 60,300)*
Percent (95% C.I.)
93.8 (± 3.5)
5.4 (± 3.3)
0.8 (± 0.9)*
18.1 (± 5.4)
81.4 (± 5.5)
0.5 (± 0.6)*
* For particular responses to questionnaire items where the percent is extreme (i.e., near 0.0% or 100.0% and the sample
size is relatively small), the 95% confidence intervals should be treated with caution. For further details see Section B6.4
of Appendix B.
N/A For particular responses to questionnaire items where the percent is equal to 0.0% or 100.0%, traditional 95% confidence
intervals based solely on sampling error are not applicable. For further details see Section B6.4 of Appendix B.
t The total number of rural domestic wells (10,509,373) and the subpopulation corresponding to rural domestic wells on
farm land (1,219,841) were estimated from the 1987 American Housing Survey administered by the U.S. Bureau of the
Census. They are treated as constants in the Survey, and hence, have no associated confidence intervals.
National Pesticide Survey: Phase I Report
-------
Appendix D: Questionnaire Data and Questionnaires D-21
Exhibit D-4
Estimated Number and Percent of Rural Domestic Welis In
Counties with High Pesticide Use
Kern
11.1
DWS
TLI
#6
II.2
DWS
WOR
#1
II.3
DWS
WOR
#4
II.4
DWS
WOR
#3
I1.5
DWS
Main
D.3
Questionnaire Item
Septic system on the property:
Septic tank
Septic field
Cesspool
Topographic setting at well:
Hilltop
Hillside
Flat valley
Depression
Plateau
Other
Don't know
Description of land area within
300 feet of well:
Roofed
Paved
Graveled
Exposed or vegetated soil
Other
Well protection at surface:
Well house or shed
Concrete pad
Covered pit
Other
Age of well:
< 5 years
5 to 10 years
11 to 20 years
> 20 years
Don't know
Estimates for Rural Domestic Wells in
Counties with High Pesticide Use
Number (95% C.I.)
1,030,000 (± 273,000)
943,000 (± 277,000)
96,400 (± 84,300)
181,000 (± 122,000)
155,000 (± 130,000)
568,000 (± 294,000)
14,600 (± 18,500)*
68,000 (± 80,700)*
169,000 (± 163,000)
0 N/A
8,190 (± 11,700)*
0 N/A
4,880 (± 9,720)*
983,000 (± 321,000)*
160,000 (± 291,000)*
183,000 (± 136,000)
65,100 (± 57,400)
198,000 (± 117,000)
732,000 (± 356,000)
140,000 (± 79,200)
141,000 (± 64,300)
385,000 (± 174,000)
275,000 (± 110,000)
216,000 (± 90,600)
Percent (95% C.I.)
89.2 (±10.3)
81.6 (± 8.0)
8.3 (± 6.7)
15.6 (± 9.2)
13.5 (± 9.8)
49.2 (±23.8)
1.3 (± 1.6)*
5.9 (± 6.5)*
14.6 (±14.2)
0.0 N/A
0.7 (± 1.0)*
0.0 N/A
0.4 (± 0.9)*
85.0 (±24.6)*
13.9 (±24.7)*
15.8 (±12.9)
5.6 (± 5.1)
17.2 (±10.2)
63.3 (±18.8)
12.1 (± 5.9)
12.2 (± 4.7)
33.3 (± 9.0)
23.7 (±10.0)
18.7 (± 5.0)
National P*»etlHri»
-------
D-22 Appendix D: Questionnaire Data and Questionnaires
Exhibit D-4 (cont'd)
Estimated Number and Percent of Rural Domestic Wells In
Counties with High Pesticide Use
Item
11.6
DWS
Main
D.4
11.7
DWS
Main
D.6
11.8
DWS
Main
D.8
DWS
Main
D.10
II. 9
DWS
WOR
#2
11.10
DWS
Main
D.17
Questionnaire Item
Well redrilled:
Yes
No
Don't know
Well depth:
< 20 feet
20 to 50 feet
51 to 100 feet
101 to 200 feet
201 to 500 feet
> 500 feet
Don't know
Well casing information:
Well cased
Not cased
Don't know casing
Cased to total depth
Cased < total depth
Don't know casing depth
Well closed at ground surface:
Yes
No
Don't know
Aquifer:
Single
Double
Don't know
Estimates for Rural Domestic Wells In
Counties with High Pesticide Use
Number (95% C.I.)
54,800 (± 29,700)
1,020,000 (± 300,000)
81,000 (± 53,500)
25,300 (± 28,400)*
181,000 (± 109,000)
190,000 (± 88,700)
277,000 (± 108,000)
243,000 (± 171,000)
29,100 (± 28,100)*
211,000 (± 131,000)
1,010,000 (± 303,000)
30,300 (± 26,600)
114,000 (± 90,800)
613,000 (± 195,000)
154,000 (± 128,000)
244,000 (± 113,000)
1,010,000 (± 321,000)
57,400 (± 77,100)*
93,100 (± 76,100)
348,000 (± 145,000)
407,000 (± 126,000)
401,000 (± 175,000)
Percent (95% C.I.)
4.7 (± 2.6)
88.3 (± 4.6)
7.0 (± 4.2)
2.2 (± 2.4)*
15.7 (± 9.1)
16.4 (± 6.9)
24.0 (± 7.3)
21.1 (±12.7)
2.5 (± 2.5)*
18.2 (± 9.7)
87.5 (± 8.0)
2,6 (± 2.4)
9.9 (± 7.3)
60.6 (±14.3)
15.2 (±10.7)
24.2 (± 8.2)
87.0 (± 9.7)
5.0 (± 6.8)*
8.1 (± 6.3)
30.1 (± 8.4)
35.2 (± 7.5)
34.7 (±11.1)
National Pesticide Survey: Phase I Report
-------
Appendix D: Questionnaire Data and Questionnaires D-23
Exhibit D-4 (cont'd)
Estimated Number and Percent of Rural Domestic Wells In
Counties with High Pesticide Use
Item
11.11
DWS
Main
B.14
11.12
DWS
Main
B.1
11.13
DWS
Main
B.2
11.14
DWS
Main
B.3
11.15
DWS
Main
B.4
DWS
Main
B.5
Questionnaire Item
Non-operating wells within 500
feet of the well:
Yes
No
Don't know
Non-farm pesticides used in the
last 3 years:
Used inside house
Not used inside house
Don't know
Non-farm pesticides used in the
last 3 years:
Used on lawn
Not used on lawn
Don't know
Non-farm pesticides used in the
last 3 years:
Used in garden
Not used in garden
Don't know
Non-farm pesticides stored for
more than 1 month in the last 3
years:
Non-farm pesticides stored
Not stored
Don't know whether stored
Stored < 100 feet from well
Stored > 100 feet from well
Don't know where stored
Estimates for Rural Domestic Wells In
Counties with High Pesticide Use
Number (95% C.I.)
269,000 (± 113,000)
801,000 (± 242,000)
85,900 (± 89,200)*
946,000 (± 277,000)
210,000 (± 95,000)
0 N/A
511,000 (± 186,000)
645,000 (± 204,000)
0 N/A
494,000 (± 156,000)
663,000 (± 228,000)
0 N/A
829,000 (± 274,000)
328,000 (± 150,000)
0 N/A
526,000 (± 190,000)
295,000 (± 152,000)
7,190 (± 14,400)*
Percent (05% C.I.)
23.3 (± 7.9)
69.3 (± 8.2)
7.4 (± 7.3)*
81.8 (± 6.5)
18.2 (± 6.5)
0.0 N/A
44.2 (± 9.3)
55.8 (± 9.3)
0.0 N/A
42.7 (± 8.7)
57.3 (± 8.7)
0.0 N/A
71.7 (±11.0)
28.3 (±11.0)
0.0 N/A
63.5 (±12.5)
35.7 (± 12.7)
0.9 (± 1.7)*
-------
D-24 Appendix D: Questionnaire Data and Questionnaires
Exhibit D-4 (cont'd)
Estimated Number and Percent of Rural Domestic Wells In
Counties with High Pesticide Use
Kern
11.16
DWS
Main
B.6
DWS
Main
B.7
11.17
DWS
Main
B.17
and
B.18
DWS
Main
C.6
11.18
DWS
Main
C.11
and
B.20
Questionnaire Item
Non-farm pesticide container
disposal in the last 3 years:
Containers disposed
No disposal
Don't know
< 100 feet from well
> 100 feet from well
Don't know how far
Property farmed:
Yes
No
Pesticides used for farming in the
last 5 years:
Pesticides used
Not used
Don't know whether used
Farm pesticides stored for more
than 1 month in the last 5 years:
Farm pesticides stored
Not stored
Don't know whether stored
Estimates for Rural Domestic Wells In
Counties with High Pesticide Use
Number (95% C.I.)
173,000 (± 94,000)
984,000 (± 298,000)
0 N/A
63,400 (± 60,000)
109,000 (± 59,700)
0 N/A
287,000 (± 97,600)
870,000 (± 322,000)
251,000 (± 95,100)
36,100 (± 30,600)
0 N/A
71,900 (± 41,600)
1,080,000 (± 323,000)
0 N/A
Percent (95% C.I.)
14.9 (± 7.4)
85.1 (± 7.4)
0.0 N/A
36.8 (±25.3)
63.2 (±25.3)
0.0 N/A
24.80 (± 9.80)
75.20 (± 9.80)
87.4 (±10.8)
12.6 (±10.8)
0.0 N/A
6.2 (± 3.7)
93.8 (± 3.7)
0.0 N/A
National Pesticide Survey: Phase I Report
-------
Appendix D: Questionnaire Data and Questionnaires D-25
Exhibit D-4 (cont'd)
Estimated Number and Percent of Rural Domestic Wells In
Counties with High Pesticide Use
Item
11.19
DWS
Main
C.14
and
B.22
DWS
Main
C.16
and
B.23
II.20
DWS
Main
C.17
and
B.24
11.21
DWS
LAQ
#1
\\.22
DWS
LAQ
#3
Questionnaire Item
Farm pesticide container disposal
in the last 5 years:
Containers disposed
No disposal
Don't know
< 100 feet from well
> 1 00 feet from well
Don't know how far
Accidental spill of farm pesticide
in the last 5 years:
Farm pesticides spilled
No spill
Don't know whether spilled
Crops farmed within 1/2 mile of
well in the last 3 years:
Yes
No
Don't know
Pasture land within 1/2 mile of
well in the last 3 years:
Yes
No
Don't know
Estimates for Rural Domestic Wells In
Counties with High Pesticide Use
Number (95% C.I.)
96,500 (± 44,900)
208,000 (± 72,400)
0 N/A
11,400 (± 14,300)*
81,500 (± 45,300)*
3,650 (± 5,520)*
13,600 (± 20,100)*
291,000 (± 105,000)*
0 N/A
1,040,000 (± 308,000)
116,000 (± 97,100)
2,140 (± 4,290)*
664,000 (± 185,000)
469,000 (± 251,000)
22,800 (± 45,700)*
Percent (95% C.I.)
31.7 (± 7.9)
68.3 (± 7.9)
0.0 N/A
11.8 (±15.2)*
84.4 (±16.2)*
3.8 (± 6.0)*
4.5 (± 6.6)*
95.5 (± 6.6)*
0.0 N/A
89.8 (± 8.1)
10.0 (± 8.1)
0.2 (± 0.4)*
57.4 (±14.0)
40.6 (±14.5)
2.0 (± 3.9)*
National Pesticide Survev: Phas* i n«m/»r+
-------
D-26 Appendix D: Questionnaire Data and Questionnaires
Exhibit D-4 (cont'd)
Estimated Number and Percent of Rural Domestic Wells In
Counties with High Pesticide Use
Item
11.23
DWS
LAQ
#4
(11.22
suHtern)
11.24
DWS
LAQ
#5
11.25
DWS
LAQ
#7
11.26
DWS
LAQ
#9
11.27
DWS
LAQ
#11j
Questionnaire Item
Pesticides used on pasture land
within 1/2 mile of well in the last 3
years:
Yes
No
Don't know
Major non-agricultural pesticide
application within 1/2 mile of well
in the last 3 years:
Yes
No
Don't know
Accidental spill of any pesticide
within 1/2 mile of well in the last 3
years:
Yes
No
Don't know
Accidental spill of any hazardous
chemical within 1/2 mile of well in
the last 3 years:
Yes
No
Don't know
Golf course within 1/2 mile of well:
Yes
No
Don't know
Estimates for Rural Domestic Wells In
Counties with High Pesticide Use
Number (95% C.I.)
353,000 (± 131,000)
311,000 (± 182,000)
0 N/A
340,000 (± 205,000)
808,000 (± 366,000)
8,700 (± 17,500)*
4,140 (± 8,250)*
1,130,000 (± 325,000)*
20,300 (± 40,900)*
3,690 (± 7,300)*
1,140,000 (± 325,000)*
8,700 (± 17,500)*
27,100 (± 32,600)*
1,130,000 (± 317,000)*
0 N/A
Percent (95% C.I.)
53.2 (±20.0)
46.9 (±20.0)
0.0 N/A
29.4 (±18.9)
69.9 (±18.8)
0.8 (± 1.5)*
0.4 (± 0.7)*
97.9 (± 4.2)*
1.8 (± 3.5)*
0.3 (± 0.6)*
98.9 (± 1.6)*
0.8 (± 1.5)*
2.3 (± 2.7)*
97.7 (± 2.7)*
0.0 N/A
National Pesticide Survey: Phase I Report
-------
Appendix D: Questionnaire Data and Questionnaires D-27
Exhibit D-4 (cont'd)
Estimated Number and Percent of Rural Domestic Wells In
Counties with High Pesticide Use
Item
11.28
DWS
LAQ
#12
II.29
DWS
LAQ
#13
Questionnaire Hern
River, canal, bay, spring, pond,
etc. within 1/2 mile of well:
Yes
No
Don't know
Irrigation within 1/2 mile of well:
Yes
No
Don't know
Estimates for Rural Domestic Wells In
Counties with High Pesticide Use
Number (95% C.I.)
1,080,000 (±337,000)*
40,900 (± 42,600)*
35,800 (± 70,400)*
234,000 (± 150,000)
923,000 (± 318,000)
0 N/A
Percent (95% C.I.)
93.4 (± 7.3)*
3.5 (± 3.6)*
3.1 (± 6.1)*
20.2 (±12.8)
79.8 (± 12.8)
0.0 N/A
* For particular responses to questionnaire items where the percent is extreme (i.e., near 0.0% or 100.0% and the sample size
is relatively small), the 95% confidence intervals should be treated with caution. For further details see Section B6.4 of
Appendix B.
N/A For particular responses to questionnaire items where the percent is equal to 0.0% or 100.0%, traditional 95% confidence
intervals based soley on sampling error are not applicable. For further details see Section B6.4 of Appendix B.
-------
D-28 Appendix D: Questionnaire Data and Questionnaires
Exhibit D-5
Estimated Number and Percent of Rural Domestic Wells In
Counties with High Vulnerability
Item
111.1
DWS
TLI
#6
III.2
DWS
WOR
#1
III.3
DWS
WOR
#4
III. 4
DWS
WOR
#3
III.5
DWS
Main
D.3
Questionnaire Item
Septic system on the property:
Septic tank
Septic field
Cesspool
Topographic setting at well:
Hilltop
Hillside
Flat valley
Depression
Plateau
Other
Don't Know
Description of land area within
300 feet of well:
Roofed
Paved
Graveled
Exposed or vegetated soil
Other
Well protection at surface:
Well house or shed
Concrete pad
Covered pit
Other
Age of well:
< 5 years
5 to 1 0 years
11 to 20 years
> 20 years
Don't know
Estimates for Rural Domestic Wells In
Counties with High Vulnerability
Number (95% C.I.)
2,810,000 (± 488,000)
2,680,000 (± 443,000)
196,000 (± 154,000)
598,000 (± 259,000)
639,000 (± 391,000)
1,010,000 (± 420,000)
25,100 (± 41,300)*
96,900 (± 124,000)*
508,000 (± 356,000)
0 N/A
92,700 (± 154,000)*
0 N/A
4,880 (± 9,720)*
2,560,000 (± 562,000)*
217,000 (± 288,000)*
1,150,000 (± 385,000)
661,000 (± 1,300,000)*
502,000 (± 984,000)*
1,290,000 (± 2,520,000)*
560,000 (± 235,000)
493,000 (± 229,000)
803,000 (± 290,000)
553,000 (± 227,000)
465,000 (± 286,000)
Percent (95% C.I.)
97.7 (± 2.0)
93.1 (± 6.3)
6.8 (± 5.5)
20.8 (± 7.6)
22.2 (±11.5)
35.0 (±14.6)
0.9 (± 1.5)*
3.4 (± 4.3)*
17.7 (±13.1)
0.0 N/A
3.2 (± 5.4)*
0.0 N/A
0.2 (± 0.3)*
89.1 (±11.7)*
7.5 (±10.1)*
40.1 (±13.6)
23.0 (±45.1)*
17.5 (±34.2)*
44.8 (±87.8)*
19.5 (± 6.9)
17.1 (± 7.6)
28.0 (± 9.6)
19.3 (± 7.6)
16.2 (± 9.5)
National Pesticide Survey: Phase I Report
-------
Appendix D: Questionnaire Data and Questionnaires D-29
Exhibit D-5 (cont'd)
Estimated Number and Percent of Rural Domestic Wells In
Counties with High Vulnerability
Item
111.6
DWS
Main
D.4
111.7
DWS
Main
D.6
111.8
DWS
Main
D.8
DWS
Main
D.10
III. 9
DWS
WOR
#2
111.10
DWS
Main
D.17
Questionnaire Item
Well redrilled:
Yes
No
Don't know
Well depth:
< 20 feet
20 to 50 feet
51 to 100 feet
101 to 200 feet
201 to 500 feet
> 500 feet
Don't know
Well casing information:
Well cased
Not cased
Don't know casing
Cased to total depth
Cased < total depth
Don't know casing depth
Well closed at ground surface:
Yes
No
Don't know
Aquifer:
Single
Double
Don't know
Estimates for Rural Domestic Wells in
Counties with High Vulnerability
Number (95% C.I.)
162,000 (± 126,000)
2,390,000 (± 466,000)
317,000 (± 270,000)
129,000 (± 118,000)*
418,000 (± 180,000)
449,000 (± 164,000)
626,000 (± 214,000)
532,000 (± 249,000)
76,700 (± 127,000)*
643,000 (± 263,000)
2,220,000 (± 435,000)
241,000 (± 173,000)
414,000 (± 184,000)
1,170,000 (± 385,000)
559,000 (± 304,000)
489,000 (± 174,000)
2,210,000 (± 469,000)
398,000 (± 227,000)
268,000 (± 164,000)
554,000 (± 251,000)
1,370,000 (± 344,000)
953,000 (± 352,000)
Percent (95% C.I.)
5.6 (± 4.3)
83.3 (±10.0)
11.1 (± 9.2)
4.5 (± 4.2)*
14.5 (± 5.9)
15.6 (± 5.8)
21.8 (± 6.0)
18.5 (± 7.7)
2.7 (± 4.5)*
22.4 (± 8.0)
77.2 (± 7.4)
8.4 (± 5.9)
14.4 (± 5.9)
52.8 (±14.2)
25.2 (±11.8)
22.0 (± 8.4)
76.8 (± 7.7)
13.8 (± 8.1)
9.3 (± 5.5)
19.3 (± 8.2)
47.6 (±10.5)
33.2 (±10.0)
-------
D-30 Appendix D: Questionnaire Data and Questionnaires
Exhibit D-5 (cont'd)
Estimated Number and Percent of Rural Domestic Wells In
Counties with High Vulnerability
Hem
111.11
DWS
Main
B.14
111.12
DWS
Main
B.1
111.13
DWS
Main
B.2
111.14
DWS
Main
B.3
111.15
DWS
Main
B.4
DWS
Main
B.5
Questionnaire Item
Non-operating wells within 500
feet of the well:
Yes
No
Don't know
Non-farm pesticides used in the
last 3 years:
Used inside house
Not used inside house
Don't know
Non-farm pesticides used in the
last 3 years:
Used on lawn
Not used on lawn
Don't know
Non-farm pesticides used in the
last 3 years:
Used in garden
Not used in garden
Don't know
Non-farm pesticides stored for
more than 1 month in the last 3
years:
Non-farm pesticides stored
Not stored
Don't know whether stored
Stored < 100 feet from well
Stored > 100 feet from well
Don't know where stored
Estimates for Rural Domestic Wells In
Counties with High Vulnerability
Number (95% C.I.)
793,000 (± 303,000)
1,880,000 (± 405,000)
201,000 (± 152,000)
2,280,000 (± 452,000)
594,000 (± 357,000)
0 N/A
1,050,000 (± 434,000)
1,820,000 (± 544,000)
0 N/A
1,260,000 (± 333,000)
1,620,000 (± 397,000)
0 N/A
1,890,000 (± 422,000)
979,000 (± 399,000)
0 N/A
1,460,000 (± 376,000)
365,000 (± 147,000)
64,300 (± 90,400)*
Percent (95% C.I.)
27.6 (± 9.0)
65.4 (± 8.6)
7.0 (± 5.5)
79.3 (±11.4)
20.7 (±11.4)
0.0 N/A
36.6 (±14.6)
63.4 (±14.6)
0.0 N/A
43.7 (± 9.5)
56.3 (± 9.5)
0.0 N/A
65.9 (±11.7)
34.1 (±11.7)
0.0 N/A
77.3 (± 7.4)
19.3 (± 6.7)
3.4 (± 4.9)*
National Pesticide Survey: Phase I Report
-------
Appendix D: Questionnaire Data and Questionnaires D-31
Exhibit D-5 (cont'd)
Estimated Number and Percent of Rural Domestic Wells In
Counties with High Vulnerability
Item
111.16
DWS
Main
B.6
DWS
Main
B.7
111.17
DWS
Main
B.17
and
B.18
DWS
Main
C.6
111.18
DWS
Main
C.11
and
B.20
Questionnaire Item
Non-farm pesticide container
disposal in the last 3 years:
Containers disposed
No disposal
Don't know
< 100 feet from well
> 100 feet from well
Don't know how far
Property farmed:
Yes
No
Pesticides used for farming in the
last 5 years:
Pesticides used
Not used
Don't know whether used
Farm pesticides stored for more
than 1 month in the last 5 years:
Farm pesticides stored
Not stored
Don't know whether stored
Estimates for Rural Domestic Wells In
Counties with High Vulnerability
Number (95% C.I.)
81,800 (± 73,100)
2,790,000 (± 503,000)
0 N/A
44,600 (± 47,600)
37,300 (± 36,900)
0 N/A
241,000 (± 134,000)
2,630,000 (± 480,000)
147,000 (± 100,000)
93,800 (± 77,800)
0 N/A
93,500 (± 71,000)
2,780,000 (± 483,000)
0 N/A
Percent (95% C.I.)
2.9 (± 2.6)
97.2 (± 2.6)
0.0 N/A
54.5 (±29.7)
45.5 (±29.7)
0.0 N/A
8.4 (± 4.6)
91.6 (± 4.6)
61.0 (±25.2)
39.0 (±25.2)
0.0 N/A
3.3 (± 2.5)
96.7 (± 2.5)
0.0 N/A
-------
D-32 Appendix D: Questionnaire Data and Questionnaires
Exhibit D-5 (cont'd)
Estimated Number and Percent of Rural Domestic Wells In
Counties with High Vulnerability
Item
111.19
DWS
Main
C.14
and
B.22
DWS
Main
C.16
and
B.23
III. 20
DWS
Main
C.17
and
B.24
111.21
DWS
LAQ
#1
Hl.22
DWS
LAQ
#3
Questionnaire Item
Farm pesticide container disposal
in the last 5 years:
Containers disposed
No disposal
Don't know
< 1 00 feet from well
> 100 feet from well
Don't know how far
Accidental spill of farm pesticide
in the last 5 years:
Farm pesticides spilled
No spill
Don't know whether spilled
Crops farmed within 1/2 mile of
well in the last 3 years:
Yes
No
Don't know
Pasture land within 1/2 mile of well
in the last 3 years:
Yes
No
Don't know
Estimates for Rural Domestic Wells In
Counties with High Vulnerability
Number (95% C.I.)
23,100 (± 30,900)
277,000 (± 200,000)
0 N/A
3,750 (± 7,460)*
18,200 (± 30,300)*
1,160 (± 2,300)*
0 N/A
300,000 (± 202,000)
0 N/A
1 ,750,000 (± 525,000)
1,100,000 (± 434,000)
15,000 (± 29,800)*
1,590,000 (± 530,000)
1,260,000 (± 429,000)
20,200 (± 41,100)*
Percent (95% C.I.)
7.7 (±11.9)
92.3 (±11.9)
0.0 N/A
16.2 (±61.0)*
78.8 (±74.4)*
5.0 (±19.0)*
0.0 N/A
100.0 N/A
0.0 N/A
60.9 (±14.2)
38.6 (±14.2)
0.5 (± 1.0)*
55.4 (±14.4)
43.9 (±14.4)
0.7 (± 1.4)*
National Pesticide Survey: Phase I Report
-------
Appendix D: Questionnaire Data and Questionnaires D-33
Exhibit D-5 (cont'd)
Estimated Number and Percent of Rural Domestic Wells In
Counties with High Vulnerability
Item
111.23
DWS
LAQ
#4
(111.22
tub-item)
111.24
DWS
LAQ
#5
111,25
DWS
LAQ
#7
111.26
DWS
LAQ
#9
111.27
DWS
LAQ
#11j
Questionnaire Item
Pesticides used on pasture land
within 1/2 mile of well in the last 3
years:
Yes
No
Don't know
Major non-agricultural pesticide
application within 1/2 mile of well
in the last 3 years:
Yes
No
Don't know
Accidental spill of any pesticide
within 1/2 mile of well in the last 3
years:
Yes
No
Don't know
Accidental spill of any hazardous
chemical within 1/2 mile of well in
the last 3 years:
Yes
No
Don't know
Golf course within 1/2 mile of well:
Yes
No
Don't know
Estimates for Rural Domestic Wells In
Counties with High Vulnerability
Number (95% C.I.)
714,000 (± 450,000)
878,000 (± 419,000)
0 N/A
705,000 (± 301,000)
2,110,000 (± 575,000)
59,700 (± 69,500)*
0 N/A
2,850,000 (± 486,000)*
28,100 (± 41,300)*
0 N/A
2,820,000 (± 490,000)*
52,600 (± 88,000)*
180,000 (± 165,000)
2,650,000 (± 477,000)
39,100 (± 79,800)*
Percent (95% C.I.)
44.9 (±22.9)
55.1 (±22.9)
0.0 N/A
24.5 (±11.6)
73.4 (±11.1)
2.1 (± 2.4)*
0.0 N/A
99.0 (± 1.4)*
1.0 (± 1.4)*
0.0 N/A
98.2 (± 3.1)*
1.8 (± 3.1)*
6.3 (± 5.6)
92.4 (± 5.8)
1.4 (± 2.8)*
-------
D-34 Appendix D: Questionnaire Data and Questionnaires
Exhibit D-5 (cont'd)
Estimated Number and Percent of Rural Domestic Wells In
Counties with High Vulnerability
Item
111.28
DWS
LAQ
#12
III.29
DWS
LAQ
#13
Questionnaire Item
River, canal, bay, spring, pond,
etc. within 1/2 mile of well:
Yes
No
Don't know
Irrigation within 1/2 mile of well:
Yes
No
Don't know
Estimates for Rural Domestic Wells In
Counties with High Vulnerability
Number (95% C.I.)
2,790,000 (± 493,000)
45,400 (± 40,600)
41,500 (± 66,800)*
543,000 (± 281,000)
2,320,000 (± 456,000)
11,000 (± 22,300)*
Percent (95% C.I.)
97.0 (± 3.4)
1.6 (± 1.5)
1.4 (± 2.3)*
18.9 (± 9.1)
80.7 (± 8.9)
0.4 (± 0.8)*
N/A
For particular responses to questionnaire items where the percent is extreme (i.e., near 0.0% or 100.0% and the sample
size is relatively small), the 95% confidence intervals should be treated with caution. For further details see Section B6.4
of Appendix B.
For particular responses to questionnaire items where the percent is equal to 0.0% or 100.0%, traditional 95% confidence
intervals based solely on sampling error are not applicable. For further details see Section B6.4 of Appendix B.
National Pesticide Survey: Phase I Report
-------
Appendix D: Questionnaire Data and Questionnaires D-35
Exhibit D-6
Estimated Number and Percent of Rural Domestic Wells In
"Cropped and Vulnerable" County Subregions
Item
IV.1
DWS
TLJ
#6
IV.2
DWS
WOR
#1
IV.3
DWS
WOR
#4
IV.4
DWS
WOR
#3
IV.5
DWS
Main
D.3
Questionnaire Item
Septic system on the property:
Septic tank
Septic field
Cesspool
Topographic setting at well:
Hilltop
Hillside
Flat valley
Depression
Plateau
Other
Don't know
Description of land area within
300 feet of well:
Roofed
Paved
Graveled
Exposed or vegetated soil
Other
Well protection at surface:
Well house or shed
Concrete pad
Covered pit
Other
Age of well:
< 5 years
5 to 10 years
11 to 20 years
> 20 years
Don't know
Estimates for Rural Domestic Wells In
'Cropped and Vulnerable* County Subregions
Number (95% C.I.)
3,370,000 (±711,000)
3,080,000 (± 672,000)
224,000 (± 149,000)
379,000 (± 222,000)
1,070,000 (± 483,000)
1,730,000 (± 539,000)
1,390 (± 2,780)*
85,500 (± 101,000)*
421,000 (± 273,000)
0 N/A
77,700 (± 103,000)*
0 N/A
16,700 (± 25,700)*
3,530,000 (± 770,000)
71,700 (± 72,000)*
986,000 (± 402,000)
445,000 (± 203,000)
517,000 (± 271,000)
2,020,000 (± 655,000)
409,000 (± 193,000)
590,000 (± 238,000)
1,010,000 (± 273,000)
918,000 (± 374,000)
764,000 (± 348,000)
Percent (95% C.I.)
91.3 (± 5.9)
83.5 (± 7.3)
6.1 (± 3.9)
10.3 (± 5.4)
29.0 (±10.5)
467.0 (±12.1)
0.0 (± 0.1)*
2.3 (± 2.8)*
11.4 (± 7.3)
0.0 N/A
2.1 (± 2.8)*
0.0 N/A
0.5 (± 0.7)*
95.5 (± 3.5)
1.9 (± 1.7)*
20.7 (± 9.2)
12.0 (± 5.6)
14.0 (± 6.9)
54.7 (±12.0)
11.1 (± 5.2)
16.0 (± 5.1)
27.4 (± 5.3)
24.7 (± 8.6)
20.7 (± 7.7)
-------
D-36 Appendix D: Questionnaire Data and Questionnaires
Exhibit D-6 (cont'd)
Estimated Number and Percent of Rural Domestic Wells In
"Cropped and Vulnerable" County Subregions
Item
IV.6
DWS
Main
D.4
IV.7
DWS
Main
D.6
IV.8
DWS
Main
D.8
DWS
Main
D.10
IV.9
DWS
WOR
#2
IV.10
DWS
Main
D.17
Questionnaire Item
Well redrilled:
Yes
No
Don't know
Well depth:
< 20 feet
20 to 50 feet
51 to 100 feet
101 to 200 feet
201 to 500 feet
> 500 feet
Don't know
Well casing information:
Well cased
Not cased
Don't know casing
Cased to total depth
Cased < total depth
Don't know casing depth
Well closed at ground surface:
Yes
No
Don't know
Aquifer:
Single
Double
Don't know
Estimates for Rural Domestic Wells In
Cropped and Vulnerable' County Subregions
Number (95% C.I.)
133,000 (± 77,600)
3,120,000 (± 726,000)
443,000 (± 302,000)
39,800 (± 47,300)*
470,000 (± 195,000)
984,000 (± 318,000)
1,040,000 (± 378,000)
382,000 (± 200,000)
23,700 (± 35,000)*
751,000 (± 270,000)
3,020,000 (± 766,000)
224,000 (± 159,000)
454,000 (± 217,000)
1,480,000 (± 427,000)
925,000 (± 389,000)
612,000 (± 246,000)
2,880,000 (± 687,000)
389,000 (± 326,000)
422,000 (± 290,000)
883,000 (± 306,000)
1,560,000 (± 529,000)
1,250,000 (± 407,000)
Percent (95% C.I.)
3.6 (± 2.0)
84.4 (± 8.1)
12.0 (± 7.7)
1.1 (± 1.3)*
12.7 (± 4.9)
26.7 (± 6.4)
28.2 (± 7.6)
10.4 (± 4.7)
0.6 (± 9.6)*
20.3 (± 6.1)
81.7 (± 7.1)
6.1 (± 4.2)
12.3 (± 6.0)
49.0 (±10.3)
30.7 (± 8.2)
20.3 (± 6.7)
78.0 (±10.4)
10.6 (± 8.2)
11.4 (± 7.8)
23.9 (± 7.1)
42.3 (± 9.7)
33.8 (± 9.1)
National Pesticide Survey: Phase I Report
-------
Appendix D: Questionnaire Data and Questionnaires D-37
Exhibit D-6 (cont'd)
Estimated Number and Percent of Rural Domestic Wells In
"Cropped and Vulnerable" County Subregions
Item
IV. 11
DWS
Main
B.14
IV. 12
DWS
Main
B.1
IV. 13
DWS
Main
B.2
IV. 14
DWS
Main
B.3
IV. 15
DWS
Main
B.4
DWS
Main
B.5
Questionnaire Item
Non-operating wells within 500
feet of the well:
Yes
No
Don't know
Non-farm pesticides used in the
last 3 years:
Used inside house
Not used inside house
Don't know
Non-farm pesticides used in the
last 3 years:
Pesticide used on lawn
Not used on lawn
Don't know
Non-farm pesticides used in the
last 3 years;
Used in garden
Not used in garden
Don't know
Non-farm pesticides stored for
more than 1 month in the last 3
years:
Non-farm pesticides stored
Not stored
Don't know whether stored
Stored < 100 feet from well
Stored > 100 feet from well
Don't know where stored
Estimates for Rural Domestic Wells In
Cropped and Vulnerable' County Subregions
Number (95% C.I.)
963,000 (± 383,000)
2,470,000 (± 573,000)
257,000 (± 195,000)
2,760,000 (± 686,000)
933,000 (± 416,000)
0 N/A
1,410,000 (± 432,000)
2,250,000 (± 590,000)
26,600 (± 54,400)*
1,410,000 (± 352,000)
2,250,000 (± 612,000)
26,600 (± 54,500)*
2,510,000 (± 571,000)
1,180,000 (± 380,000)
0 N/A
1,880,000 (± 462,000)
554,000 (±211,000)
74,800 (± 93,600)*
Percent (95% C.I.)
26.1 (± 8.2)
67.0 (± 8.5)
7.0 (± 5.0)
74.7 (± 9.7)
25.3 (± 9.7)
0.0 N/A
38.3 (± 8.5)
61.0 (± 8.3)
0.7 (± 1.5)*
38.2 (± 6.9)
61.1 (± 7.1)
0.7 (± 1.5)*
68.0 (± 6.7)
32.0 (± 6.7)
0.0 N/A
74.9 (± 7.2)
22.1 (± 7.0)
3.0 (± 3.7)*
-------
D-38 Appendix D: Questionnaire Data and Questionnaires
Exhibit 0-6 (cont'd)
Estimated Number and Percent of Rural Domestic Wells In
"Cropped and Vulnerable" County Subregions
Item
IV. 16
DWS
Main
B.6
DWS
Main
B.7
IV. 17
DWS
Main
B.17
and
B.18
DWS
Main
C.6
IV. 18
DWS
Main
C.11
and
B.20
Questionnaire Item
Non-farm pesticide container
disposal in the last 3 years:
Containers disposed
No disposal
Don't know
< 100 feet from well
> 100 feet from well
Don't know how far
Property farmed:
Yes
No
Pesticides used for farming in the
last 5 years:
Pesticides used
Not used
Don't know whether use
Farm pesticides stored for more
than 1 month in the last 5 years:
Farm pesticides stored
Not stored
Don't know whether stored
Estimates for Rural Domestic Wells in
Cropped and Vulnerable" County Subregions
Number (95% C.I.)
169,000 (± 165,000)
3,520,000 (± 755,000)
0 N/A
32,800 (± 24,800)*
137,000 (± 163,000)*
0 N/A
392,000 (± 119,000)
3,300,000 (± 738,000)
362,000 (± 110,000)
29,900 (± 28,500)
0 N/A
151,000 (± 79,100)
3,530,000 (± 767,000)
10,000 (± 20,700)*
Percent (95% C.I.)
4.6 (± 4.3)
95.4 (± 4.3)
0.0 N/A
19.4 (±36.3)*
80.6 (±36.3)*
0.0 N/A
10.6 (± 2.7)
89.4 (± 2.7)
92.4 (± 6.8)
7.6 (± 6.8)
0.0 N/A
4,1 (± 2.0)
95.6 (± 2.0)
0.3 (± 0.6)*
National Pesticide Survey: Phase I Report
-------
Appendix D: Questionnaire Data and Questionnaires 0-39
Exhibit D-6 (cont'd)
Estimated Number and Percent of Rural Domestic Wells In
"Cropped and Vulnerable" County Subregions
Item
IV. 19
DWS
Main
C.14
and
B.22
DWS
Main
C.16
and
B.23
IV.20
DWS
Main
C.17
and
B.24
IV.21
DWS
LAQ
#1
IV.22
DWS
LAQ
#3
Questionnaire Item
Farm pesticide container disposal
in the last 5 years:
Containers disposed
No disposal
Don't know
< 100 feet from well
> 100 feet from well
Don't know how far
Accidental spill of farm pesticide
in the last 5 years:
Farm pesticides spilled
No spill
Don't know whether spilled
Crops farmed within 1/2 mile of
well in the last 3 years:
Yes
No
Don't know
Pasture land within 1/2 mile of
well in the last 3 years:
Yes
No
Don't know
Estimates for Rural Domestic Wells In
Cropped and Vulnerable" County Subregions
Number (95% C.I.)
119,000 (± 74,200)
344,000 (± 107,000)
0 N/A
11,400 (± 14,300)*
107,000 (± 73,600)*
1,160 (± 2,300)*
10,900 (± 15,400)*
452,000 (± 144,000)*
0 N/A
2,940,000 (± 654,000)
672,000 (± 326,000)
79,000 (± 109,000)*
2,510,000 (± 718,000)
1,080,000 (± 364,000)
97,600 (± 153,000)*
Percent (95% C.I.)
25.8 (±12.3)
74.2 (± 12.3)
0.0 N/A
9.S (±13.3)*
89.5 (±13.7)*
1.0 (± 2.1)*
2.4 (± 3.3)*
97.6 (± 3.3)*
0.0 N/A
79.7 (± 8.1)
18.2 (± 7.3)
2.1 (± 3.0)*
68.0 (± 9.9)
29.3 (± 9.3)
2.6 (± 4.2)*
-------
D-40 Appendix D: Questionnaire Data and Questionnaires
Exhibit D-6 (cont'd)
Estimated Number and Percent of Rural Domestic Wells In
"Cropped and Vulnerable" County Subregions
Item
IV.23
DWS
LAQ
#4
(IV.22
sub-Item)
IV.24
DWS
LAQ
#5
IV.25
DWS
LAQ
#7
IV.26
DWS
LAQ
#9
IV.27
DWS
LAQ
#11j
Questionnaire Item
Pesticides used on pasture land
within 1/2 mile of well in the last 3
years:
Yes
No
Don't know
Major non-agricultural pesticide
application within 1/2 mile of well
in the last 3 years:
Yes
No
Don't know
Accidental spill of any pesticide
within 1/2 mile of well in the last 3
years:
Yes
No
Don't know
Accidental spill of any hazardous
chemical within 1/2 mile of well in
the last 3 years:
Yes
No
Don't know
Golf course within 1/2 mile of well:
Yes
No
Don't know
Estimates for Rural Domestic Wells In
"Cropped and Vulnerable* County Subregions
Number (95% C.I.)
867,000 (± 411,000)
1,640,000 (± 579,000)
0 N/A
556,000 (± 282,000)
3,020,000 (± 738,000)
113,000 (± 131,000)*
10,100 (± 14,500)*
3,490,000 (± 789,000)*
190,000 (±261,000)*
3,690 (± 7,300)*
3,530,000 (± 795,000)*
161,000 (± 261,000)*
168,000 (± 144,000)
3,460,000 (± 756,000)
67,000 (± 102,000)*
Percent (95% C.I.)
34.5 (±13.5)
65.5 (±13.5)
0.0 N/A
15.1 (± 7.2)
81.9 (± 7.9)
3.1 (± 3.6)*
0.3 (± 0.4)*
94.6 (± 7.0)*
5.1 (± 7.0)*
0.1 (± 0.2)*
95.5 (± 7.1)*
4.4 (± 7.1)*
4.6 (± 3.8)
93.6 (± 4.5)
1.8 (± 2.8)*
National Pesticide Survey: Phase I Report
-------
Appendix D: Questionnaire Data and Questionnaires 0-41
Exhibit 0-6 (cont'd)
Estimated Number and Percent of Rural Domestic Wells In
"Cropped and Vulnerable" County Subregions
Item
IV.28
DWS
LAQ
#12
IV.29
DWS
UQ
#13
Questionnaire Item
River, canal, bay, spring, pond,
etc. within 1/2 mile of well:
Yes
No
Don't know
Irrigation within 1/2 mile of well:
Yes
No
Don't know
Estimates for Rural Domestic Wells in
Cropped and Vulnerable" County Subregions
Number (95% C.I.)
3,530,000 (± 793,000)
139,000 (± 134,000)
18,400 (± 26,300)*
743,000 (± 403,000)
2,910,000 (± 761,000)
38,500 (± 55,700)*
Percent (95% C.I.)
95.7 (± 3.7)
3.8 (± 3.7)
0.5 (± 0.7)*
20.1 (±10.3)
78.8 (±11.1)
1.0 (± 1.5)*
N/A
For particular responses to questionnaire items where the percent is extreme (i.e., near 0.0% or 100.0% and the sample
size is relatively small), the 95% confidence intervals should be treated with caution. For further details see Section B6.4
of Appendix B.
For particular responses to questionnaire items where the percent is equal to 0.0% or 100.0%, traditional 95% confidence
intervals based solely on sampling error are not applicable. For further details see Section B6.4 of Appendix B.
National Pesticide Survey: Phase I Report
-------
D-42 Appendix D: Questionnaire Data and Questionnaires
Exhibit D-7
Estimated Number and Percent of Rural Domestic Wells In
Counties with High Pesticide Use
and High Vulnerability
Item
V.1
DWS
TLI
#6
V.2
DWS
WOR
#1
V.3
DWS
WOR
#4
V.4
DWS
WOR
#3
V.5
DWS
Main
D.3
Questionnaire Item
Septic system on the property:
Septic tank
Septic field
Cesspool
Topographic setting at well:
Hilltop
Hillside
Flat valley
Depression
Plateau
Other
Don't know
Description of land area within
300 feet of well:
Roofed
Paved
Graveled
Exposed or vegetated soil
Other
Well protection at surface:
Well house or shed
Concrete pad
Covered pit
Other
Age of well:
< 5 years
5 to 1 0 years
11 to 20 years
> 20 years
Don't know
Estimates for Rural Domestic Wells In Counties
with High Pesticide Use and High Vulnerability
Number (95% C.I.)
312,000 (± 105,000)
306,000 (± 99,600)
35,900 (± 70,300)
27,500 (± 32,500)*
26,000 (± 39,800)*
190,000 (± 151,000)
0 N/A
14,700 (± 29,100)*
83,700 (± 110,000)*
0 N/A
4,930 (± 9,760)*
0 N/A
4,880 (± 9,720)*
318,000 (± 115,000)*
14,600 (± 24,400)*
85,100 (± 85,200)*
23,000 (± 37,500)*
47,500 (± 64,200)*
197,000 (± 131,000)*
67,100 (± 44,600)
56,200 (± 30,500)
123,000 (± 61,600)
39,500 (± 35,600)
56,600 (± 44,800)
Percent (95% C.I.)
91.1 (± 9.8)
89.4 (± 8.7)
10.5 (±12.6)
8.0 (±10.2)*
7.6 (±11.6)*
55.6 (±35.7)
0.0 N/A
4.3 (± 8.5)*
24.5 (±34.2)*
0.0 N/A
1.4 (± 2.9)*
0.0 N/A
1.4 (± 2.9)*
92.9 (± 8.4)*
4.3 (± 7.4)*
24.9 (±26.9)*
6.7 (±11.4)*
13.9 (±17.5)*
57.5 (±30.9)*
19.6 (± 8.1)
16.4 (± 9.1)
35.9 (±18.3)
11.6 (±10.3)
16.5 (± 9.1)
National Pesticide Survey: Phase I Report
-------
Appendix D: Questionnaire Data and Questionnaires D-43
Exhibit D-7 (cont'd)
Estimated Number and Percent of Rural Domestic Wells In
Counties with High Pesticide Use
and High Vulnerability
Item
V.6
DWS
Main
D.4
V.7
DWS
Main
D.6
V.8
DWS
Main
D.8
DWS
Main
D.10
V.9
DWS
WOR
#2
V.10
DWS
Main
D.17
Questionnaire Item
Well redrilled:
Yes
No
Don't know
Well depth:
< 20 feet
20 to 50 feet
51 to 100 feet
101 to 200 feet
201 to 500 feet
> 500 feet
Don't know
Well casing information:
Well cased
Not cased
Don't know casing
Cased to total depth
Cased < total depth
Don't know casing depth
Well closed at ground surface:
Yes
No
Don't know
Aquifer:
Single
Double
Don't know
Estimates for Rural Domestic Wells In Counties
with High Pesticide Use and High Vulnerability
Number (95% C.I.)
14,700 (± 11,500)
302,000 (± 107,000)
25,700 (± 21,200)
16,700 (± 25,800)*
96,600 (± 74,800)
74,500 (± 40,400)
74,300 (± 40,400)
17,000 (± 26,000)*
0 N/A
62,900 (± 46,600)
280,000 (± 113,000)
21,900 (± 24,200)*
39,800 (± 41,400)*
189,000 (± 87,200)
10,700 (± 13,700)*
81,100 (± 54,200)
266,000 (± 126,000)*
33,800 (± 66,100)*
42,200 (± 56,300)*
94,100 (± 79,900)
156,000 (± 51,000)
91,500 (± 62,300)
Percent (95% C.I.)
4.3 (± 4.2)
88.2 (± 2.6)
7.5 (± 4.3)
4.9 (± 7.3)*
28.2 (±17.6)
21.8 (± 9.3)
21.7 (±14.4)
5.0 (± 8.0)*
0.0 N/A
18.4 (±11.2)
82.0 (±15.6)
6.4 (± 7.4)*
11.6 (±11.7)*
67.3 (±12.7)
3.8 (± 5.3)*
28.9 (± 14.9)
77.8 (±22.8)*
9.9 (±19.9)*
12.3 (±15.4)*
27.5 (±17.7)
45.7 (±19.1)
26.8 (±14.9)
National Pesticide Survev: Phase I Report
-------
D-44 Appendix D: Questionnaire Data and Questionnaires
Exhibit D-7 (cont'd)
Estimated Number and Percent of Rural Domestic Wells In
Counties with High Pesticide Use
and High Vulnerability
Item
V.11
DWS
Main
B.14
V.12
DWS
Main
B.1
V.13
DWS
Main
B.2
V.14
DWS
Main
B.3
V.15
DWS
Main
B.4
DWS
Main
B.5
Questionnaire Item
Non-operating wells within 500
feet of the well:
Yes
No
Don't know
Non-farm pesticides used in the
last 3 years:
Used inside house
Not used inside house
Don't know
Non-farm pesticides used in last 3
years:
Used on lawn
Not used on lawn
Don't know
Non-farm pesticides used in last 3
years:
Used in garden
Not used in garden
Don't know
Non-farm pesticides stored for
more than 1 month in the last 3
years:
Non-farm pesticides stored
Not stored
Don't know whether stored
Stored < 1 00 feet from well
Stored > 1 00 feet from well
Don't know where stored
Estimates for Rural Domestic Wells In Counties
with High Pesticide Use and High Vulnerability
Number (95% C.I.)
82,700 (± 41,400)
241,000 (± 84,800)
18,700 (± 18,100)
300,000 (± 106,000)
41,700 (± 26,20)
0 N/A
117,000 (± 56,900)
225,000 (± 121,000)
0 N/A
152,000 (± 53,800)
190,000 (± 95,300)
0 N/A
227,000 (± 65,200)
115,000 (± 67,800)
0 N/A
177,000 (± 60,500)
43,100 (± 22,000)
7,190 (± 14,400)*
Percent (95% C.I.)
24.2 (± 8.4)
70.4 (± 0.1)
5.5 (± 4.8)
87.8 (± 6.2)
12.2 (± 6.2)
0.0 N/A
34.3 (±18.8)
65.8 (±18.8)
0.0 N/A
44.5 (±13.7)
55.5 (±13.7)
0.0 N/A
66.4 (±10.9)
33.6 (±10.9)
0.0 N/A
77.9 (± 8.0)
19.0 (±11.1)
3.2 (± 6.3)*
National Pesticide Survey: Phase I Report
-------
Appendix D: Questionnaire Data and Questionnaires D-45
Exhibit D-7 (cont'd)
Estimated Number and Percent of Rural Domestic Wells In
Counties with High Pesticide Use
and High Vulnerability
Item
V.16
DWS
Main
B.6
DWS
Main
B.7
V.17
DWS
Main
B.17
and
B.18
DWS
Main
C.6
V.18
DWS
Main
C.11
and
B.20
Questionnaire Item
Non-farm pesticide container
disposal in the last 3 years:
Containers disposed
No disposal
Don't know
< 100 feet from well
> 1 00 feet from well
Don't know how far
Property farmed:
Yes
No
Pesticides used for farming in the
last 5 years:
Pesticides used
Not used
Don't know whether used
Farm pesticides stored for more
than 1 month in the last 5 years:
Farm pesticides stored
Not stored
Don't know whether stored
Estimates for Rural Domestic Wells In Counties
with High Pesticide Use and High Vulnerability
Number (95% C.I.)
35,900 (± 36,100)
306,000 (± 112,000)
0 N/A
17,400 (± 29,100)*
18,600 (± 24,800)*
0 N/A
28,700 (± 20,600)
313,000 (± 117,000)
26,100 (± 21,200)*
2,590 (± 3,860)*
0 N/A
8,810 (± 12,800)
333,000 (± 117,000)
0 N/A
Percent (95% C.I.)
10.5 (± 9.8)
89.5 (± 9.8)
0.0 N/A
48.3 (±78.9)*
51.7 (±78.9)*
0.0 N/A
8.4 (± 6.2)
91.6 (± 6.2)
91.0 (±16.5)*
9.0 (±16.5)*
0 N/A
2.6 (± 3.7)
97.4 (± 3.7)
0.0 N/A
National Pesticide Survev: Phase I Reoort
-------
D-46 Appendix D: Questionnaire Data and Questionnaires
Exhibit D-7 (cont'd)
Estimated Number and Percent of Rural Domestic Wells In
Counties with High Pesticide Use
and High Vulnerability
Item
V.19
DWS
Main
C.14
and
B.22
DWS
Main
C.16
and
B.23
V.20
DWS
Main
C.17
and
B.24
V.21
DWS
LAQ
#1
V.22
DWS
LAQ
#3
Questionnaire Item
Farm pesticide container disposal
in the last 5 years:
Containers disposed
No disposal
Don't know
< 1 00 feet from well
> 100 feet from well
Don't know how far
Accidental spill of farm pesticide
in the last 5 years:
Farm pesticides spilled
No spill
Don't know whether spilled
Crops farmed within 1/2 mile of
well in the last 3 years:
Yes
No
Don't know
Pasture land within 1/2 mile of
well in the last 3 years:
Yes
No
Don't know
Estimates for Rural Domestic Wells In Counties
with High Pesticide Use and High Vulnerability
Number (95% C.I.)
8,210 (± 7,940)
20,500 (± 13,700)
0 N/A
3,750 (± 7,460)*
3,310 (± 5,150)*
1,160 (± 2,300)*
0 N/A
28,700 (± 20,600)
0 N/A
291,000 (± 134,000)
51,300 (± 39,400)
0 N/A
176,000 (± 74,500)
166,000 (± 89,600)
0 N/A
Percent (95% C.I.)
28.7 (±12.8)
71.4 (±12.8)
0.0 N/A
45.6 (±92.3)*
40.3 (±80.1)*
14.1 (±35.5)*
0.0 N/A
100.0 N/A
0.0 N/A
85.0 (±13.7)
15.0 (±13.7)
0.0 N/A
51.5 (±17.8)
48.5 (±17.8)
0.0 N/A
National Pesticide Survey: Phase I Report
-------
Appendix 0: Questionnaire Data and Questionnaires D-47
Exhibit D-7 (cont'd)
Estimated Number and Percent of Rural Domestic Wells In
Counties with High Pesticide Use
and High Vulnerability
Item
V.23
DWS
LAQ
#4
(V.22
sub-item)
V.24
DWS
LAQ
#5
V.25
DWS
LAQ
#7
V.26
DWS
LAQ
#9
V.27
DWS
LAQ
#11 j
Questionnaire Item
Pesticides used on pasture land
within 1/2 mile of well in the last 3
years:
Yes
No
Don't know
Major non-agricultural pesticide
application within 1/2 mile of well
in the last 3 years:
Yes
No
Don't know
Accidental spill of any pesticide
within 1/2 mile of well in the last 3
years:
Yes
No
Don't know
Accidental spill of any hazardous
chemical within 1/2 mile of well in
the last 3 years:
Yes
No
Don't know
Goff course within 1/2 mile of well:
Yes
No
Don't know
Estimates for Rural Domestic Wells In Counties
with High Pesticide Use and High Vulnerability
Number (95% C.I.)
38,900 (± 47,300)*
137,000 (± 90,500)*
0 N/A
166,000 (± 140,000)
176,000 (± 154,000)
0 N/A
0 N/A
342,000 (± 119,000)
0 N/A
0 N/A
342,000 (± 119,000)
0 N/A
4,870 (± 9,710)*
337,000 (± 118,000)*
0 N/A
Percent (95% C.I.)
22.1 (±29.5)*
77.9 (±29.5)*
0.0 N/A
48.5 (±41.2)
51.5 (±41.2)
0.0 N/A
0.0 N/A
100.0 N/A
0.0 N/A
0.0 N/A
100.0 N/A
0.0 N/A
1.4 (± 2.9)*
98.6 (± 2.9)*
0.0 N/A
National
Curw.
-------
D-48 Appendix D: Questionnaire Data and Questionnaires
Exhibit D-7 (cont'd)
Estimated Number and Percent of Rural Domestic Wells In
Counties with High Pesticide Use
and High Vulnerability
item
V.28
DWS
LAQ
#12
V.29
DWS
LAQ
#13
Questionnaire Item
River, canal, bay, spring, pond,
etc. within 1/2 mile of well:
Yes
No
Don't know
Irrigation within 1/2 mile of well:
Yes
No
Don't know
Estimates for Rural Domestic Wells In Counties
with High Pesticide Use and High Vulnerability
Number (95% C.I.)
331,000 (± 111,000)*
11,200 (± 15,000)*
0 N/A
110,000 (± 99,600)
233,000 (± 79,900)
0 N/A
Percent (95% C.I.)
96.7 (± 4.0)*
3.3 (± 4.0)*
0.0 N/A
32.0 (±23.2)
68.0 (±23.2)
0.0 N/A
N/A
For particular responses to questionnaire items where the percent is extreme (i.e., near 0.0% or 100.0% and the sample
size is relatively small), the 95% confidence intervals should be treated with caution, because the statistical assumptions
are not justifiable in this case. For further details see Section B6.4 of Appendix B.
For particular responses to questionnaire items where the percent is equal to 0.0% or 100.0%, traditional 95% confidence
intervals based soley on sampling error are not applicable. For further details see Section B6.4 of Appendix B.
National Pesticide Survey: Phase I Report
-------
National Pesticide Survey
Appendix E: Survey Analytes and
Analytical Chemistry Tables
-------
Appendix E: Survey Analytes and
Analytical Chemistry Tables
Introduction
Appendix E presents data supplementing the description of the analytes and methods of chemical
analysis of NPS well water samples presented in Chapter 4. The tables in this appendix contain data on the
following:
Survey Analytes;
Minimum Quantification Limits;
Quality Control Failures;
Laboratory Control Standards Precision and Accuracy;
Laboratory Control Standards vs Spiked Field Samples; and
False Negative Rates.
The analytical results are listed for each analysis method and analyte, and for both the contract and EPA
laboratories.
Survey Analytes
Exhibit E-l provides information about each Survey analyte. The following categories of information
are presented. If information is not available for a particular analyte," " appears under the appropriate
category.
Analyte (CAS Registry Number) - The Survey analyte is listed along with its Chemical
Abstracts Service (CAS) Registry Number. CAS is a unique identifier assigned by the
American Chemical Society's Chemical Abstracts Service to catalog chemical compounds.
CAS Registry Numbers are not available for all Survey analytes.
Common and/or Trade Names - The common name is the name frequently used for the
chemical. The trade name is the name under which a product is sold. Many of the Survey
analytes are contained in products with several different trade names.
Type -- Pesticides are chemicals used to kill, repel, or mitigate pest organisms. Principal
types are: herbicides (weeds), insecticides (insects), fungicides (fungi), acaricides (mites,
spiders, and ticks), nematocides (nematodes), and rodenticides (rodents).
Principal Use -- The particular pests that are killed or controlled by each pesticide, and the
types of crops and plants to which it is applied.
Regulatory Status - Pesticide manufacture and use are regulated under the Federal
Insecticide, Fungicide, and Rodemicide Act (FTFRA). EPA registers pesticides for specific
uses after weighing risks and benefits for each use. The nine categories used to describe the
regulatory status of the NPS analytes are:
Reregistration (R) - EPA is reevaluating for reregistration existing pesticide active
ingredients originally registered prior to current scientific and regulatory standards.
Special Review (S) - EPA is aware that this registered pesticide may pose a potential
safety problem and is conducting a Special Review, which is an intensive and
National Pesticide Survey: Phase I Report
-------
E-2 Appendix E: Survey Analytes and Analytical Chemistry Tables
systematic examination of the risks and benefits of the pesticide. Based on this
review, EPA may choose one of several regulatory options, including modifications
to the terms and conditions of the registration such as restricting use, changes in
packaging, requiring prescribed protective equipment, label warnings, changes in
application rates or frequencies, and prohibition against certain types of application
methods; or cancellation of some or all uses.
Cancellation (C) - EPA has canceled the registration of this pesticide because when
used according to widespread and commonly recognized practice, the pesticide
generally causes unreasonable adverse effects on the environment, or the registrant(s)
elected not to support reregistration of this pesticide.
In Process (P) - EPA and/or registrants are involved in activities to determine
whether this existing pesticide active ingredient will enter the reregistration process.
Not Registered in the United States. (N) - EPA has not registered this pesticide.
Degradate (D) - This is a degradate of a registered active ingredient, not an active
ingredient itself.
Contaminant (T) - This is a contaminant of a registered active ingredient, not a
degradate or an active ingredient itself.
New Chemical (W) - EPA registered this pesticide active ingredient based on current
scientific and regulatory standards.
Nitrate (##) is not a pesticide and is not subject to regulation under FIFRA.
Nitrate in public water supplies is regulated under the Safe Drinking Water Act.
NFS Analytical Method -- The eight different analytical methods used to determine whether
the water samples collected for the Survey contained any of the 127 analytes are as follows:
NPS Method 1 - Gas Chromatography with a Nitrogen-Phosphorous
Detector
(46 nitrogen-phosphorus containing pesticides)
NPS Method 2 - Gas Chromatography with an Electron Capture Detector
(29 chlorinated pesticides)
NPS Method 3 - Gas Chromatography with an Electron Capture Detector
(17 chlorinated acid pesticides)
NPS Method 4 - High Performance Liquid Chromatography with an
Ultraviolet Detector
(18 carbonates and related compound pesticides)
NPS Method 5 - Direct Aqueous Injection HPLC with Post-Column
Derivatization
(10 N-methylcarbamoyloxime and N-methylcarbamate
pesticides)
National Pesticide Survey: Phase I Report
-------
Appendix E: Survey Analytes and Analytical Chemistry tables E-3
NPS Method 6 - Gas Chromatography with a Nitrogen-Phosphorous
Detector
(Ethylene thiourea)
NPS Method 7 - Microextraction and Gas Chromatography
(5 pesticides, including EDB and DBPC)
NFS Method 8 - Purge and Trap Gas Chromatography (volatile organic
chemicals) Not Used
NFS Method 9 - Automated Cadmium Reduction and Colorimetric
Detection
(Nitrate and nitrite, measured as nitrogen (N))
Detected in NFS - If an analyte was detected at or above the NFS Minimum Reporting Limit
or if a qualitative analyte was detected and confirmed to be present during laboratory analysis,
a "yes" is listed. If an analyte was not detected, a "no" appears in this column. However,
analytes that were not detected may be present in the Survey wells at levels below the NFS
Minimum Reporting Limit.
EPA conducted qualitative analyses for 15 of the 127 analytes. An asterisk (*) appears in this
column to signify the qualitative analytes. In some cases these analytes may be detected, but
a precise concentration can not be determined for these analytes. In other cases, due to the
unreliability of the results of the laboratory method for that analyte and/or the instability of
the analyte, it cannot be determined if the analyte is present or not.
NPS Minimum Reporting Limit - The NFS Minimum Reporting Limit (MRL) is the
minimum concentration level at which an analyte was reported as having been detected. All
analytes except nitrate are reported in micrograms per liter (/tg/L), the equivalent of parts per
billion (ppb). The analytical method for nitrate analyzed the combined presence of nitrate
and nitrite in collected samples and reported the result as nitrogen (N) in mg/L, the
equivalent of parts per million (ppm). For the 15 analytes classified as qualitative, ""
appears in this column to signify that there is no designated NPS Minimum Reporting Limit.
Lifetime Health Advisory Level -- The Lifetime Health Advisory Level (HAL) represents the
maximum concentration of a contaminant in water that may safely be consumed over an
average human lifetime. Lifetime Health Advisory Levels are based on health effects that
were found in humans or animals given high doses of the chemical in laboratory studies.
Health Advisories include a margin of safety to protect human health and should be regarded
as guidelines. Health Advisories are based on available scientific data on non-cancer toxic
effects. Advisories do not exist for all Survey analytes.
Maximum Contaminant Level ~ The Maximum Contaminant Level (MCL), established
pursuant to the Safe Drinking Water Act, is the maximum permissible level of a contaminant
that is allowed in a public water supply. The date in parenthesis indicates that the MCL was
proposed, and provides the date of proposal.
Risk Specific Dose at ID"* Cancer Risk - The Risk Specific Dose (RSD) represents the
concentration of a chemical in drinking water which if consumed for a lifetime would result
in an excess lifetime cancer risk of one in one million. This assumes that all of the exposure
to the chemical is from the drinking water source and a 70 kilogram (150 pound) adult
consumes approximately two liters per day of the contaminated water for 70 years.
National Pesticide Survey: Phase I Report
-------
E-4 Appendix E: Survey Analytes and Analytical Chemistry Tables
Minimum Quantification Limits
Exhibit E-2 presents the Minimum Quantification Limits (MQLs) for each analyte and analysis
method and for both contract and EPA laboratories. The NFS analytic methods were designed to identify both
the presence and concentration of analytes in water samples, whenever possible. For 112 of the 127 analytes,
MQLs were established above which a measured concentration was reported. Analytes detected above the
MQL were reported with quantification (i.e., at measured concentration). The MQL was used to determine
the minimum reporting limit (MRL), established at one-half the MQL. Analytes detected at a concentration
between one-half the established MQL and the MQL were reported as detected without a measured
concentration.
For 15 analytes that were either unstable or lacked reliable method performance, no reporting limit
was established. The confirmed presence of these analytes is reported as "positive detection" without
quantification.
Quality Control Failures
Exhibit E-3 presents the numbers of analyses deleted from the Survey due to failure to meet quality
control guidelines. In addition to the quality control procedures employed in the laboratories, a series of
quality control measures also were implemented for the sample collection and shipping operations. Although
in some cases the judgment of the EPA chemist performing final data review did override the formal quality
control procedures, generally data that failed to meet the established quality control limits were not used in
the analysis of Survey results.
Laboratory Control Standards Precision and Accuracy (10 Times MQL)
Exhibit E-4 presents the results of field samples, fortified at 10 times the MQLs for methods 1-7 and
10 mg/L as nitrogen for method 9, for each analysis method and analyte. To determine the precision and
accuracy of chemical analyses, field samples were fortified with the method analytes at the primary laboratories
(spiked samples). Approximately 40 field samples were fortified at each of three concentrations: 2, 5, and
10 times the MQL for methods 1-7, and 2 times the MQL, 5 and 10 mg/L as nitrogen, for method 9. The
results shown in Exhibit E-4 indicate that excellent precision and accuracy were obtained over the course of
the Survey.
Laboratory Control Standards vs. Spiked Field Sample Results
(Spiking Level 10 Times MQL)
Exhibit E-5 presents the results of analyses of laboratory control standards, fortified at 10 times the
MQL, conducted at both the primary and referee laboratories for each analysis method and analyte. These
results were used to assess the relative precision and accuracy obtained at each laboratory. The data shown
in Exhibit E-5 indicate that similar precision and accuracy were obtained at both the primary and referee
laboratories.
False Negative Rates
Exhibit E-6 presents the results of the comparison of the EPA referee laboratory analyses of each
chemical in samples in which this chemical was not observed by the corresponding primary laboratory.
Duplicate samples were collected at about 10 percent of the field sites for analysis at the referee laboratories.
Taking differences in MQL into account (EPA laboratory MQLs were different from contract laboratory
MQLs for many analytes), an analyte detected in the EPA laboratory analysis was not observed at the contract
laboratory in only two cases out of 200 samples. The largest false negative rate of these two cases - atrazine
- was only 0.98 percent. Thus, there was a maximum possible false negative rate of approximately 1 percent
or less for each analyte.
National Pesticide Survey: Phase I Report
-------
Appendix E: Survey Anarytes and Ana.yt.ca, Chemistry Tables E-5
CB
<
5-S
"x.o
W§
o
Q.
"5
p
si
U**
eeds end grasses
ns, peanuts, rice,
large seeded
Co
In
and
legu
s
Controls annu
weeds In corn,
and peanuts
a
o
11
I:-?
c-i
i §i.
|lf!
fill
o £ J §
Insecticide
Acarlclde
Nematoclde
I
§
I
ll
arb aulf
-88-4)
«.
S
b
(See
arb sul
-87-3)
2)
S
m
b
Controls weeds In
pineapple, sugarcane,
bananas, plantains, co
and potatoes
1
Formerly an ex
herbicide
Atraton
(1610-17-9)
*)
*!
I
m
l* cartain
orghum,
ple, and
912-2
|
Control* weed* In
wheat, lentils, pea
beets, barley, and
arban
01-27-9)
ckroaches, flies,
and turf Insect*
i
CO C-
f
I
National Pesticide Survey: Phase I Report
-------
E-6 Appendix E: Survey Analytes and Analytical Chemistry Tables
(0
T" 2
ui o
.*'§
.Q 0)
!E °-
X T5
UJ g
po
^*3
CO
Analyte
(CAS Befltatry
corn,
, pea
soybe
Be
I
8
ontrols
Controls weeds and
non-crop areas, and
weeds In citrus and
Bromacll
(314-40-9)
1
1
1
1
R
b
o
-
Z
nual grasses and
teded,
i rice and other
S s
£* g
H"H
(SIS§
I
1
|
1
||
ii
i
i
i
«
§
1
8
b
i
-
IE
assy weeds and
i corn
Controls gn
nut grass Ir
I
|
i
Butylate
(200*41 -5)
I
I
^
«
I
1
8
b
o
Z
in
IE
lects In forests,
mentals, shade
angeland
Controls Int
lawns, orna
trees, and r
o
2
j:
c
1
n
s£
1
1
9 tL
t
«.
9
1
2
z
n
0)
Of
tects, mKes,
on corn and
Controls ln<
nematodes
other crops
. £
2 4 "S
ITS!
ill
c t
if O
i ^
1
1
1
1
V.
i
n
O
^^
3
SG.
T
«
I
o
a
f
«
Q
i
A
c
m
5
11
(See C
Carboluran, p
(1563-38-8)
phenol,
1
ley,
ling
Controls sm
oats, wheat,
O
Amiben
Veglben
National Pesticide Survey: Phase I Report
-------
Appendix E: Survey Analytes and Analytical Chemistry Tables E-7
CO
t
"3
o
"^ 0)
T- T3
li o
.«'§
J3 O
SO-
X "Si
UJ
,v
l«
ni
t.
i
if.
l
controls other I
as fire ants a
irground cable
«t!
CM i
O
I
If
I
ls mttes on
, and vegeta
I
R
C)
."o "a
??s
E S rf
I
!l
I
CM
S
i .3
Iff
IrfS
II
|o|
6 o if i
1
Controls an
broadleaf w
cropland
li
II
s?
!i
31
I
g«
I
ols w
and
l
OL<
5 S §
Jill
si
Control
alfalfa,
peanut
£
I
National Pesticide Survey: Phase I Report
-------
E-8 Appendix E: Survey Anatytes and Analytical Chemistry Tables
(D
^ C
li o
s!
IE °-
* -x
uj «
o
i
Controls annual
turf, ornamental
vegetables
Dact
A
1-32
Q.
8
i
(Sea DCPA)
Dacthal acid
metabolites
A acid
bolites
O
iders
i
1
1
1
1
«
6
0
M
U
(See 4,4-DDD)
ff
4
I
S
g
V
2.
V ^*
|
1
1
-------
Append* E: Survey Analytes and An.lyt.ca. Chemtetry Tables E-9
I
CO
3 >
T."0
uj .o
s "§
XI O
CO
i
A §
f*f
ijl
&s
t
if
il
i
s
*
o
b
a
P
2
I
9
Ifs
5 - =
15:
III
if*
I
V)
ll
0 C
if?
II
I
n
ei
i
Hi
I
1
I
I
National Pesticide Survey: Phase I Report
-------
E-10 Appendix E: Survey Analytes and Analytical Chemistry Tables
fl;
(0
0
a
0)
S 2*
c $
^ o>
T 2
LU .y
!S o
Ii
UJ g
and
heat,
Controls broadl
grassy weeds I
barley, and ban
1-1
Is|
^§:-
15*11
>l|l
:n;
2 8 «-|
111!
EndoauHan 1
( - )
(See
i
I
I
I
I
suHan
-07^)
Wfc.
on
and
In
ols Ins
grains,
g
areas
S
(See
I
I
C
Nematocl
Insectici
sects on bananaa
bbage, and corn
ide
Controla Ins
Is an additiv
gasoline
fung
r flo
rose
pota
and
Li
. 0»
of turf,
n, and
Controla di
beans, cor
sorghum
i
I
!
i
S
₯
National Pesticide Survey: Phase I Report
-------
Appendix E: Survey Analytes and Analytical Chemistry Tables E-11
0)
*" 0)
t- T3
*O
A 0)
Z Q-
X
UJ g
o
'§
I**
h*
hi
60
i
1
Ils
111
lil,
air:
II
il
1
1
1
1
8
o
Z
a
st
(See Fenamlpho
o
i.
E
if
1
U1
i
i
|
u.
1
|
1
f
! _
i
i
i
2
5
o
Z
Q
f
(See Fenamlpho
5
&
1
1
i
i
§
1
1
L
i
i_
i
,
t
6
1
E
ri
1-
& >
Protects against
mildew on apple
and roses
|
!
c S
ill
0
1 «
II
1
1
1
i
1
b
Z°
tz
«r
|
52
Controls annual
and broadleaf wi
I
1
o 9
II
is-
h.
i
5
'
1
1
1
8
b
*
%
j'
s|
OB
Controls annual
weeds In cotton
1
s
X
1
T
Is
11
O
I
o
i
i
(Sea
I
I
<
I
ll
!|
*
efS
I
§
i
O
National Pesticide Survey: Phase I Report
-------
E-12 Appendix E: Survey Analytes and Analytical Chemistry Tables
(0
Q)
fQ
fl)
^ 0)
i- T3
ii 'o
.±r'^
n o
£-
UJ «
o
1
1
1
1
200|ig/L
^
«.
;
d
1
E
S.
i
\i\
IP
s
Controls
conifers,
plneappli
1
1
Z
1
>
|«?
*i5 in
3 C
ll
1
1
,
^
M.
$
d
i
E
E :
80
1
|l
1 £
Controls
carrots, c
,
1
z
t
§f>
fc >
3 g
c Q
D 2.
i
l
,
l
0.
0
=£
O
0
S
1
I
1
1
o
li.
I
«
on cherries,
ird repellant
a bi
zo
sectic
-77-5)
§i
.
§
I
8
d
ntrols Ins
ade trees
a variety of
1
1
4
8
8
d
o
Z
E
|f
^
« 3
"D *
III
8
1
i
0
ll
Btolachlor
1218-45-2)
z e.
l
l
d
1
to
o'
o
z.
E
TJ c -O
sifs
Illl
Controls gi
broadleaf \
soybeans,
peas, and
.§
i
1 .
*li
atribuzin
1087-64-9)
ZOt
1
1
1
1
8
d
o
O
'E'
I
(See Metril
^
|
f
I
1
1
etrlbuzln DA
5045-02-4)
S G
1
1
I
t
1
O
!
is
I
g
I
1
i
j-
z _
a
I
II
National Pesticide Survey: Phase I Report
-------
AppendbcE: Survey Analytes and Ana.yt.ca. Chemistry Tables E-13
ill!
(0
I
(0
>»
.£ S
= I
.0 o
IE O-
uj|
o
?
»« i
'I'
a, £
._ TJ
£ S
f*
ols
s and
leav
rice
Mollnat
(8212-8
i
8
o'
Controls annual gra
broadleaf weed*
mlde
9-7)
Na
1 5
C
In
Inorganic fe
N
1
!
lon
a v
aa ap
on p
Oagradate of
pesticides, c
,
such
uitoe
of Insect
and mosqui
and apples
ill
ill
4-
(1
S
on co
, and
Controls Ins
stone fruits,
cranberries
and
i!
||e
tf
II
I
o
b
o
o
II
II
I!
National Pesticide Survey: Phase I Report
-------
E-14 Appendix E: Survey Anaiytes and Analytical Chemistry Tables
ft
"5
ir
!n o
'E Q-
x =
UJ g
o
1
ls ann
, and
r bee
In s
tom
Pebulate
(1114-71-2)
I
I
I
1
8
d
£
CM
CC
Controls a broad range of
Insects In cotton
Insecticide
Ambush
Perthrlne
cls-Permethrin
(61949-78-6)
I
I
I
1
o
CM
i
CM
CC
(See cls-Permethrin)
f
5
(See clt-Par
o *
II
c
trana-Permethrl
(61949-77-7)
I
flk «£.
i in
§w
t
1
§
1
s?
d
^
CO
C
Controls broadleaf and
woody plants In pastures
and rangeland
Herbicide
§
D
5
ll
.0 5>
I
1
1
8
1
K5
6
-
CC
Controls perennials,
broadleaf weeds, and
grasses In non-crop areas
Herbicide
Qesagram
Prometon
(161 0-1 84)
1
1
I
1
o
d
o
-
cc
Controls weeds In cotton,
peas, canots, and
vegetables
Herbicide
Qesagard
Caparol
f-f
1
1
1
8
i
t
-
cc
Controls weeds and grass In
lettuce, legumes, and trees
Herbicide
j
Pronamlde
(23950-58-5)
1
1
I
<
ri
i
»
0
(See Pronamlde)
*"
8
g
i
£
1
«s»
(See PronamWe
5 *.
1 ?
« e-
Pronamtde me
(N-(1,1-
dlmethylacetor
8
Propachlor
(1916-16-7)
.30
»g/L
Controls annual
weeds and gras
m
Mllogard
National Pesticide Survey: Phase I Report
-------
Appendix E: Survey Ana«ytes and Analytical Chemistry fables E-15
0)
p
₯<
ifr
8e/>
^
VS
LU .O
s w
£ O
2 °-
uj 1
o
1 1
1 1
\t
1 »
|§
\t
1 »
In
1 "^
b
i «
1 °
Controls weeds In alfalfa.
lettuce, spinach, sugar
beets, lentils, and peas
1
I
1
IPC
Beet-Kleen
1
Propham
(122-12-9)
1
*l
- t.
t
*
*
s
0
f
^
o:
Controls annual grasses and
weeds In crops, especially
com, and fruit such as
citrus, asparagus, and nuts
1
I
Mti
il JJ
SImazlne
(122-34-9)
1
1
1
£
*
§
**
6
Z
Z
Controls broadleaf weeds In
rice
I
!
Slmehyn
(1014-70-6)
1
1
1
s
«l
s
0
Z
^
(C
Controls Insects on corn,
cotton, vines, and fruM
i
!
!
II
of
il
1
1
1
a
a
£
0
Z
*
Z
Controls weeds on rice,
corn, peas, and peanuts
£
i
&
i
Swep
(1918-16.9)
1
I
1
E
*
8
6
Z°
CO
O
Controls woody plants In
Industrial areas
1
f
t
1
.
2.4.5-T
(Trtchforophenc-
xyacetlc acid)
(93-76-5)
!
!
1
§
1
8
0
J
cr
Controls vegetative weeds In
non-crop and rangeland
1
|
It
Tebuthluron
(34014-18-1)
1
1
1
s
1
i
£
^
cc
Controls annual and
perennial weeds In sugar
cane, alfalfa, apples,
peaches, citrus, pecans, and
mint
j
*
i
07
TerbacH
(5902-51-2)
I
/
i
0)
6
I
«
^-
K
Controls sol) Insects and
nematodes on com,
vegetables, and sorghum
I
I
Counter
Terbufo*
(13071-79-9)
1
1
I
1
s
0
Z
^
U
Controls weeds In winter
wheat and barley
1
C
II
Terbutryn
(886-504))
)
_r 6f
If
s!
1
8
1
S
0
£
-------
E-16 Appendix E; Survey Analytes and Analytical Chemistry Tables
(0
0)
CO
^<
s>
£ S
c ^
8^
^ 0)
T- 2
ill 'o
±:1
.a o
Z°-
X TS
U g
.0
*3
(0
-V
If*
Eii
*L
2:
I
I
III
n a
li
-
2 *
III
o 8
IS
li
2
O
a>
f~u
.1-
CD CD 00
5-
§J
« J
II
c
i-
s jf
o «
Hi
IM
»-£ S.
2*5
I
a>
o
"
f
5
I
s
s
s
National Pesticide Survey: Phase I Report
-------
Appendix E: Survey Analytes and Analytical Chemistry tables E-17
Exhibit E-2
Minimum Quantification Limits (MQL)
Analyte
METHOD 1
Alachlor
Ametryn
Atraton
Atrazine
Bromacil
Butachlor
Butylate
Carboxin
Chlorpropham
Cycloate
Diazinon
Dichlorvos
Diphenamid
Disulfoton
Disulfoton sulfone
Disulfoton sulfoxide
EPIC
Ethoprop
Fenamiphos
Fenarimol
Fluridone
Hexazinone
Merphos
Methyl paraoxon
Metolachlor
Contract Lab
MQL (pg/L)
1.0
0.26
0.34
0.24
2.2
1.5
0.60
1.0
0.70
0.40
Qualitative
0.24
0.43
Qualitative
Qualitative
Qualitative
0.30
0.12
0.30
0.40
1.8
0.27
Qualitative
0.30
1.5
EPA Lab
MQL frig/L)
1.5
0.90
0.35
0.30
16
1.8
1.2
6.0
4.0
0.80
Qualitative
2.4
1.0
Qualitative
Qualitative
Qualitative
0.60
0.30
2.2
0.60
3.8
0.60
Qualitative
12
4.4
National Pesticide Survey: Phase I Report
-------
E-18 Appendix E: Survey Analytes and Analytical Chemistry Tables
Exhibit E-2 (continued)
Minimum Quantification Limits (MQL)
Analyte
Metribuzin
Mevinphos
MGK264
Molinate
Napropamide
Norflurazon
Pebulate
Prometon
Prometryn
Pronamide
Propazine
Simetryn
Simazine
Stirofos
Tebuthiuron
Terbacil
Terbufos
Terbutryn
Triademefon
Tricyclazole
Vernolate
METHOD 2
Aldrin
alpha-Chlordane
gamma-Chlordane
Chlorneb
Chlorobenzilate
Contract Lab
MQL 0/g/L)
0.36
0.30
2.1
0.36
0.50
0.36
0.38
0.29
0.20
Qualitative
0.20
0.10
0.75
0.36
0.45
3.5
Qualitative
0.30
0.32
1.2
0.37
0.12
0.12
0.12
1.4
Qualitative
EPA Lab
MQL Qjg/L)
0.80
2.3
4.0
0.79
0.80
0.60
0.80
0.30
0.80
Qualitative
0.30
0.90
0.29
8.0
0.40
16
Qualitative
0.79
0.40 -
8.0
1.2
0.050
0.0050
0.010
0.16
Qualitative
National Pesticide Survey: Phase I Report
-------
Appendix E: Survey Analytes and Analytical Chemistry fables E-19
Exhibit E-2 (continued)
Minimum Quantification Umits (MQL)
Analyte
Chlorothalonil
DCPA
4,4-DDD
4,4-DDE
4,4-DDT
Dieldrin
Endosulfan I
Endosulfan II
Endosulfan sutfate
Endrin
Endrin aldehyde
Etridiazole
alpha-HCH
beta-HCH
delta-HCH
gamma-HCH
Heptachlor
Heptachlor-epoxide
Hexachlorobenzene
Methoxychlor
cis-Permethrin
trans-Permethrin
Propachlor
Trifluralin
Contract Lab
MQL 0/g/L)
0.12
0.12
0.25
0.12
0.30
0.12
0.12
0.25
0.25
0.25
0.25
0.25
0.12
0.12
Qualitative
0.085
0.12
0.12
0.12
0.60
1.8
3.9
1.3
0.25
EPA Lab
MQL fc/g/L)
0.070
0.015
0.015
0.025
0.18
0.10
0.025
0.025
0.040
0.010
0.020
0.070
0.010
0.010
Qualitative
0.0050
0.055
0.020
0.010
0.41
1.8
4.4
0.55
0.015
National Pesticide Survey: Phase I Report
-------
E-20 Appendix E: Survey Analytes and Analytical Chemistry Tables
Exhibit E-2 (continued)
Minimum Quantification Limits (MQL)
Analyte
METHOD 3
Acifluorfen
2,4-DB
Bentazon
Chloramben
2,4-D
Dalapon
DCPA acid metabolites
Dicamba
3,5-Dichlorobenzoic acid
Dichlorprop
Dinoseb
5-Hydroxy Dicamba
4-Nitrophenol
PCP
Picloram
2,4,5-T
2,4,5-TP
METHOD 4
Atrazine dealkylated
Barban
Carbofuran phenol
Cyanazine
Diuron
Fenamiphos sulfone
Fenamiphos sulfoxide
Fluometuron
Contract Lab
MQL fr/g/L)
Qualitative
2.0
0.50
Qualitative
0.50
Qualitative
0.20
0.20
0.60
0.50
2.5
0.20
Qualitative
0.20
1.0
0.20
0.20
4.4
3.8
42
4.7
0.63
57
9.4
0.95
EPA Lab
MQL (jug/L)
Qualitative
3.2
2.0
Qualitative
1.6
Qualitative
0.20
0.40
0.80
1.0
1.0
3.2
Qualitative
0.20
1.6
0.40
0.40
3.1
10
18
4.9
0.52
62
40
1.3
National Pesticide Survey: Phase I Report
-------
Appendix E: Survey Analytes and Analytical Chemistry Tables E-21
Exhibit E-2 (continued)
Minimum Quantification Limits (MQL)
Analyte
3-Ketocarbofuran phenol
Linuron
Metribuzin DA
Metribuzin DADK
Metribuzin DK
Neburon
Pronamide metabolite
Propanil
Propham
Swep
METHOD 5
Aldicarb
Aldicarb sulfone
Aldicarb sulfoxide
Baygon
Carbaryl
Carbofuran
3-Hydroxycarbofuran
Methiocarb
Methomyl
Oxamyl
METHOD 6
ETU
Contract Lab
MQL (//g/L)
1.8
0.95
1.2
Qualitative
Qualitative
0.60
6.2
0.60
11
0.30
1.4
1.2
1.7
1.9
1.2
2.3
2.1
2.9
1.2
1.9
9.0
EPA Lab
MQL fr/g/L)
2.2
0.56
0.90
Qualitative
Qualitative
0.62
4.6
0.56
10
0.87
2.6
2.0
2.4
3.1
1.7
3.2
3.4
5.0
1.8
2.7
7.5
National Pesticide Survey: Phase I Report
-------
E-22 Appendix E: Survey Analytes and Analytical Chemistry Tables
Exhibit E-2 (continued)
Minimum Quantification Limits (MQL)
Analyte
METHOD 7
EDB
DBCP
1 ,2-Dichloropropane
cis-1 ,3-Dichloropropene
trans-1 ,3-Dichloropropene
Contract Lab
MQL fcig/L)
0.020
0.020
1.5
0.020
0.19
EPA Lab
MQL 0/g/L)
0.042
0.027
7.5
0.29
0.36
Analyte
METHOD 9
Nitrate
Contract Lab
MQL (mg/L)
0.30
EPA Lab
MQL (mg/L)
0.30
National Pesticide Survey: Phase I Report
-------
Appendix E: Survey Analytes and Analytical Chemistry Tables E-23
Exhibit E-3
Quality Control Failures
Analyte
METHOD 1
Alachlor
Ametryn
Atraton
Atrazine
Bromacil
Butachlor
Butyfate
Carboxin
Chlorpropham
Cycloate
Diazinon
Dichlorvos
Diphenamid
Disulfoton
Disulfoton sulfone
Disuffoton sulfoxide
EPTC
Ethoprop
Fenamiphos
Fenarimol
Fluridone
Hexazinone
Merphos
Methyl paraoxon
Metolachlor
Total No.
Deleted
112
134
151
124
99
157
142
150
125
119
92
117
96
92
92
92
110
126
122
120
125
132
92
132
135
No. Laboratory
Related
54
76
93
66
41
99
84
92
67
61
34
59
38
34
34
34
52
68
64
62
67
74
34
74
77
No. Sampling
Related
58
58
58
58
58
58
58
58
58
58
58
58
58
58
58
58
58
58
58
58
58
58
58
58
58
National Pesticide Survey: Phase I Report
-------
E-24 Appendix E: Survey Analytes and Analytical Chemistry Tables
Exhibit E-3 (continued)
Quality Control Failures
Analyte
Metribuzin
Mevinphos
MGK264
Molinate
Napropamide
Norflurazon
Pebulate
Prometon
Prometryn
Pronamide
Propazine
Simetryn
Simazine
Stirofos
Tebuthiuron
Terbacil
Terbutos
Terbutryn
Triademefon
Tricyclazole
Vernolate
METHOD 2
Aldrin
alpha-Chlordane
gamma-Chlordane
Chlorneb
Chlorobenzilate
Total No.
Deleted
102
100
136
136
136
123
150
96
92
92
123
110
153
109
100
127
92
119
99
117
116
82
78
78
76
76
No. Laboratory
Related
44
42
78
78
78
65
92
38
34
34
65
52
95
51
42
69
34
61
41
59
58
25
21
21
19
19
No. Sampling
Related
58
58
58
58
58
58
58
58
58
58
58
58
58
58
58
58
58
58
58
58
58
57
57
57
57
57
National Pesticide Survey: Phase I Report
-------
Appendix E: Survey Analytes and Analytical Chemistry tables E-25
Exhibit E-3 (continued)
Quality Control Failures
Analyte
Chlorothalnil
DCPA
4,4'-DDD
4,4'-DDE
4,4'-DDT
Dieldrin
Endosulfan 1
Endosulfan II
Endosulfan sulfate
Endrin
Endrin aldehyde
Etridiazole
alpha-HCH
beta-HCH
delta-HCH
gamma-HCH
Heptachlor
Heptachlor epoxide
Hexachlorobenzene
Methoxychlor
cis-Permethrin
trans-Permethrin
Propachlor
Trifluralin
METHOD 3
Acifluorfen
2,4-DB
Total No.
Deleted
119
78
78
78
82
76
76
76
78
82
76
80
84
78
76
82
78
76
80
84
80
86
76
82
69
83
No. Laboratory
Related
62
21
21
21
25
19
19
19
21
25
19
23
27
21
19
25
21
19
23
27
23
29
19
25
11
25
No. Sampling
Related
57
57
57
57
57
57
57
57
57
57
57
57
57
57
57
57
57
57
57
57
57
57
57
57
58
58
National Pesticide Survey: Phase I Report
-------
E-26 Appendix E: Survey Analytes and Analytical Chemistry Tables
Exhibit E-3 (continued)
Quality Control Failures
Analyte
Bentazon
Chloramben
2,4-D
Dalapon
DCPA acid metabolites
Dicamba
3,5-Dichlorbenzoic acid
Dichloroprop
Dinoseb
5-Hydroxy Dicamba
4-Nitrophenol
PCP
Picloram
2,4,5-T
2,4,5-TP
METHOD 4
Atrazine dealkylated
Barban
Carbofuran phenol
Cyanzine
Diuron
Fenamiphos suffone
Fenamiphos sulfoxide
Fluometuron
3-Ketocarbofuran phenol
Linuron
Metribuzin DA
Total No.
Deleted
89
69
98
69
71
87
82
94
108
91
69
120
100
82
69
81
66
80
70
71
93
68
72
65
79
102
No. Laboratory
Related
31
11
40
11
13
29
24
36
50
33
11
62
42
24
11
22
7
21
11
12
34
9
13
6
20
43
No. Sampling
Related
58
58
58
58
58
58
58
58
58
58
58
58
58
58
58
59
59
59
59
59
59
59
59
59
59
59
National Pesticide Survey: Phase I Report
-------
Appendix E: Survey Analytes and Analytical Chemistry tables E-27
Exhibit E-3 (continued)
Quality Control Failures
Analyte
Metribuzin DADK
Metribuzin DK
Neburon
Pronamide metabolite
Propanil
Propham
Swep
METHOD 5
Aldicarb
Aldicarb sulfone
Aldicarb sulfoxide
Baygon
Carbaryl
Carbofuran
3-Hydroxycarbofuran
Methiocarb
Methomyl
Oxamyl
METHOD 6
ETU
METHOD 7
EDB
DBCP
1 ,2-Dichloropropane
cis-1 ,3-Dichloropropene
trans-1 ,3-Dichloropropene
Total No.
Deleted
65
65
65
65
71
73
68
60
61
61
60
60
60
60
60
60
60
79
67
67
67
67
67
No. Laboratory
Related
6
6
6
6
12
14
9
0
1
1
0
0
0
0
0
0
0
16
9
9
9
9
9
No. Sampling
Related
59
59
59
59
59
59
59
60
60
60
60
60
60
60
60
60
60
63
58
58
58
58
58
National Pesticide Survey: Phase I Report
-------
E-28 Appendix E: Survey Analytes and Analytical Chemistry Tables
Exhibit E-3 (continued)
Quality Control Failures
Analyte
METHOD 9
Nitrate
Total No.
Deleted
58
No. Laboratory
Related
1
No. Sampling
Related
57
National Pesticide Survey: Phase I Report
-------
Appendix E: Survey Analytes and Analytical Chemistry fables E-29
Exhibit E-4
Laboratory Control Standards Precision and Accuracy (10 Times MQL)
Analyte
METHOD 1
Alachlor
Ametryn
Atraton
Atrazine
Bromacil
Butachlor
Butylate
Carboxin
Chlorpropham
Cycloate
Diazinon
Dichlorvos
Diphenamid
Disulfoton
Disulfoton sulfone
Disulfoton suffoxide
EPTC
Ethoprop
Fenamiphos
Fenarimol
Fluridone
Hexazinone
Merphos
Methyl paraoxon
Contract Lab
Average
Recovery
97
96
108
98
96
96
82
92
98
94
64
97
96
77
92
96
92
97
91
98
96
96
104
102
% Relative
Standard
Deviation
5.2
6.5
7.4
5.3
7.2
6.1
10.4
11.3
6.5
5.4
60.6
5.3
7.0
25.4
10.7
9.8
4.5
5.3
13.6
7.6
9.7
9.8
32.8
8.8
EPA Lab
Average
Recovery
92
90
91
93
91
92
84
87
94
91
48
93
94
67
79
76
89
90
87
93
94
93
122
84
% Relative
Standard
Deviation
6.4
7.2
6.2
5.9
7.1
8.0
8.2
9.7
5.6
5.5
72.1
10.0
5.7
37.4
35.4
17.3
13.0
7.8
13.7
6.8
10.5
9.3
15.4
15.5
National Pesticide Survey: Phase I Report
-------
E-30 Appendix E: Survey Analytes and Analytical Chemistry Tables
Exhibit E-4 (continued)
Laboratory Control Standards Precision and Accuracy (10 Times MQL)
Analyte
Metolachlor
Metribuzin
Mevinphos
MGK264
Molinate
Napropamide
Norflurazon
Pebulate
Prometon
Prometryn
Pronamide
Propazine
Simetryn
Simazine
Stirofos
Tebuthiuron
Terbacil
Terbufos
Terbutryn
Triademefon
Tricyclazole
Vemolate
METHOD 2
Aldrin
alpha-Chlordane
Contract Lab
Average
Recovery
97
95
97
90
95
98
98
90
96
96
37
97
96
97
99
96
98
47
96
98
97
89
84
96
% Relative
Standard
Deviation
6.0
5.4
5.4
9.4
5.1
8.3
7.8
6.2
5.6
5.9
92.4
4.9
6.7
5.8
8.0
6.6
7.1
75.0
6.1
6.5
9.9
6.4
12.3
13.7
EPA Lab
Average
Recovery
93
89
88
91
91
90
91
89
92
92
65
93
90
94
92
86
90
36
92
93
92
88
80
86
% Relative
Standard
Deviation
6.0
6.9
13.9
6.0
5.9
9.9
10.1
5.8
6.2
6.6
68.5
5.9
7.4
5.8
12.5
8.1
11.5
89.5
6.6
6.2
10.5
6.4
11.5
8.4
National Pesticide Survey: Phase I Report
-------
Appendix E: Survey Anaiytes and Analytical Chemistry tables E-31
Exhibit E-4 (continued)
Laboratory Control Standards Precision and Accuracy (10 Times MQL)
Analyte
gamma-Chlordane
Chlomeb
Chlorobenzilate
Chlorothaloni!
DCPA
4,4-DDD
4,4-DDE
4,4-DDT
Dieldrin
Endosuffan I
Endosutfan II
Endosulfan sulfate
Endrin
Endrin aldehyde
Etridiazole
alpha-HCH
beta-HCH
defta-HCH
gamma-HCH
Heptachlor
Heptachlor-epoxide
Hexachlorobenzene
Methoxychlor
cis-Permethrin
trans-Permethrin
Contract Lab
Average
Recovery
100
95
97
74
103
99
97
95
94
99
93
97
97
93
102
95
100
95
112
97
93
91
108
109
106
% Relative
Standard
Deviation
14.6
9.9
44.7
16.1
14.1
20.2
19.6
21.0
13.3
14.1
15.3
19.1
16.4
16.6
12.9
10.5
13.1
27.6
10.4
13.1
11.1
25.2
22.4
24.8
22.9
EPA Lab
Average
Recovery
86
106
96
111
90
99
85
103
98
97
98
86
99
94
106
97
88
94
97
77
97
79
89
99
89
% Relative
Standard
Deviation
10.9
12.1
17.3
13.8
6.3
18.5
12.4
20.4
15.4
12.2
19.1
19.5
14.1
17.0
20.0
11.8
9.2
10.5
9.5
9.9
11.4
18.4
17.0
18.4
15.8
National Pesticide Survey: Phase I Report
-------
E-32 Appendix E: Survey Analytes and Analytical Chemistry Tables
Exhibit E-4 (continued)
Laboratory Control Standards Precision and Accuracy (10 Times MQL)
Analyte
Propachlor
Trifluralin
METHOD 3
Acifluorfen
2,4-DB
Bentazon
Chloramben
2,4-D
Dalapon
DCPA acid metabolites
Dicamba
3,5-Dichlorobenzoic acid
Dichlorprop
Dinoseb
5-Hydroxy Dicamba
4-Nitrophenol
POP
Picloram
2,4,5-T
2,4,5-TP
METHOD 4
Atrazine dealkylated
Barban
Carbofuran phenol
Cyanazine
Contract Lab
Average
Recovery
102
103
69
98
100
10
97
105
94
95
94
99
81
87
96
86
84
96
98
85
91
93
94
% Relative
Standard
Deviation
8.4
16.2
41.3
10.7
8.2
119.3
9.1
19.0
8.8
9.3
9.7
10.1
24.4
13.9
19.1
22.7
18.7
8.6
8.0
11.0
7.5
11.5
7.0
EPA Lab
Average
Recovery
100
81
51
90
92
53
93
49
93
89
73
91
40
94
75
73
87
99
89
94
101
100
100
% Relative
Standard
Deviation
10.7
15.7
24.5
4.1
3.8
30.3
4.0
29.1
4.5
7.2
9.3
3.9
30.9
3.4
15.1
7.5
7.9
8.5
6.0
5.6
2.9
5.2
3.8
National Pesticide Survey: Phase I Report
-------
Appendix E: Survey Analytes and Analytical Chemistry Tables E-33
Exhibit E-4 (continued)
Laboratory Control Standards Precision and Accuracy (10 Times MQL)
Analyte
Diuron
Fenamiphos sutfone
Fenamiphos sutfoxide
Fluorneturon
3-Ketocarbofuran phenol
Linuron
Metribuzin DA
Metribuzin DADK
Metribuzin DK
Neburon
Pronamide metabolite
Propanil
Propham
Swep
METHOD 6
ETU
Contract Lab
Average
Recovery
92
94
93
89
90
91
83
20
34
88
97
93
92
94
91
% Relative
Standard
Deviation
8.0
13.4
12.2
7.4
6.2
7.5
11.9
13.9
14.7
7.8
7.9
8.7
6.7
10.3
5.9
EPA Lab
Average
Recovery
100
102
100
101
98
102
97
.
.
99
100
100
99
100
96
% Relative
Standard
Deviation
3.6
3.8
4.3
2.6
5.7
4.1
6.3
.
.
1.9
4.2
3.5
2.3
3.6
9.3
National Pesticide Survey: Phase I Report
-------
E-34 Appendix E: Survey Analytes and Analytical Chemistry Tables
Exhibit E-5
Laboratory Control Standards vs. Spiked Field Sample Results
(Spiking Level 10 Times MQL)
Analyte
METHOD 1
Alachlor
Ametryn
Atraton
Atrazine
Bromacil
Butachlor
Butylate
Carboxin
Chlorpropham
Cycloate
Diazinon
Dichiorvos
Diphenamid
Disulfoton
Disutfoton sutfone
Disutfoton sulfoxide
EPIC
Ethoprop
Fenamiphos
Fenarimol
Fluridone
Hexazinone
Merphos
Methyl paraoxon
Spiked Samples
Average
Recovery
98
98
110
100
98
97
88
89
99
94
87
97
99
83
88
91
94
99
95
101
97
97
80
109
% Relative
Standard
Deviation
3.6
9.7
12.1
4.2
5.9
10.0
6.0
7.6
5.6
5.5
14.3
5.9
5.6
13,1
10.5
10.6
3.7
2.9
16.2
10.4
11.7
10.6
27.2
6.8
Lab Control Standard
Average
Recovery
97
96
108
98
96
96
82
92
98
94
64
97
96
77
92
96
92
97
91
98
96
96
104
102
% Relative
Standard
Deviation
5.2
6.5
7.4
5.3
7.2
6.1
10.4
11.3
6.5
5.4
60.6
5.3
7.0
25.4
10.7
9.8
4.5
5.3
13.6
7.6
9.7
9.8
32.8
8.8
National Pesticide Survey: Phase I Report
-------
Appendix E: Survey Analytes and Analytical Chemistry tables E-35
Exhibit E-5 (continued)
Laboratory Control Standards vs. Spiked Field Sample Results
(Spiking Level 10 Times MQL)
Analyte
Metolachlor
Metribuzin
Mevinphos
MGK264
Molinate
Napropamide
Norflurazon
Pebulate
Prometon
Prometryn
Pronamide
Propazine
Simetryn
Simazine
Stirofos
Tebuthiuron
Terbacil
Terbufos
Terbutfyn
Triademefon
Tricyclazole
Vernolate
METHOD 2
Aldrin
alpha-Chlordane
Spiked Samples
Average
Recovery
98
95
98
88
96
101
99
91
97
96
27
98
96
97
104
95
101
45
98
99
104
90
80
92
% Relative
Standard
Deviation
9.5
6.2
6.2
11.9
3.8
10.2
9.2
6.0
7.2
6.9
73.7
6.0
8.3
8.1
5.7
7.0
9.1
74.9
4.0
8.5
10.7
6.1
12.0
11.2
Lab Control Standard
Average
Recovery
97
95
97
90
95
98
98
90
96
96
37
97
96
97
99
96
98
47
96
98
97
89
84
96
% Relative
Standard
Deviation
6.0
5.4
5.4
9.4
5.1
8.3
7.8
6.2
5.6
5.9
92.4
4.9
6.7
5.8
8.0
6.6
7.1
75.0
6.1
6.5
9.9
6.4
12.3
13.7
National Pesticide Survey: Phase I Report
-------
E-36 Appendix E: Survey Analytes and Analytical Chemistry Tables
Exhibit E-5 (continued)
Laboratory Control Standards vs. Spiked Field Sample Results
(Spiking Level 10 Times MQL)
Analyte
gamma-Chlordane
Chlorneb
Chlorobenzilate
Chlorothalonil
DCPA
4,4-DDD
4,4-DDE
4,4-DDT
Dieldrin
Endosulfan I
Endosulfan II
Endosulfan sutfate
Endrin
Endrin aldehyde
Etridiazole
alpha-HCH
beta-HCH
delta-HCH
gamma-HCH
Heptachlor
Heptachlor-epoxide
Hexachlorobenzene
Methoxychlor
cis-Permethrin
trans-Permethrin
Spiked Samples
Average
Recovery
97
94
108
75
109
92
89
92
91
96
90
94
94
92
105
92
100
95
111
96
91
96
118
103
96
% Relative
Standard
Deviation
11.1
11.6
18.1
15.7
10.0
17.7
16.1
21.1
12.3
13.5
18.0
16.9
17.2
15.0
13.4
8.3
10.7
11.4
12.2
8.2
11.0
9.2
30.2
25.9
27.4
Lab Control Standard
Average
Recovery
100
95
97
74
103
99
97
95
94
99
93
97
97
93
102
95
100
95
112
97
93
91
108
109
106
% Relative
Standard
Deviation
14.6
9.9
44.7
16.1
14.1
20.2
19.6
21.0
13.3
14.1
15.3
19.1
16.4
16.6
12.9
10.5
13.1
27.6
10.4
13.1
11.1
25.2
22.4
24.8
22.9
National Pesticide Survey: Phase I Report
-------
Appendix E: Survey Analytes and Analytical Chemistry fables E-37
Exhibit E-5 (continued)
Laboratory Control Standards vs. Spiked Field Sample Results
(Spiking Level 10 Times MQL)
Analyte
Propachlor
Trifluralin
METHOD 3
Acffluorfen
2,4-DB
Bentazon
Chloramben
2,4-D
Dalapon
DCPA acid metabolites
Dicamba
3,5-Dichlorobenzoic acid
Dichlorprop
Dinoseb
5-Hydroxy Dicamba
4-Nitrophenol
PCP
Picloram
2,4,5-T
2,4,5-TP
METHOD 4
Atrazine dealkylated
Barban
Carbofuran phenol
Cyanazine
Spiked Samples
Average
Recovery
100
108
76
100
101
11
98
100
97
98
95
98
84
90
89
81
82
99
100
83
91
95
94
% Relative
Standard
Deviation
9.4
16.1
50.1
12.9
10.1
120.1
12.9
19.2
8.8
6.4
10.5
10.3
25.3
10.9
36.3
34.4
15.9
8.4
7.9
7.5
8.8
10.9
6.5
Lab Control Standard
Average
Recovery
102
103
69
98
100
10
97
105
94
95
94
99
81
87
96
86
84
96
98
85
91
93
94
% Relative
Standard
Deviation
8.4
16.2
41.3
10.7
8.2
119.3
9.1
19.0
8.8
9.3
9.7
10.1
24.4
13.9
19.1
22.7
18.7
8.6
8.0
11.0
7.5
11.5
7.0
National Pesticide Survey: Phase I Report
-------
E-38 Appendix E: Survey Analytes and Analytical Chemistry Tables
Exhibit E-5 (continued)
Laboratory Control Standards vs. Spiked Field Sample Results
(Spiking Level 10 Times MQL)
Analyte
Diuron
Fenamiphos sutfone
Fenamiphos sutfoxide
Fluometuron
3-Ketocarbofuran phenol
Linuron
Metribuzin DA
Metribuzin DADK
Metribuzin DK
Neburon
Pronamide metabolite
Propanil
Propham
Swep
METHOD 5
Aldicarb
Aldicarb sutfone
Aldicarb sufoxide
Baygon
Carbaryl
Carbofuran
3-Hydroxycarbofuran
Methiocarb
Methomyl
Oxamyl
Spiked Samples
Average
Recovery
94
94
95
91
89
91
79
17
89
98
95
93
95
100
101
100
99
97
99
99
94
102
98
% Relative
Standard
Deviation
5.2
6.8
7.8
7.2
5.9
7.1
10.6
.
7.5
3.5
3.9
5.4
7.4
7.0
5.3
6.9
4.3
5.6
4.0
4.0
4.7
4.8
4.3
Lab Control Standard
Average
Recovery
92
94
93
89
90
91
83
20
34
88
97
93
92
94
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
% Relative
Standard
Deviation
8.0
13.4
12.2
7.4
6.2
7.5
11.9
13.9
14.7
7.8
7.9
8.7
6.7
10.3
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
National Pesticide Survey: Phase I Report
-------
Appendix E: Survey Analytes and Analytical Chemistry Tables E-39
Exhibit E-5 (continued)
Laboratory Control Standards vs. Spiked Field Sample Results
(Spiking Level 10 Times MQL)
Analyte
METHOD 6
ETU
METHOD 7
EDB
DBCP
1 ,2-Dichloropropane
cis-1 ,3-Dichloropropene
trans- 1 ,3-Dichloropropene
METHOD 9
Nitrate
Spiked Samples
Average
Recovery
93
97
95
96
97
102
103
% Relative
Standard
Deviation
7.4
7.7
10.1
10.5
8.8
10.5
5.9
Lab Control Standard
Average
Recovery
91
NA
NA
NA
NA
NA
NA
% Relative
Standard
Deviation
5.9
NA
NA
NA
NA
NA
NA
National Pesticide Survey: Phase I Report
-------
E-40 Appendix E: Survey Analytes and Analytical Chemistry Tables
Exhibit-6
False Negative Rates
Analyte
METHOD 1
Alachlor
Ametryn
Atraton
Atrazine
Bromacil
Butachlor
Butylate
Carboxin
Chlorpropham
Cycloate
Diazinon
Dichlorvos
Diphenamid
Disutfotbn
Disulfoton sulfone
Disulfoton sutfoxide
EPIC
Ethoprop
Fenamiphos
Fenarimol
Fluridone
Hexazinone
Merphos
Methyl paraoxon
Metolachlor
Metribuzin
No. of
Duplicates
111
108
105
108
111
103
108
97
112
110
112
101
112
112
112
112
100
110
100
109
97
100
112
107
105
110
No. of False
Negatives
0
0
0
1
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
% of False
Negatives
<0.89
<0.92
<0.94
0.93
<0.89
<0.96
<0.92
<1.0
<0.88
<0.90
<0.88
<0.98
<0.88
<0.88
<0.88
<0.88
<0.99
<0.90
<0.99
<0.91
<1.0
<0.99
<0.88
<0.93
<0.94
<0.90
National Pesticide Survey: Phase I Report
-------
Appendix E: Survey Analytes and Analytical Chemistry tables E-41
Exhibit-6 (continued)
False Negative Rates
Analyte
Mevinphos
MGK264
Molinate
Napropamide
Norflurazon
Pebulate
Prometon
Prometryn
Pronamide
Propazine
Simetryn
Simazine
Stirofos
Tebuthiuron
Terbacil
Terbufos
Terbutryn
Triademefon
Tricyclazole
Vernolate
METHOD 2
AJdrin
alpha-Chlordane
gamma-Chlordane
Chlorneb
Chlorobenzilate
Chtorothalonil
No. of
Duplicates
112
111
104
90
93
108
109
108
112
107
107
105
102
103
109
112
108
111
109
112
119
118
120
117
120
113
No. of False
Negatives
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
% of False
Negatives
<0.88
<0.89
<0.95
<1.1
<1.1
<0.92
<0.91
<0.92
<0.88
<0.93
<0.93
<0.94
<0.97
<0.96
<0.91
<0.88
<0.92
<0.89
<0.91
<0.88
<0.83
<0.84
<0.83
<0.85
<0.83
<0.88
National Pesticide Survey: Phase I Report
-------
E-42 Appendix E: Survey Analytes and Analytical Chemistry Tables
Exhibit-6 (continued)
False Negative Rates
Analyte
DCPA
4,4-DDD
4,4-DDE
4,4-DDT
Dieldrin
Endosulfan 1
Endosulfan II
Endosulfan sulfate
Endrin
Endrin aldehyde
Etridiazole
alpha-HCH
beta-HCH
delta-HCH
gamma-HCH
Heptachlor
Heptachlor-epoxide
Hexachlorobenzene
Methoxychlor
cis-Permethrin
trans-Permethrin
Propachlor
Trifluralin
METHOD 3
Acifluorfen
2,4-DB
Bentazon
No. of
Duplicates
120
120
120
118
118
120
120
116
120
118
113
118
120
120
120
120
120
119
120
117
119
120
117
133
132
132
No. of False
Negatives
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
% of False
Negatives
<0.83
<0.83
<0.83
<0.84
<0.84
<0.83
<0.83
<0.85
<0.83
<0.84
<0.88
<0.84
<0.83
<0.83
<0.83
<0.83
<0.83
<0.83
<0.83
<0.85
<0.83
<0.83
<0.85
<0.75
<0.75
<0.75
National Pesticide Survey: Phase I Report
-------
Appendix E: Survey Analytes and Analytical Chemistry fables E-43
Exhibit-6 (continued)
False Negative Rates
Analyte
Chloramben
2,4-D
Dalapon
DCPA acid metabolites
Dicamba
3,5-D/chlorobenzoic acid
Dichlorprop
Dinoseb
5-Hydroxy Dicamba
4-Nrtrophenol
PCP
Picloram
2,4,5-T
2,4,5-TP
METHOD 4
Atrazine dealkylated
Barban
Carbofuran phenol
Cyanazine
Diuron
Fenamiphos sulfone
Fenamiphos sulfoxide
Fluometuron
3-Ketocarbofuran phenol
Linuron
Metribuzin DA
Metribuzin DADK
No. of
Duplicates
133
125
133
133
128
132
132
127
133
132
128
121
128
133
114
130
122
131
129
123
129
120
125
126
121
134
No. of False
Negatives
0
0
0
1
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
% of False
Negatives
<0.75
<0.79
<0.75
0.75
<0.78
<0.75
<0.75
<0.78
<0.75
<0.75
<0.78
<0.82
<0.78
<0.75
<0.87
<0.76
<0.81
<0.76
<0.77
<0.81
<0.77
<0.83
<0.79
<0.79
<0.82
<0.74
National Pesticide Survey: Phase I Report
-------
E-44 Appendix E: Survey Analytes and Analytical Chemistry Tables
Exhibit-6 (continued)
False Negative Rates
Analyte
Metribuzin DK
Neburon
Pronamide metabolite
Propanil
Propham
Swep
METHOD 5
Aldicarb
Aldicarb sulfone
Aldicarb sutfoxide
Baygon
Carbaryl
Carbofuran
3-Hydroxycarbofuran
Methiocarb
Methomyl
Oxamyl
METHOD 6
ETU
METHOD 7
EDB -
DBCP
1 ,2-Dichloropropane
cis-1 ,3-Dichloropropene
trans-1 ,3-Dichloropropene
METHOD 9
Nitrate
No. of
Duplicates
134
129
130
127
132
129
88
88
87
88
88
88
88
88
88
87
126
82
82
82
82
82
119
No. of False
Negatives
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
% of False
Negatives
<0.74
<0.77
<0.76
<0.78
<0.75
<0.77
<1.1
<1.1
<1.1
<1.1
<1.1
<1.1
<1.1
<1.1
<1.1
<1.1
<0.79
<1.2
<1.2
<1.2
<1.2
<1.2
<0.83
National Pesticide Survey: Phase I Report
------- |