CONTENTS FOR ATTACHMENTS 1 AND 2
ATTACHMENT 1: Documentation for the Continuing Survey of Food Intakes by
Individuals (CSFII)
1. CONTENTS OF THE FOOD COMMODITY INTAKE DATABASE CD-ROM
1.1. TREE DIAGRAM OF FILES ON CD-ROM
1.2. FOOD COMMODITY INTAKE DATABASE FOR THE 1994-96
CONTINUING SURVEY OF FOOD INTAKES BY INDIVIDUALS ITS
SUPPLEMENTAL CHILDREN'S SURVEY (CSFII 1994-96, 1998) . . .
1.3. SUMMARY DESCRIPTIONS OF FOOD COMMODITY INTAKE
DATABASE FILES
1.3.1. Documentation Files
1.3.2. EPA Commodity Files
1.3.3. Food Commodity Intake Data Files
1.3.4. Supporting Data Files
1.4. SUGGESTED CITATIONS, ACQUISITION, CONTACTS AND
DISCLAIMERS
2. ESSENTIAL INFORMATION
3. METHODOLOGY FOR TRANSLATING THE 1994-96 CONTINUING
SURVEY OF FOOD INTAKES BY INDIVIDUALS AND
ITS SUPPLEMENTAL CHILDREN'S SURVEY (CSFII 1994-96, 1998)
INTO INTAKES OF EPA-DEFINED FOOD COMMODITIES
3.1. FOOD COMMODITY INTAKE DATABASE
3.2. CONTINUING SURVEY OF FOOD INTAKES BY INDIVIDUALS
(CSFII)
3.3. EPA LIST OF FOOD COMMODITIES
3 .4. TRANSLATING CSFII TO EPA FOOD COMMODITIES
3.5. DEFINING MIXTURES IN TERMS OF INGREDIENTS
3 .6. ASSIGNING FOODS AND INGREDIENTS TO EPA COMMODITIES
3.6.1. Matching USDA Food Codes to EPA Commodities
3.6.2. Converting Weights of Foods and Ingredients
3.6.3. CSFII Food-Code-to-Commodity Translation File
3 .7. SPECIAL REQUIREMENTS FOR ASSIGNING FOODS AND
INGREDIENTS TO EPA FOOD COMMODITIES
3.7.1. Alcoholic Beverages
3.7.2. Baby Foods
3.7.3. Chips
3.7.4. Coffee and Tea
3.7.5. Egg, Dried
l
-------
CONTENTS FOR ATTACHMENTS 1 AND 2
3.7.6. Fats and Oils
3.7.6.1. Vegetable Oils
3.7.6.2. Margarine and Butter
3.7.7. Fish
3.7.8. Fruit
3.7.9. Gelatin, Dry
3.7.10. Legumes
3.7.11. Meat
3.7.11.1. Beef
3.7.11.2. Game
3.7.11.3. Goat
3.7.11.4. Kidney
3.7.11.5. Meat, Not Further Specified
3.7.11.6. Pork
3.7.11.7. Rabbit
3.7.11.8. Sheep
3.7.12. Meat and Poultry, Processed
3.7.13. Milk
3.7.14. Poultry
3.7.14.1. Chicken
3.7.14.2. Turkey
3.7.14.3. Poultry Other Than Chicken and Turkey
3.7.14.4. Mixtures Described as Containing Chicken or Turkey
3.7.14.5. Mixtures Described as Containing Meat and/or Poultry
3.7.15. Spices and Herbs
3.7.16. Sweeteners, Carbohydrate
3.7.17. Vegetables
3.7.18. Water
3.7.19. Commodities Not Consumed
3.7.20. Special Requirement To Include Commodities Consumed in Very
Small Amounts
3.8. NOTES ABOUT REVIEWING THE FOOD-CODE-TO-COMMODITY
TRANSLATION FILE
3.9. LIMITATIONS OF THE FOOD COMMODITY INTAKE DATABASE
3.10. REFERENCES
Table 1. CSFII food codes representing commercial baby foods
Table 2. EPA food commodities not appearing in the intake database . . .
4. CHARACTERISTICS AND FORMATS OF THE FOOD COMMODITY INTAKE
DATABASE
4.1. INTRODUCTION
ii
-------
CONTENTS FOR ATTACHMENTS 1 AND 2
4.2. DATABASE STRUCTURE
4.2.1. Documentation Files (Files in Directory "\document")
4.2.2. EPA Commodity Files (Files in Directory "\epa_comm")
4.2.3. Food Commodity Intake Data Files (Files in Directory "\intake")
4.2.4. Supporting Data Files (Files in Directory "\support")
4.3. GENERAL NATURE OF THE DATA
4.3.1. Data File Characteristics
4.3.2. List of Key Fields
4.3.3. Sampling Weights
4.4. DATA MIL FORMATS
4.4.1. Introduction to File Formats for ASCII Delimited Files
4.4.2. Formats for the Food Commodity Intake Data Files
4.4.2.1. Format of the Commodity Intakes File (comm9498.txt) .
4.4.2.2. Format for the Files of Commodity Intakes by Cooked
Status, Food Form and Cooking Method (ffcm9498.txt)
4.4.2.3. File Format for Sample Person Data (smpl9498.txt) ....
4.4.3. Formats for the Supporting Files
4.4.3.1. Food Code-to-Commodity Translation File (fc_comm.txt)
4.4.3.2. Food Code Descriptions (fcdesc.txt)
4.4.3.3. Food Code Include Statements (fcincl.txt)
4.4.3.4. Food Code Outline (fcscheme.txt)
4.4.3.5. Descriptions of Recipe Modifications (moddesc.txt) ....
4.5. MISCELLANEOUS NOTES
4.5.1. Responding Sample Persons With No Foods Reported for a Day .
4.5.2. Unreported Body Weights
4.5.3. Lower Limit for Reporting Intakes
4.5.4. Using This Database in Conjunction With the Continuing Survey
of Food Intakes by Individuals and the Supplemental Children's
Survey (CSI-'II 1994-96, 1998)
5. CONTROL STATISTICS FOR THE SMPL9498.TXT DATA FILE, ALL
RECORDS,UNWEIGHTED
APPENDIX A: EPA'S USE OF FOOD CONSUMPTION DATA IN ASSESSING
DIETARY RISK FROM PESTICIDES
APPENDIX B: CSFII 1994-96, 1998 METHODOLOGY (SECTION 3 FROM CSFII
1994-96, 1998 CD-ROM DOCUMENTATION)
APPENDIX C: CSFII 1994-96, 1998 SAMPLING WEIGHTS AND STATISTICAL
NOTES (EXCERPTS FROM SECTIONS 5 AND 6, CSFII 1994-96, 1998
CD-ROM DOCUMENTATION)
in
-------
CONTENTS FOR ATTACHMENTS 1 AND 2
ATTACHMENT 2: The EPA Food Commodity Vocabulary (Source: CSFII
Documentation)
APPENDIX A: THE MSI I LIST
APPENDIX B: THE HERB LIST
APPENDIX C: THE SPICE LIST
APPENDIX D: SELECTED REFERENCES FOR THE EPA FOOD COMMODITY
VOCABULARY
iv
-------
ATTACHMENT 1
Documentation for the Continuing Survey
of Food Intakes by Individuals (CSFII)
Attachment 1-1
-------
1. CONTENTS OF THE FOOD COMMODITY INTAKE DATABASE CD-ROM
1.1 Tree diagram of files on CD-ROM
These files comprise the Food Commodity Intake Database for the
1994-96 Continuing Survey of Food Intakes by Individuals and its
Supplemental Children's Survey (CSFII 1994-96, 1998)
Directory and
File Names
[CD-ROM drive]:
\readme.txt
\document
\sectionl.wpd
\section2.wpd
\section3.wpd
\section4.wpd
\section5.wpd
\appenda.wpd
\appendb.wpd
\appendc.wpd
\epa_comm
\csffcm.wpd
\foodvoc.wpd
\intake
\comm9498.txt
\ffcm9498.txt
\smpl9498.txt
\Support
\fc comm.txt
\fcdesc.txt
\fcincl.txt
\fcscheme.txt
\moddesc.txt
Title or Description
Drive letter for CD-ROM
Readme file
Documentation files
Contents, tree diagram, data
file list
Essential information
Methodology for translating food
intakes into EPA food
commodities
Data file characteristics and
formats
Control counts
Appendix A
Appendix B
Appendix C
EPA Commodity Files
EPA procedures for assigning EPA
cooked status, food form, and
cooking method values to USDA
food codes
EPA Food commodity vocabulary
Food Commodity Intake Data Files
Commodity intakes for the
Continuing Survey of Food
Intakes by Individuals (CSFII
1994-96, 1998)
Commodity intakes by cooked
status, food form, and cooking
method for CSFII 1994-96, 1998
Sample person data for CSFII
1994-96, 1998
Supporting Data Files
CSFII 1994-96, 1998 foods linked
to EPA commodities (grams of
commodities per 100 grams food)
Food code descriptions
Food code include statements
(extensions of food code
descriptions)
Food code outline
Descriptions of recipe
Attachment 1-2
-------
modifications
1.2 Food Commodity Intake Database for the 1994-96 Continuing
Survey of Food Intakes by Individuals its Supplemental
Children's Survey (CSFII 1994-96, 1998)
Table of Database Files
Documentation Files
sectionl.wpd
section2.wpd
section3.wpd
section4.wpd
section5.wpd
appenda.wpd
appendb.wpd
appendc.wpd
EPA documents
csffcm.wpd
foodvoc.wpd
Intake Data Files
comm94 98.txt
ffcm9498.txt
smpl9498.txt
Supporting Data Files
fc comm.txt
fcdesc.txt
fcincl.txt
fcscheme.txt
moddesc.txt
Table of contents, File names
Essential information
Methodology for translating food
intakes into EPA food commodity
intakes
Data file characteristics and
formats
Control counts
Appendix A
Appendix B
Appendix C
EPA procedures for assigning EPA
cooked status, food form, and
cooking method values to USDA
food codes
EPA Food commodity list
Commodity intakes for CSFII
1994-96, 1998
Commodity intakes by cooked
status, food form, and cooking
method for CSFII 1994-96, 1998
Sample person data for CSFII
1994-96, 1998
CSFII 1994-96, 1998 foods linked to
EPA commodities (grams of
commodities per 100 grams food)
Food code descriptions
Food code include statements
(extensions of food code
descriptions)
Food code outline
Descriptions of recipe
modifications
1.3 Summary descriptions of Food Commodity Intake Database Files
Attachment 1-3
-------
This CD-ROM contains the Food Commodity Intake Database. Summary
descriptions of files included on this CD-ROM are presented
below. Complete information on data file characteristics and
formats are found in \document\section4.wpd.
1.3.1 Documentation files (files in directory \document)
Eight documentation files (5 documentation sections and 3
appendices) are included as WordPerfect files. They explain the
methodology used to develop the Food Commodity Intake Database,
data file characteristics and formats, and other information
important for using this database.
1.3.2 EPA Commodity files (files in directory \epa_comm)
The Environmental Protection Agency (EPA) provided two
WordPerfect files:
EPA Food Commodity Vocabulary - "foodvoc.wpd"
This file contains a list of EPA commodities used in this
project. The list includes the name of each EPA commodity, its
EPA code, and a description of the weight basis for the
commodity.
PROCEDURES FOR CODING CSFFCM - "csffcm.wpd"
EPA assigned codes for cooked status, food form, and cooking
method to each CSFII food code. This file discusses the
procedures used by EPA to assign those codes, which were then
used to create the "Commodity Intakes by Cooked Status, Food
Form, and Cooking Method" file described below.
1.3.3 Food Commodity Intake Data Files (files in directory
\intake)
These files include (a) two commodity intake data files that were
created by translating the CSFII 1994-96, 1998 food intakes into
EPA-defined food commodities, and (b) one file containing
characteristics of sample persons in the survey.
COMMODITY INTAKE DATA - "comm9498.txt"
The Commodity Intake File contains one record per EPA-defined
commodity, per person, per day. Each record contains the daily
total intake of a commodity, aggregated from the foods reported
by the sample person. If the sample person provided 2 days of
intake data, records for daily averages of each commodity are
also present. Data are presented as grams of commodity consumed
per kilogram of body weight. Records are not present for
commodities not consumed by the sample person.
COMMODITY INTAKES BY "COOKED STATUS, FOOD FORM, AND COOKING
Attachment 1-4
-------
METHOD" - "ffcm94 98.txt"
This file further defines the CSFII commodity intake data by the
EPA codes for "Cooked Status," "Food Form," and "Cooking Method"
(CSFFCM). They contain one record per commodity per unique
CSFFCM, per person, per day. Each record contains the daily
total intake of a "commodity-CSFFCM" combination, aggregated from
the foods reported by the sample person. Records of daily
averages for each Commodity-CSFFCM combination are present, if
the sample person provided 2 days of intake data. Data are
presented as grams consumed per kilogram of body weight. Records
are not present for Commodity-CSFFCM combinations not consumed by
the sample person. For an explanation of "cooked status, food
form, and cooking method" codes, see directory \epa_comm on this
CD-ROM.
SAMPLE PERSON DATA - "smpl9498.txt"
The Sample Person data file contains demographic, sampling
weight, and other respondent and household information. One
record per responding sample person is present. Data in these
files were extracted from the CSFII 1994-96, 1998 Household and
Sample Person data records.
1.3.4 Supporting Data Files (files in directory \support)
Five data files provide supporting documentation for the
translation of the CSFII 1994-96, 1998 intake data into EPA food
commodities.
FOOD-CODE-TO-COMMODITY TRANSLATION FILE - "fc_comm.txt"
The Food-Code-to-Commodity Translation file documents the
assignment of USDA food codes to EPA commodities as used in this
translation of food intakes into commodity intakes. Data are
expressed as the amount of each EPA commodity contained within a
100 gram portion of each specific CSFII food code. A separate
record exists for each commodity assignment for a food code. For
a discussion about the assignment of food codes to commodities,
see section 3 on Methodology.
FOOD CODE DESCRIPTIONS - "fcdesc.txt"
The Food Code Descriptions file provides descriptions for each
food code used in the CSFII 1994-96, 1998. Descriptions are
usually generic in nature except for certain breakfast cereals,
infant formulas, and candies. The file includes descriptions up
to 200 characters long, as well as abbreviated descriptions up to
60 characters long.
FOOD CODE INCLUDE STATEMENTS - "fcincl.txt"
Food Include Statements are names of specific foods associated
with a particular generic food description in the Food Code
Attachment 1-5
-------
Descriptions file.
FOOD CODE OUTLINE - "fcscheme.txt"
The food code scheme identifies the meanings associated with the
first, second, and third (and sometimes fourth) digits of the 8-
digit food codes.
DESCRIPTIONS OF RECIPE MODIFICATIONS - "moddesc.txt"
In the CSFII, some database recipes were modified when a
responding sample person supplied specific information about
certain food ingredients. This file contains descriptions of
those modifications.
1.4 Suggested citations, acquisition, contacts and disclaimers
Suggested citation for this database:
U.S. Environmental Protection Agency, Office of Pesticide
Programs and U. S. Department of Agriculture, Agricultural
Research Service. 2 000. Food Commodity Intake Database,
Version 2.1. CD-ROM.
Suggested citations when identifying the original sources of data
for this database:
U.S. Environmental Protection Agency. 2 000. The EPA Food
Commodity Vocabulary, Master List dated June 15, 2000.
U.S. Department of Agriculture, Agricultural Research
Service. 2 0 00. Continuing Survey of Food Intakes by
Individuals 1994-96, 1998. National Technical Information
Service, 5285 Port Royal Road, Springfield VA
22161; (703)487-4650 . CD-ROM. NTIS Accession no. PB2000-
500027.
Acquisition
Copies of the Food Commodity Intake Database may be purchased
from the National Technical Information Service, 5285 Port Royal
Road, Springfield, VA 22161; telephone 1-800-553-6847.
Contacts
U.S. Environmental Protection Agency(7509C)
Ariel Rios Building
12 0 0 Pennsylvania Avenue, N.W.
Washington, DC 20460
FAX: 703-305-0871
Disclaimers
Mention of trade names, commercial products, or companies in this
Attachment 1-6
-------
database is solely for the purpose of providing specific
information and does not imply recommendation or endorsement by
the U.S. Department of Agriculture over others not mentioned.
The United States Department of Agriculture (USDA) prohibits
discrimination in its programs on the basis of race, color,
national origin, sex, religion, age, disability, political
beliefs, and marital or familial status. (Not all prohibited
bases apply to all programs.) Persons with disabilities who
require alternative means for communication of program
information (Braille, large print, audiotape, etc.) should
contact the USDA Office of Communications at (202) 720-2791.
To file a complaint, write the Secretary of Agriculture, U.S.
Department of Agriculture, Washington, DC 20250, or call (202)
720-7327 (voice) or (202) 720-1127 (TDD). USDA is an equal
employment opportunity employer.
Attachment 1-7
-------
Food Commodity Intake Database, CSFII 1994-96, 1998
2 ESSENTIAL INFORMATION
The notes in this section provide information to help use this
database effectively, although they do not substitute for reading
the remainder of the documentation.
The Food Commodity Intake Database is based on:
(a) Food consumption surveys conducted by the U.S. Department of
Agriculture (USDA)--the 1994-96 Continuing Survey of Food Intakes
by Individuals and its Supplemental Children's Survey (CSFII
1994-96, 1998), and
(b) A list of commodities prepared by the U.S. Environmental
Protection Agency (EPA)--the Food Commodity Vocabulary, dated
June 15, 2 000.
The Food Commodity Intake Database was developed for the purpose
of estimating human exposures to pesticide residues through
intakes of foods and beverages. For information about how to use
these data for dietary risk assessment, users should contact the
EPA Office of Pesticide Programs, Division of Health Effects,
Ariel Rios Building, 1200 Pennsylvania Avenue, NW(7509C),
Washington, DC 20460, FAX 7 03-305-0871.
Food commodity intakes are included for each sample person from
the surveys. Intakes are expressed as grams of commodity
consumed, per kilogram of body weight, per day. Two days of
intakes, plus daily averages, are present for most sample
persons.
Data files are also included which indicate how commodities were
consumed in terms of their "cooked status, food form, and
cooking method." Codes for these parameters were assigned to the
USDA food codes by EPA.
The smallest amount of commodity consumption represented in this
database is 0.000001 grams, per kilogram of body weight, per day.
Amounts smaller than 0.000001 were set to this amount. This
expression is not intended to suggest that data can be estimated
this precisely. It was set to this small amount at the request
of EPA to reduce the likelihood that small amounts would round to
zero.
Water consumption is not included in this database at this time.
Intakes of water, as defined in the EPA food commodity list, are
being prepared by the EPA Office of Water.
A Food-Code-to-Commodity Translation File is included in this
database to document how each USDA food code was assigned to the
EPA commodities. Data are expressed as grams of commodity per
10 0 grams of food.
Attachment 1-8
-------
The amounts of commodities assigned to one USDA food code do not
always equal 100. This occurs because (a) some food constituents
were not assigned to commodities, for example, salt was excluded,
and (b) conversions were sometimes necessary in order to express
the commodities as defined by EPA; for example, rice as consumed
was converted to the dry weight of the grain.
The smallest amount of a commodity identified in the Food-Code-
to-Commodity Translation File is 0.001 grams per 100 grams of
food. Commodity amounts lower than this limit, but greater than
zero, were set to 0.001 grams. Three decimal places were
selected for expression of the commodity amounts to accommodate
an EPA requirement that the presence of commodities in very small
amounts be represented in this database. It is not intended to
suggest that data can be estimated this precisely.
Several other special requirements were requested by EPA to guide
the translation of USDA food codes to commodities. These
requirements are discussed in Section 3.
Users may observe some unexpected findings in the Food-Code-to-
Commodity Translation File. For example, the following five
commodities are present in small amounts in many foods:
"Cassava," "corn, field, starch," "Potato, flour," "Rice, flour,"
and "Wheat, flour." This occurs because modified food starch is
frequently listed as an ingredient in processed foods. When the
source of the starch was not identified, it was assigned to those
commodities. Other examples of unexpected findings can be found
in section 3.8, Notes about reviewing the Food-Code-to-Commodity
Translation File. Reasons for unexpected findings can be found in
discussions on matching USDA food codes to EPA commodities
(section 3.6.1), converting weights of foods and ingredients
(sections 3.6.2), and special requirements for assigning foods
and ingredients to EPA commodities (section 3.7).
Not all commodities were consumed.
Data users are encouraged to read section 3, Methodology for
Translating the CSFII 1994-96, 1998 into intakes of EPA-defined
commodities. Limitations of the data are discussed in section
3.9.
Attachment 1-9
-------
Food Commodity Intake Database, CSFII 1994-96, 1998
3 Methodology for translating the 1994-96 Continuing Survey of
Food Intakes by Individuals and its Supplemental Children's
Survey (CSFII 1994-96, 1998) into intakes of EPA-defined
commodities
3.1 Food Commodity Intake Database
The Food Commodity Intake Database contains intake data and
related information for approximately 548 commodities listed in
the Environmental Protection Agency's (EPA) Food Commodity
Vocabulary (16). The database is the result of cooperative
efforts by the U.S. Department of Agriculture (USDA),
Agricultural Research Service (ARS), and EPA's, Health Effects
Division. It was developed for use in estimating human exposures
to pesticide residues through intakes of foods and beverages. A
discussion of EPA's use of food consumption data in assessing
dietary risk from pesticides is provided in Appendix A.
The database contains several different types of data files
(intake data, supporting data, documentation). A complete list
of the data files is located in section 1.
The intake data are based on the 1994-96 Continuing Survey of
Food Intakes by Individuals (CSFII 1994-96) and its Supplemental
Children's Survey (CSFII 1998) (8), conducted by ARS.
Commodities listed in the Food Commodity Vocabulary were selected
and defined by EPA. Because the survey data are expressed in
terms of foods as eaten, conversions of the data were sometimes
necessary in order to express them in terms of commodities as
defined by EPA. For example, consumption of cooked rice needed
to be expressed in terms of the EPA commodity, "Rice, dry weight
of grain." Data conversions and other steps used to develop the
commodity intake data are referred to, collectively, as a
translation. The data sets and processes used in the translation
are identified below. For information about how to use these
data for dietary risk assessments, users should contact the EPA
Office of Pesticide Programs, Division of Health Effects, Ariel
Rios Building, 1200 Pennsylvania Avenue, NW (7509C), Washington,
DC 20460.
3.2 Continuing Survey of Food Intakes by Individuals (CSFII)
The CSFII is a series of U. S. Department of Agriculture surveys
designed to measure the kinds and amounts of foods eaten by
Americans. The CSFII 1994-96 was conducted between January 1994
and January 1997 with a target population of noninstitutionalized
individuals in all 50 states and Washington, D.C. In each of the
3 survey years, data were collected for a nationally
representative sample of individuals of all ages. The CSFII 1998
was a survey of children ages 0 through 9 which was a supplement
to the CSFII 1994-96. It used the same sample design as the
Attachment 1-10
-------
CSFII 1994-96 and was intended to be merged with CSFII 1994-96 to
increase the sample size for children. The merged surveys are
designated as CSFII 1994-96, 1998.
In the CSFII 1994-96, 1998, dietary intakes were collected
through in-person interviews using 24-hour recalls on 2
nonconsecutive days. A total of 21,662 individuals provided data
for the first day; of those individuals, 20,607 provided data for
a second day. Over 11,000 of the sample persons represent
children up to 18 years of age.
Several sets of sampling weights are available for use with the
intake data, and all sets have been included in this database.
Using appropriate weights, data can be analyzed in the following
ways: (1) one year alone, i.e., 1994, 1995, 1996, or 1998; (2) 3
years combined for the CSFII 1994-96; or (3) 4 years combined for
CSFII 1994-96, 1998. It is recommended that all 4 years be
combined in order to provide an adequate sample size for
children. For an explanation of sample weighting, see section 5
of the documentation on the survey CD-ROM (8). Selections from
this section are also included here for easy reference
(\document\appendc.wpd).
Complete results from the surveys in microdata form are available
on CD-ROM (7,8) . Documentation on the CD-ROMs includes survey
instruments, detailed descriptions of survey and data processing
methods, sample design, and calculation of weights. The survey
instruments, as well as a Design and Operation Report for CSFII
1994-96 (5), can also be found on the Food Surveys Research Group
web site: http://www.barc.usda.gov/bhnrc/foodsurvey/home.htm.
Several sections of the documentation from the CSFII 1994-96,
1998 CD-ROM are included here for easy reference. Section 3, a
discussion of survey methodology, is located in Appendix B;
excerpts from sections 5 and 6 on sampling weights and
statistical notes are located in Appendix C in \document.
Three different data files (record types) from CSFII 1994-96,
1998 were used during the translation of the survey data to EPA
commodities:
1. Record Type 15 (Household Level Data)
2. Record Type 25 (Sample Person Data)
3. Record Type 30 (Food Intake Data)
In addition, information was used from the following three ARS
technical databases (7a,7b,7c):
1. Food Coding Database
2. Recipe Database
3. Survey Nutrient Database
Attachment 1-11
-------
These databases are maintained by ARS for processing and
analyzing food consumption data, and the versions used for a
survey are released along with the survey data. Primarily,
databases from the CSFII 1994-96 CD-ROM were used for this
translation (7c). However, if a food was reported for the first
time during 1998, it may not appear on the earlier databases.
For those foods, entries can be found on the CSFII 1994-96, 1998
CD-ROM (8).
A total of 5,831 different foods were reported as consumed in the
CSFII 1994-96, 1998. Each food is identified by a unique 8-digit
food code in the CSFII data (Record type 30). An additional
3,913 variations of these foods were also reported and identified
by a unique 6-digit modification code. Appendix B, section 3.3.3
(\document\appendb.wpd) contains a discussion about food codes
and descriptions. A complete list of food codes and modification
codes are included in the directory \support.
3.3 EPA list of food commodities
EPA provided ARS with the EPA Food Commodity Vocabulary for this
project. It is a list of over 540 commodity items for use in
estimating human exposure to pesticide residues through food
consumption. The list, which includes foods, beverages, and
water, is included as a WordPerfect document on this CD-ROM
(\epa_comm\foodvoc.wpd). It lists the name of each EPA
commodity, its EPA code, and a description of the weight basis
for the commodity. For example, the code for the commodity
"Apple, juice" is 11000100 and the weight basis is "weight of
juice at single strength (or standard dilution)." Many foods
have separate entries and codes on the list to represent baby
foods. For example, "Apple, juice, baby food," has a separate
entry with a unique code, 11000101.
A discussion by EPA of the basis for selecting the various
commodities is included in the EPA Food Commodity Vocabulary. In
general, items on the list (other than water) can be categorized
by the following different types:
Commodity type
Example
Food (no further description)
Food with distinguishing characteristic
Food part
Food, processing specific
Food component
Food consumed as baby food
Blackberry
Apple, peeled fruit
Turnip, tops
Peach, dried
Milk, fat
Carrot, baby food
Some foods, for example, peach, may be included on the list as
several different commodity types:
Attachment 1-12
-------
EPA Code
Commodity Description
12002600
12002601
12002610
12002611
12002620
12002611
Peach
Peach, baby food
Peach, dried
Peach, dried, baby food
Peach, juice
Peach, juice, baby food
3.4 Translating CSFII to EPA food commodities
Four major steps were involved in translating the CSFII 1994-96,
1998 to EPA commodities. 1) Many of the foods reported in the
survey were mixtures, and the first step was to define those
foods in terms of their ingredients and amounts. 2) Next, each
food or ingredient was assigned to its appropriate EPA commodity,
or commodities, including converting amounts to meet EPA's weight
basis requirements when necessary. The assignments were
documented in a Food-Code-to-Commodity Translation File.
3) The Food-Code-to-Commodity Translation File was then used in
conjunction with the food intake data from CSFII 1994-96, 1998 to
produce commodity intake estimates. During this step, each
record type 3 0 (food intake data record) was translated to
represent consumption of specific EPA commodities. 4) Finally,
consumption of each EPA commodity was totaled into a daily amount
for each sample person and expressed in terms of grams consumed
per kilogram of body weight. Two-day averages of commodity
intakes were calculated for sample persons providing two days of
food intake.
The lower limit for representing intake of a commodity was
0.000001 grams per kilogram of body weight per day. Any daily
amount less than .000001 grams, but greater than zero, was set to
this lower limit of daily consumption on the commodity intake
data files. The purpose of expressing the commodity intake data
to six decimal places was not to imply precision to that level.
Expression to six decimal places was requested by EPA in order to
ensure that intakes of very small amounts would not be
overlooked. See section 3.7.20 for discussion of EPA's special
requirement to include commodities consumed in very small
amounts. See section 1.3 for a description of the Food Commodity
Intake Data file.
Additional details regarding preparation of the Food-Code-to-
Commodity Translation file are provided below.
3.5 Defining mixtures in terms of ingredients
During processing of CSFII intake data, recipes had been
identified for many of the food mixtures for the purpose of
estimating nutrient content. These recipes appear on the CSFII
CD-ROMs in the Recipe Database (7c). A discussion of the Recipe
Attachment 1-13
-------
Database can be found in Appendix B, section 3.3.5
(\document\appendb.wpd). The recipes are considered
"representative," meaning they were not exact for every sample
person, nor were they constructed for the purpose of estimating
consumption of ingredients. Nevertheless, the recipes had been
carefully developed and refined over many years for USDA food
consumption surveys; and prior to the CSFII 1994-96, 1998 they
had been reviewed and updated to reflect the food marketplace at
that time.
These recipes were used as the starting point for the translation
project. EPA identified several situations where changes to the
recipes would provide a broader representation of commodity
sources. Such changes were incorporated when ingredients were
assigned to the commodities. These situations are discussed
below in Section 3.7, "Special Requirements for Assigning Foods
and Ingredients to EPA Commodities."
When a recipe ingredient was also a mixture (e.g., bread crumbs),
its ingredient amounts were estimated for this project using
procedures described in Appendix B, section 3.3.5
(\document\appendb.wpd). Ingredient amounts in brand-name
products were estimated based on label information using
procedures previously described by Marcoe and Haytowitz (1).
During CSFII 1994-96, 1998 food coding, a recipe could be
modified to match more closely the foods eaten by sample persons.
There were three main purposes for recipe modifications: to
record the type of fat (for example, butter); the type of milk
(for example, skim milk); and the amount of water or milk used to
dilute foods (for example, condensed soup). Modified recipes are
included in the recipe database and were used in this
translation.
3.6 Assigning foods and ingredients to EPA commodities
3.6.1 Matching USDA food codes to EPA commodities
Most CSFII food codes were matched to one or more commodity items
on EPA's list. Food codes which represented mixtures were
assigned based on their recipe ingredients. Ingredients which
were not assigned to commodities included water, which was
assigned to commodities by EPA's Office of Water, and several
ingredients which were not included on EPA's list: salt,
leavening, artificial sweeteners, pectin, added nutrients, and
some forms of alcohol.
Commodity assignments for brand-name foods were based on
ingredients listed on labels from the marketplace at the time of
the CSFII 1994-96, 1998. CSFII food codes frequently represented
more than one brand of a product. Assignments were based on
ingredient listings of national brands, and frequently
Attachment 1-14
-------
represented a blending of ingredient lists for two or more
brands.
Modified food starch is an ingredient in many processed foods.
When the type of modified food starch was not specified on the
label, it was assigned as 95% to "Corn, field, starch," and the
remaining 5% was assigned in equal amounts to the following
commodities: "Cassava," "Potato, flour," "Rice, flour," and
"Wheat, flour." Maltodextrin was assigned to commodities in the
same proportions as modified food starch. Soya lecithin was
assigned to "Soybean, oil." Carrageenan, alginate, and agar were
assigned to "Seaweed."
3.6.2 Converting weights of foods and ingredients
As stated above, conversions were made as needed to conform to
the weight basis of the commodity as defined on the EPA list of
commodities. For example, the EPA commodity "Corn, field, syrup"
is based on the weight of the syrup; therefore, when corn syrup
solids was an ingredient within a mixture, the weight was
converted to the syrup form (100 grams of corn syrup solids = 131
grams of corn syrup). Factors used in this type of adjustment
are based on differences between total solids of the consumed
foods or ingredients and total solids for the commodities, and
were derived by dividing the percent total solids of the food or
ingredient by the percent total solids of the corresponding
commodity. Values for total solids were derived from the 1994-96
Survey Nutrient Database (7b) as follows: 100 - percent moisture
content = percent total solids.
Since grain commodities are defined as the dry weight of grain or
flour, they frequently required weight adjustments to account for
the amount of moisture which is absorbed during cooking. For
example, rice absorbs moisture when it is steamed, so the weight
of cooked rice was adjusted downward to eliminate the weight of
the water.
Sometimes grain ingredients, such as flour, actually lose
moisture during food processing. For example, ready-to-eat
breakfast cereals may lose moisture during processing, so the
weight of flour ingredients were adjusted upward to approximate
their original weights before the moisture loss.
Examples of how weights were adjusted to account for differences
in total solids between foods as consumed and foods as defined in
EPA's commodity list:
Attachment 1-15
-------
Example 1:
Food or ingredient
Total solids
EPA-defined Commodity
Total solids
Conversion factor
Rice, cooked
31.56 percent
Rice, dry weight of grain
88.38 percent
31.56/88.38 = 0.357
Example 2:
Food or ingredient
solids
EPA-defined Commodity
42.15 percent
Orange juice concentrate Total
Orange juice (weight of single
strength juice)
11.90 percent
42.15/11.90 = 3.542
Total solids
Conversion factor
When the weight basis of a commodity was identified as "edible
portion as consumed," the amount was assigned according to the
form in which the food was consumed. For example, if a carrot
was consumed raw, it was assigned according to its raw weight.
If it was consumed cooked, assignment was made according to the
cooked weight. If the carrot was consumed cooked, but it appeared
in the recipe ingredient list as a raw carrot, its weight was
adjusted to the cooked form before assignment. Adjustment
factors embedded within the recipes were used for this type of
adjustment. These factors were based on data from Agriculture
Handbook No. 102, Food Yields Summarized by Different Stages of
Preparation (2); Agricultural Handbook No. 8, Composition of
Foods...Raw, Processed, Prepared (6), or unpublished data from
ARS food laboratories.
3.6.3 CSFII Food-Code-to-Commodity Translation File
A CSFII Food-Code-to-Commodity Translation File was prepared to
identify the commodity assignments (excluding water) for each
food code from the CSFII 1994-96, 1998. Some food codes have
more than one set of commodity values on this database. This
occurs when a modification was made to the food code's
corresponding recipe during the survey. (See section 3.2 above.)
To identify one set of commodity values in the database, use the
8-digit food code in conjunction with the 6 digit modification
code. The translation file is included on this CD-ROM in
\support.
In this file, information is expressed as the amount in grams of
each commodity per 100 grams of a CSFII food. However, the
amounts of all commodities associated with one food code, plus
modification code, may not total exactly to 100 grams. This
occurs for several reasons: 1) Commodities are not always
expressed in the same forms as foods are consumed; for example,
the commodity form for white rice is "dry form of grain." As a
Attachment 1-16
-------
result, 100 grams of food code 56205010, "Rice, white, cooked,"
translate to only 35.7 grams of the EPA commodity, "Rice, white,
dry form of grain." See section 3.6.2 above for examples of how
weights of foods and ingredients were adjusted during the
translation. Additional examples of adjustments include: 100
grams of frozen orange juice concentrate translate to 354 grams
of single strength orange juice, 100 grams of corn syrup solids
translate to 131 grams of corn syrup; 100 grams of dry egg powder
translate to 393 grams of raw egg; and 100 grams of cooked
macaroni translate into 3 7.9 grams of wheat flour. Anytime
grains, concentrated juices, corn syrup solids, or dry egg powder
are identified as an ingredient within a CSFII food, the total of
commodities for the CSFII food code will likely add to more or
less than 100 on the Food-Code-to-Commodity translation file. 2)
Water as a commodity was excluded from this database. For
example, the weight of water used to reconstitute canned
condensed soups is not represented in this data file. When water
is an ingredient in a CSFII food, commodities will add to less
than 100 on the Food-Code-to-Commodity Translation file. 3) Some
food constituents were not assigned to commodities: salt,
leavening, artificial sweeteners, pectin, added nutrients, and
some forms of alcohol. Anytime these foods are ingredients in a
CSFII food, commodities will add to less than 100 on the Food
Code-to-Commodity Translation file.
The smallest amount of a commodity that was included in this
database was 0.001 grams per 100 grams of the CSFII food.
Commodity amounts lower than this limit, but greater than zero,
were set to 0.001 grams. Three decimal places were selected for
expression of the commodity amounts to accommodate an EPA
requirement that the presence of commodities in very small
amounts be represented in this database. See section 3.7.19 for
discussion of EPA's special requirement to include commodities
consumed in very small amounts.
3.7 Special requirements for assigning foods and ingredients to
EPA food commodities
In addition to converting amounts of foods and ingredients to
account for weight bases specified in the EPA food commodity
list, several special requirements were established to guide the
assignment of foods and ingredients to commodities. These
requirements, which are listed below, were based either on
descriptions of foods within the EPA food commodity list, or on
communications between EPA and ARS.
3.7.1 Alcoholic beverages
Rum was assigned to the commodity "Sugar, cane, sugar," adjusted
for the difference in total solids.
Beer was assigned 9.6 percent to the commodity "Barley, flour"
Attachment 1-17
-------
0.7 percent to "Hop," and 4.1 percent to "Rice, flour." The
remainder was considered non-commodities, or was water and was
excluded from this database.
Wine was assigned to the commodity "Grape, wine and sherry,"
except rice wine was assigned to the commodity "Rice, white, dry
form of grain," adjusted for the difference in total solids.
Other forms of alcohol were not required by EPA to be assigned to
commodities.
3.7.2 Baby foods
Commodity codes are specific for commercial baby foods. All
ingredients from food codes identified as baby foods were
assigned to commodities specified as "from baby food." CSFII
food codes representing commercial baby foods are listed in Table
1.
3.7.3 Chips
The potato portion of regular potato chips was assigned to the
commodity, "Potato, chips." The potato from restructured potato
chips was assigned to "Potato, dry." Corn from corn chips was
assigned to the commodity, "Corn, field, meal." Apple from apple
chips was assigned to "Apple, dried." Banana from banana chips
was assigned to "Banana, dried." The fat ingredients from all
these types of chips were assigned to commodity oils as described
in the section 3.7.6 on fats and oils.
3.7.4 Coffee and tea
Commodities for coffee and tea are:
Coffee, roasted bean
Coffee, instant
Tea, dried (leaf)
Tea, instant
Prepared coffee and tea were assigned to their dry commodity
forms, based on amounts typically recommended for preparation.
Coffee not specified as brewed or made from instant was assigned
90% "Coffee roasted bean," and 10 percent "Coffee, instant" (4).
Tea not specified as brewed from leaves or made from instant was
assigned 86 percent "Tea, dried (leaf)" and 14 percent "Tea,
instant" (13). Herbal teas were assigned to the commodity,
"Herbs, other."
3.7.5 Egg, dried
Dried egg ingredients (whole, yolk, or white) were assigned to
the commodities for their fresh forms, adjusted for differences
Attachment 1-18
-------
in total solids.
3.7.6 Fats and oils
3.7.6.1 Vegetable oils
When CSFII food descriptions represented one specific brand name,
oil ingredients listed on the labels were assigned to their
corresponding commodities; for example, coconut oil was assigned
to the commodity "Coconut, oil."
When a label listed more than one type of oil, for example,
coconut and/or palm oil, and no information was available on the
proportion of each oil in the food product, assignment was made
according to their relative proportions in the food supply. When
labels from more than one brand were used to identify
ingredients, all identified oils were used in the assignments.
Information on relative proportions of certain oils in the food
supply was used in situations as described below. These relative
proportions are based on unpublished data from the USDA Center
for Nutrition Policy and Promotion's Food Supply Database. Their
data were based upon information from the USDA's Economic
Research Service and Foreign Agricultural Service (3,10,14).
Percent
Oil
74 .6
Soybean
8 .8
Corn
6 . 6
Cottonseed
3 . 7
Rapeseed (canola
3 . 6
Olive
2 .4
Peanut
0.3
Sunflower
0 . 001
Safflower
0 . 001
Sesame
When specific oil ingredients could not be determined,
assignments were based on composites from the above list of oils,
based on relative proportions of each oil. Judgments were made
for each recipe regarding the most appropriate oils to include,
based on information from popular and ethnic cookbooks, and
consultation among food and nutrition specialists.
Two composites were used in assigning shortening to oil
commodities. One was for commercial foods (soybean and
cottonseed oils) and the other for home-prepared foods (soybean
and rapeseed oils) . Proportions of each oil within the
composites were based on the above food supply data. For foods
Attachment 1-19
-------
that could be either commercial or home prepared, assignments
were made 50 percent to each composite. Oil sources for
shortening were based on information from the Nutrient Data
Laboratory(9).
3.7.6.2 Margarine and Butter
In most instances, the oil ingredients in margarine and
margarine-like spreads were assigned to commodities according to
the following proportions (3) :
Percent Oil
81.7 Soybean
10.2 Corn
5.7 Rapeseed (canola)
0.8 Cottonseed
0 . 8 Safflower
0.8 Sunflower
However, when labels of manufactured foods identified a specific
type of margarine as an ingredient, e.g., corn oil margarine, the
oil ingredient was assigned to the specific oil source.
In the Survey Recipe Database, recipes frequently contained
either margarine or butter as a "default" ingredient (specific
source of fat not known by the survey respondent). In those
cases, the butter or margarine ingredient was divided into 7 0
percent regular margarine and 3 0 percent butter before assignment
to EPA commodities, based on the proportion of those foods in the
food supply (3). Recipe modifications, when the respondent
specifically named margarine or butter as the ingredient, were
not changed.
Butter was assigned to milk commodities as discussed in section
3.7.13 .
3.7.7 Fish
EPA commodities for fish include:
Fish, freshwater finfish
Fish, freshwater finfish, farm raised
Fish, saltwater finfish, tuna
Fish, saltwater finfish, other
Fish, shellfish, crustacean
Fish, shellfish, mollusc
Species of fish and seafood were assigned to their appropriate
EPA commodities. Catfish was assigned 100 percent to the
commodity, "Fish, freshwater finfish, farm raised." Trout was
assigned 50 percent to the commodity "Fish, freshwater, finfish,"
Attachment 1-20
-------
and 50 percent to "Fish, freshwater finfish, farm raised" (9).
No other species reported in the CSFII 1994-96, 1998 were
assigned to the "farm raised" category.
3.7.8 Fruit
Fruit juices and nectars were assigned to commodities for their
respective juices, when such commodities existed, adjusted to the
weight of single-strength juice. When juice was not listed as a
separate commodity for a fruit, juice was assigned to the regular
form of the fruit. Sweeteners from sweetened juice and juice
drinks were assigned to the appropriate commodities for
carbohydrate sweeteners.
Fruits consumed dried were assigned to commodities for the dried
forms. Re-hydrated or stewed dried fruits were required by EPA
to be assigned to the succulent forms.
The recipe for food code 62101050, "Fruit mixture, dried (mixture
includes three or more of the following: apples, apricots, dates,
papaya, peaches, pears, pineapples, prunes, raisins)," was
assigned to all 9 commodities listed in the description, in equal
proportions.
Sweeteners in canned fruit were assigned to the appropriate
commodities for carbohydrate sweeteners.
In the CSFII 1994-96, 1998, yogurt items containing fruit were
coded under umbrella codes described as "Yogurt, fruit variety."
The kind of fruit was not specified. For this project, the fruit
ingredients were assigned in equal proportions to the commodities
"Strawberry," "Blueberry," "Cherry," "Peach," "Banana," and
"Raspberry." The yogurt portion was assigned to milk as
described in section 3.7.13.
In CSFII ice cream and ice milk were coded as either "chocolate,"
or "flavors other than chocolate." For ice cream or ice milk
coded as "flavors other than chocolate," an amount of .001 grams
per 100 grams of food was assigned to each of the following
commodities: "Strawberry," "Blueberry," "Cherry," "Peach,"
"Banana," and "Raspberry."
3.7.9 Gelatin, dry
Dry gelatin as an ingredient in nonmeat foods was assigned in
equal proportions to the commodities "Beef, meat byproducts" and
"Pork, meat byproducts." As an ingredient in beef-containing
foods, e.g., beef broth, dry gelatin was assigned to "Beef, meat
byproducts" only. As an ingredient in pork-containing foods, it
was assigned to "Pork, meat byproducts" only.
3.7.10 Legumes
Attachment 1-21
-------
Sprouted or cooked legumes (dry beans and peas) were assigned to
their dry forms.
Beans described as white beans were assigned 50 percent to the
commodity "Bean, navy" and 50 percent to the commodity "Bean,
great northern."
3.7.11 Meat
3.7.11.1 Beef. EPA commodities for beef are:
Beef, fat
Beef, kidney
Beef, liver
Beef, meat
Beef, meat byproducts
Beef, meat, dried
Each beef item, other than kidney, liver, and byproducts, was
separated into two components. The total fat component, as
specified in the Survey Nutrient Database (7c), was assigned to
the commodity "Beef, fat." After subtracting total fat, the
remainder of an item was assigned to the commodity "Beef, meat."
Beef kidney, liver, and byproducts were assigned to their
respective commodities, which included both their fat and nonfat
components. See section 3.7.11.4 regarding kidney. Beef
byproducts are defined in the EPA list of commodities, located in
the \epa_comm.
Dried chip beef and beef jerky, including both fat and nonfat
components, were assigned to the commodity "Beef, meat, dried."
Veal, bison, and buffalo items were included in beef commodities.
3.7.11.2 Game. Bear, caribou, deer, frog, and squirrel were
assigned to the EPA commodity, "Meat, game." Other game meats
(armadillo, beaver, elk, groundhog, moose, snake, opossum, and
raccoon) listed in the description of this commodity were not
reported in CSFII 1994-96, 1998.
3.7.11.3 Goat. EPA commodities for goat are:
Goat, fat
Goat, kidney
Goat, liver
Goat, meat
Goat, meat byproducts
No goat kidney, liver or meat byproducts were reported in the
CSFII 1994-96, 1998. Other goat items were separated into two
components. The total fat component, as specified in the Survey
Attachment 1-22
-------
Nutrient Database (7b), was assigned to the commodity "Goat,
fat." After subtracting total fat, the remainder was assigned to
the commodity "Goat, meat."
3.7.11.4 Kidney. The CSFII 1994-96, 1998 did not distinguish
between different types of kidney. Kidney was assigned 56
percent to "Beef, kidney," 43 percent to "Pork, kidney," and 1
percent to "Sheep, kidney" (4).
3.7.11.5 Meat, not further specified. Some foods were described
as containing meat without being specific for the type of meat,
for example, "Meat Pie, Not Further Specified." The Survey
Recipe Database entries for those foods usually contained no more
than one meat ingredient. For most of those items, the meat
ingredient was assigned as 56 percent beef, 43 percent pork, and
1 percent lamb. Recipes for processed meats and pizzas were not
changed. These proportions were based on USDA food supply data
for 1991-1995 (4).
For foods described as containing meat and/or poultry, see
section 3 . 7.14.5.
3.7.11.6 Pork. EPA commodities for pork are:
Pork, fat
Pork, kidney
Pork, liver
Pork, meat
Pork, meat byproducts
Pork, skin
Pork products, other than pork kidney, liver, byproducts, and
skin, were separated into two components. The total fat
component, as specified in the Survey Nutrient Database (7b), was
assigned to the commodity "Pork, fat." After subtracting total
fat, the remainder of the pork was assigned to the commodity
"Pork, meat." Pork includes both fresh and cured forms. Pork
kidney, liver, skin, and byproducts were assigned to their
respective commodities. See section 3.7.11.4 regarding Kidney.
Pork byproducts are defined in the EPA commodity list in
\epa_comm.
3.7.11.7 Rabbit. The commodity "Rabbit, meat" includes both fat
and nonfat components. It also includes both wild and domestic
forms of rabbit.
3.7.11.8 Sheep. EPA commodities for sheep are:
Sheep, fat
Sheep, kidney
Sheep, liver
Sheep, meat
Attachment 1-23
-------
Sheep, meat byproducts
In CSFII 1994-96, 1998, food descriptions for sheep products were
identified as "lamb (including mutton)These products, other
than kidney and byproducts, were separated into two components.
The total fat component, as specified in the Survey Nutrient
Database (7b), was assigned to the commodity "Sheep, fat." After
subtracting total fat, the remainder was assigned to the
commodity "Sheep, meat." See section 3.7.11.4 regarding Kidney.
Sheep byproducts are defined in the EPA commodity list, located
in \epa_comm.
3.7.12 Meat and Poultry, Processed
CSFII foods and ingredients for processed meats were assigned to
commodities in proportions that reflected the amount of meat,
fat, organ meats, and other ingredients allowed by the USDA meat
and poultry standards (11, 12, 12a, 12b).
3.7.13 Milk
EPA commodities for milk are:
Milk, fat
Milk, nonfat solids
Milk, water
Milk, lactose
Foods assigned to milk commodities included dry, fluid, and
concentrated forms of milk, cheese, yogurt, cream, and butter.
Goat milk was included, but not soy milk. The milk items were
divided into fat, nonfat solids, and water based on the 1994-96
Survey Nutrient Database (7b) and were assigned to the
appropriate commodities. The nutrient database entries for total
fat and moisture were used to determine proportions of the
commodities "Milk, fat" and "Milk, water." "Milk, nonfat solids"
was the remainder after subtracting total fat and moisture. For
example, commodity assignments for 100 grams of whole milk were
3.34 grams to "Milk, fat," 87.99 grams to "Milk, water," and 8.67
grams (100-(3.34 + 87.99)) to "Milk, nonfat solids."
Lactose identified as a separate ingredient in a commercial baby
food was assigned to the commodity "Milk, lactose." Lactose as
an ingredient most often appeared in infant formulas. The lactose
from non-baby-foods and lactose inherent in milk and milk
products was included in the commodity "Milk, nonfat solids."
3.7.14 Poultry
3.7.14.1 Chicken. EPA commodities for chicken include:
Chicken, fat
Attachment 1-24
-------
Chicken, meat
Chicken, skin
Chicken, liver
Chicken, byproducts
The total fat components of both chicken flesh and chicken skin
were assigned to the commodity "Chicken, fat." The nonfat
component of chicken flesh was assigned to "Chicken, meat." The
nonfat component of chicken skin was assigned to the commodity
"Chicken, skin." Chicken neck was assigned to "Chicken,
byproducts."
The ratio of flesh to skin for different poultry pieces was
obtained from Agricultural Handbook No. 8 (6). The amount of
total fat present in different poultry pieces was based on the
1994-96 Survey Nutrient Database (7b).
3.7.14.2 Turkey. EPA commodities for turkey are the same as for
chicken (fat, meat, skin, liver, and byproducts). Assignments of
fat and nonfat components of turkey flesh and skin were made the
same as for chicken. The commodity, "Turkey, byproducts,"
included turkey neck. For turkey items described as dark meat, a
portion was assigned to turkey byproducts to represent turkey
neck.
3.7.14.3 Poultry other than chicken and turkey. EPA commodities
for "Poultry, other" are the same as for chicken (fat, meat,
skin, liver, and byproducts). Assignments of fat and nonfat
components of the poultry pieces were similar to assignments for
chicken. Dove, duck, partridge, pheasant, pigeon, quail, and
squab were assigned to the "Poultry, other" commodities. Emu,
goose, guinea hen, and ostrich were also listed by EPA as
"Poultry, other," but those foods were not reported in the
surveys.
3.7.14.4 Mixtures described as containing chicken or turkey. In
the CSFII 1994-96, 1998 several food mixtures were described as
containing either chicken or turkey, for example, "Chicken or
turkey teriyaki." The Survey Recipe Database entries usually
contained chicken, but not turkey, as an ingredient for those
foods. Therefore, for most of these foods the chicken ingredient
was assigned to commodities for both chicken and turkey in the
proportions of 77 percent chicken and 23 percent turkey. This
distribution was based on chicken and turkey food supply data for
the years 1990-1994 (3).
3.7.14.5 Mixtures described as containing meat and/or poultry.
In the CSFII 1994-96, 1998 several food mixtures were described
as containing either meat or poultry, for example, "Tamale with
meat and/or poultry." Assignment of meat and/or poultry
ingredients in most of these foods were handled in two steps.
First, they were divided into 50 percent meat and 50 percent
Attachment 1-25
-------
poultry. Then, meat was divided into 56 percent beef, 43 percent
pork, and 1 percent lamb. (See section 3.7.11.8, Meat, not
further specified.) Poultry was divided into 77 percent chicken
and 2 3 percent turkey (3).
3.7.15 Spices and herbs
The EPA Food Commodity Vocabulary lists some spices and herbs as
separate commodities. Others are grouped together in "Spices,
other" and "Herbs, other." Complete listings of spices and herbs
included in these commodities can be found in the Food Commodity
Vocabulary's Appendices B and C in \epa_comm\foodvoc.wpd. Spices
and herbs usually were not included in CSFII recipes, but were
added before commodity assignments were made.
3.7.16 Sweeteners, carbohydrate
EPA commodities for carbohydrate sweeteners are:
Beet, sugar, molasses
Beet, sugar
Corn, field, syrup
Honey
Maple sugar
Maple syrup
Sorghum syrup
Sugarcane, molasses
Sugarcane, sugar
Sugar and molasses were not identified by source (sugarcane vs.
sugar beet) in the CSFII 1994-96, 1998. Therefore, sugar and
molasses were assigned to the commodities "Sugarcane, sugar" and
"Beet, sugar," according to the proportions of 56 percent
sugarcane and 44 percent sugar beet. These proportions reflect an
estimate of refined sugar use in the continental United States
from 1990-94 (15). Sugar included table sugar, confectioner's
sugar, light brown sugar, dark brown sugar, raw sugar, and
caramelized sugar.
Dry corn syrup solids, dextrose, and fructose were assigned to
the commodity of "Corn, field, syrup," adjusting from the dry
forms to liquid based on differences in total solids. Sugarcane
syrup was assigned to the commodity "Sugarcane, sugar," adjusting
from the liquid to dry form.
Corn syrup, maple sugar, maple syrup, sorghum syrup, and honey
were assigned directly to their respective commodities. No
adjustments were necessary.
Attachment 1-26
-------
3.7.17 Vegetables
Dried vegetables and vegetable juices were assigned to
commodities for their respective dried or juice forms when such
commodities were listed. If dried vegetables or vegetable juices
were not present in the commodities list, they were assigned to
their respective fresh vegetable forms. No adjustments were made
to their weights.
3.7.18 Water
The EPA Food Commodity Vocabulary includes the following entries
for water:
Water, dilution, source not specified
Water, tapwater- direct (drinking)
Water- indirect (cooking)
Water, bottled water
Water, commercial beverage
Water in foods and beverages is being assigned to the above
categories by the EPA Office of Water. The information is not a
part of this database at this time.
3.7.19 Commodities not consumed
Several commodities on EPA's list were not consumed at all during
the CSFII 1994-96, 1998 and do not appear in the intake database.
Table 2 lists these commodities. Several commodities were
included with food codes for similar foods. For example, the
description for food code 53304000 is "Pie, blueberry, two crust
(include huckleberry)." In this latter situation, EPA requested
that a small portion of the food code be assigned to the
"included" commodity, if the commodity was not covered in any
other food-code-to-commodity translation. Since huckleberry had
received no other assignment, a small portion of food code
53304000 was assigned to the commodity "Huckleberry." In
accordance with EPA's request, assignment was 0.0 01 grams of
"Huckleberry" per 100 grams of pie (the smallest amount the Food
Code-to-Commodity Translation File could accept).
3.7.2 0 Special requirement to include commodities consumed in
very small amounts
EPA requested that data processing procedures be established to
ensure that commodities consumed in very small amounts be
represented in the resulting database files. While overstating
precision is a concern, the possibility of small amounts rounding
to zero was an even greater concern to EPA. Therefore, data in
the Food-Code-to-Commodity Translation File are presented to 3
decimal places, allowing an estimate as small as 0.001 to be
Attachment 1-27
-------
included. Data in the commodity intake file are presented to 6
decimal places, allowing intake estimates as small as 0.000001
gram per kilogram of body weight to be included. These
expressions are not intended to suggest that data can be
estimated to these small amounts with such precision.
3.8 Notes about reviewing the Food-Code-to-Commodity Translation
File
When reviewing the Food-Code-to-Commodity Translation File, some
unexpected findings may be observed. Reasons for these findings
can be found in the above discussions on matching USDA food codes
to EPA commodities (section 3.6.1), converting weights of foods
and ingredients (section 3.6.2), and special requirements for
assigning foods and ingredients to EPA commodities (section 3.7).
Examples of unexpected findings are listed below:
1. The following five commodities are present in small amounts
in many foods: "Cassava," "Corn, field, starch," "Potato, flour,"
"Rice, flour," and "Wheat, flour." This occurs because modified
food starch is frequently listed as an ingredient in processed
foods. When the source of the starch was not identified, it was
assigned to these commodities.
2. "Barley, flour," is found in small amounts in many foods.
This occurs when malt was listed as an ingredient.
3. "Beef, byproducts" and Pork, byproducts" are found in non-
meat foods. Presence of these commodities in non-meat foods
usually results from the presence of gelatin as an ingredient in
the food. For example, gelatin is an ingredient in marshmallows.
Therefore, when marshmallows are used in other foods, meat
byproducts occur in the commodity assignments for those foods.
4. "Beef, byproducts" are found in ground beef and in foods
containing ground beef as an ingredient. Because the Standards
of Identity for ground beef allow for inclusion of beef cheeks, a
small proportion of ground beef was assigned to beef byproducts.
5. Small amounts of "Soybean, oil" may be found in foods not
expected to have a fat ingredient. This results from the presence
of soya lecithin as an ingredient.
6. Many foods were assigned to nine different oil commodities
(per EPA requirement), including small amounts of "Olive, oil"
and "Sesame, oil." This occurs because a composite of oils was
used when the specific type of oil was unknown, such as in home
prepared foods (see section 3.7.6).
7. Fruit juices are present in concentrates at greater than 100
percent. This occurs because, upon assignment to a commodity,
Attachment 1-28
-------
the juice concentrate was adjusted to the corresponding weight of
the reconstituted product. Thus, 100 grams of orange juice
concentrate translates to 354.2 grams of the commodity "Orange,
juice."
3.9 Limitations of the Food Commodity Intake Database
Users of the Food Commodity Intake Database should understand the
nature of food survey data, as well as the assumptions required
for translating survey food codes into EPA-defined commodities.
This information should be considered in the interpretation of
analyses based on the Food Commodity Intake Database.
--The food intake surveys were based on ability of sample
persons, with assistance of probing by trained interviewers, to
recall the foods they ate during the 24-hour period prior to the
day of their interview.
--Amounts of foods consumed were estimated, involving two steps.
First, the sample persons provided a description of the volumes,
using standard survey measurement aids or label information.
Then, the volumes were converted to grams using a database of
weight/volume relationships.
--Ingredient amounts for commercial products were estimated.
Ingredients and amounts for other food mixtures were based on
representative recipes, usually from popular cookbooks but from
regional or specialty cookbooks when necessary. The recipes
were not specific for each sample person.
--Sometimes ingredients were proportioned among commodities based
on food supply information, for example, oils and sweeteners (see
section 3.7 above).
--Body weights were self reported. Missing body weights were
assigned default values (see section 4.5.2).
Attachment 1-29
-------
3.10 References
(1) Marcoe, K.K., and D.B. Haytowitz. 1993. Estimating nutrient
values of mixed dishes from label information. Food Technology
47 (4) : 69-75.
(2) Matthews, R.H., and Y.J. Garrison. 1995. Food Yields
Summarized by Different Stages of Preparation. U.S. Department of
Agriculture, Agricultural Research Service, Agriculture Handbook
No. 102.
(3) Putnam, J.J. and J.E. Allshouse. 1996. Food Consumption,
Prices, and Expenditures, 1996: Annual Data 1970-94. U.S.
Department of Agriculture, Economic Research Service, Statistical
Bulletin No. 928.
(4) Putnam, J.J. and J.E. Allshouse. 1997. Food Consumption,
Prices, and Expenditures, 197 0-95. U.S. Department of
Agriculture, Economic Research Service, Statistical Bulletin No.
939.
(5) Tippett, K.S., and Y.S. Cypel (eds.). 1998. Design and
Operation: The Continuing Survey of Food Intakes by Individuals
and the Diet and Health Knowledge Survey, 1994-96. U.S.
Department of Agriculture, Agricultural Research Service,
Nationwide Food Surveys Report No. 96-1.
(6) U.S. Department of Agriculture, Agricultural Research
Service. 1976-92. Composition of Foods...Raw, Processed,
Prepared. Agriculture Handbook No. 8, Revised Sections 1-22.
(7) U.S. Department of Agriculture, Agricultural Research
Service. 1998. 1994-96 Continuing Survey of Food Intakes by
Individuals and 1994-96 Diet and Health Knowledge Survey.
National Technical Information Service, 5285 Port Royal Road,
Springfield VA 22161;(703)487-4650. CD-ROM: Accession no. PB98-
500457.
(7a) . Food Coding Database for the 1994-1996 Continuing
Survey of Food Intakes by Individuals. On CD-ROM: 1994-96
Continuing Survey of Food Intakes by Individuals and 1994-96 Diet
and Health Knowledge Survey. NTIS Accession no. PB98-500457.
(7b) . Survey Nutrient Database for the 1994-1996 Continuing
Survey of Food Intakes by Individuals. On CD-ROM: 1994-96
Continuing Survey of Food Intakes by Individuals and 1994-96 Diet
and Health Knowledge Survey. NTIS Accession no. PB98-500457.
(7c) . Survey Recipe Database for the 1994-1996 Continuing
Survey of Food Intakes by Individuals. On CD-ROM: 1994-96
Continuing Survey of Food Intakes by Individuals and 1994-96 Diet
and Health Knowledge Survey. NTIS Accession no. PB98-500457.
Attachment 1-30
-------
(8) U.S. Department of Agriculture, Agricultural Research
Service. 2 0 00. Continuing Survey of Food Intakes by Individuals
1994-96, 1998. National Technical Information Service, 5285 Port
Royal Road, Springfield VA 22161;(703)487-4650. CD-ROM. NTIS
Accession no. PB2000-500027.
(9) U.S. Department of Agriculture, Agricultural Research
Service, Nutrient Data Laboratory. 1998. Unpublished Data.
Personal communication between Nutrient Data Laboratory staff and
Food Surveys Research Group staff.
(10) U.S. Department of Agriculture, Economic Research Service.
July 1995. Oil Crops Yearbook.
(11) U.S. Department of Agriculture, Food Safety and Inspection
Service, Inspection Operations, Regulatory Programs. 1996. Food
Standards and Labeling Policy Book.
(12) U.S. Department of Agriculture, Food Safety and Inspection
Service. 1998. Animal and animal products: Approval of
substances for use in the preparation of products. Code of
Federal Regulations, Title 9, Pt. 318.7.
(12a) . Animals and animal products: Definitions and standards
of identity or composition. Code of Federal Regulations, Title 9,
Pts. 319.1-319.881.
(12b) . Animals and animal products: Poultry products
inspection regulations, Subpart P, Definitions and standards of
identity or composition. Code of Federal Regulations, Title 9,
Pts. 381.155-381.174.
(13) U.S. Department of Agriculture, Foreign Agricultural
Service. 1995. Tropical Products: World Markets and Trade,
Circular Series FTROP 3-95, page 21, Table: "Tea: Sales by U.S.
Retail Stores, 1995 with Comparison".
(14) U.S. Department of Agriculture, Foreign Agricultural
Service, Oilseeds and Products Division, Import and Export Data.
1998. Personal communication between Foreign Agricultural Service
and Center for Nutrition Policy and Promotion staff.
(15) U.S. Department of Agriculture, National Agricultural
Statistics Service. 1997. Agricultural Statistics 1997, page II-
20, Table 2-33.
(16) U.S. Environmental Protection Agency. 2000. The EPA Food
Commodity Vocabulary, Master List dated June 15, 2000.
Table 1. CSFII food codes representing commercial baby foods
Attachment 1-31
-------
Food code
or food code
prefix Foods
117 * Infant formulas
133 Milk desserts
14220000 Cottage cheese, with fruit
20000070 Meat, not specified as to type
20000090 Meat sticks, not specified as to type of meat
217 Beef
228 Pork
234 Lamb or veal
247 Poultry
2518 Variety meats
276 Meal, poultry, or fish mixtures with nonmeat items
34 Eggs
53203050 Cookie, fruit
53203100 Cookie
53242250 Cookie, teething
54350000 Crackers
54408100 Pretzel
578 Cereals
585 Grain mixtures
57 Fruits, fruit juices, and fruit mixtures
76 Vegetables and mixtures mostly vegetable
* All food codes with "117" in the first three digits.
Attachment 1-32
-------
Table 2. EPA food commodities not appearing in the intake
database
Acerola
Bean, broad, succulent
Broccoli raab
Buckwheat, flour
Burdock
Canistel
Cardoon
Celeriac
Celtuce
Cherimoya
Chickpea, flour
Citrus, oil
Crabapple
Cress, garden
Cress, upland
Elderberry
Feij oa
Fennel, Florence
Filbert, Oil
Goat, meat byproducts
Goat, kidney
Goat, liver
Hickory nut
Horse, meat
Jaboticaba
Jackfruit
Kohlrabi
Kumquat
Lemongrass
Spanish Lime
Loganberry
Longan
Loquat
Lychee, dried
Mamey apple
Parsley, turnip rooted
Pawpaw
Peppermint, oil
Pummelo
Quince
Quinoa
Radish, Oriental, tops
Radish, tops
Rape greens
Rye, flour *
Salsify, roots
Salsify, tops
Sapote
Sorghum
Soursop
Spearmint
Spearmint, oil
Sugar apple
Tree Tomato
Water, dilution, source NS**
Water, tapwater-direct
(drinking) **
Water-Indirect (cooking) **
Water, bottled water **
Water, commercial beverage**
Almond, oil-babyfood
Almond-babyfood
Arrowroot, flour-babyfood
Basil, fresh leaves-babyfood
Coriander, leaves-babyfood
Coriander, seed-babyfood
Corn, field, meal-babyfood
Egg, white (solids)-babyfood
Ginger-babyfood
Guar, seed-babyfood
Guava-babyfood
Honey-babyfood
Papaya-babyfood
Passionfruit, juice-babyfood
Passionfruit-babyfood
Peach, dried-babyfood
Pepper, bell-babyfood
Plum, prune, dried-babyfood
Potato, tuber, w/peel-babyfood
Sesame, oil-babyfood
Sesame, seed-babyfood
Soybean, soy milk-babyfood
Triticale, flour-babyfood
Turkey, liver-babyfood
* Rye flour translated to EPA
Commodity "Rye, Whole grain"
**Water translated by EPA
Office of Water
Attachment 1-33
-------
4 CHARACTERISTICS AND FORMATS OF THE FOOD COMMODITY INTAKE
DATABASE
4.1 Introduction
The Food Commodity Intake Database contains several different
data files related to intake of over 500 EPA-defined commodities.
It is based on food consumption surveys conducted by the U.S.
Department of Agriculture's Agricultural Research Service. The
surveys are the 1994-96 Continuing Survey of Food Intakes by
Individuals (CSFII 1994-96) and its Supplemental Children's
Survey (CSFII 1998) . The database was developed for use in
estimating human exposure to pesticide residues through food and
beverage intake.
The general file structure and information concerning the use of
the data are discussed in Sections 4.2 and 4.3. Section 4.4
contains the data file formats, which provide detailed
descriptions of all of the fields included in each file. The
field number, name, width, and type are given, along with a full
description of the field and its applications, valid values for
the field, and, if applicable, a reference to the question number
on the original CSFII questionnaire. Section 4.5 contains
miscellaneous notes about the data files.
4.2 Database Structure
The Commodity Intake Database is made up of 18 data files. These
data files are classified into four categories as follows: (1) 8
Documentation Files, (2) 2 EPA Commodity Files, (3) 3 Commodity
Intake Data Files, and (4) 5 Supporting Data Files. Records in
the different data files can be linked or connected through the
key fields that are described Section 4.3.2. (See section 1
"\document\sectionl.wpd" for a tree diagram and summary
description of data files on this database.)
4.2.1 Documentation Files (files in directory "\document")
Each of the 5 documentation sections and the 3 appendices listed
in the Table of Contents is contained in a separate WordPerfect
file, as indicated below.
Sectionl.wpd Table of contents, File names
Section2.wpd Essential information
Section3.wpd Methodology for translating food intakes
into EPA commodity intakes
Section4.wpd Data file characteristics and formats
Section5.wpd Control counts
Appenda.wpd Appendix A. EPA's use of food consumption
Attachment 1-34
-------
data in assessing dietary risk from
pesticides
Appendb.wpd Appendix B. CSFII 1994-96, 1998
Methodology (Section 3 from CSFII 1994-
96, 1998 CD-ROM documentation)
Appendc.wpd Appendix C. CSFII 1994-96, 1998 Sampling
Weights and Statistical Notes (excerpts
from sections 5 and 6 of the CSFII 1994-
96, 1998 CD-ROM documentation)
4.2.2 EPA Commodity Files (files in directory "\epa_comm")
The Environmental Protection Agency (EPA) provided 2 files. Both
are included as WordPerfect documents:
foodvoc.wpd EPA commodity list
csffcm.wpd Procedures for assigning EPA cooked
status, food form, and cooking method
values to USDA food codes
4.2.3 Commodity intake data files (files in directory "\intake")
The commodity intake data files include 2 files that were created
by translating the CSFII 1994-96, 1998 food intakes into EPA
commodities, and one file of sample person data. These files,
which are in ASCII delimited format, include:
comm9498.txt Commodity intakes for CSFII 1994-96, 1998
ffcm9498.txt Commodity intakes by cooked status, food
form, and cooking method for CSFII 1994-
96, 1998
smpl9498.txt Sample person data for CSFII 1994-96,
1998
4.2.4 Supporting data files (files in directory "\support")
Five data files provide supporting documentation for the
translation of the CSFII 1994-96, 1998 intake data into EPA
commodities. These files, which are in ASCII delimited format,
are:
fc comm.txt
fcdesc.txt
fcincl.txt
fcscheme.txt
moddesc.txt
CSFII 1994-96, 1998 food codes linked to
EPA commodities (grams of commodities per
10 0 grams of food)
Food code descriptions
Food code include statements (extensions
of food code descriptions)
Food code outline
Descriptions of recipe modifications
Attachment 1-35
-------
4.3 General nature of the data
The data files that make up the Commodities Intake Database are
in the following formats:
Directory
Format
\Document
\EPA_Comm
\lntake
\Support
WordPerfect
WordPerfect
ASCII delimited
ASCII delimited
ASCII delimited fields are separated by the caret (*") symbol and
alpha-numeric/character fields are enclosed in tilde (~) marks.
The majority of the data fields are numeric, but there are some
alpha-numeric text fields. The numeric fields include explicit
decimal points, and do not include leading zeros.
Some fields have "missing" values due to a nonspecific or absent
response, or due to the lack of the necessary data for
calculations. These fields are not left blank, but are filled
with codes to indicate the separate categories of "refused,"
"don't know," and "not ascertained," or sometimes
"indeterminable." The usual convention for a one column field is
a '7' for "refused," '8' for "don't know," and '9' for "not
ascertained." Two-column fields have values of '97,' '98,' and
'99' to represent these categories of "missing" values.
Continuous fields may also have codes for "missing" values.
When a blank or null value is found in a field, it means that a
response or calculation for that field does not apply to a
particular situation. For example, the two-day sampling weight
fields will be blank for sample persons who have not provided two
days of intake. In general, if a skip pattern dictates that a
question should not be asked of a respondent, the corresponding
field on the record will be blank. Blank or Null fields appear
in the data files as two delimiters side by side (e.g., **) .
4.3.1 Data File Characteristics
The ASCII delimited files have variable record lengths. The five
ASCII delimited data files have the following record counts.
File Total number of records
Attachment 1-36
-------
comm94 98.txt
3,173,197
ffcm9498.txt
5,781,938
fcincl.txt
5, 652
smpl9498.txt
21,662
fc comm.txt
99,011
fcdesc.txt
5, 845
moddesc.txt
3 , 902
4.3.2 List of Key Fields
Specific fields in each data file have been designated as "key"
fields to (1) ensure that each record is unique, and (2) allow
data in different files to be linked or joined. Following is a
list of fields designated as key fields:
HHID
Household identification number
SPNUM
Sample person number
DAYCODE
Day of intake identifier
COM_CODE
EPA commodity code
CKD_STAT
EPA cooked status code
FD_FORM
EPA food form code
CKG_METH
EPA cooking method code
FOODCODE
USDA food code
MODCODE
Recipe modification code
4.3.3 Sampling weights
Note: For an in-depth discussion of weighting for the CSFII 1994-
96, 1998, see Appendix C in \document\appendc.wpd.
The "smpl9498.txt" data file contains several different sampling
weights. By using appropriate weights: (a) data for each survey
year (1994, 1995, 1996, or 1998) can be used separately; (b) data
for the 3 years of CSFII 1994-96 can be combined; or (c) data for
all 4 years of the surveys can be combined. In addition, weights
are present for using only one day of intake for each sample
person; other weights provide for the use of only sample persons
who provided 2 days of intake.
Attachment 1-37
-------
Use the annual sampling weight fields, in conjunction with the
year field, when data are to be analyzed for one survey year
(1994, 1995, 1996, or 1998). Annual sampling weight fields are:
WTA_DAY1 The annual day 1 sampling weight for all responding
CSFII 1994-96, 1998 sample persons. This annual
sampling weight is used whenever the sample of
interest includes sample persons who provided the
first day of intake data regardless of whether they
provided the second day.
WTA_2DAY The annual 2-day sampling weight for all CSFII 1994-
96, 1998 sample persons with two days of intake.
This annual sampling weight is used whenever the
sample of interest includes only sample persons who
provided 2 days of intake data.
The 3-year sampling weight fields for CSFII 1994-96 are:
WT3_DAY1 The 3-year day 1 sampling weight for all responding
CSFII 1994-96 sample persons. This 3-year sampling
weight is used whenever the sample of interest
includes sample persons who provided the first day of
intake data regardless of whether they provided the
second day.
WT3_2DAY The 3-year 2-day sampling weight for all CSFII 1994-
96 sample persons with two days of intake. This
3-year sampling weight is used whenever the sample of
interest includes only sample persons who provided 2
days of intake data.
The sample weight field for combining all 4 years of the CSFII
1994-96, 1998 are:
WT4_DAY1 The 4-year day 1 sampling weight for all responding
CSFII 1994-96, 1998 sample persons. This 4-year
sampling weight is used whenever the sample of
interest includes sample persons from both surveys
who provided the first day of intake data regardless
of whether they provided the second day.
WT4_2DAY The 4-year 2-day sampling weight for all CSFII 1994-
96, 1998 sample persons with two days of intake.
This 4-year sampling weight is used whenever the
sample of interest includes only sample persons from
both surveys who provided 2 days of intake data.
4.4 Data File Formats
Attachment 1-38
-------
4.4.1 Introduction to the File Formats for ASCII delimited files
Seven data files on the Food Commodity Intake Database are stored
in ASCII delimited formats (files in directories "\intake" and
"\support"). Fields are separated by the caret {*) symbol and
alpha-numeric/character fields are enclosed in tilde (~) marks.
Blank fields appear as two delimiters side by side (i.e., **) .
This section on file formats describes the contents of each of
the 7 data files. The fields in each file are listed and for
each field, the following information is provided: field number,
field name, field width, field type, and description.
The field number identifies the position or order in which the
field can be found on the data record.
The field name will be no longer than eight characters and
will always be referred to in the file formats in uppercase
letters.
The width is the maximum number of columns allocated to the
field including, where appropriate, an explicit decimal point.
All alpha-numeric fields are encased in tildes (~), which are
not counted in the field width.
The type of the field is either 'N' for numeric or 'A' for
alpha-numeric or character. If a numeric field has a
fractional part the number of decimal places follows the N.
For example, 'N2' indicates a field with two decimal places.
The description includes allowed values and their meanings and
skip patterns dictated by specific field values.
4.4.2 Formats for the food commodity intake data files
4.4.2.1 Format of the Commodity Intakes file (comm94 98.txt!
Field Field Width
Number Name /Type Description/Application/Values
HHID 5 N Household identification number.
10001 - 52999 = HHID
SPNUM 2 N Sample person (SP) number.
1 - 23 = SP number
Attachment 1-39
-------
DAYCODE
N Day 1 / Day 2 / Average indicator.
Note: There is one record per EPA
commodity per SP per day of
intake. Where two days were
reported there is also a record
containing daily averages.
1 = Day 1
2 = Day 2
4 = Average of day 1 and day 2
COM CODE 8
N
EPA commodity code.
Note: See "foodvoc.wpd" for codes and
complete descriptions of each
COM CODE.
COM AMT 10 N6
Total (or average) amount of the
EPA commodity consumed during that
day.
Amount in grams per kilogram body
weight
4.4.2.2 Format for the files of commodity intakes by cooked
status, food form and cooking method (ffcm9498.txt)
Field Field Width
Number Name /Type Description/Application/Values
HHID
N Household identification number.
10001 - 52999 = HHID
SPNUM
N Sample person (SP) number.
1 - 23 = SP number
DAYCODE
N Day 1 / Day 2 / Average indicator.
Note: There is one record for each EPA
commodity and Cooked Status-Food
Form-Cooking Method
combination(COM-FFCM), per SP,
per day of intake. Where two
days were reported there is also
a record containing daily
Attachment 1-40
-------
averages.
1 = Day 1
2 = Day 2
4 = Average of day 1 and day 2
COM_CODE 8 N EPA commodity code.
Note: See "foodvoc.wpd" for possible
codes and complete descriptions
of each COM CODE.
CKD_STAT 1 N EPA cooked status identifier.
0 = Not applicable
1 = Uncooked
2 = Cooked
3 = Processed oil
6 = Frozen meal
7 = Salad
8 = Sandwich
9 = Not specified as to cooked
or uncooked
FD_FORM 1 N EPA food form identifier.
0 = not applicable
1 = Fresh
2 = Frozen
3 = Dried
4 = Canned
5 = Cured, pickled, smoked,
salted
8 = Other process forms not
listed above
9 = NS as to form or multiple
forms
CKG_METH 1 N EPA cooking method identifier.
0 = None or Not applicable
1 = Baked
2 = Boiled
3 = Fried
4 = Fried or baked
5 = Boiled or baked
8 = Not specified as to cooking
method or multiple cooking
methods
9 = Not specified as to whether
Attachment 1-41
-------
further cooking occurred
CFFCMAMT 10 N6
Total amount consumed of the commodity
and cooked status-food form-cooking
method combination.
Amount in grams per kilogram body
weight
4.4.2.3 File Format for Sample Person Data (smpl9498.txt)
Where appropriate, the description includes references to the
original question number from the CSFII questionnaires. For
example "H52" indicates Household question number 52.
Questionnaire symbols are:
S = Screener questionnaire
H = Household questionnaire
DA = Individual Intake Questionnaire (day 1)
DB = Individual Intake Questionnaire (day 2)
Field
Number
Field
Name
Width
/Type
Description/Application/Values
HHID
5 N
SPNUM 2 N
REGION 1 N
URB
1 N
HHSIZE 2 N
Household identification number.
10001 - 52999 = HHID
Sample person (SP) number.
1 - 23 = SP number
Region of the United States.
1 = Northeast
2 = Midwest
3 = South
4 = West
Urbanization; Metropolitan Statistical
Area(MSA) status.
1 = MSA, central city
2 = MSA, outside central city
3 = Non-MSA
H Household size; count of household
members.
1 - 23 = Count
Attachment 1-42
-------
6 INCOME 6 N H52. During the previous calendar
year, approximately how much income
from all sources did you and other
household members have before taxes?
(Please give me your best estimate.)
Note: annual incomes have been imputed for
households that could not or would not provide
a response to this question. See documentation
for the CSFII 1994-96, 1998 for an explanation
of the methods employed. See INCREP for the
original response to H52. See IMPFLAG for the
method of imputation employed.
0 - 99999 = Dollars
100000 = $100,000 or more
7 INCREP 1 N H52. Type of original response to H52.
* 1 = Value of INCOME is the actual
amount reported.
* 5 = No household interview
* 6 = Not a household in the
previous calendar year
7 = Refused
8 = Don't know
9 = Not ascertained
* Skip INCCODE.
8 INCCODE 1 A H53. Please tell me which letter on this
card best represents your combined
household income before taxes for the
previous calendar year.
Note: H53 is only asked of households
that could not or would not answer H52.
Applies if: INCREP >= 7
A = Under $50 00
B = $5,000 - $9,999
C = $10,000 - $14,999
D = $15,000 - $19,999
E = $20,000 - $24,999
F = $25,000 - $29,999
G = $30,000 - $34,999
H = $35,000 - $39,999
Attachment 1-43
-------
I = $40,000 - $44,999
J = $45,000 - $49,999
K = $50,000 - $59,999
L = $60,000 - $74,999
M = $75,000 - $99,999
N = $100,000 and over
7 = Refused
8 = Don't know
9 = Not ascertained
Blank = Not applicable
IMPFLAG 1 N Annual income imputation flag.
1 = Not imputed, value of INCOME
is the actual amount
reported.
2 = Imputed, value based on H53
(INCCODE)
3 = Imputed, value based on
monthly income
4 = Imputed, value based on
regression equation
5 = Imputed, based on segment
level mean income
10 PCTPOV 3 N Annual income expressed as a percentage
of the poverty threshold. Based on
INCOME (using imputed values) and HHSIZE.
0 - 299 = Percentage of the poverty
threshold
300 = 300% or more
11 AGE 2 N Age of household member in years.
Note: Age at time of day 1 intake.
0 = Under 1 year old
* 1 - 89 = Age in years
* 90 = 90 or older
* Skip AGE_M.
12 AGE_M 2 N Age of household member in months. Valid
only for children 11 months old or
younger.
Note: Age at time of day 1 intake.
Applies if: AGE = 0
0 = Less than one month old
Attachment 1-44
-------
1 - 11 = Months of age
Blank = Not applicable
13 SEX 1 N Sex of household member.
1 = Male
2 = Female
14 WGT_SP 3 N DA30. How much do you weigh without
shoes?
1 - 995 = Pounds
997 = Refused
998 = Don't know
999 = Not ascertained
15 WGT_KG 3 N1 Weight of sample person in kilograms.
1 - 452 = Kilograms
The amount of each commodity consumed is
reported in grams per kilogram body weight.
The CSFII 1994-96, 1998 reports each sample
persons body weight (WGT_SP) in pounds, as
reported by the sample person. These body
weights were converted from pounds to kilograms
using the following formula:
Weight in Kilograms = Weight in pounds
2 .2046
Note: default body weights were used for sample
persons whose WGT_SP = '997', '998', or '999'. See
section 4.5.2 about Unreported Body Weights.
16 REL_REF 2 N S8. What is your relationship to the
reference person?
0 =
Reference person
1 =
Spouse
2 =
Natural or adopted child;
step child
3 =
Grandchild
4 =
Parent
5 =
Brother or sister
6 =
Other relative
7 =
Foster child
8 =
Partner; roommate;
girlfriend; boyfriend
9 =
Roomer or boarder
10 =
Employee
11 =
Guest
Attachment 1-45
-------
12 = Other unrelated
17 RACE 1 N S9. Which of the groups on this card
best describes your race?
1 = White
2 = Black
3 = Asian, Pacific Islander
4 = American Indian, Alaskan
native
5 = Other
18 ORIGIN 1 N S10. Do any of these groups (from a
card) represent your national origin?
1 = Mexican, Mexican American,
Chicano
2 = Puerto Rican
3 = Cuban
4 = Other Spanish / Hispanic
5 = None of the above
19 PL_STAT 1 N Pregnant / lactating status.
Note: From questions H26, H27, H29 and
H31. Also, these questions were only
asked of households with certain
characteristics as identified at
screening.
1 = Pregnant
2 = Lactating
3 = Pregnant and lactating
4 = Not pregnant or lactating
5 = Not female 10-55
2 0 BF_STAT 1 N Breast-feeding status.
Note: From questions H2 9 and H3 0. Also,
these questions were only asked of
households with children 3 years old or
less identified at screening.
1 = Breast-feeding
2 = Not breast-feeding
3 = Over 3 years old
21 H20_C00K 2 N H18. What is the main source of the
water used for cooking in your home?
1 = Community water supply
2 = Well or rain cistern
Attachment 1-46
-------
(household's)
3 = Spring (household's or public)
4 = Bottled water (purchased)
96 = Other
98 = Don't know
99 = Not ascertained
22 H20_BEVR 2 N H19. What is the main source of the
water used in your home for preparing
beverages such as coffee, tea, juices,
and baby formula?
1 = Community water supply
2 = Well or rain cistern
(household's)
3 = Spring (household's or public)
4 = Bottled water (purchased)
96 = Other
98 = Don't know
99 = Not ascertained
23 H20_DRNK 2 N H20. What is the main source of plain
drinking water in your home?
1 = Community water supply
2 = Well or rain cistern
(household's)
3 = Spring (household's or public)
4 = Bottled water (purchased)
96 = Other
98 = Don't know
99 = Not ascertained
24 D1_H20_0 3 N DA15. How many fluid ounces of plain
drinking water, that is, tap water or any
bottled water that is not carbonated,
with nothing added to it, did you drink
yesterday - day 1?
* 0 = None
1 - 995 = Fluid ounces
998 = Don't know
999 = Not ascertained
* Skip D1_H20_H - D1_H20_A
25 D1_H20_H 1 N DA16. How much of this plain drinking
water came from your home? Would you say
all, most some, or none - day 1?
Applies if: D1_H20_0 > 0
Attachment 1-47
-------
* 1 = All
2 = Most
3 = Some
4 = None
8 = Don't know
9 = Not ascertained
* Skip D1_H20_A
26 D1_H20_A 1 N DA17. What was the main source of plain
drinking water that did not come from
your home? Was it tap water, water from
a drinking fountain, bottled water, or
something else - day 11
Applies if: D1_H20_H > 1
1 = Tap water / drinking
fountain
2 = Bottled water
6 = Other
8 = Don't know
9 = Not ascertained
Blank = Not applicable
27 D2_H20_0 3 N DB13. How many fluid ounces of plain
drinking water, that is, tap water or any
bottled water that is not carbonated,
with nothing added to it, did you drink
yesterday - day 2?
Applies if: C0MP_D2 = 1
* 0 = None
1 - 995 = Fluid ounces
998 = Don't know
999 = Not ascertained
Blank = Not applicable
* Skip D2_H20_H - D2_H20_A
28 D2_H20_H 1 N DB14. How much of this plain drinking
water came from your home? Would you say
all, most some, or none - day 2?
Applies if: D2_H20_0 > 0
* 1 = All
2 = Most
3 = Some
4 = None
Attachment 1-48
-------
8 = Don't know
9 = Not ascertained
Blank = Not applicable
* Skip D2_H20_A
29 D2_H20_A 1 N DB15. What was the main source of plain
drinking water that did not come from
your home? Was it tap water, water from
a drinking fountain, bottled water, or
something else - day 2?
Applies if: D2_H20_H > 1
1 = Tap water/drinking fountain
2 = Bottled water
6 = Other
8 = Don't know
9 = Not ascertained
Blank = Not applicable
3 0 C0MP_D1 1 N Is there complete Day 1 intake data for
this individual?
1 = Yes
31 COMP_D2 1 N Is there complete Day 2 intake data for
this individual?
1 = Yes
* 2 = No
* Skip WT3_2DAY, WTA_2DAY.
32 WT3_DAY1 8 N 3-year day 1 sample weight for CSFII
1994-96. Read section 4.3.3 on sampling
weights.
Applies to day 1 records with 1994,
1995, or 1996 in YEAR field.
1 - 99999999 = Weight
Blank = Not applicable
33 WT3_2DAY 8 N 3-year 2-day sample weight for CSFII
1994-96. Read section 4.3.3 on sampling
weights.
Applies to day 1 and day 2 records
with 1994, 1995, or 1996 in YEAR
Attachment 1-49
-------
field and 1 C0MP_D2 field.
1 - 99999999 = Weight
Blank = Not applicable
34 WTA_DAY1 8 N Annual day 1 sample weight. Read section
4.3.3 on sampling weights.
Use in conjunction with YEAR field
Applies if: C0MP_D1 = 1
1 - 99999999 = Weight
Blank = Not applicable
35 WTA_2DAY 8 N Annual 2-day sample weight. Read section
4.3.3 on sampling weights.
Use in conjunction with YEAR field
Applies if: C0MP_D2 = 1
1 - 99999999 = Weight
Blank = Not applicable
36 WT4_DAY1 8 N 4-year day 1 sample weight. Read section
4.3.3 on sampling weights.
Applies if: C0MP_D1 = 1
1 - 99999999 = Weight
Blank = Not applicable
37 WT4_2DAY 8 N 4-year 2-day sample weight. Read section
4.3.3 on sampling weights.
Applies if: C0MP_D2 = 1
1 - 99999999 = Weight
Blank = Not applicable
3 8 YEAR 4 N Year of the survey.
1994 = 1994 sample
1995 = 1995 sample
1996 = 1996 sample
1998 = 1998 sample
3 9 D1_NREC 2 N Day 1: number of food records.
0 - 99 = Number
Attachment 1-50
-------
40 D2_NREC 2 N Day 2: number of food records.
Applies if: C0MP_D2 = 1
0 - 99 = Number
Blank = Not applicable
4.4.3 Formats for the supporting files
4.4.3.1 Food Code-to-commodity Translation File - (fc_comm.txt!
Field Field Width
Number Name /Type Description/Application/Values
1
FOODCODE
8
N
Food code.
2
MODCODE
6
N
Modification code.
3
COM_CODE
8
N
EPA commodity code.
4
COM_AMT
10
N3
Total amount in grams of the EPA
commodity in 10 0 grams of the food
4.4.3.2 Food code descriptions (fcdesc.txt)
Field Field Width
Number Name /Type Description/Application/Values
1 FOODCODE 8 N Food code.
2 DESCR 200 A Description of food.
3 ABBRDESC 60 A Abbreviated description of food.
4.4.3.3 Food code include statements (fcincl.txt)
Field Field Width
Number Name /Type Description/Application/Values
1 FOODCODE 8 N Food code.
2 SEQUENCE 2 N Sequence number.
3 INCLUDE 8 0 A Include statement
4.4.3.4 Food Code outline (fcscheme.txt!
ASCII text file
Attachment 1-51
-------
4.4.3.5 Descriptions of recipe modifications (moddesc.txt!
Field Field Width
Number Name /Type Description/Application/Values
1 MODCODE 6 N Recipe modification code number.
2 MODDESC 240 A Recipe modification description
4.5 Miscellaneous Notes
4.5.1 Responding sample persons with no foods reported
for a day
There are sample persons who completed an individual intake
interview but reported consuming no foods or beverages for that
day. The "smpl9498.txt" fields providing the number of foods
reported for a day, D1_NREC and D2_NREC, will have a value of 0
in such cases. Such sample persons do not have records in the
commodity intake ("comm94 98.txt") nor the cooked status-food
form-cooking method commodity intake ("ffcm9498.txt") data files
for that day.
4.5.2 Unreported body weights
The amount of each commodity consumed is reported in grams per
kilogram body weight. The CSFII, 1994-96 reports each sample
persons body weight (WGT_SP) in pounds, which was converted into
kilograms. There are some sample persons who completed an
individual intake interview but failed to provide a body weight.
Where a body weight was not reported, the following default
weights were used:
Gender and Age
Body weight (Kg)
Children < 6 months of age
6 . 0
Males 6-11 months of age
9.4
1
year
11.8
2
years
13 . 6
3
years
15 . 7
4
years
17 . 8
5
years
19 . 8
6
years
23 . 0
7
years
25 .1
8
years
28.2
9
years
31 .1
10
years
36.4
11
years
40.3
12
years
44 . 2
Attachment 1-52
-------
13
years
49 . 9
14
years
57 .1
15
years
61. 0
16
years
67 .1
17
years
66 . 7
> 17
years
70 . 0
Females 6-11 months of age 8.8 Kg
1 year 10.8
1 year 10.8
2 years 13.0
3 years 14 . 9
4 years 17.0
5 years 19.6
6 years 22.1
7 years 24.7
8 years 2 7.9
9 years 31.9
10 years 36.1
11 years 41.8
12 years 46.4
13 years 50.9
14 years 54.8
15 years 55.1
16 years 58.1
17 years 59.6
> 17 years 60.0
Body weights used as default weights, in instances in which body
weights were not reported, were provided by EPA. The reference
weights for children less than 6 months of age are derived from
Table 7-1 of the Exposure Factors Handbook, U.S. Environmental
Protection Agency, National Center for Environmental Assessment,
Office of Research and Development, Washington, DC, (EPA/600/C-
99/001), February 1999. Table 7-1 is derived from: Hamill PTV,
TA Dried, CL Johnson, R. Reed, AF Riche, WE Moore. Physical
Growth: National Center for Health Statistics Percentiles. Am J
Cain Nut 32:607-629, 1979. The weights of children 6 months
through the 17th year are from Table 7-3, Body Weights of
Children, adapted from the National Center for Health Statistics
(NCHS), 1987: Anthropometric reference data and prevalence of
overweight, United States, 1976-1980. Data from the National
Health and Nutrition Examination Survey (NHANES II), Series 11,
No. 238, Hyattsville, MD: U.S. Department of Health and Human
Services, Public Health Service, NCHS. DHHS Publication No.
(PHS) 87-1688. The weights provided for those 18 year of age and
older are those traditionally used in risk assessments by EPA's
Office of Pesticide Programs, Health Effects Division for adult
males and females.
4.5.3 Lower limit for reporting intakes
The commodity intake data files ("comm9498.txt" and
Attachment 1-53
-------
"ffcm9498.txt") contain the amount of each EPA commodity consumed
by each sample person providing at least one day of dietary
intake in the CSFII 1994-96, 1998. Commodities not consumed by a
sample person are not included in the database. The smallest
amount that could be included in these intake files is 0.000001
grams per kilogram body weight. When an amount of a commodity
consumed from a food was less than this amount, this lower limit
amount of 0.000001 grams per kilogram body weight was assigned,
as requested by EPA.
4.5.4 Using this database in conjunction with the Continuing
Survey of Food Intakes by Individuals and the Supplemental
Children's Survey (CSFII 1994-96, 1998)
The Food Commodity Intake Database was created by translating the
CSFII 1994-96, 1998 food and beverage intakes into EPA
commodities. The household identification number (HHID) and
sample person number (SPNUM)uniquely identify each sample person
and provide linkage with the survey data. These identifiers,
along with the Day code (DAYCODE), provide linkage to the
survey's daily intake records.
Attachment 1-54
-------
5. Control statistics for the SMPL9498.txt data file,
all records, unweighted
This file contains descriptive statistics for variables in the
SMPL9498.txt file. These statistics are not population
estimates. They are provided for checking purposes. The SAS
MEANS procedure was used to generate this listing which includes
a count of records with non-missing values (N) for each variable,
the unweighted mean of all values, and the minimum, maximum, and
sum of each variable. They are unweighted and were computed
using all values of each variable including values such as '998'
indicating a "don't know" response.
Variable
Mean
Label
N
HHID
Household ID
21662
:6262 . 0
SPNUM
Sample
person
number
21662
1 . 8
REGION
Region
21662
2 . 6
URB
Urbanization
21662
1 . 9
HHSIZE
Household size
21662
3 . 8
INCOME
Annual
income
: total
21662
18624 . 9
INCREP
Annual
income
: actual report
21662
2 .5
IMPFLAG
Annual
income
: imputation flag
21662
1.4
PCTPOV
Annual
income
: percent of poverty
21662
205 . 0
AGE
Age in
years
21662
25 .4
AGE M
Age in
months
1551
5.5
SEX
Sex
21662
1.5
WGT SP
Weight
of SP
21662
138 . 9
WGT KG
Weight
of SP
in Kilograms
21662
47 . 2
REL REF
Relationship
to reference person
21662
1 . 6
RACE
Race
21662
1.5
Attachment 1-55
-------
ORIGIN
4 . 6
PL_STAT
4 . 8
BF_STAT
2 . 7
H20_C00K
2 .4
H20_BEVR
3 . 0
H20_DRNK
3 .3
D1_H20_0
29.3
D1_H20_H
1 . 8
D1_H20_A
1 . 6
D2_H20_0
29.5
D2_H20_H
1 . 9
D2_H20_A
1 . 8
COMP_Dl
1 . 0
COMP_D2
1 . 0
WT3_DAY1
16263 . 9
WT3_2DAY
17114 .1
WTA_DAY1
38123 .3
WTA_2 DAY
40075.0
WT4_DAY1
12090 .2
WT4_2DAY
12709.1
YEAR
1995 . 8
D1_NREC
14 .4
D2_NREC
14 . 0
Hispanic origin
Pregnant/lactating status
Breastfeeding status
Source of water: cooking
Source of water: beverages
Source of water: drinking
Day 1: amount of water
Day 1: water from home
Day 1: away from home water
Day 2: amount of water
Day 2: water from home
Day 2: away from home water
Day 1 flag
Day 2 flag
Final 3-year day 1 weight
Final 3-year two day weight
Final annual day 1 weight
Final annual two day weight
Final 4-year day 1 weight
Final 4-year two day weight
Year of survey
Day 1: number of food records
Day 2: number of food records
21662
21662
21662
21662
21662
21662
21662
17359
6513
20607
16260
5906
21662
21662
16103
15303
21662
20607
21662
20607
21662
21662
20607
Variable Label
Maximum
Minimum
Attachment 1-56
-------
0
0
0
0
0
0
0
0
0
0
0
0
0
8
0
0
0
0
0
0
0
0
0
0
0
Household ID
Sample person number
Region
Urbanization
Household size
Annual income: total
Annual income: actual report
Annual income: imputation flag
Annual income: percent of poverty
Age in years
Age in months
Sex
Weight of SP
Weight of SP in Kilograms
Relationship to reference person
Race
Hispanic origin
Pregnant/lactating status
Breastfeeding status
Source of water: cooking
Source of water: beverages
Source of water: drinking
Day 1: amount of water
Day 1: water from home
Day 1: away from home water
Attachment 1-57
-------
0
0
0
0
0
0
0
0
0
0
0
0
0
0
Day 2: amount of water
Day 2: water from home
Day 2: away from home water
Day 1 flag
Day 2 flag
Final 3-year day 1 weight
Final 3-year two day weight
Final annual day 1 weight
Final annual two day weight
Final 4-year day 1 weight
Final 4-year two day weight
Year of survey
Day 1: number of food records
Day 2: number of food records
Attachment 1-58
-------
Variable Label
Sum
HHID
568887640.0
SPNUM
38281.0
REGION
57357 . 0
URB
41798.0
HHSIZE
81549.0
INCOME
836693653.0
INCREP
55063.0
IMPFLAG
30647.0
PCTPOV
4441342.0
AGE
550082.0
AGE_M
8598.0
SEX
32337.0
WGT_SP
3008008.0
WGT_KG
1021680.0
REL_REF
35644.0
RACE
33251.0
ORIGIN
100481.0
PL_STAT
104004.0
BF_STAT
58905.0
H20_C00K
51221.0
H20_BEVR
65725.0
H20_DRNK
71879 . 0
D1_H20_0
634554.0
Household ID
Sample person number
Region
Urbanization
Household size
Annual income: total
Annual income: actual report
Annual income: imputation flag
Annual income: percent of poverty
Age in years
Age in months
Sex
Weight of SP
Weight of SP in Kilograms
Relationship to reference person
Race
Hispanic origin
Pregnant/lactating status
Breastfeeding status
Source of water: cooking
Source of water: beverages
Source of water: drinking
Day 1: amount of water
Attachment 1-59
-------
D1_H20_H
31896.0
D1_H20_A
10513.0
D2_H20_0
607609.0
D2_H20_H
30141.0
D2_H20_A
10739.0
C0MP_D1
21662.0
COMP_D2
22717.0
WT3_DAY1
261897277.0
WT3_2DAY
261897260.0
WTA_DAY1
825826029.0
WTA_2 DAY
825825998.0
WT4_DAY1
261897244.0
WT4_2DAY
261897236.0
YEAR
43231966.0
D1_NREC
311153.0
D2_NREC
287676.0
Day 1: water from home
Day 1: away from home water
Day 2: amount of water
Day 2: water from home
Day 2: away from home water
Day 1 flag
Day 2 flag
Final 3-year day 1 weight
Final 3-year two day weight
Final annual day 1 weight
Final annual two day weight
Final 4-year day 1 weight
Final 4-year two day weight
Year of survey
Day 1: number of food records
Day 2: number of food records
Attachment 1-60
-------
Commodity Intake Database
Appendix A
EPA's Use of Food Consumption Data in Assessing Dietary Risk from
Pesticides (Provided by EPA)
EPA is responsible for regulating the nature and amount of
pesticide residues in food under the Federal Food, Drug and
Cosmetic Act which authorizes EPA to set a legal and enforceable
tolerance, or an exemption from the requirement of a tolerance.
EPA's Office of Pesticide Programs regulates pesticides to ensure
that their use does not pose unreasonable risks to human health
and that pesticide residues found in food are safe. Assessing
the amount of pesticide in or on the foods that we eat - both for
fresh, raw foods such as lettuce or processed foods such as
frozen french fries - is a complex process that requires data
from numerous sources along with an understanding of risk
analysis and risk management methods.
The dietary risk posed by a pesticide in food can be expressed
as a function of exposure and toxicity. The dietary exposure is
derived from the amount of pesticide residue that is present in
and on food (the residue) and the types and amounts of food in a
person's diet (i.e., food consumption). Toxicity is expressed as
a reference dose to which a person can be safely exposed over
time.
Dietary exposure to pesticides in foods is estimated by
considering pesticide residues in foods and the amount of food
consumed. In an attempt to conserve limited resources, EPA
assesses dietary exposure in a tiered approach proceeding from
conservative to more refined assumptions as the risk management
situation dictates. Dietary exposure estimates based on
tolerance level residues (farm-gate residues) reflect a
Theoretical Maximum Residue Contribution (TMRC) which
overestimate actual dietary exposure. To better estimate dietary
exposure, EPA developed a process by which pesticide tolerance
data (40 CFR 158.240) and compliance monitoring data are refined
to reflect pesticide residues in food as consumed (dinner-plate
residues). The best estimate of pesticide residues in food, as
consumed, is termed the Anticipated Residue (AR) estimate. When
estimating ARs, EPA uses all available data. It should be noted
that since data sets vary in quality, considerable scientific
judgement is required to derive anticipated residue estimates.
For many years, EPA has used food consumption data collected by
USDA through various large, nationally representative surveys for
its dietary risk assessments. These surveys have sampled
thousands of households to learn about what, and how much, people
eat. The Commodity Intake Database is a result of cooperative
work by EPA and USDA in taking information on the foods people
Attachment 1-61
-------
reported eating in the 1994-96 Continuing Survey of Food Intakes
by Individuals (CSFII) and the Supplemental Children's Survey
(SCS) and converting "foods" to the "agricultural commodity"
terms and quantities used in EPA regulation of pesticide residues
in foods. "Agricultural commodity" is a term used by EPA to mean
plant or animal parts consumed by humans as food; when such items
are raw or unprocessed, they are referred to as "raw agricultural
commodities." Many food items contain more than one commodity,
as, for example, an apple pie may contain the commodities apples,
flour, fat, sugar and spices. (See also Section 3 of the
Documentation of the Commodity Intake Database for more
information on the methods used to translate foods to
agricultural commodities).
Finally, toxicity of a pesticide is brought into the risk
assessment equation. EPA toxicologists review study data on
possible toxicity associated with exposure to a pesticide (often
animal studies) and establish reference doses that are estimates
of the level of either one-day exposure (acute) or daily exposure
over a life-span (chronic) that are believed to have no
significant harmful effects on humans. Safety factors are
incorporated into these reference doses to account for the
differences in response that may occur when animal data are used
to estimate possible human response, and also to account for the
variation of response to the pesticide that may occur within the
human population.
Having information on pesticide residues, the number of various
commodities treated with a given pesticide, the quantity of the
commodities consumed by the population in the United States, in
conjunction with toxicity data expressed in a reference dose,
allows EPA to complete dietary risk assessments used to
determine that pesticide residues found in food are within safe
limits.
Attachment 1-62
-------
Commodity Intake Database
Appendix B
3 METHODS IN THE CONTINUING SURVEY OF FOOD INTAKES BY INDIVIDUALS
1994-96, 1998
The methods used in the Supplemental Children's Survey to the 1994-96
Continuing Survey of Food Intakes by Individuals (CSFII 1998) were
identical to those used in the CSFII 1994-96. The CSFII 1998 sample
design shared most basic features with the CSFII 1994-96, but differed
in a few respects outlined in documentation section 3.1.2, "CSFII 1998
sample design."
3.1 Sample Design
3.1.1 CSFII/DHKS 1994-96 sample design
The primary goal of the sample design for the CSFII/DHKS 1994-96 was to
obtain nationally representative samples of noninstitutionalized persons
residing in households in the United States for each of 40 analytic
domains defined by sex, age (10 age groups), and income level (a
"low-income" group and an "all-income" group) that were aimed to meet
specified precision levels for estimates of mean day-1 saturated fat and
iron intakes. Excluded were persons who lived in group quarters or
institutions, who resided on military installations, or who were
homeless. The specific precision goals were that the coefficients of
variation (CVs) for mean saturated fat and iron intakes should be 3
percent or less for each of the 20 all-income sex-age domains and 5
percent or less for each of the 20 low-income sex-age domains. These
precision goals were translated into 3-year target sample sizes. In
addition, the sample design specified that one day-1 intake respondent
20 years of age or older be selected for the DHKS from each household
with at least one day-1 intake respondent age 20 or over. For the
CSFII/DHKS 1994-96, a single sample was selected that met precision
requirements by income level, in contrast to past CSFII/DHKS surveys
where a separate sample of low-income persons was also chosen in
addition to the basic general sample.
The sample selection process was designed by Westat, Inc., a private
research firm in Rockville, MD, under contract to ARS. The sample for
the CSFII/DHKS 1994-96 was derived from a Westat, Inc., master sample.
This master sample, which was in existence prior to the award of the
contract for the CSFII/DHKS 1994-96, is a stratified, multistage area
Attachment 1-63
-------
probability sample. The sampling frame was organized using estimates of
the U.S. population in 1990 (USDC/BOC 1993). The stratification plan
took into account geographic location, degree of urbanization, and
socioeconomic characteristics.
At the first stage of sampling, the entire United States was divided
into primary sampling units (PSU's) consisting of Metropolitan
Statistical Areas (MSA's) (see section 3.6, "Glossary," below),
counties, or groups of counties. Because of its size, the New York MSA
was divided into three PSU's. For the same reason, the Los Angeles and
Chicago MSA's were each divided into two PSU's. Apart from these, each
of the other MSA's constituted a single PSU. Some counties outside MSA's
were grouped to form PSU's containing at least 15,000 people. A total of
1,404 PSU's was created, and 62 PSU's were selected for use in the
CSFII/DHKS 1994-96, as described below.
The 24 PSU's with the largest populations were included with certainty.
The remaining (noncertainty) PSU's were then assigned to 1 of 38 strata
of approximately equal size (in terms of 1990 population), and one PSU
was selected from each stratum with probability proportional to the 1990
population. Stratification factors included region of the country (four
census regions) (see section 3.6, "Glossary," below); whether or not the
PSU was an MSA and the population size of the MSA; percentage of the
population that was black or Hispanic; and per capita income. Among the
noncertainty strata, 26 were MSA strata and 12 were non-MSA strata.
The second stage was the selection from each PSU of 36 area segments
consisting of blocks or groups of blocks. Area segments were chosen with
probability proportional to size. The CSFII/DHKS 1994-96 was designed so
that data collection would be spread evenly over the 3 years of the
survey and over the quarters of the year. From each sampled PSU, twelve
segments were subsampled for each of the 3 years of the survey, three
segments for each quarter of the year. Addresses of all dwelling units
in the subsampled area segments were then listed in accordance with 1990
Census listing rules and consistent with the 1990 Census definition of a
housing unit (see section 3.6, "Glossary," entry for "Dwelling unit").
In the third stage, listed dwelling units in the selected area segments
were drawn into the sample from the listings. For the three years of the
CSFII/DHKS 1994-96, a sample of 34,016 dwelling units in all was
designated for screening. Calculation of the number of dwelling units to
be screened took into account the sample sizes needed to achieve the
desired levels of precision specified by ARS prior to contract award,
the percentages of individuals in each sex-age group living in
households at or below 130 percent of the Federal poverty guidelines
(DHHS 1996), a projected figure for vacant dwelling units, and a safety
factor allowing for random sampling variation. Sample households were
screened to identify appropriate numbers of sample persons in specified
sex-age groups.
The last sampling stage involved selection of individuals from the
sampled households. As described in the first paragraph of this section
(section 3.1.1), the CSFII 1994-96 was designed to obtain sample sizes
for the sex-age groups that would produce estimates with equivalent
coefficients of variation over the sex-age groups, both for the total
Attachment 1-64
-------
population and for the low-income population. To obtain the desired
numbers of individuals, sex-age subgroups were sampled at different
rates. This procedure was implemented at the screening stage of the
survey. The age groups used were 1 to 2 years, 3 to 5 years, 6 to 11
years, 12 to 19 years, 20 to 29 years, 30 to 39 years, 40 to 49 years,
50 to 59 years, 60 to 69 years, and 70 years and over. The approach used
to select persons for the intake interviews was to designate subsets of
households within which persons meeting specified sex-age/income
criteria would be included in the study. For example, for a
predesignated subset of households in the dwelling unit sample, only
children between the ages of 1 and 2 years and low-income males between
the ages of 50 and 59 years were to be included in the sample. Sampled
households were assigned to the various subsets in a random fashion to
ensure the unbiased selection of sample persons for the study. In
addition, all infants under 1 year of age in households that contained
at least one sample person 1 year or older were included in the sample.
To facilitate the selection of sample persons in the field, each
screening questionnaire carried a sampling message specifying the
characteristics of the persons to be included in the sample. The
proportion of households receiving a particular message was determined
to satisfy the target sampling rates for the various sex-age/income
domains. After completing the listing of household members, the
interviewer identified which, if any, of the household members fell into
the sex and age groups that had been predetermined for that household.
The interviewer had no discretion as to whom to include. In the CSFII
1994-96, a total of 20,126 individuals was initially selected into the
sample.
Respondents for the DHKS 1994-96 were selected from among sample persons
20 years of age and over who had completed the day-1 intake interview in
the CSFII 1994-96. Only one DHKS respondent per household was selected
in households with eligible participants. In households with more than
one CSFII participant 20 years of age or over, one of the participants
was selected randomly with probability assigned to maintain
distributions of all-income and low-income individuals in the six
sex-age groups age 20 years and over in the DHKS that conformed
approximately to the corresponding distributions of individuals in the
CSFII. In the DHKS 1994-96, a total of 7,842 individuals was selected
into the sample.
For more detailed information on the CSFII/DHKS 1994-96 sample design,
see Tippett and Cypel (eds.) 1997, which is included on Disk 1 in
\pdffiles\dor.pdf.
3.1.2 CSFII 1998 sample design
The CSFII 1998 had its roots in the Food Quality Protection Act of 1996,
which required the Secretary of Agriculture to provide the Environmental
Protection Agency (EPA) with information on food consumption patterns of
a statistically valid sample of infants and children. This requirement
followed a report entitled Pesticides in the Diets of Infants and
Children (NAS/NRC 1993) that concluded that current food consumption
data for children did not provide sufficient sample sizes for adequate
estimation of dietary exposure to pesticide residues. In response to the
Attachment 1-65
-------
1996 mandate, the Agricultural Research Service (ARS) of the U.S.
Department of Agriculture (USDA) conducted the CSFII 1998 as a
supplement to the CSFII/DHKS 1994-96. CSFII 1998 data used in
conjunction with CSFII/DHKS 1994-96 data, with appropriate weights (see
documentation section 5, "Sampling Weights"), meet the requirement for a
larger sample of children.
The goal of the sample design for the CSFII 1998 was to obtain
nationally representative samples of noninstitutionalized persons 9
years of age or younger residing in households in the United States for
each of 28 analytic domains defined by sex, age (7 age groups), and
income level (a "low-income" group and an "all-income" group). The age
groups used were under 1 year, 1 year, 2 years, 3 years, 4 years, 5 to 6
years, and 7 to 9 years.
A complex multistage area probability sample design that incorporated
the same primary and second stage sampling units developed for the
CSFII/DHKS 1994-96 was used to select children for the CSFII 1998. The
same 62 PSU's that were selected for the CSFII/DHKS 1994-96 were used
for the CSFII 1998. The PSU's were selected with probabilities
proportional to the 1990 population. From each PSU, the 24 area segments
used in the last 2 years of the CSFII/DHKS 1994-96 were used for the
CSFII 1998. Those 24 segments were selected because they were the
segments with the most up-to-date listing information.
Dwelling units (DU's) were selected from the area segments using listing
information from the CSFII 1994-96 along with quality control procedures
referred to as the "missed structure" and "missed dwelling unit"
procedures. In preparation for the CSFII/DHKS 1994-96, interviewers had
listed over 210,000 DU's within the 1,488 area segments included in the
CSFII 1998. DU's that had been selected for the CSFII/DHKS 1994-96 were
excluded from the CSFII 1998 sample. A sample of 65,519 DU's (i.e., an
average of 44 DU's per sample segment) was drawn for the CSFII 1998 from
the existing area segment listings. An additional 2,905 DU's were added
to the sample through quality control procedures referred to as the
"missed structure" and "missed DU" procedures. Thus, 68,424 DU's were
selected for the CSFII 1998.
Each sampled dwelling unit was screened to determine whether it
contained children who were eligible for the survey. From the DU's with
children 9 years of age or younger, a sample of eligible children was
selected by a probability sampling process designed to achieve the
target sample sizes. Finally, to increase the number of 3-year-old girls
in the sample, a special "supplemental" sample was selected and fielded
in the fourth quarter of the study. The sampling procedures described
above resulted in the initial selection into the sample of 6,413
children (including 2,100 low-income children).
For more detailed information on the CSFII 1998 sample design, see
"Sample Design -- Supplemental Children's Survey to the 1994-96
Continuing Survey of Food Intakes by Individuals (CSFII 1998)", which is
included on Disk 1 in \pdffiles\98_samp.pdf.
3.2 Data collection
Attachment 1-66
-------
3.2.1 CSFII/DHKS 1994-96 and CSFII 1998
The CSFII 1998 methods were nearly identical to those used in the CSFII
1994-96. A few questions were omitted because they were considered
inappropriate for interviews with or about children, for example,
questions about alcoholic beverages and smoking. See indented sections
below regarding differences between CSFII 1998 and CSFII 1994-96.
Data were collected by Westat, Inc. Prior to data collection, listers
visited every sample address in person to determine by visual inspection
whether that location represented a dwelling unit (see section 3.6,
"Glossary," below). An introductory letter and a brochure describing the
survey were mailed to each dwelling unit 1 week before the initial
in-person contact by the interviewer. In all materials for respondents,
the survey was referred to as the "What We Eat in America" survey rather
than by the official survey name. To contact individuals in the dwelling
units, interviewers made at least four visits before referring the case
to a supervisor. In a number of difficult cases, contact attempts
exceeded the level of effort required by the contract in order to
complete the interview. In cases where a dwelling unit was determined to
contain a household but the household could not be contacted after four
visits, interviewers were instructed to ask two neighbors for
information on the number of household members and their sexes and ages
as well as on the time household members were most likely to be home.
At each dwelling unit, the interviewer attempted a screening interview
to determine whether any members of the household were eligible to
participate in the survey. Any household member 18 years of age or older
was an acceptable respondent for the screening questionnaire (screener).
However, it was recommended that interviewers attempt to conduct this
portion of the survey with either the main meal planner/preparer (see
section 3.6, "Glossary," below) or a person knowledgeable about
household characteristics such as income because those persons were the
preferred respondents for the household questionnaire, which typically
followed the screener. It was not necessary for the respondent(s)
completing the screener and/or household questionnaire to be sample
persons (see section 3.6, "Glossary," below). If a household member (see
section 3.6, "Glossary," entry for "Household") refused to complete the
screener, the interviewer was instructed to ask the household member for
information on the number of household members and their sexes and ages
so that the number of eligible respondents could be determined. (The
number of eligible respondents was important for calculating the
response rates provided in documentation section 4, "Response Results.")
At the beginning of the screening interview, the interviewer reminded
the respondent about the letter and brochure that had been sent and
provided new ones if the respondent did not remember. During the
interview, information was collected on the number of persons living in
the household; the first name of the person or one of the persons who
owned or rented the home (reference person); the first name of the
reference person's spouse, if any; and the first name, race, ethnicity
(Hispanic or non-Hispanic), date of birth, age, sex, and relationship to
the reference person of any other people living in the household,
including friends, relatives, roomers, boarders, employees, and
Attachment 1-67
-------
household members who were away from home at the time of the interview
but who usually lived there.
One screener question asked whether the total income of all household
members from all sources during the previous year was more or less than
an amount specific to the household's size. That question was part of
the strategy for meeting the low-income sample size goals discussed in
documentation section 3.1, "Sample Design."
In the CSFII 1994-96, the screener income question was asked only when
the household included individuals in sex and age groups specified in
the sampling message for that dwelling unit. In the CSFII 1998, the
question was asked in all households.
The maximum income level used, where necessary, during the screening
process to determine the household's eligibility for inclusion in the
low-income group corresponded to 13 0 percent of the Federal poverty
guidelines (DHHS 1998), which are based on household size and income.
This income level was selected because it is the same as one of the
income criteria used to determine whether nonelderly households are
eligible to participate in the Food Stamp Program. Not all households
meeting the income criteria are eligible for food stamps; other
criteria, such as asset limitations, must also be met. The CSFII 1994-96
and CSFII 1998 screened households for income level only, not for food
stamp eligibility.
At households where one or more sample persons were selected, the
interviewer administered the household questionnaire--a series of
questions about the educational level and employment status of household
members 15 years of age and older, household income, food assistance
program participation, food expenditures, and some other food-related
practices. During the household interview, the interviewer asked the
respondent to identify the "female head of household" and the "male head
of household"; this question was included for the benefit of researchers
who wish to make historical comparisons involving those variables.
Interviewers made up to three visits after screening to complete the
household questionnaire before referring the case to a supervisor.
Interviewers' visits were scheduled in a manner designed to ensure that
at least 10 percent of day-1 food intake interviews took place on each
day of the week. A label specified 3 days of the week that would be
acceptable for collecting day-1 food intake information from that
attached to the survey materials for each household. Repeated in-person
visits were made as necessary to attempt to complete day-1 intakes with
sample persons on the scheduled days of the week. In some cases, when
repeated visits had been made on different scheduled days and at
different times, interviewers were permitted to change the day of the
week in order to obtain an interview. In households with more than one
sample person, if one of the sample persons was not at home when the
interviewer visited, the protocol required the interviewer to make up to
three additional visits in an attempt to obtain a day-1 intake for that
sample person. Often the number of visits required by the contract was
exceeded in order to obtain the interview. An extensive range of
strategies was employed in order to convert refusals, sometimes
involving efforts by two or more interviewers.
Attachment 1-68
-------
Day-1 intakes were to be collected in person. Before conducting the
day-1 interview, the interviewer told the sample person that her or his
participation would involve two in-person interviews (and possibly, for
one sample person in the household, the DHKS interview by telephone). At
the conclusion of the day-1 interview, the interviewer notified the
sample person that she or he would be returning in a few days to conduct
another interview.
According to the survey protocol, the day-2 interview was to be
conducted 3 to 10 days after the day-1 interview but not on the same day
of the week. In the CSFII 1994-96, less than 1 percent of day-2
interviews were conducted sooner than 3 days after the day-1 interview,
20 percent were conducted more than 10 days after the day-1 interview,
and 1 percent were conducted on the same day of the week as the day 1
intake exactly 1 week later. In the CSFII 1998, less than 1 percent of
day-2 interviews were conducted sooner than 3 days after the day-1
interview, 17 percent were conducted more than 10 days after the day-1
interview, and 2 percent were conducted on the same day of the week as
the day 1 intake exactly 1 week later. Five percent of day-2 interviews
in the CSFII 1994-96 and 16 percent in the CSFII 1998 were conducted by
telephone, with supervisory permission. Sample persons interviewed by
telephone were asked to report food quantities using the measuring
guides that had been used in the day-1 interview (described below) and
given to the household.
The day-1 and day-2 questionnaires were very similar. Both included a
1-day dietary recall using a multiple-pass method in order to maximize
the sample person's ability to remember what she or he ate and drank
[Tippett and Cypel (eds.) 1997, DeMaio et al. 1993, Guenther et al.
1995] .
The 1-day recall began with the sample person being asked to report
everything eaten or drunk the previous day between midnight and
midnight. The interviewer did not interrupt the sample person during
this initial listing of the day's intake. The sample person was invited
to add any other items remembered as the interview progressed. Then, for
each food and drink listed, the interviewer asked the name of the eating
occasion and the time it began.
For the CSFII 1998, the introduction was revised to delete references to
coffee and alcoholic beverages, and the category "alcoholic beverage
break" was deleted from the card the interviewer handed the respondent
as an aid in naming the eating occasion. These changes were made to both
day-1 and day-2 questionnaires.
The interviewer used a Food Instruction Booklet (FIB) to probe for a
complete description of every food item and the amount eaten. Under each
appropriate category of food/drink listed in the FIB, there was a list
of the questions (probes) the interviewer was required to ask in order
to collect enough detail for the food to be coded. Probes varied with
the type of food or beverage being recalled. Some examples of FIB probes
are "What was the brand name?" and "Were they regular, reduced calorie,
high fiber, or something else?" When appropriate, questions were asked
about the use of salt ("Was salt used in cooking or preparing the
[food]?") and fat ("Was any kind of fat or oil used in cooking or
Attachment 1-69
-------
preparing the [food]?") in food preparation and about additions ("Did
you add anything to the [food]?"). The interviewer was directed to ask
for ingredients in some categories (for example, soups; tacos, burritos,
enchiladas, and fajitas; sandwiches; salads; and mixed dishes,
casseroles, and stews). Interviewers were required to use the FIB to
obtain a detailed description of every food item recalled by the sample
person, including additions remembered as the result of questions asked
in describing another food. The FIB also suggested the types of measures
(weight, volume, or size) appropriate for the food.
For the CSFII 1998, the FIB was refined to reflect some changes in food
products since 1996, as well as changes in food terminology. For
example, food label regulation changes for milk that went into effect in
January 1998 narrowed the use of the term "lowfat" from 1-percent or 2-
percent milk to only 1-percent. The regulations also introduced the term
"reduced-fat" for 2-percent milk. As a consequence, the term "low-fat"
for milk was deleted from FIB probes, and respondents were asked to
specify the percent fat in the milk they used.
Measuring guides used to aid the sample person in estimating amounts
were household measuring cups (1/4 cup, 1/3 cup, % cup, and 1 cup) and
spoons (1/4 teaspoon, % teaspoon, 1 teaspoon, and 1 tablespoon); a
12-inch ruler with 1/8-inch increments marked; "thickness sticks," a set
of 8 small rectangular pieces of hard plastic, each 1/8 inch in
thickness; a laminated card printed with concentric circles 1 inch to 6
inches in diameter, two perpendicular 6-inch rulers, pictures of a fish
filet and chicken parts, and diagrams specifying the dimensions to be
measured or estimated when describing and quantifying various shapes.
The cups and spoons could also be used to measure the capacity of
tableware. One additional measuring guide, a 2-cup measuring cup, could
be used only when the sample person referred to a bowl or cup in her or
his home. The sample person could then fill the bowl or cup with water
to represent the amount eaten or drunk, and the interviewer could
measure the volume of water by pouring it into the 2-cup measure.
After each item on the initial list of the day's intake had been
described and quantified, the interviewer reviewed for the sample person
all the foods listed for each eating occasion and probed for additional
foods eaten before the first eating occasion listed, in between listed
occasions, and after the last occasion listed. Then, for each food or
drink reported, the interviewer asked where it was obtained and whether
it was eaten at home or not. For foods eaten away from home, the sample
person was also asked whether the food or drink had ever been in the
home before it was eaten; this question was included for the benefit of
researchers choosing to make historical comparisons involving the
variable "food from the home supply."
Additional questions asked on both day 1 and day 2 pertained to whether
the sample person's intake on the previous day had been usual or unusual
and why, how much plain drinking water the sample person drank on the
previous day and whether it came from home or another source, and how
many hours of television or videos the sample person watched on the
previous day. Further questions in the day-1 questionnaire included the
type of salt usually used by the sample person and frequency of use at
the table; whether the sample person was on a diet and, if so, the type
Attachment 1-70
-------
and source of the diet; whether the sample person considered herself or
himself to be vegetarian; frequency of vitamin or mineral supplement use
and type of supplement; use of fish oil and fiber supplements; whether
the sample person ever had a blood cholesterol check; self-reported
height and weight (without shoes); self-assessed health status; food
allergies; physician-diagnosed medical conditions; frequency of vigorous
exercise; cigarette smoking status and number of cigarettes smoked per
day; and consumption (ever or never) of alcoholic beverages during the
past 12 months. The day-2 interview contained an additional question on
the consumption (ever or never) of 28 foods during the past 12 months.
For the CSFII 1998, questions on exercise, smoking, and consumption of
alcoholic beverages were removed from the questionnaires.
Proxy interviews were conducted routinely for child sample persons under
6 years of age and any other sample persons (including adults) who could
not report for themselves due to physical or mental limitations; proxy
interviews were not permitted for any other reason. Proxy interviews
were not considered to be an acceptable substitute for an in-person
interview with adult sample persons who were difficult for the
interviewer to reach or who were nonrespondents. Child sample persons 6
to 11 years of age (6 to 9 years of age in CSFII 1998) were asked to
provide their own food intake data assisted by an adult household member
(referred to as the assistant). The preferred proxy or assistant was the
person responsible for preparing the sample person's meals. If the
sample person, proxy, or assistant could not provide enough descriptive
or quantitative information about the foods eaten, it was sometimes
necessary to seek that information from another caregiver such as a
babysitter or school cafeteria personnel. It was permissible for any
number of caregivers to contribute intake data for a sample person.
The first use of Spanish-language questionnaires in the CSFII and DHKS
was in 1994-96. Interviewers who were bilingual in English and Spanish
were provided with questionnaires and survey materials translated into
standard Spanish and received an extra day of training in their use. The
Spanish questionnaires reduced the number of language barrier cases and
provided a standardized translation of the questionnaire content. They
also minimized the need for interpreters, a practice that raises
concerns about consistency of interpretation and interview length. If a
sample person spoke neither English nor Spanish, a family member or
neighbor 16 years of age or older was permitted to serve as an
interpreter.
Spanish questionnaires were used in 2.8 percent of CSFII 1994-96
interviews (excluding screeners) and 4.4 percent of CSFII 1998
interviews.
The CSFII 1994-96 and CSFII 1998 used in-kind incentives. The
interviewer told the screener respondent that each participating
household would receive a gift. A set of measuring cups and spoons was
given to the screener respondent after the screener was completed and
the household was found to contain any sample person(s). An insulated
nylon sack was given to each sample person prior to the collection of
the intake, and at the conclusion of the day-2 interview each responding
Attachment 1-71
-------
sample person received a travel-type beverage mug as a thank-you gift
for participating.
Average questionnaire administration time in the CSFII 1994-96 was about
7 minutes for the screener, 19 minutes for the household questionnaire,
32 minutes for the day-1 intake, and 29 minutes for the day-2 intake.
Average questionnaire administration time in the CSFII 1998 was about 7
minutes for the screener, 20 minutes for the household questionnaire, 32
minutes for the day-1 intake, and 3 0 minutes for the day-2 intake.
3.2.2 Diet and Health Knowledge Survey 1994-96
The DHKS was conducted only with respondents 20 years of age and older
and so was not part of the CSFII 1998. This section is included because
DHKS 1994-96 data are included in this release.
The Diet and Health Knowledge Survey was conducted as a telephone
follow-up to the CSFII 1994-96. According to survey design, telephone
contact was to be initiated 2 to 3 weeks after the day-2 intake. For
households without telephones or with unlisted numbers not provided to
interviewers, in-person interviews were the designated mode of contact.
When all sample persons in a household either had completed a day-1
intake or had been judged to be day-1 nonrespondents, the DHKS
respondent was randomly selected by a computerized process from among
eligible CSFII sample persons 20 years of age and over who had provided
a day-1 intake. Sample persons were not eligible if their intake(s) had
been completed by proxy, nor were any proxies allowed to complete the
DHKS. Due to these criteria, not all households had a DHKS respondent.
The interviewer scheduled an appointment for the telephone interview
when the selected DHKS respondent had completed a day-2 intake. The same
interviewer who administered the CSFII typically administered the DHKS.
This continuity of interviewers maintained any rapport established
between interviewer and respondent and was expected to have a beneficial
effect on the response rate. Interviewers operating out of their own
homes administered the questionnaire from a hard copy without computer
assistance.
The interviewer mailed a DHKS reminder card 3 to 5 days prior to the
scheduled interview. In addition to the appointment date and time, this
card contained a list of response categories for selected questions in
the DHKS questionnaire. During the interview, the respondent was
directed to look at the set of response categories applicable to a
particular question, thus reducing the need for the interviewer to
repeat the response options. The card served both as an appointment
reminder and as a means of improving the flow of the interview.
The first telephone contact was attempted on the scheduled day and time;
if this attempt was unsuccessful, additional calls were made as needed
at different times of the day and on different days of the week to reach
respondents. The survey protocol required at least six telephone
attempts at each number (as needed to obtain the interview), followed by
four in-person visits. In a number of difficult cases, contact attempts
exceeded the required level of effort in order to complete the
Attachment 1-72
-------
interview. Overall, the DHKS interview in 1994-96 took an average of 30
minutes to complete; it took longer to complete the DHKS in person (34
minutes, on average).
The telephone interview began with a request to speak to the person with
whom the appointment had been made. The interviewer identified herself
or himself and reminded the respondent that during the CSFII she or he
had been told she or he would be recontacted later by telephone to
answer a few more questions about food and nutrition issues. The DHKS
respondent's name and age were verified at this time.
The gift that was provided at the end of CSFII day 2 also served as an
incentive to complete the DHKS. Pretests and interviewer debriefings
suggested that interest in the questionnaire content was also a
motivating factor in completing the interview for some respondents.
Of all DHKS 1994-96 interviews, 84 percent were completed by telephone
and 16 percent in person. The primary reasons for conducting interviews
in person were that the household did not have a telephone or that
limitations were posed by respondents' physical conditions (e.g., hard
of hearing, feeble). Another reason was language barrier cases where an
interpreter was needed.
In 1994-96, 74 percent of DHKS interviews were completed between 2 and 3
weeks after the last CSFII interview, as contractually specified.
Interviews completed earlier than 2 weeks or later than 3 weeks were
considered mistimed. Four percent of cases were completed earlier than 2
weeks due to reasons such as prior knowledge of extended periods of
absence from the household (e.g., hospitalization, travel) and
interviewer error. In 22 percent of cases, the length of time between
the CSFII and the DHKS interviews was extended beyond 3 weeks because
numerous contacts were required to complete the interview. These
mistimings often centered on broken appointments where respondents were,
for example, too busy or not at home at the scheduled time. Refusal
conversion efforts also contributed to mistimings; some cases required
intensive, prolonged efforts on the part of two or more interviewers to
complete the interview.
In the DHKS 1994-96, a Spanish version of the questionnaire was
available for use by bilingual interviewers. It served to reduce the
number of language barrier cases and provided a standardized translation
of the questionnaire content. The Spanish questionnaire also minimized
the need for interpreters, a practice that raises concerns about
consistency of interpretation and interview length. In 1994-96, 147 DHKS
interviews (2.6 percent) were conducted using the Spanish questionnaire.
In 1994-96, there were 61 cases (1.1 percent of DHKS interviews) where
bilingual interviewers and telephones were not available or the
respondent spoke a foreign language other than Spanish, interpreters
were used. In these in-person interviews, the interpreters were required
to be 16 years of age or older.
The content of the DHKS 1994-96 questionnaire was governed by a need for
data on knowledge and attitudes about the Dietary Guidelines for
Americans (USDA/DHHS 1990), food labeling issues, and dietary behaviors
related to fat intake. Information from the DHKS can contribute to the
Attachment 1-73
-------
research base needed to develop food guidance materials and identify
strategies for targeting nutrition education efforts. Thus, the data
collected include self-perceptions of the adequacy of intake levels of
nutrients and other dietary components, awareness of diet-health
relationships, perceived importance of following dietary guidance for
specific nutrients and other dietary components, behaviors related to
fat intake and food safety, knowledge about food sources of fats and
cholesterol, and self-perceptions about weight status. Also asked in the
DHKS 1994-96 was a new series of questions regarding food labels. It
covered use of various sections of the food label, use of specific
information on the nutrient panel, frequency of using food labels when
buying specified categories of food, ease of understanding food label
information, and level of confidence in food label information.
3.3 Data Processing
3.3.1 Food coding and editing
The food intake data for the CSFII 1998 were coded and edited using
Survey Net, the same computer-assisted food coding and data management
system used with the CSFII 1994-96. Survey Net was developed
cooperatively by ARS and the University of Texas-Houston Health Science
Center's School of Public Health, and was tailored specifically to the
questions, quality control needs, and data processing needs of the CSFII
1994-96. A general-use version of the software, the Food Intake Analysis
System (FIAS), is available to researchers interested in using ARS
survey food coding and nutrient databases. [For FIAS program and price
information contact the University of Texas-Houston Health Science
Center, School of Public Health, P.O. Box 20186, Houston, Texas 77225.
Phone: (713) 500-9775. Fax: (713) 500-9329.]
Survey Net is a multilevel software system used by both the survey
contractor and ARS. It operates on a computer network with multiple
users accessing a set of central databases. These include (1) a food
coding database containing food descriptions and food measures with
their corresponding gram weights, (2) a predefined recipe database,
and(3) the Survey Nutrient Database. All three databases are available
with their documentation in the \TSF98 directory on Disk 2.
Westat's food coders used Survey Net to match descriptions of foods
eaten by sample persons to foods listed in the food coding database.
Coders entered partial or complete words or phrases from the sample
person's descriptions of foods to retrieve food codes containing the
same terms. Once a matching food description was found and selected,
Survey Net provided a list of common household measures (such as 1 cup
or 1 small piece) appropriate for that food. Coders selected the measure
corresponding to the sample person's description of the amount eaten.
When descriptions of foods or quantities not present in the food coding
database were encountered, they were entered as "unknowns" for ARS to
resolve later.
A recipe modification feature of Survey Net allowed coders to view the
predefined recipes which list ingredients and amounts for every food
code in the Food coding database, and to modify the recipes to match
Attachment 1-74
-------
more closely the foods eaten by sample persons. Recipes were modified
primarily by deleting or substituting ingredients. Modified recipes were
numbered for reference purposes and are included with the recipe
database on the CD-ROM. Recipe modification numbers appear in the field
MODCODE in record type 30 (rt30.dat).
There were three main purposes for recipe modifications: to record the
specific type of fat, the type of milk, and the dilutions of foods.
Recipes for foods such as vegetables, eggs, pasta, rice, and hot cereals
were modified to reflect the type of fat (such as oil, margarine,
margarine spreads, or butter) used in cooking. Recipes for foods such as
puddings, soups, and beverages were modified to reflect the type of milk
(such as whole, lowfat, 2-percent, 1-percent, or skim) used in their
preparation. Some foods commonly modified for both type of fat and type
of milk were scrambled eggs and omelets, and macaroni and cheese.
Recipes for foods such as soups, infant formulas, and beverages were
modified to reflect dilutions with amounts of milk or water that
differed from label directions. For example, the survey recipe for
orange juice was modified if one can of frozen concentrate was mixed
with four cans of water, instead of three cans of water.
Another aspect of the flexibility of food coding in the CSFII 1994-96
and CSFII 1998 is the use of combination codes, whose development and
auxiliary use in analyses are discussed in detail in documentation
section 3.3.8, "Combination codes." Combinations were often instances of
one food being added to another, such as margarine to toast or gravy to
potatoes. For some types of food made up of several components that are
relatively easy to describe and quantify separately (such as sandwiches
and salads) as well as for some mixed dishes, two or more food codes
linked together in a food combination present a more precise picture of
what was actually eaten by respondents than if a single food code is
used.
Each food in the combination was coded separately and assigned the same
combination type number (COMBTYPE) and sequence number (COMBNUM) in
record type 30 fields (rt30.dat) separate from the food code. There were
11 combination types: beverage, cereal, bread/baked product, salad,
sandwich, soup, frozen meal, ice cream/frozen yogurt, vegetable, fruit,
and other mixture. Two-digit sequence numbers (01 and so on) linked the
foods in a particular combination with each other and distinguished them
from foods in other combinations. For example, a sample person might
have cereal with milk in the morning and again in the afternoon. All the
components of these two combinations would be assigned the combination
type number for a cereal combination. The morning cereal with milk would
be assigned one sequence number, and the afternoon cereal with milk
would be assigned a different sequence number.
Survey Net's capabilities include a "copy foods" feature that allowed
entries from a particular eating occasion, day, or sample person to be
copied to a different eating occasion or day for the same person or to
the food intake of another sample person in the same household. Survey
Net also automatically performed gram weight checks of food quantities
entered against maximum and minimum values established by ARS for each
food. This weight check allowed coders to correct entry errors
immediately. Coders recorded any questions regarding their food and
Attachment 1-75
-------
quantity selections in a notepad within Survey Net, which coding
supervisors then reviewed and answered.
3.3.2 Processing of intakes by ARS
Westat electronically transmitted all coded intakes to ARS. All entries
in each intake requiring review or resolution by ARS were highlighted in
Survey Net's food summary screens. These included all "unknowns" (those
foods or quantities that could not be coded by Westat coders); newly
created recipe modifications; and notepad entries of questions and
explanations of coding decisions. Feedback was provided to Westat on
reviewed intakes.
As the final step in Survey Net processing, the nutritive value of each
food eaten was calculated using the weight of the food and data from the
Survey Nutrient Database. Where recipes had been modified, nutritive
values reflected those modifications.
3.3.3 Food coding database
As mentioned previously, three databases are used in Survey Net. These
include a food coding database (food descriptions, food measures, and
gram weights of those measures); a recipe database; and a nutrient
database.
The food coding database for CSFII 1998 contained 7,321 food codes, each
bearing a complete description of the food and, if relevant, the
preparation method. Each food code consists of 8 digits used to classify
foods into groups for study. The first digit in the food code identifies
one of nine major food groups: (1) milk and milk products; (2) meat,
poultry, fish, and mixtures; (3) eggs; (4) legumes, nuts, and seeds; (5)
grain products; (6) fruits; (7) vegetables; (8) fats, oils, and salad
dressings; and (9) sugars, sweets, and beverages. The second, third, and
(sometimes) fourth digits of a food code identify increasingly more
specific subgroups within the nine major food groups. The remaining
digits are used for identification of particular foods within a
numerical sequence.
Documentation section 12.1, "Food Coding Scheme," provides an outline of
the major food groups and subgroups identified by the first 1 to 3
digits of the food code. Documentation section 12.2, "Food Codes and
Abbreviated Descriptions," provides a list of the complete 8-digit food
codes with abbreviated descriptive information about each code. Below
are examples of the information found in documentation section 12.2.
CODE NUMBER ABBREVIATED FOOD DESCRIPTION
28141010 Chicken, fried, pot, veg, dessert (froz meal, lg meat)
53105260 Cake, choc, devil's food/fudge, w/icing, homemade
More detailed food descriptive information is available on Disk 2 in the
following files -- Food Description File (\tsf98\fcdb\cbdes.txt), Food
Includes File (\tsf98\fcdb\cbincl.txt), Subcode Descriptions File
(\tsf98\fcdb\cbsubdes.txt), and Subcode Includes File
(\tsf98\fcdb\cbsubinc.txt). For example, information from the Food
Attachment 1-76
-------
Descriptions and Food Includes files is provided below for both food
items listed above.
COMPLETE FOOD DESCRIPTION
Chicken, fried, with potatoes, vegetable, dessert
(frozen meal, large meat portion)
(Include Banquet Extra Helping Fried Chicken
Dinner; Swanson Hungry Man Fried Chicken Dinner)
Cake, chocolate, devil's food, or fudge, with icing,
coating, or filling, made from home recipe or purchased
ready-to-eat
(Include chocolate, devil's food, or fudge, NS from
home recipe, from mix or bought RTE;
Jack-in-the-Box Double Fudge Cake)
Sample persons varied in their knowledge of foods as well as in their
ability to recall or describe foods eaten. Thus, the descriptions of
foods provided by sample persons varied from very specific to very
general. Also, sample persons could not always provide details regarding
food preparation (such as the method of cooking or whether the food was
cooked with or without fat); the original form of the food (such as
fresh, frozen, dry, or canned); or the ingredients in a mixture.
Generally, foods reported with complete descriptions were assigned codes
that preserved the identity or name of the food and the amount of detail
specified. However, if the description of a food was general, such as
"bread," "juice," or "beef," a "not further specified" (NFS) code was
assigned. (See documentation section 3.3.5, "Recipe database.") In other
cases, foods were reported with descriptions lacking only one detail.
These foods were placed in codes providing as much detail as given and
noting the one lacking detail as "not specified" in the code
description, e.g., "chicken breast, fried, no coating, not specified as
to skin eaten."
Identification by brand names is widespread in the food coding database.
Several types of survey codes are specific to brands in the description
of the code or in the weights provided. Codes may be unique to a
particular brand if warranted, such as for breakfast cereals that differ
in fortification levels, or they may encompass several brands of similar
foods, such as cheese crackers. When appropriate, measures and their
gram weight equivalents are specified by brand.
The guidelines used to decide if a new code is needed for a brand name
food are the same as for other foods. A new code may be created for one
or more of the following reasons: (1) no code presently exists for a
food similar to the food reported, (2) the reported food contains either
sizable amounts or intentionally reduced amounts of one or more
nutrients, (3) the food is likely to be reported again, or (4) the form
or type of food is of special interest to data users. Special effort is
made to incorporate ethnic foods and foods modified to be lower in fat,
sodium, or sugar.
CODE NUMBER
28141010
53105260
Attachment 1-77
-------
3.3.4 Food measures and weights
Prior to the CSFII 1994-96, the food coding database's list of food
measures and their corresponding weights in grams were examined for
consistency by a Weights and Measures Team that included members from
both ARS and the National Center for Health Statistics, U.S. Department
of Health and Human Services. Cubic inch weights of many meats and fluid
ounce weights of beverages were reviewed and revised if necessary. Cup
weights for breakfast cereals and fluid ounce weights for infant
formulas were updated based on new information from the manufacturers.
Dimensions were added to the measure description for many fresh fruits
and vegetables. New foods and ethnic foods were prepared and weighed in
a USDA food laboratory and added to the database. Brand-specific and
household measures as needed were also added to the list. There are
presently over 30,000 weights for measures of foods in the food coding
database.
3.3.5 Recipe database
The purpose of the recipe database is to provide information for use
during generation of the Survey Nutrient Database. It contains a recipe
entry for each unique food code in the food coding database. These
entries include ingredients and their amounts, as well as information
for determining changes that may occur in nutrients during cooking.
Foods that are not mixtures, e.g., whole milk, are represented as
single-ingredient recipes. Ingredients are identified with codes linking
them to the Primary Data Set of nutrient values (see documentation
section 3.3.6.2, "Primary Data Set"). The recipe database also serves as
public documentation for how nutrient values were calculated for each
survey food code. Recipes are considered "representative," meaning they
are not exact for every sample person nor were they developed to
determine the intake of specific food ingredients. A variety of popular,
regional, and specialty cookbooks were consulted to aid in constructing
representative recipes. Recipes for many of the commercially available
mixtures were estimated from label information (Marcoe and Haytowitz
1993) .
In preparation for the CSFII 1994-96, recipes for "Not Further
Specified" (NFS) food codes were reviewed. These NFS codes are used when
sample persons are unable to provide further detail about a food. For
example, the "Milk, NFS" code is used when sample persons do not give
the fat content of the milk they drank. The present recipe for "Milk,
NFS" is a composite of whole milk, 2-percent milk, 1-percent milk, and
skim milk in proportions that reflect milk production statistics. The
"Milk, NFS" recipe is revised each year to reflect the most current
production data. Recipes for other NFS codes may be based on composites,
as for milk, or they may be based on the form of food most frequently
consumed in the food group in question. For example, the recipe for
"Bread, NFS" is white bread.
Attachment 1-78
-------
3.3.6 Survey Nutrient Database
About the Survey Nutrient Database
The Survey Nutrient Database is maintained specifically for use with
nationwide food surveys (Perloff et al. 1990). It is updated once a year
when a nationwide food survey is under way. Its source of nutrient
values is the Primary Data Set of nutrient values maintained in the ARS
Nutrient Data Laboratory (see "Primary Data Set" below).
The Survey Nutrient Database includes values for food energy and the
following nutrients and food components: protein, total fat, saturated
fatty acids, monounsaturated fatty acids, polyunsaturated fatty acids,
19 individual fatty acids, cholesterol, total carbohydrate, dietary
fiber, vitamin A (as international units and as retinol equivalents),
carotenes, vitamin E, vitamin C, thiamin, riboflavin, niacin, vitamin
B-6, folate, vitamin B-12, calcium, phosphorus, magnesium, iron, zinc,
copper, sodium, potassium, alcohol, moisture (water), selenium*,
caffeine*, and theobromine*. Values for the three items with asterisks
(*) were added to the database for the first time with this release.
The Survey Nutrient Database contains two files of nutrient values: (1)
The Survey Nutrient Values, Set 1, which includes data for each unique
survey food code from the food coding database (see documentation
section 3.3.3, "Food coding database" above); and (2) the Survey
Nutrient Values, Set 2, which is identical to Set 1 of the Survey
Nutrient Values with the following exception: In recipes where salt is
considered an optional ingredient, it was removed from the recipe before
the nutrients were calculated.
Both Set 1 and Set 2 of the Survey Nutrient Values were used during the
last step of Survey Net processing when the nutritive value for each
consumed food was calculated. If the sample person indicated salt was
used in cooking the food, or if she or he did not know, data were
selected from Set 1. If salt was not used, data were selected from Set
2 .
Primary Data Set
The Primary Data Set of nutrient values is maintained by the ARS
Nutrient Data Laboratory in support of the National Nutrition Monitoring
and Related Research Program. These nutrient values are used to create
the Survey Nutrient Database. The Primary Data Set is updated each year
when a nationwide survey is being conducted. The main source of data for
this version of the Primary Data Set (1998) was Release No. 11 of the
USDA Nutrient Database for Standard Reference (USDA/ARS 1996), the same
as used for the CSFII 1994-96. Unpublished data collected by the
Nutrient Data Laboratory were also used as needed, especially for new
products and for foods that recently changed. The most notable changes
were to folate values as discussed below. As the survey was conducted,
data for new foods were added as they were reported by sample persons,
Attachment 1-79
-------
and the final number of foods in the data set was 3,067. New values in
the Primary Data Set can be identified by the "date added/modified"
field [see 1998 formats document accompanying the Technical Support
Files (on Disk 2 in \tsf98\formats\formats.txt)].
Selenium, caffeine, and theobromine values were added to the Primary
Data Set for this release. The selenium content of plants, in particular
cereal grains, is strongly influenced by the quantity of biologically
available selenium in the soil in which they grow and, hence, their
geographical origin (Holden et al. 1991). Values for major dietary
contributors of selenium are based on laboratory analyses of food
samples drawn from retail outlets according to nationwide sampling
plans, in order to provide average values appropriate for national food
surveys (Holden et al. 1991, Gebhardt et al. 1990) .
Most of the values for major contributors of nutrients are supported by
laboratory analyses (Matthews 1991). Nutrient values not available from
laboratory analyses were imputed by Nutrient Data Laboratory
nutritionists from data for other forms of the food or from data for
similar foods (Gebhardt 1992). For each value in the Primary Data Set, a
source code is present that indicates whether the value is analytical or
imputed.
Folate values in this version of the Primary Data Set were updated to
reflect regulations that became effective on January 1, 1998, requiring
the addition of folic acid to enriched cereal grain products subject to
standards of identity (DHHS/FDA 1996).These products include flour,
cornmeal and grits, farina, rice, macaroni, noodles, bread, rolls, and
buns. Folic acid may continue to be added (with some restrictions on
amounts) to breakfast cereals, infant formulas, medical foods, food for
special dietary use, and meal replacement products. For the most part,
values in this data set were calculated based on enrichment levels
specified in the regulations, since analytical values were not yet
available. For those foods where the enrichment level is given as a
range, the midpoint was used to set the value. Food items containing any
of these products as ingredients, such as baked products made with
enriched flour, were also updated.
The state of analytical methodology for measuring nutrients in foods has
been evaluated by Beecher and Matthews (1990), and they reported that
adequate methodology for folate is lacking. The current microbiological
method approved by the Association of Official Analytical Chemists
International applies only to foods that contain the free forms of the
vitamin. Data generated by ARS for use in food composition databases
were obtained by a modified method using enzymes to release bound forms.
Recent research on determining the folate content of high-protein and
high-carbohydrate foods indicates that additional improvements in
methodology are needed (Martin et al. 1990) .
Data users should note that values for carotenes are those used by ARS
in arriving at the values for total vitamin A and are not solely
beta-carotene. Also, the values for vitamin E (quantified as
alpha-tocopherol equivalents) are based on somewhat limited data.
Attachment 1-80
-------
Recipe calculations
Entries in the recipe database identify the Primary Data Set (PDS)
item(s) used to derive the Survey Nutrient Values, Set 1 and Set 2. As
mentioned in the recipe database discussion, some survey food codes have
a one-to-one correspondence with items in the Primary Data Set and are
represented by single ingredient recipes, such as the following:
Survey food code:
111-12110, Milk, cow's fluid, 2% fat
Recipe ingredient:
PDS Number PDS item Amount
01079 Milk, 2% Fat, with Vit A 100 grams
However, many survey food codes require multiple ingredients, for
example:
Survey food code:
423-01010, Peanut butter sandwich
Recipe ingredients:
PDS Number PDS item Amount
16098 Peanut butter 24.0 grams
18069 Bread, white 52.0 grams
The retention factor method (Powers and Hoover 1989) was used for
calculating the nutrient content of recipes. Perloff has described how
this method is used for generating values in the Survey Nutrient
Database, including how factors estimating changes in nutrients due to
cooking or processing are used in the calculations (Perloff 1985) .
Factors for calculating moisture and fat changes are stored in each
recipe. Factors for estimating losses in 18 vitamins and minerals are
stored in a separate data file, the Nutrient Retention Factors File,
which is accessed during the recipe calculation procedure. The presence
of special codes in the recipe entries indicate when the retention
factors are used. Retention factors for selenium and vitamin E are not
available.
3.3.7 Multi-year databases
The nutrient intake data for the CSFII 1998 were calculated using the
1998 values from the multi-year food coding, nutrient, and recipe
databases that are included only on Disk 2. Some foods changed between
the CSFII 1994-96 and the CSFII 1998. For example, folic acid is now
added to enriched grain products. In such cases, both the Primary Data
Set and the Survey Nutrient Database contain multiple records for the
different nutrient levels in the food. Multiple records also exist for
some food weights and recipes. Multiple records do not exist for
modified recipes.
All records in the multi-year food coding, nutrient, and recipe
databases have start- and end-date fields indicating the time period
Attachment 1-81
-------
when each record was available for coding. These date fields can be used
to extract a single-year version from the multi-year database.
3.3.8 Combination codes
Rationale for and development of combination codes
A notable feature available on the CSFII 1994-96 and 1998 combined data
set is combination codes. Data users can find combination code data in
record type 30 (rt30.dat) fields COMBNUM (positions 104-5) and COMBTYPE
(positions 106-7) .
Combination codes were developed for two distinct purposes. First, a
greater level of specificity in coding is possible when sufficient
detail about the foods that make up a combination is collected. For some
foods, two or more food codes linked together in a food combination
present a more precise picture of what was actually eaten by respondents
than if a single food mixture code is used. Second, the use of
combination codes provides insight into patterns of food
consumption--what types and amounts of foods are eaten together and what
types and amounts of foods are eaten as separate items. This information
is helpful in answering questions about not only what people are eating,
but how they are eating it and how much. For example, do adults and
children consume milk differently? Do adults get more of their milk from
drinking it as a beverage, or from adding it to another food, such as
coffee or cereal?
Recognition of the need for a way to express food combinations through
multiple food codes began with the NFCS 1977-78. For the NFCS 1977-78,
three "partition codes" were developed to indicate foods that were part
of a sandwich, part of a salad, or part of a frozen meal, as shown in
table 3-1 on the next page. Approximately 12 percent of all foods were
assigned one of these partition codes.
Attachment 1-82
-------
Table 3-1. Use of partition codes and combination codes, NFCS
1977-78 through CSFII 1996
1977-78 1985-86 1987-88 1989-91
NFCS# CSFII* NFCS# CSFII##
Percent
Partition code
type:
Sandwich
10 .7
13 . 0
12 . 6
13 .1
Salad
1. 0
4.2
4 . 0
3 . 9
Frozen meal
* *
* *
* *
NA
Mixture
NA
1 . 7
1 . 7
2.2
Soup
NA
. 1
. 1
.2
Beverage
NA
NA
8.3
9.3
Missing
. 1
* *
* *
NA
Single item
88.2
CO
I-1
o
73 .3
71.3
1994 1995 1996
CSFII CSFII CSFII
Percent
Combination type:
Sandwich
13 .8
13 . 7
14 . 8
Salad
5 .1
5 . 1
5.2
Frozen meal
* *
* *
. 0
Other mixture
5 . 0
5.3
5.3
Soup
.6
.6
. 5
Beverage
7.4
8.3
8 . 1
Cereal
6 . 1
6.3
5.9
Baked product
7.2
7.3
7.2
Ice cream
.4
.4
. 5
Vegetable
3 .5
3 .8
3.8
Fruit
.4
. 5
.4
Single item
50 . 5
48 .8
48 . 5
#Basic sample.
*Women and children, basic and low-income samples.
##Combined basic and low-income samples.
** Calculated value is <0.1%.
The number of partition codes and the utilization of these codes
increased gradually through the years. In the CSFII 1985-86, partition
codes were added for mixtures and for soups, and 19 percent of all foods
were assigned a partition code. In the NFCS 1987-88, a partition code
was added for beverages with additions (for example, coffee with cream
and sugar) or with multiple ingredients (for example, "health shakes,"
that is, milk- or juice-based drinks with fruit, cereals, and other
ingredients pureed together), and the percent of foods assigned a
partition code increased to 27 percent.
Attachment 1-83
-------
The increased use of partition codes was also due to the concurrent
increase in the number of nutrients in the Survey Nutrient Database,
such as fiber. Greater specificity in reporting and coding of foods was
necessary in order for appropriate nutrient values to be assigned. The
use of partition codes allowed this information to be coded and at the
same time avoided having to add unmanageable numbers of new food codes.
In the CSFII 1994-96, nearly one-half of all foods items were reported
in combination. This near-doubling in the number of foods that were part
of a combination is attributable in part to two major changes in the way
mixture information was collected and coded. First, there was an
expansion of the concept of "partition codes" with the addition of five
more combination codes for cereal, baked product, ice cream, vegetable,
and fruit combinations. These codes were used to code the ingredients in
selected mixtures, as well as to link accompanying food items with the
foods they were combined with "at the table," such as cream cheese on a
bagel, margarine on a baked potato, or banana or berries on cereal.
Second, the Food Instruction Booklet (FIB) was revised to standardize
the collection of details about additions to foods and about mixtures,
thus enabling greater specificity in food coding.
There were no changes in combination codes between CSFII 1994-96
and CSFII 1998.
Data collection and coding of combinations in the CSFII 1994-96
The FIB is described in documentation section 3.2.1, "CSFII/DHKS 1994-96
and CSFII 1998." Under each category of food/drink in the FIB, there was
a set of questions (probes) the interviewer was required to ask in order
to collect enough detail for the food to be coded. For the CSFII
1994-96, major changes made to the FIB include not only more food
Attachment 1-84
-------
categories, but also more standardized probes, including probes about
ingredients of foods and any additions to foods.
For the CSFII 1994-96, interviewers and coders were trained on how to
record and code combinations. Following instructions in the FIB, the
interviewers recorded ingredients of mixtures such as sandwiches and
salads and placed brackets around them to identify them as one food item
eaten. They also used brackets to link foods added together "at the
table," such as the cream added to coffee and the jam spread on toast.
Coders used this information to code the foods as eaten in combination.
If insufficient information was available to code separately all the
food items included in a salad or sandwich (for example, when detailed
descriptions or amounts of ingredients were not given), the coder would
attempt to find a close single-code match for the combination in the
food coding database.
If enough information was available to code a combination as two or more
separate food items, all food codes for that combination were assigned
both a combination type number and a sequence number. The coder chose
the combination type from a list of categories provided by ARS (see
table 3-2 below). Each combination was assigned a sequence number which
served to distinguish that particular combination from other
combinations consumed by that sample person on that intake day. The
combination type and sequence number are labeled as COMBTYPE and
COMBNUM, respectively, on record type 30 (rt30.dat).
Attachment 1-85
-------
Table 3-2. Combination types (and type numbers)--foods with additions
or foods in combination
Beverage (01)--
* Coffee/tea with: milk, cream/cream substitute; sugar/sugar
substitute
* Water with: lemon; lime; fruit juice
* Infant formula with: instant baby cereal added to formula
* All milkshake/float ingredients coded separately
* All beverage/mixed drink ingredients coded separately
Cereal (02) - -
* Ready-to-eat breakfast cereals with: milk; sugar/sugar
substitute; fruit
* Cooked cereals such as oatmeal, cream of wheat, grits with:
milk; sugar/sugar substitute; fruit; margarine/butter; gravy
* Several breakfast cereals in a mixture coded separately
* Instant baby cereal with: formula, milk, water, beverage added
Bread/baked product (03)—
* Toast, rolls, buns, bagels, biscuits, muffins, sweet breads,
pancakes (including potato), waffles with: margarine/butter;
jam/jelly; cheese/cream cheese; whipped cream; syrup; fruit; gravy
* Cakes, pies, brownies, cookies with: ice cream; whipped cream;
fruit
* Crackers with: meat; cheese; dip; peanut butter; jam/jelly;
margarine/butter
* Nacho chips/corn chips with: cheese; dip; refried beans, etc.
(nacho supremes)
* Rice cakes with: peanut butter; jelly; cheese, etc.
* Tortilla with salsa
Salad (04)- -
* All salad ingredients coded separately and/or additions
* Green leafy salads, pasta salads, fruit salads, potato salad,
taco salad, egg salad
* Salad dressing added to salad
Sandwich (05)--
* All sandwich ingredients coded separately and/or additions
* "Filled" tacos, enchiladas and burritos
* Hamburger, hot dogs with ingredients coded separately and/or
additions
* Quesadilla
--continued
Attachment 1-86
-------
Table 3-2 -- continued
Soup (06)- -
* Soup with: crackers; cheese; croutons; green spring onions
* All soup ingredients coded separately and/or additions
Frozen meal (07)—
* Frozen meal with: catsup, tartar sauce, margarine/butter
* All frozen meal ingredients coded separately
Ice cream/frozen yogurt (08) —
* Ice cream or frozen yogurt with: syrup; toppings; fruit; nuts;
whipped cream; candy; cookies
* All ingredients of a sundae coded separately
Vegetables (09) —
* French fries with: catsup; gravy; steak sauce; vinegar; dressing
* Potato chips with: dip
* Potatoes with: gravy; sour cream; toppings; butter/margarine added
* Beans, legumes with: sauce; margarine/butter
* Vegetables (not specified as salad) with: margarine/butter; sauce;
dip; dressing
* Vegetables in a mixture coded separately
Fruit (10)- -
* Fruit with: whipped topping; sugar; milk/cream; syrup; honey
* Fruits in a mixture (not specified as salad) coded separately
Other mixtures (99)—
* Rice with: butter; gravy; sauce
* Pasta/spaghetti with: butter; gravy; sauce
* Meat, poultry, fish with: gravy; sauce; onions
* Eggs with: catsup, salsa
* Pizza with: grated cheese
* Yogurt (not frozen) with: nuts, fruit, cereal, etc.
* Foods/mixtures of foods that do not fit in other combination
categories
Attachment 1-87
-------
Examples of analyses using combination codes
The presence of combination codes in the CSFII 1994-96 and 1998 combined
data set may be useful in planning analyses, especially concerning
salads, sandwiches, or foods combined "at the table," such as cereal and
milk or corn chips and salsa. Examples 1 and 2 below illustrate how
using combination codes can affect frequencies and mean food and
nutrient intakes. Example 3 illustrates how combination codes can also
provide insight into food consumption patterns.
Two-day intake data from the CSFII 1994 were used for all examples. The
estimates are unweighted.
Example 1 (using combination types to measure frequency and mean intake
of specific food mixtures)--To fully account for all reports of a food
mixture such as a sandwich or salad, consideration must be made of the
different ways that foods may have been recorded and coded. Depending on
how a food was reported, it may have been coded as a single item or as
multiple items linked via a combination type and sequence number.
Including both ways of reporting in an analysis requires familiarity
with the food coding database, but it can give a more complete picture
of the consumption of that food.
For instance, suppose the research objective was to determine
consumption of hamburgers and cheeseburgers. All of the hamburgers and
cheeseburgers that were coded as a single item received codes in the
range 275-10210 through 275-10690 in the CSFII 1994 food coding
database. The number of reports, mean intakes by sex-age group, and
sources of hamburgers and cheeseburgers coded as a single item are
presented in tables 3-3 and 3-4.
Table 3-3. Number of reports and mean intake of hamburgers and
cheeseburgers, single-code items only*, CSFII 1994
(unweighted)
Sex and age
(years)
Number of
individuals
Number of
reports
Mean intake
per report
(gm)
Children < 6
1,140
150
96
Children 6-11
506
67
134
Teens 12-19
529
108
207
Women 2 0 +
1, 541
120
188
Men 20+
1, 547
215
220
Total
5,263
660
*Includes hamburgers or cheeseburger codes in the range 275-10210
through 275-10690 regardless of whether that food was eaten in
combination with another food or not.
Attachment 1-88
-------
Table 3-4. Places where hamburgers and cheeseburgers were
obtained, single-code items only*, CSFII 1994
(unweighted)
Restau-
Sex and age Store** rant
(years)
Fast School
food cafeteria Other
Number
Children <6
Children 6-11
Teens 12-19
Women 2 0+
Men 20+
6
4
0
3
1
6
3
2
3
3
208
135
52
96
111
5
0
0
8
1
2
0
5
3
3
Total
14
17
602
14
13
*Includes hamburgers or cheeseburger codes in the range 275-10210
through 275-10690 regardless of whether that food was eaten in
combination with another food or not.
**Includes prepared sandwiches or sandwich ingredients purchased
from stores .
Attachment 1-89
-------
It is not surprising that most of the hamburgers and cheeseburgers coded
as a single item were from fast food places, because the preferred
method given in the FIB for reporting standardized items such as fast
food sandwiches from national chains was as a single item. Nonfast-food
(or nonstandardized) hamburgers and cheeseburgers were more commonly
coded as multiple food items linked with a combination code, because the
FIB specified probes for the ingredients of nonstandardized sandwiches.
Using only the single-item food codes does not consider those hamburgers
and cheeseburgers that were coded as multiple food items linked with a
combination code. One way to expand the definition of hamburgers and
cheeseburgers would be to include all sandwich combinations (COMBTYPE =
05) containing at least one code from the range 215-00100 through
215-40100 (ground beef) and one code from the range 510 through
518 (yeast breads and rolls). Other ingredients might also be part
of these combinations. For example, this group would include a report of
a sandwich with ground beef, lettuce, tomato, and ketchup on a kaiser
roll.
The numbers of reports and amounts resulting from adding combinations of
food items eaten as hamburgers and cheeseburgers to hamburgers and
cheeseburgers coded as a single item appear in tables 3-5 and 3-6. The
number of reports of hamburger and cheeseburger consumption is nearly
double that shown in table 3-3, and the distribution is less dominated
by the fast food sandwiches, as expected.
Table 3-5. Number of reports and mean intake of hamburgers and
cheeseburgers, single-code items and combinations,
CSFII 1994 (unweighted)
Sex and age
(years)
Number of
individuals
Number of Mean intake
reports per report
(gm)
Children < 6
Children 6-11
Teens 12-19
Women 2 0 +
Men 20+
1,140
506
529
1, 541
1, 547
233
160
216
272
430
207
183
220
107
148
Total
5,263
1,311
Attachment 1-90
-------
Table 3-6. Places where hamburgers and cheeseburgers were
obtained, single-code items and combinations, CSFII
1994 (unweighted)
Restau- Fast School
Sex and age Store* rant food cafeteria Other
(years)
Number
Children <6
52
12
144
12
13
Children 6-11
50
9
60
33
8
Teens 12-19
43
6
125
29
13
Women 2 0 +
91
27
135
5
14
Men 20+
107
30
251
5
37
Total
343
84
715
84
85
*Includes prepared sandwiches or sandwich ingredients purchased
at stores.
Example 2 (using combination codes to aggregate food groups for nutrient
analyses)--Similarly, assessments of the nutrient contributions of
specific foods can be affected if the food was often eaten as part of a
mixture that was coded in combination with other foods as well as
separately. Lettuce can serve as an illustration of this type of
situation. There is a series of codes in the 1994 food coding database
for lettuce-based salads coded as a single item (751-43000 through
751-46000 and 751-48000). An example of these lettuce-based salads is
751-43000 (lettuce, salad with assorted vegetables including tomatoes
and/or carrots, no dressing). The nutrient contribution of this food
group is shown in table 3-7.
Attachment 1-91
-------
Table 3-7. Number of reports, mean intake, and nutrient
contribution of lettuce-based salads, single-code
salads only*, CSFII 1994 (unweighted)
Mean
Number salad
of intake
re-
per
Ener-
Pro-
Carbo-
Sex and age
ports
report
gy
tein
hydrate
Fat
(years)
(gm)
(kcal)
(gm)
(gm)
(gm)
Children <6
24
38
7
.4
1.4
. 1
Children 6-11
21
57
12
. 7
2 .1
.3
Teens 12-19
15
85
34
1. 9
3 .2
1. 7
Women 2 0 +
46
115
36
1.8
4.3
1.6
Men 20+
38
140
57
3 . 0
5.3
2 . 9
*Includes the nutrients from all lettuce-based salad codes
(751-43000 through 751-46000 and 751-48000) regardless of whether
that food was eaten in combination with another food or not.
Using only the single-code salads has two deficiencies that can be
corrected by the use of combination codes. First, it can be noted from
examination of the food coding database that the lettuce-based salads
coded as a single item do not include salad dressing. This is because
the FIB specified probes for salad dressing in order to obtain as much
information as possible about the type and amount of salad dressing
eaten. Salad dressing is always linked to salad via a combination type
(04, salad) and sequence number. Consequently, if only single-code
salads are considered, the contribution of lettuce-based salads to total
fat intake is underestimated. Second, restricting the analysis to
single-code lettuce-based salads misses any salad-type combinations with
lettuce coded simply as lettuce (751-13000) .
When all lettuce-containing salad combinations (COMBTYPE = 04) are added
to all single-code lettuce-based salads (this time incorporating any
other ingredients linked to them via combination type 04 and sequence
number), the nutrient contributions are considerably different, as shown
in table 3-8. Not surprisingly, the contribution of lettuce-based salads
to nutrient intake, most notably energy and fat, is dramatically
increased when mixtures linked by combination codes are included. Mean
salad intakes also increased markedly, and the number of reports of
salads increased five- to fourteen-fold across the sex-age groups.
Attachment 1-92
-------
Table 3-8. Number of reports, mean intake, and nutrient
contribution of lettuce-based salads, single-code and
combination salads, CSFII 1994 (unweighted)
Mean
Number salad
of intake
re-
per
Ener-
Pro-
Carbo-
Sex and age
ports
report
gy
tein
hydrate
Fat
(years)
(gm)
(kcal)
(gm)
(gm)
(gm)
Children <6
144
77
77
1.5
4 .1
6.3
Children 6-11
109
101
101
1.6
5 . 5
8.5
Teens 12-19
106
170
197
4 . 9
9.2
16 .3
Women 2 0 +
594
179
179
4.2
9.9
14 .3
Men 20+
541
205
210
4 .4
11.4
17.2
Example 3 (using combination codes to examine food consumption
patterns)--The manner in which individuals consume foods, that is,
separately or together with other foods, may be determined by using
combination codes. It can be expected that population subgroups vary in
their consumption patterns. For instance, children consume milk
primarily as a single-code item whereas adults more often consume milk
in combination with another food, such as coffee or cereal, as shown in
table 3-9 on the next page. Although nearly one-third (32.6 percent) of
all reports of milk by women were milk consumed as part of a beverage
combination (such as in coffee), the largest percentage of the total
quantity (in grams) that was consumed by women was provided by milk
consumed as a single item (57.1 percent). Milk added to cereal made a
substantial contribution to total milk consumption for all sex-age
groups.
Limitations of combination codes
While combination codes may be used to identify foods eaten together,
disaggregation of combinations is not sufficient to enable researchers
to look at the total intake of a specific food. For example, a
researcher who wished to look at the total intake of tomatoes from all
sources could not arrive at that number by combining tomatoes reported
separately with those that were reported as part of a combination. That
method of analysis would miss tomatoes that are included as ingredients
in many single-code mixtures such as 283-10220 (chili beef soup) and
581-30010 (lasagna with meat and/or poultry).
Attachment 1-93
-------
Table 3-9. Milk consumption by combination type, CSFII 1994
(unweighted)
Sex and age
(years)
Combination type
Single
item
Beverage
combina-
tion
Cereal
combina-
tion
Other
combina-
tions
Number of reports
Children <12
3 ,383
200
1,665
15
Teens 12-19
471
27
283
3
Women 2 0+
788
739
719
18
Men 20+
959
641
764
24
All ages
5,601
1,607
3,431
60
Percent of all milk reports
Children <12 64.3 3.8 31.6 0.3
Teens 12-19 60.1 3.4 36.1 0.4
Women 20+ 34.8 32.6 31.8 0.8
Men 20+ 40.2 26.8 32.0 1.0
All ages 52.4 15.0 32.1 0.6
Percent of total quantity consumed
Children <12 70.5 4.0 25.3 0.2
Teens 12-19 67.1 2.5 30.0 0.4
Women 20+ 57.1 11.1 31.2 0.6
Men 20+ 61.1 8.4 29.6 0.9
All ages 65.6 6.1 27.9 0.5
3.4 Quality Control
At every step during the development and execution of the CSFII/DHKS
1994-96 and CSFII 1998, quality control has been one of ARS' primary
concerns. During the process of CSFII/DHKS 1994-96 questionnaire
development, ARS solicited input from the Continuing Survey Users Group,
which is made up of representatives from 13 Federal agencies, as well as
other Federal users. The CSFII intake questionnaire underwent cognitive
testing by the Census Bureau's Center for Survey Methods Research, and
the "multiple-pass" approach used for the first time in 1994 was
developed to optimize the completeness of intake data collected [Tippett
and Cypel(eds.) 1997, DeMaio et el. 1993, Guenther et al. 1995]. The
DHKS questionnaire was revised and expanded with input from members of
the Continuing Survey Users Group and an in-house DHKS working group. It
Attachment 1-94
-------
was then pretested for comprehensibility and flow by ARS in
collaboration with the Census Bureau's Demographic Surveys Division. ARS
staff and the contractor revised the Food Instruction Booklet used in
conjunction with the intake section of the CSFII questionnaire,
expanding the booklet to standardize probing by interviewers and ensure
the collection of adequate detail for food coding.
All field supervisors, interviewers, and coders attended extensive
training sessions. All sessions were scripted for consistency and were
monitored by ARS staff. Bilingual interviewers attended an additional
day of training in the use of Spanish language questionnaires. Practice
interviews were reviewed by supervisors, and telephone retraining was
conducted when necessary. Detailed instruction manuals were provided to
supervisors, interviewers, and coders.
Electronic communications permitted close tracking by the contractor and
ARS of assigned cases in the field, their completion status, and
documents in various stages of processing. Electronic delivery of survey
data facilitated the timely resolution of such issues as errors in
sample person selection or clarification by the interviewer of data
received by the home office.
Survey Net, a computer-assisted food coding system (see documentation
section 3.3.1, "Food coding and editing," for additional information on
Survey Net) developed under a cooperative agreement between ARS and the
University of Texas-Houston Health Science Center, School of Public
Health, provided efficiency and accuracy in on-line coding of foods and
editing. Edit checks were built into the system to reduce data entry of
erroneously high or low food amounts and to catch some of the most
common reporting, recording, and coding errors.
A pilot study duplicating the planned survey design on a small scale was
conducted from April to June of 1993. The pilot study tested the
questionnaires, data collection methods, field management procedures,
data entry and processing procedures, and survey management software
slated for implementation in the CSFII 1994-96. This experience provided
an excellent opportunity to further refine the quality of survey
instruments and improve the efficiency of survey operations.
As a result of the pilot study, interviewer training was lengthened to 7
full days to allow more thorough coverage of survey procedures.
Modifications that had been made to the questionnaires and data
collection procedures were judged to be effective in reducing respondent
burden and facilitating the collection of high-quality data. Interviewer
field notebooks and debriefing after the pilot study provided feedback
resulting in further revision of the questionnaires. Survey management
software programs used by the contractor and ARS were found to be
effective tools for monitoring survey activities and improving the
efficiency of survey operations.
ARS data processing activities were reviewed by a panel of outside
experts in November 1994. The panel's primary recommendation was that
ARS scale back its exhaustive review of the data by prioritizing tasks
and streamlining the mechanics of data processing. Quality control
Attachment 1-95
-------
procedures described in this section reflect ARS' implementation of the
panel's recommendations.
Achieving acceptable response rates in the CSFII/DHKS 1994-96 was a
priority for ARS. By contract, Westat, Inc., was required to meet
specified response rate requirements for each questionnaire (screener,
household, individual intake, and DHKS).
Many steps were taken to monitor interviewer performance. These included
partial reinterviews of 10 percent of each interviewer's cases to
validate contact of households, audiotaping of at least one intake
interview and two DHKS interviews per interviewer per year, and
in-person observations.
Interviewers were instructed to edit their own work as soon as possible
after the interview to identify and correct errors in recording and to
permit (with supervisory permission) retrieval of any missing
information from the respondents. Completed questionnaires were reviewed
within 2 days of receipt at the Westat central office to determine
whether they met ARS minimum criteria. If not, callbacks were made to
obtain missing information. Reviews sometimes led to telephone
mini-retrainings of interviewers. Field staff memos and a quarterly
newsletter provided all interviewers with answers to questions raised
during training and in the field, as well as feedback on problem areas
detected in data review by the contractor and ARS.
Food coders were required to pass a certification test developed by ARS
before they were allowed to code survey data. Initially, 100 percent of
each food coder's work was verified by blind double-coding with
resolution of any differences. At the supervisor's discretion, this
adjudication process was applied to less of the coder's work; 10 percent
of the food coder's work continued to be verified routinely. Problems in
the food coding process were discussed at biweekly food coding meetings.
ARS monitored coder performance by occasionally observing food coders at
work, by periodically attending coder meetings and refresher trainings,
and by comparing information recorded on the questionnaire to coded
entries.
Accuracy of nonfood data entry was verified by routine 100 percent
independent double entry with resolution of differences by coding
supervisors. Nonfood data were edited for reasonableness, logic, and
consistency; supervisors resolved discrepancies.
ARS verified the accuracy of weekly data delivery by checking each
hard-copy document received against an electronic list of documents. At
least 10 percent of all food intake questionnaires were reviewed for
accuracy in coding and data entry. In addition, all foods and food
amounts that could not be coded by the contractor (i.e., "unknowns")
were reviewed and coded by ARS food coding staff. Other food codes and
amounts flagged by the contractor as questionable were reviewed for
accuracy. All recipe modifications (see documentation section 3.4.1,
"Food coding and editing") done by Westat were reviewed by ARS coding
staff.
Attachment 1-96
-------
A series of reviews was conducted on food intake data. Values of food
variables falling outside reasonable parameters were flagged, checked
against information recorded on the questionnaire, and corrected if in
error. ARS reviewed audiotaped intake interviews for proper interviewing
techniques. Any problems in interviewer or coder performance detected by
ARS were brought to the attention of the contractor.
ARS review of nonfood data was also extensive, encompassing over 3 0
specific edit checks for reasonableness, consistency, and logic. Values
falling outside of reasonable or expected parameters were checked
against information recorded on the questionnaire and corrected if in
error.
All screeners from eligible households were reviewed to confirm that
proper sampling procedures had been followed. Sampling errors were
immediately brought to the contractor's attention.
All household questionnaires and DHKS questionnaires were reviewed to
ensure that proper interviewing and coding procedures had been followed.
Any interviewer or coder problems were summarized in periodic reports to
Westat. Also, audiotaped DHKS interviews were reviewed by ARS, and
general feedback was provided to the contractor.
The accuracy of the Survey Nutrient Database was also a priority for
ARS. Numerous quality control checks were performed on various
components of the Survey Nutrient Database, such as nutrient values for
new or updated codes in the Primary Data Set, the recipe file, and the
file of weights for household measures. Final nutrient values in the
1994-96 Survey Nutrient Database were confirmed by a series of
comparisons to earlier Survey Nutrient Databases, with subsequent review
of values falling outside of reasonable parameters. After food codes
were aggregated by type of food, averages of nutrients from those foods
were subjected to many of the same rigorous outlier checks conducted for
Primary Data Set codes.
Every nutrient intake value (daily total) from each responding sample
person's intake was tested for reasonableness against parameters for
individuals of that age and sex. In addition to detecting errors in
coding of foods or amounts, this provided an additional quality check of
the nutrient database.
Attachment 1-97
-------
3.5 Glossary
Age--Calculated from date of birth, if given. Otherwise, age as given by
respondent. For responding sample persons (see "Responding sample
person"), this is the age as of the day-1 intake; for others, this is
the age on the day of screening.
Alpha-tocopherol equivalent--See "Vitamin E."
Assistant--Person who assisted in the dietary recall for a sample person
age 6 to 9 years in CSFII 1998 and age 6 to 11 years in CSFII 1994-96.
Black--See "Race."
Breast-fed child--A child 3 years of age or younger at the time of the
household interview who was identified by the household respondent as
being breast fed currently. Breast- fed sample persons were included in
the weighting process, and the survey data set includes information on
breast-fed children as discussed in section 7.6.2, "Breast-fed
Children."
Calcium conversion factor--A factor that expresses the amount of calcium
in 100 grams of a given milk product (that is, any food code beginning
with "1") as a proportion of the amount of calcium in 100 grams of fluid
whole cow's milk. For example, the calcium conversion factor for Cheddar
cheese was calculated by dividing the amount of calcium in 100 grams of
Cheddar cheese (721 milligrams) by the amount of calcium in 100 grams of
fluid whole cow's milk (119.4 milligrams), resulting in a calcium
conversion factor of 6.04. Used in calculation of calcium equivalent as
described below.
Calcium equivalent --The amount, expressed in grams, of whole fluid cow's
milk that has the same quantity of calcium as the reported milk product.
Derived by multiplying the amount of the milk product eaten, expressed
in grams, by the calcium conversion factor (see "Calcium conversion
factor" above.) For example, the calcium equivalent of 2 ounces (57
grams) of Cheddar cheese is calculated by multiplying 57 grams x 6.04
(the calcium conversion factor for Cheddar cheese) = 344 grams. Thus,
the amount of calcium in 57 grams of cheddar cheese is equal to the
amount of calcium in 344 grams of whole fluid milk. Intakes of total
milk and milk products may be compared among population groups using
calcium equivalents to take into account the different calcium densities
of milk products subgroups (for example, fluid milk and cheese) that may
be used in varying proportions by the population groups. The calcium
equivalent is present on record type 30 (rt30.dat) in the field CALEQ.
Carotenes--Beta-carotene and other provitamin-A carotenoids. See
"Vitamin A."
Central city--See "Urbanization."
Combination--Foods combined together and consumed as a unit that were
coded using two or more food codes; identified by the record type 30
(rt30.dat) fields COMBNUM and COMBTYPE. For more discussion of
combinations, see sections 3.3.1, "Food coding and editing," and 3.3.8,
"Combination codes."
Attachment 1-98
-------
Dietary fiber--Total dietary fiber including both the insoluble fraction
(cellulose, hemicellulose, and lignin) and the soluble fraction (for
example, gums in cereal grains and pectin in fruits and vegetables).
Dietary intake--See "Food intake."
Dwelling unit--House, apartment, room, or group of rooms occupied as
separate living quarters, when the occupants do not live and eat with
any other person in the structure and when there is direct access from
the outside or through a common area or hall. Synonymous with "housing
unit" as described in the definition of "households" for the 1990 Census
(Baugher and Lamison-White 1996) .
Eating occasion--Any report of eating or drinking by a sample person.
Each change in time of eating reported on the questionnaire was
considered to be a separate eating occasion.
Educational level--For each household member 15 years of age or older,
the household respondent was asked to name the highest grade of formal
schooling completed, starting with "kindergarten or less" and continuing
in 1-grade or 1-year increments to "5 or more years of college." Formal
schooling does not include trade or vocational schooling, company
training, or tutoring, unless credit is given which would be accepted at
a regular school or college. High school equivalency (GED) was
considered equal to completing grade 12.
Employment status--For each household member 15 years of age or older,
the household respondent was asked whether the person worked during the
week preceding the interview and, if so, how many hours. "Work" includes
any full-time or part-time activity for which money, goods, or services
were received. Employment includes active duty in the armed forces. An
individual was also "employed" if he or she had a job but was not
actually at work that week. Full-time status equals 35 hours or more
worked during the week; part-time status equals 1 to 34 hours. See
Attachment 1-99
-------
discussion of the field EMP_STAT in section 9.3, "Additional
Documentation on Calculated Variables."
Ethnic origin--The screener respondent reported whether or not each
household member was of Mexican/Mexican-American/Chicano, Puerto Rican,
Cuban, or other Spanish or Hispanic origin.
Exercise--Sample persons 12 years of age or older were asked "How often
do you exercise vigorously enough to work up a sweat?"
Female head of household--Person indicated as such by the household
respondent. (Included for purposes of historical comparison.)
Folate--Total folate content; includes naturally occurring folate and
added folic acid. Folate values have been updated to reflect the
regulation requiring enriched grain products to include added folic acid
beginning January 1998.
Food intake--All beverages (except plain water with nothing in it) and
foods ingested. Does not include inedible parts of foods (such as bones,
rinds, and seeds); uneaten portions of food; or vitamin, mineral, or
other supplements.
Health status--Self-appraised.
Height--Self-reported.
Home food supply--Foods and beverages ingested at home (including food
obtained away from home and carried home to be eaten) and food items
carried from home and eaten elsewhere, such as those in picnics and
packed lunches. (Included for purposes of historical comparison.) See
the file formats for record type 30 (rt30.dat) fields EATHOME and
EVERHOME.
Household--All persons who regularly share a house, an apartment, a
room, or a group of rooms used as separate living quarters. Household
membership is based on the place where a person usually lives or sleeps
for 6 or more months per year and where the person is free to return at
any time. Includes persons temporarily absent, such as those who were in
the hospital or traveling; students who live away from the sampled
dwelling unit in dormitories or sorority or fraternity housing while
attending school, who are scheduled to return to the household at the
end of the term, and who use the sampled dwelling unit as their
permanent address; domestic or other employees who usually live and
sleep at the sampled dwelling unit; boarders or roomers who usually live
and sleep at the sampled dwelling unit; and persons temporarily visiting
the dwelling unit who have no usual place of residence elsewhere, such
as a visitor who is house hunting. Excludes former household members who
live in institutions, nursing homes, convents, etc.; persons working
abroad; and members of the armed forces stationed elsewhere. Excludes
students who live in an off-campus dwelling unit while attending school,
persons who take their meals in the household but usually lodge or sleep
elsewhere, domestic or other employees who live in an adjacent but
separate dwelling unit, and persons temporarily visiting the household
who have a usual place of residence elsewhere to which they are free to
Attachment 1-100
-------
return at any time. Excludes noninstitutional group quarters of nine or
more unrelated persons living and eating together.
Household income--Household respondent's estimate of the total income
from all sources, before taxes, of all household members for the
calendar year prior to the interview. Includes income of
roomers/boarders. Excludes income of live-in employees. See discussion
of the field INCOME in section 9.3, "Additional Documentation on
Calculated Variables."
Household member--See "Household."
Household respondent--Person who answered the household questionnaire,
usually either the main meal planner/preparer or a person knowledgeable
about household characteristics such as income; not necessarily a sample
person.
Household size--Number of individuals in a household.
Income--Both monthly and annual household income were collected. See
section 9.3, "Additional Documentation on Calculated Variables," for a
discussion of income and for information on imputed incomes.
Key field--A frequently-used field (variable) included in all record
types (data files). See section 7.4.1 for a list of the key fields.
Lactating female--A female household member 10 to 55 years of age
identified by the household respondent as currently breast-feeding a
child 3 years of age or less.
Main meal planner/preparer--Person who usually plans and/or prepares the
household's meals or does the major food shopping. This person was the
preferred household respondent, proxy, and assistant.
Male head of household--Person indicated as such by the household
respondent. (Included for purposes of historical comparison.)
Metropolitan Statistical Area--A geographic area consisting of a large
population nucleus together with adjacent communities that have a high
degree of economic and social integration with that nucleus; defined by
the Federal Office of Management and Budget for use in the presentation
of statistics by agencies of the Federal government (USDC/BOC and APDU
1993) .
Midwest--See "Region."
Niacin--Nicotinic acid and nicotinamide present in foods. Does not
include potential niacin that could be converted from dietary
tryptophan, a niacin precursor, in the body.
Nonmetropolitan areas--See "Urbanization."
Nonrespondent--Sample person who did not complete an interview.
Northeast-- See "Region."
Attachment 1-101
-------
Nutrient intake--Nutrient content of all foods and beverages (except
plain water with nothing in it) ingested. Excludes vitamin, mineral, and
other supplements.
One-day dietary recall--A recall of beverages and foods ingested during
the day preceding the interview--the 24 hours from 12:00 a.m. (midnight)
to 11:59 p.m.
Percentage of poverty level--Household income for the previous calendar
year expressed as a percentage of the Federal poverty thresholds
(Baugher and Lamison-White 1996) adjusted for inflation. See discussion
of the field PCTPOV in section 9.3, "Additional Documentation on
Calculated Variables."
Poverty level--See "Percentage of poverty level."
Pregnant female--Female household member 10 to 55 years of age
identified by the household respondent as currently pregnant.
Proxy--Knowledgeable adult who completed the dietary recall for children
under 6 years of age and other sample persons unable to report for
themselves due to physical or mental limitations or because of illness.
Proxy interviews were not substituted for in-person interviews with
adult sample persons who were difficult for the interviewer to reach or
who were nonrespondents .
Race--The screener respondent reported the race of each household member
as white, black, Asian/Pacific islander, American Indian/Alaskan native,
or some other race.
Recommended Dietary (or Energy) Allowances (RDA or REA)- - Levels of
nutrient (or energy) intake considered by the Food and Nutrition Board
of the National Academy of Sciences to be adequate to meet the known
nutritional needs of practically all healthy individuals (NRC/FNB 1989).
In a population group whose usual intake approximates or exceeds the
RDA, the likelihood of deficiency is small (NRC/FNB 1989).
Region--An area of the United States as defined by the U.S. Department
of Commerce for the 1990 Census of Population. The four census regions
and their States are as follows:
(1) Northeast: Connecticut, Maine, Massachusetts, New
Hampshire, New Jersey, New York, Pennsylvania, Rhode Island,
Vermont;
(2) Midwest: Illinois, Indiana, Iowa, Kansas, Michigan, Minnesota,
Missouri, Nebraska, North Dakota, Ohio, South Dakota,
Wisconsin;
(3) South: Alabama, Arkansas, Delaware, District of Columbia,
Florida, Georgia, Kentucky, Louisiana, Maryland, Mississippi,
North Carolina, Oklahoma, South Carolina, Tennessee, Texas,
Virginia, West Virginia;
(4) West: Alaska, Arizona, California, Colorado, Hawaii, Idaho,
Attachment 1-102
-------
Montana, Nevada, New Mexico, Oregon, Utah, Washington,
Wyoming.
Responding sample person--Household member who was selected to
participate in the individual intake component of the survey and who
provided at least 1 day of dietary intake data.
Retinol equivalents--See "Vitamin A."
Sample person--Household member selected to participate in the
individual intake component of the survey.
Sampling weights --Weights required in analysis to compensate for
variable probabilities of selection, differential nonresponse rates, and
possible deficiencies in the sampling frame. See section 5, "SAMPLING
WEIGHTS."
Screening respondent --Household member 18 years of age or older who
answered the screening questionnaire (screener).
Source of food--The place where each food or beverage (or most of the
ingredients of a mixed item) was obtained, for example, from a store,
restaurant, vending machine, or Meals on Wheels; as a mail order
purchase; or as a gift from someone else. This information was provided
by the sample person, proxy, or assistant.
South--See "Region."
Suburban areas--See "Urbanization."
Supplements --Vitamins and minerals ingested in a form other than in food
or beverage. Not included in food and nutrient intake data.
Urbanization--Based on Metropolitan Statistical Areas (MSA's) defined by
the Federal Office of Management and Budget (OMB) using information and
recommendations provided by the U.S. Bureau of the Census. The three
levels of urbanization are as follows:
(1) MSA, central city: All OMB-designated central cities, as
defined by their corporate city limits, located in 1990 MSA's.
These are primarily the urban cores of the MSA's. Although
some MSA's contain no central city, most MSA's contain one or
more.
(2) MSA, outside central city: The remaining counties or county
equivalents located in MSA's.
(3) Non-MSA: All counties or county equivalents that were located
outside of 1990 MSA's.
Vitamin A--Vitamin A activity derived from both preformed vitamin A
(retinol) and provitamin A carotenoids. Values are expressed as
international units (IU) and as micrograms of retinol equivalents (RE).
One IU equals 0.3 micrograms of retinol, 0.6 micrograms of
beta-carotene, or 1.2 micrograms of other carotenoids having vitamin A
Attachment 1-103
-------
activity. One RE equals 1 microgram of retinol, 6 micrograms of
beta-carotene, or 12 micrograms of other provitamin A carotenoids.
Vitamin E--Vitamin E activity derived from alpha-, beta-, and
gamma-tocopherol and alpha-tocotrienol. Values are expressed as
milligrams of alpha-tocopherol equivalents. One alpha-tocopherol
equivalent equals 1 milligram of alpha-tocopherol, 2 milligrams of
beta-tocopherol, 10 milligram of gamma-tocopherol, or 3.3 milligrams
alpha-tocotrienol.
Weight --Self-reported.
Weighting factors--See "Sampling weights."
West--See "Region."
White--See "Race."
Attachment 1-104
-------
3.6 References
Baugher, E. and L. Lamison-White. 1996. Poverty in the United States:
1995. U.S. Bureau of the Census, Current Population Reports, Series
P60-194 (see appendix A).
Beecher, G.R., and R.H. Matthews. 1990. Nutrient composition of foods.
In: M.L. Brown, ed., Present Knowledge in Nutrition, 6th ed., pp.
430-443. International Life Sciences Institute Press, Washington, DC.
DeMaio, T.M., S. Ciochetto, and W.L. Davis. 1993. Research on the
Continuing Survey of Food Intakes by Individuals. In: 1993 Proceedings
of the Section on Survey Research Methods, papers presented at American
Statistical Association, San Francisco, August 8-12, 1993, and American
Association for Public Opinion Research, St. Charles, IL, May 20-23,
1993, vol. II, pp. 1021-1026. American Statistical Association,
Alexandria, VA.
DHHS (U.S. Department of Health and Human Services). 1998. 1998 Poverty
Guidelines. Federal Register 63 (36) :9235-9238 .
DHHS/FDA (U.S. Department of Health and Human Services, Food and Drug
Administration). 1996. Food standards: Amendment of standards of
identity for enriched grain products to require addition of folic acid.
Final rule. Code of Federal Regulations, Title 21, Parts 136, 137, and
139. Federal Register 61(44--March 5, 1996):8781-8797.
Federal Register. October 22, 1990. Public Law 101-445. National
Nutrition Monitoring and Related Research Act of 1990.
Gebhardt, S.E. 1992. Imputing nutrient values. In: 17th National
Nutrient Data Bank Conference Proceedings, Baltimore, MD, June 7-10,
1992, pp. 143-153. International Life Sciences Institute Press,
Washington, DC.
Gebhardt, S.E., J.M. Holden, and C.S. Davis. 1990. Preliminary report: A
nationwide study of the selenium content of U.S. foods. In: M.R.
Stewart, ed., Proceedings of the 15th National Nutrient Databank
Conference, Blacksburg, VA, June 3-6, 1990, pp. 43-52. The CBORD Group,
Inc., Ithaca, NY.
Guenther, P.M., T.J. DeMaio, L.A. Ingwersen, et al. 1995. The
multiple-pass approach for the 24-hour recall in the Continuing Survey
of Food Intakes by Individuals (CSFII) 1994-96. International Conference
on Dietary Assessment Methods, Boston, January 23, 1995.
Holden, J.M., S. Gebhardt, C.S. Davis, and D.G. Lurie. 1991. A
nationwide study of the selenium contents and variability in white
bread. Journal of Food Composition and Analyses 4:183-195.
Marcoe, K.K., and D.B. Haytowitz. 1993. Estimating nutrient values of
mixed dishes from label information. Food Technology 47(4) :69-75.
Attachment 1-105
-------
Martin, J.I., W.O. Landen, Jr., A.M. Soliman, et al. 1990. Application
of a tri-enzyme extraction for total folate determination in foods.
Journal of the Association of Official Analytical Chemists 73:805-808.
Matthews, R.H. 1991. Current HNIS nutrient data research. In: 16th
National Nutrient Databank Conference Proceedings, San Francisco, June
17-19, 1991, pp. 129-132. The CBORD Group, Ithaca, NY.
NAS/NRC (National Academy of Sciences, National Research Council). 1993.
Pesticides in the diets of infants and children. National Academy Press,
Washington, DC.
NRC/FNB (National Research Council, Food and Nutrition Board). 1989.
Recommended Dietary Allowances, 10th ed. National Academy Press,
Washington, DC.
Perloff, B.P. 1985. Recipe calculations for NFCS database. In: S. Murphy
and D. Rauchwarter, eds., Proceedings of Tenth National Nutrient Data
Bank Conference, San Francisco, July 22-24, 1985, pp. 11-21. National
Technical Information Service, Springfield, VA.
Perloff, B.P., R.L. Rizek, D.B. Haytowitz, et al. 1990. Dietary intake
methodology II: USDA's nutrient data base for nationwide dietary intake
surveys. Journal of Nutrition 120:1530-1534.
Powers, P.M., and L.W. Hoover. 1989. Calculating the nutrient
composition of recipes with computers. Journal of the American Dietetic
Association 89:224-232.
Tippett, K.S., and Y.S. Cypel (eds.). 1997. Design and operation: The
Continuing Survey of Food Intakes by Individuals and the Diet and Health
Knowledge Survey 1994-96. U.S. Department of Agriculture, Agricultural
Research Service, Continuing Survey of Food Intakes by Individuals
1994-96, Nationwide Food Surveys Rep. No. 96-1. (Included on Disk 1 in
\pdffiles\dor.pdf.)
USDA/ARS (U.S. Department of Agriculture, Agricultural Research
Service). 1996. USDA Nutrient Database for Standard Reference, Release
11. ARS Nutrient Data Laboratory Home Page
chttp://www.nal.usda.qov/fnic/foodcomp>.
USDA/DHHS (U.S. Departments of Agriculture and Health and Human
Services). 1990. Nutrition and your health, Dietary Guidelines for
Americans. Home and Garden Bulletin No. 232, 3rd ed.
USDC/BOC (U.S. Department of Commerce, Bureau of the Census). 1993.
Current Population Survey, March 1992. Conducted for the Bureau of Labor
Statistics. Machine readable data file.
USDC/BOC and APDU (U.S. Department of Commerce, Bureau of the Census,
and Association of Public Data Users). 1993. A guide to State and local
census geography. U.S. Department of Commerce, Bureau of the Census,
1990 CPH-I-18.
Attachment 1-106
-------
FOOD COMMODITY INTAKE DATABASE
Appendix C
[Note: This Appendix includes information from the documentation
section, Chapters 5 and 6, of the following CD-ROM: U.S. Department of
Agriculture, Agricultural Research Service. 2000. Continuing Survey of
Food Intakes by Individuals 1994-96, 1998. NTIS Accession No. PB2000-
500027. ]
5. SAMPLING WEIGHTS
5.1 Introduction to Sampling Weights Discussion
In general, the analysis of data from surveys having complex
designs requires the use of sampling weights to compensate for
variable probabilities of selection, differential nonresponse
rates, and possible deficiencies in the sampling frame. The
CSFII/DHKS 1994-96 data set release contained sets of sampling
weights appropriate for use in the analysis of the annual data
sets as well as sampling weights appropriate for the analysis
of the 3 years combined. In addition to the sampling weights
provided with the 1994-96 (3-year) release, this combined
CSFII 1994-96, 1998 (4-year) release provides sampling weights
appropriate for use in the analysis of the 4-year data set
and sampling weights for use in the analysis of the CSFII 1998
data separately. Table 5-1 provides counts of children in the
combined 1994-96, 1998 data set. Tables 5-2 and 5-3 summarize
the sampling weight sets.
Guidance in the choice of appropriate sampling weights and
in the application of the reporting guidelines followed by the
USDA in the preparation of general statistical reports is
provided in Section 5.2 below. Sections 5.3 through 5.5
document the construction of the weights. Section 5.6
discusses variance estimation procedures appropriate with the
analysis of data from this data set.
Sampling weights appropriate for the analysis of the combined
CSFII 1994-96 data at the household level were made available
in the spring of 1999. Those 3-year household weights have been
included with this release. Section 5.7 provides the
documentation that accompanied the original release of the
household weights.
Although the Diet and Health Knowledge Survey (DHKS) was not
administered for CSFII 1998, the DHKS data records and
sampling weights from 1994-96 have been included with this
release on record type 50. Furthermore, sampling weights
designed for the analysis of household-level data from both
the 3-year and 4-year data sets have been included with
this release.
5-1
Attachment 1-107
-------
Table 5-1. Number of children providing intakes in
CSFII 1994-96 and CSFII 1998
1994-96
1998
Total
Under 1 year old
376
1, 175
1, 151
1 year old
711
373
1, 084
2 years old
705
402
1,107
3 years old
492
1,344
1, 836
4 years old
511
1,348
1,859
5 years old
475
409
884
6 years old
256
343
599
7 years old
233
71
304
8 years old
236
53
289
9 years old
258
41
299
0-9 years old
4 , 253
5, 559
9, 812
Table 5-2. Final sampling weights* provided with the combined
CSFII 1994-96, 1998 data release
Annual#
1994-96
(3-year)
1994-96, 1998
(4-year)
Available on
record types
One day
of intake
WTA DAY1
WT3 DAY1
WT4 DAY1
RT2 0 ,
RT3 0 ,
RT4 0
RT25 ,
RT3 5,
Two days
of intake
WTA 2DAY
WT3 2DAY
WT4 2DAY
RT2 0 ,
RT3 0 ,
RT4 0
RT25 ,
RT3 5,
Household
level
Not
provided
WT3 HH
WT4 HH
RT15
DHKS@
WTA DHK
WT3 DHK
Not
provided
RT5 0
DHKS with
two days
of intake@
WTA DHK2
WT3 DHK2
Not
provided
RT5 0
Columns 3, 4, and 5 give the names of the sampling weights
where weights are available. Jackknife replicate weights for
variance estimation are also provided for each of these sets
of sampling weights (see section 5.6.2, "Estimation of
Sampling Errors").
These weights are appropriate for separate analysis of
years 1994, 1995, 1996, or 1998.
DHKS sampling weights are only applicable for 1994, 1995,
and 1996. The DHKS was not administered for the CSFII 1998.
5-2
Attachment 1-108
-------
Table
5-3. Summary
combined
of sampling weights included
CSFII 1994-96, 1998 release
in the
Ages
0-19
Sample
size
Sum of
weights
CV*
VIF#
Day 1:
1994
1995
1996
1998
2,298
1, 981
1, 952
5, 559
76,641,610
77,498,715
78,316,471
40,134,208
64 .83
69.98
59 . 52
209.16
1.42
1.49
1.35
5.37
4-year
11,790
77,485,571
111 .23
2 .24
3-year
6,231
77,485,604
62 . 06
1.39
2-day:
1994
1995
1996
1998
2 , 223
1, 904
1, 853
5,304
76,641,600
77,498,713
78,316,485
40,134,206
76 . 07
80 .44
73 .44
213.58
1.58
1.65
1. 54
5 . 56
4-year
11,284
77,485,611
122 . 57
2.50
3-year
5, 980
77,485,635
74 .29
1. 55
Ages 2 0+
Sample Sum of CV* VIF#
size weights
Day 1: 1994 3,291 182,865,657 58.47 1.34
1995 3,345 184,451,592 69.42 1.48
1996 3,236 185,917,776 51.72 1.27
3-year 9,872 184,411,673 59.68 1.36
2-day: 1994 3,088 182,865,609 70.04 1.49
1995 3,168 184,451,679 86.34 1.75
1996 3,067 185,917,706 63.91 1.41
3-year 9,323 184,411,625 73.78 1.54
* CV is the population coefficient of variation for the sampling
weights (standard deviation / mean) expressed as a percentage
# The variance inflation factor, VIF = 1 + (CV / 100) **2
-- continued
5-3
Table 5-3. Summary of sampling weights included in the
combined CSFII 1994-96, 1998 release -- continued
All ages
Sample Sum of CV* VIF#
size weights
Day 1: 1994
5, 589
259,507,267
65.80
1.43
Attachment 1-109
-------
1995
1996
1998
5,326
5, 188
5, 559
261,950,307
264 , 234,247
40, 134,208
72 . 14
56 . 76
209.16
1.52
1.32
5.37
4-year
21,662
261,897,244
91.40
1. 84
3-year
16,103
261,897,277
64 . 05
1.41
2-day: 1994
1995
1996
1998
5,311
5, 072
4, 920
5,304
259,507,209
261,950,392
264 , 234,191
40,134,206
77 .59
87 . 73
69.11
213 .58
1.60
1. 77
1.48
5 . 56
4-year
20,607
261,897,236
104.52
2 .09
3-year
15,303
261,897,260
77 . 74
1.60
Household
Sample
size
Sum of
weights
CV*
VIF#
4-year
12,364
98,574,787
85.67
1. 73
3-year
8, 067
98 , 574,761
45.88
1.21
* CV is the population coefficient of variation
weights (standard deviation / mean) expressed
for the sampling
as a percentage
# The variance inflation factor,
, VIF = 1 + (CV /
100)**2
Attachment 1-110
-------
5.2 Guidance for Sampling Weights and Reporting
5.2.1 Sampling weight guidance
As noted above, it is generally necessary to use sampling weights
in the analysis of data from surveys having complex designs.
This data release contains a variety of sets of sampling weights
designed to be used in various situations. The choice of which
sampling weight to use was straightforward with the
CSFII/DHKS 1994-96 release. Day 1 weights are used whenever day 1
intakes are analyzed and generally whenever analyzing CSFII data
at the person level. The 2-day weights need to be used when a
subset of the sample is used that is restricted to 2-day
respondents. The 3-year weights are generally used if all 3 years
of data are being analyzed. The annual weights are generally used
if the individual years are analyzed separately. However, results
do not tend to change very much if the annual and 3-year sampling
weights are used interchangeably because sampling procedures and
the target population were the same in each of 1994, 1995, and 1996.
With the CSFII 1998 the situation changes somewhat. Because
only children 9 years old or younger were targeted in 1998 and
relatively few of those children were in the age group 7-9 years,
the weights constructed for use with the CSFII 1998 and the
combined CSFII 1994-96, 1998 sample have several features that
should be noted. Among these features are:
1) The CSFII 1998 weights are more variable than the other
annual weights. This is mainly due to the unequal
distribution of ages in the 1998 sample as seen in Table 5-1.
It should be noted that the weights for a subset of the
CSFII 1998 sample that is more equally distributed across
ages, such as children 1-5 years or children 7-9 years,
are considerably less variable.
2) The combined CSFII 1994-96, 1998 weights are more variable
than the CSFII 1994-96 weights for children 0-9 years old.
This is due to the more variable CSFII 1998 weights and to the
difference in distribution of ages between the two samples.
3) For convenience, there are sampling weights for adults 20
years and older in the set of 4-year weights. These are
exactly the same weights found in the 3-year weight set.
Adults were not sampled in 1998.
4) Although no data was collected for persons 10-19 years in
the CSFII 1998, the 4-year weights for these persons are
slightly different than the 3-year weights. This is
because the final calibration process for the 4-year
weights was done for persons 6-19 years as a group. The
calibration adjustments necessary for the 4-year weights
for 6-19 year olds differed from the adjustments
necessary for the 3-year weights due to the inclusion of
children 6-9 years from CSFII 1998.
5-5
It will be the USDA's convention to use the 4-year combined
CSFII 1994-96, 1998 weights whenever a statistical presentation
uses data from the CSFII 1994-96, 1998 data set and displays
statistics for children 9 years and under. For statistical
presentation of data for persons 10-19, years USDA also
recommends the usage of the 4-year combined weights for the
reasons explained in item (4) above.
Furthermore, the USDA recommends caution in analyzing the
CSFII 1998 by itself. Unlike the annual samples of
CSFII 1994-96, the CSFII 1998 is a supplemental sample,
Attachment 1-111
-------
designed to be merged with the CSFII 1994-96 in order to
increase the overall sample size of children of certain ages.
The CSFII 1998 sampling weights provide some calibration of
the CSFII 1998 sample to the population of 0 to 9 year olds
but the fact that there are proportionately fewer children
7 to 9 years in the sample than children of other ages might
affect analyses of groups that include both children 7 years
or older and younger children.
5.2.2 Reporting guidance
It is the USDA's convention to follow guidelines derived from a
report of the Life Sciences Research Office (FASEB/LSRO 1995)
in identifying or flagging estimates of means, percentages, and
percentiles presented in general reports that might be less
statistically reliable than other estimates due to small cell
size or high relative variability. The guidelines for
determining such estimates take into account the complex sample
design of a survey and the procedures used to weight the data by
specifying the use of a broadly calculated design effect.
The design effect is a measure of the variability introduced
into an estimate by these factors.
Each estimate has a unique design effect. A "broadly
calculated" design effect might be an average of design
effects among related statistics or population groups. For
the convenience of having a single measure of this type of
variability, it is the USDA's convention to use a variance
inflation factor (VIF) in this role in the presentation of
general statistical tables. A VIF is solely a function of the
sampling weights. Variance inflation factors for the
CSFII 1994-96 and CSFII 1998 sampling weight sets are
presented in table 5-3 above.
5-6
Attachment 1-112
-------
Prior to the release of data from the CSFII 1998, the USDA
has used by convention a single VIF, derived from the weights
of individuals of all ages, in the presentation of statistics
from USDA survey data. This convention will be changed for
the analysis of data from the combined CSFII 1994-96, 1998
sample. Whenever a statistical presentation is based on data
for persons under 20 years of age from both CSFII 1994-96 and
CSFII 1998, a VIF based on the 4-year weights on persons 0-19
years will be used in applying the reporting guidelines. If
statistics for adults are also provided, a VIF based on the
weights of persons 20 years and older will be used. If
statistics for persons from both groups are presented, for
example, a table showing statistics for various age groups
including an all-ages group, the VIF for persons 0-19 years
will be used. The VIFs that would be used in such reports are:
Day 1,
0-19 :
2 .24
Day 1,
20+ :
1.36
Day 1,
all:
2 .24
2-day,
0-19 :
2 .50
2-day,
20+ :
1 . 54
2-day,
all:
2 .50
The reporting guidelines generally followed are:
1) An estimated mean is flagged when it is based on a cell
size of less than 30 times the average design effect
(VIF) or when its coefficient of variation (cv) is
equal to or greater than 30 percent. The cv is the ratio
of the estimated standard error of the mean to the
estimated mean, expressed as a percentage. Note that
the cv statistic refereed to here is relative to the
estimate of the mean, hence the use in the numerator of
the standard error rather than the standard deviation as
used in the calculation of the (population) coefficient of
variation shown in Table 5-3.
2) An estimated proportion (percent) that falls above
25 percent and below 75 percent is flagged when it is
based on a cell size of less than 30 times the average
design effect (VIF) or when the cv is equal to or greater
than 3 0 percent.
An estimated proportion of 25 percent or lower or
75 percent or higher is flagged when the smaller of np
and n(l-p) is less than 8 times the average design
effect, where "n" is the cell size on which the estimate
is based and "p" is the proportion expressed as a fraction.
5-7
Attachment 1-113
-------
3) Estimated percentiles are flagged according to rules that
parallel the cell size rules applied to proportions
(guideline 2) . Estimated percentiles inside the 25 to 75
range are flagged when the cell size is less than 30 times
the average design effect. Estimates of the 25 and lower
percentiles are flagged when the cell size is less than
8 times the average design effect divided by p, where p
is the level of the percentile expressed as a fraction.
Estimates of the 75 and higher percentiles are flagged when
the cell size is less than 8 times the average design
effect divided by 1 - p.
5.3 CSFII 1998 (Annual) Sampling Weights
5.3.1 CSFII 1998 weighting design
The approach used in weighting the CSFII 1998 data followed
the approach used in weighting the 1994, 1995, and 1996
person-level data. These annual data sets were weighted
separately in the following steps. First, a base weight
equal to the reciprocal of the probability of selection was
assigned to each sample person. The base weights were then
adjusted for nonresponse within weighting classes defined by
variables that were determined to be correlated with response
rates. Finally, the nonresponse-adjusted weights were ratio
adjusted to population estimates from the March Current
Population Survey (CPS) of the appropriate year
(USDC/BOC 1994, 1995, 1996, 1998) to compensate for random
variation in the observed sample counts and possible
undercoverage of certain groups in the area sample frame.
Two sets of weights were constructed for the CSFII 1998:
a set for sample persons who completed the day-1 interview
and a set for sample persons who provided 2 days of intake.
5-8
Attachment 1-114
-------
5.3.2
Base weights
The base weight associated with a sample person is the
reciprocal of the overall probability of including that
person in the survey. For the CSFII 1998, sample persons
were selected through a complex multistage sample design
involving the selection of primary sampling units (PSUs),
area segments within PSUs, households within segments,
and finally persons (sample persons) within households.
Consequently, the following components were required to
calculate the overall probabilities of selection:
1. The probability of selecting the PSU.
2. The probability of selecting the segment within the PSU.
3. The probability of selecting the household within the
segment.
4. The probability of selecting an eligible sample person
from within the household.
For any sample person, the product of these four factors is
the probability of being selected for the CSFII.
5.3.3 CSFII 1998 nonresponse adjustments
Not all sample persons completed an intake interview. To
compensate for this, the following procedures were used to
adjust the sample person base weights. First the weights
were adjusted for screening nonresponse. These adjustments
were made within classes created by grouping segments by
census region, MSA status, minority status (percent of the
population that was black or Hispanic), and quarter of the year
of field operations. Within each class, the base weight of
each eligible sample person was increased by a factor
corresponding to the screener nonresponse rate within the
class.
These screener nonresponse-adjusted weights were then
adjusted again to account for person nonresponse. A different
set of weighting classes was used for this adjustment.
These classes were defined by income level, age, sex, census
region, MSA status, quarter of the year of field operations,
and minority status of the segment. The result of this step
was a set of nonresponse-adjusted base weights for responding
sample persons.
5-9
Attachment 1-115
-------
5.3.4 CSFII 1998 population adjustments
Lastly, the nonresponse-adjusted weights were calibrated using
an iterative process called "raking ratio weighting" so that
the sum of the final weights equaled the corresponding
1998 March CPS population totals (USDC/BOC 1998) within cells
defined by the following variables:
1. Sex
2. Age group (0-2, 3-5, 6-9)
3. Home ownership
4. Season of intake (winter, spring, summer, fall)
5. Day of week of intake
6. Census region
7. MSA status (metropolitan/nonmetropolitan)
8. Household income as percentage of poverty level (using
the appropriate poverty thresholds)
9. Household received food stamps in past 12 months
10. Number of persons in the household 18 and older
11. Presence in household of children under 6 years
12. Presence in household of children 6 to 17 years
13. Presence of female head of household 40 years or younger
and no one in the household under 18 years
14. Employment status (for children this was the status of
the female head, or if there was no female head, the male
head of household)
15. Race (black or nonblack)
16. Ethnic origin (Hispanic or non-Hispanic)
Table 5-4 shows the adjustments necessary for calibration for
the weighting class age 0-5 years and Table 5-6 shows the same for
the weighting class age 6-9 years. Column 1 provides the number
of children with the various characteristics. Column 2 provides
the weighted percentages of the persons within the weighting class
in each of the categories using the nonresponse-adjusted sampling
weights. Column 3 shows the target percentage from the CPS, which
is also the weighted percentage for the sample using the final,
calibrated weights.
5-10
Attachment 1-116
-------
Table 5-4. Children 5 years and younger: Unweighted sample
sizes, weighted percentage distributions
following nonresponse adjustments, and
population targets, day 1, CSFII 1998
Variable
Sample
size
Nonresponse
adj ustment
Population
targets*
Total
Age/sex
Male
0-2
3-5
Female
0-2
3-5
Home ownership
Home owned
Home not owned
Season of intake
Winter
Spring
Summer
Fall
Day of week of intake
Sunday
Monday
Tuesday
Wednesday
Thursday
Friday
Saturday
Census region
Northeast
Midwest
South
West
Number
5,051
950
1, 573
1,000
1, 528
2,828
2 , 223
1, 166
1,240
1,667
978
961
786
771
638
580
787
528
906
1, 115
1,661
1,369
Percent
100.0 100.0
39.5
21.9
18 .3
20.3
60 .1
39.9
25 . 7
27 .1
29.6
17 . 6
18 . 9
15.8
15 .3
12 . 7
11.2
15.6
10 .5
17 . 9
22 .4
32 .8
27 . 0
25 . 1
26 . 1
23 .8
25 . 0
58 .3
41.7
25 . 0
25 . 0
25 . 0
25 . 0
14 .3
14 .3
14 .3
14 .3
14 .3
14 .3
14 .3
17 .3
23 . 9
34.3
24 . 5
MSA status
MSA (metropolitan) 4,134
Non-MSA 917
80.6
19.4
81.7
18 .3
--continued
5-11
Attachment 1-117
-------
Table 5-4. Continued.
Variable
Sample
size
Nonresponse
adj ustment
Population
targets*
Number
-Percent-
Household income as
percentage of poverty
level
0-75% 895
76-130% 924
131-300% 1,656
Over 300% 1,576
Household received food
stamps in past 12 months
Yes 1,073
No 3,978
Presence in household
of persons 18 and older
Exactly 1 655
Exactly 2 3,656
Other than 1 or 2 740
14 . 9
15 .3
34 . 9
35 . 0
17 .4
82 .6
11 . 7
74 . 5
13 .8
14 . 6
13 .4
34 .8
37.3
17 . 8
82 .2
13 .5
74 . 0
12 . 5
Presence in household
of children 6-17
Children 6-17 2,326
No children 6-17 2,725
44 . 8
55 .2
44 .5
55 .5
Employment status of
female head of household
(or male head if there
is no female head)
Have job 2,644 53.4 58.4
Do not have job 2,407 46.6 41.6
--continued
5-12
Attachment 1-118
-------
Table 5-4. Continued.
Variable
Sample
size
Nonresponse Population
adjustment targets*
Number
Percent
Race
Black
Non-black
749
4,302
86.4
13 .6
84 .2
15.8
Ethnic origin
Hispanic
Non-Hispanic
901
4 , 150
16 .1
83 . 9
17 . 7
82 .3
* Calculated using 1998 Current Population Survey data
(USDC/ BOC 1998) except for the variables "season of
intake" and "day of week of intake." Since the goal
of the CSFII was to estimate behavior on an average
day, each day of the week received an equal value of
14.3 percent, and each season received a value of
25 percent.
5-13
Attachment 1-119
-------
Table 5-5. Persons 6 to 9: Unweighted sample sizes,
weighted percentage distributions following
nonresponse adjustments, and population targets,
day 1, CSFII 1998
Variable
Sample
size
Nonresponse
adj ustment
Population
targets*
Total
Number
508
Percent
100.0 100.0
Sex
Male
Female
279
229
56 .8
43 .2
50 . 9
49.1
Home ownership
Home owned
Home not owned
304
204
66.2
33 .8
65.2
34 .8
Season of intake
Winter 134
Spring 126
Summer 156
Fall 92
Day of week of intake
Weekend (Fri - Sun) 228
Weekday (Mon - Thr) 280
Census region
Northeast 77
Midwest 110
South 187
West 134
23 .5
25 . 9
31.1
19.5
45.4
54 . 6
12 . 6
23 .6
33 .6
30.2
25 . 0
25 . 0
25 . 0
25 . 0
42 . 9
57 .1
18 .6
23 . 0
34 .4
24 . 0
MSA status
MSA (metropolitan) 411
Non-MSA 97
82 .1
17 . 9
80.7
19.3
--continued
5-14
Attachment 1-120
-------
Table 5-5. Continued.
Variable
Sample
size
Nonresponse
adj ustment
Population
targets*
Number
-Percent-
Household income as
percentage of poverty
level
0-75% 101
76-130% 85
131-300% 149
Over 300% 173
16 .1
11. 5
33 . 0
39.4
13 .8
12 .1
35.5
38.5
Household received food
stamps in past 12 months
Yes
No
99
409
17 . 8
82 .2
16 .2
83 .8
Presence in household
of persons 18 and older
Exactly 1 75
Exactly 2 365
Other than 1 or 2 68
12 .6
74 . 6
12 . 8
18 . 0
69.7
12 .3
Presence in household
of children under 6
Children under 6
No children under 6
257
251
45 . 5
54 . 5
41.4
58 .6
Employment status of
female head of household
(or male head if there
is no female
Have job
Do not have job
278
230
54 .3
45 . 7
63 .7
36.3
--continued
5-15
Attachment 1-121
-------
Table 5-5. Continued.
Variable
Sample
size
Nonresponse Population
adjustment targets*
Number
Percent
Race
Black
Non-black
77
431
85.3
14 . 7
16 .4
83 .6
Ethnic origin
Hispanic
Non-Hispanic
427
81
87.3
12 . 7
15 .3
84 . 7
* Calculated using 1998 Current Population Survey data
(USDC/ BOC 1998) except for the variables "season of
intake" and "day of week of intake." Since the goal of
the CSFII was to estimate behavior on an average day,
each day of the week received an equal value of 14.3
percent, and each season received a value of 25 percent.
5-16
Attachment 1-122
-------
5.4 CSFII 1994-96, 1998 (4-Year) Combined Person-Level
Sampling Weights
5.4.1 Introduction to person-level sampling weights discussion
Although the CSFII 1998 was a nationally representative sample
of children 9 years of age and younger, it was primarily
intended to serve as a supplement to the sample of children in
the CSFII 1994-96. A composite estimation approach was used
to combine the CSFII 1994-96 and CSFII 1998 samples. Under
this approach, the combined estimator xcomp, is considered to
be a linear combination of the corresponding CSFII 1994-96 and
CSFII 1998 estimates, i.e.,
xcomp = a * x[94-96] + (1 - a) * x[1998],
where a is a constant between 0 and 1.
Assuming that x[94-96] and x[1998] are both unbiased estimates,
the composite estimate, x[4-year], will also be unbiased for any
value of a. The approximately optimal value of a, i.e., the value
that minimizes the variance of x[4-year], is a function of the
effective sample sizes of the CSFII 1994-96 and the CSFII 1998:
a = eff[94-96] / (eff[94-96] + eff[1998])
where eff[94-96] = n[94-96] / (1 + cv[94-96:w] **2) ,
n [94-96] = the actual CSFII 1994-96 sample size,
cv[94-96:w]**2 = the square of the coefficient of variation
(expressed as a percentage) of the CSFII weights, and
eff[1998] is similarly defined with the CSFII 1998 sample
size and weights.
The factors a and (1 - a) are known as compositing factors and
were computed by sex and age group for the person-level weights.
5.4.2 Day 1 person-level weights
The nonresponse-adjusted day 1 CSFII weights described in
section 5.3.3 were recalibrated to the corresponding 1994-96 CPS
population totals. This was done so that the CSFII weights
would be consistent with the previously computed CSFII 1994-96
(3-year) weights. The recalibration of the CSFII weights was
done separately for (1) children age 5 years or younger and
(2) children 6-9 years of age. The procedures used for
calibration were exactly the same as those described in
section 5.3.4 except that the 1994-96 CPS totals were used as
control totals.
Next, the compositing factors a and (1 - a) were computed using
the CSFII 1994-96 weights and the recalibrated CSFII weights by
sex / age groups. Table 5-6 shows the day 1 compositing factors.
5-17
Attachment 1-123
-------
Next, the CSFII 1998 sample was combined with the CSFII 1994-96
sample by applying the appropriate CSFII compositing factor
(1 - a) to each recalibrated CSFII 1998 day 1 weight and by
applying the appropriate CSFII compositing factor a to each
CSFII 1994-96 day 1 weight. This was done for all children
age 9 years or younger in the combined sample.
Finally, the penultimate combined weights described in the above
paragraph were calibrated one final time to the March 1994-96
CPS totals along the dimensions used in the original calibration
of the CSFII 1994-96 day 1 weights. This final calibration
process was done separately for children 5 years of age and
younger
¦ and
for persons
6 - 19
years of
age.
Table 5
-6 .
Compositing factors
for children age 9
and under
completing
¦ the CSFII day
1 Intake
1994-96
1998
1994-96
1998
Total
compositing
composit:
sample
sample
sample
factor
factor
Sex
Age
size
size
size
(a)
(1 - a)
Male
< 1
184
576
760
0 .22
0 .78
1
362
174
536
0 .70
0.30
2
353
200
553
0 .67
0.33
3
251
687
938
0.28
0 . 72
4
244
670
914
0.27
0 .73
5
246
216
462
0 . 54
0.46
6
125
184
309
0.45
0 . 55
7-9
383
95
478
0 .83
0 .17
Female
< 1
192
599
791
0 .22
0 .78
1
349
199
548
0 .67
0.33
2
352
202
554
0 .67
0.33
3
241
657
898
0.32
0 .68
4
267
679
946
0.30
0 .70
5
229
192
421
0 .59
0.41
6
131
159
290
0 .50
0 .50
7-9
344
70
414
0 .86
0 .14
Total
4 ,253
5, 559
9
,812
5.4.3 Two-day person-level weights
The procedure followed in constructing the day 1 combined
weights was followed in constructing the combined two-day
weights. The two-day CSFII 1998 weights were recalibrated
to the 1994-96 CPS population totals, compositing factors
were computed based on both the recalibrated CSFII 1998 two-day
weights and the CSFII 1994-96 two-day weights by sex and age
groups, penultimate combined weights were created by applying
the appropriate compositing factors to the appropriate weights,
and a final raking procedure was used to calibrate the
penultimate weights.
5-18
Attachment 1-124
-------
5.5 CSFII 1994-96, 1998 (4-Year) Combined Household-Level
Sampling Weights
5.5.1 Introduction to household-level sampling weights discussion
To permit calculation of household-level estimates for items
collected in the household interview (e.g., amount and sources of
income, general information about food shopping practices, the
amounts spent on food, source of cooking and drinking water,
food stamp eligibility), a set of household weights for analysis
of the combined CSFII 1998 and CSFII/DHKS 1994-96 data sets were
computed. The procedures followed in constructing these
household-level weights were similar to those used in
constructing the CSFII 1998 and combined CSFII 1994-96, 1998
person-level weights. First, household-level weights were
constructed for the CSFII 1998 by adjusting a base weight for
nonresponse and then calibrating the nonresponse adjusted weights
to population totals. Secondly, a compositing approach was used
to combine the CSFII 1998 and the CSFII 1994-96. The
construction of the CSFII 1994-96 household weights as documented
for the release of that sampling weight set is included in this
section as section 5.7.1.
5.5.2 CSFII 1998 Household Base Weights
The first step was to assign a base weight to each responding
CSFII 1998 household that is equal to the reciprocal of the
probability of retaining the household for the household
interview. For the CSFII 1998 (and also for the CSFII 1994-96),
only those households with eligible SPs were eligible for the
household interview. Thus, the probability of including a
household in the study was equal to the probability that any of
its members was selected for the intake interviews. Under the
procedures used to select persons for the CSFII 1998, the
probability of selecting a household for the household
interview is equal to maximum probability of selection of the
SPs in the household. Hence, the base weight for the I-th
sampled household was computed from the formula:
w = min {w[1] , w[2] , ..., w [n] },
where w[l], w[2], ..., w[n] are the corresponding base weights
of the SPs in the household. In general, the household base
weights varied by quarter, as well as within quarter depending
on the composition of the household. In particular, households
with children under 1 or 3-4 years of age had considerably
smaller weights (larger probabilities of selection) than
households where the only children were between 7 and 9 years
of age.
5-19
Attachment 1-125
-------
5.5.3 Household-level nonresponse Adjustments
The procedures used for nonresponse adjustments followed those
used in constructing the CSFII 1994-96 household weights and
were essentially as follows. First, the base weights were
adjusted for screening nonresponse within classes defined by
Census region, MSA status, minority status (percent of the
population that was black or Hispanic), and quarter of field
operations. Within each class, the base weight of each
eligible sample person was increased by a factor corresponding
to the screener nonresponse rate within the class.
Next, the screener-adjusted weights were adjusted to account
for household nonresponse. The weighting classes used for
this adjustment were defined by income level, Census region,
MSA status, and minority status of the segment. Note that for
the purpose of weighting, those households that contained at
least one sample person who completed at least one intake
interview were considered to be "respondents" regardless of
whether a household interview was completed.
5.5.4 Household-level population adjustments
Lastly, the nonresponse-adjusted weights were calibrated using
the same iterative process called "raking ratio weighting" used
in calibrating the person-level weights so that the sum of the
final weights equaled the corresponding 1994-96 March CPS
population totals (USDC/BOC 1994, 1995, 1996). Since the
CSFII 1998 was restricted to households with children 9 years
of age or younger (i.e., households without children 9 years
or younger had no chance of selection for the CSFII 1998), the
totals were only for households with children 9 years of age
or younger. Cells defined by the following variables were used:
1. Home ownership and age of the head of household
2. Season of household interview (winter, spring, summer, fall)
3. Day of week of household interview
4. Census region
5. MSA status (metropolitan/nonmetropolitan)
6. Household income as percentage of poverty level (using the
appropriate poverty thresholds)
7. Household received food stamps in past 12 months
8. Presence in household of persons 18 and older
9. Presence in household of children under 6 years
10. Presence in household of children 6 to 17 years
11. Presence of female head of household 40 years or younger
and no one in the household under 18 years
12. Employment status of the head of household
13. Race (black or nonblack) of the head of household
14. Ethnic origin (Hispanic or non-Hispanic) of the head of
household
15. Household size
5-20
Attachment 1-126
-------
5.5.5 Combined CSFII 1994-96, 1998 household samples
The same compositing approach used to combine the person-level
samples was used in combining the household samples.
Compositing factors were computed using the CSFII 1994-96
household weights and the CSFII 1998 household weights by
income / household composition groups. Table 5-7 shows the
household compositing factors. Note that the choice of the
household composition grouping "households with children 7-9
years of age only" followed from the design of the CSFII 1998,
which selected a proportionately small group of 7-9 year olds,
resulting in some large CSFII 1998 household weights for
such households. Using this group for compositing purposes
reduced the impact of these large weights when the samples
were combined.
Next, the CSFII 1998 sample was combined with the
CSFII 1994-96 sample by applying the appropriate CSFII 1998
compositing factor (1 - a) to each CSFII 1998 household weight
and by applying the appropriate CSFII 1994-96 compositing
factor a to each CSFII 1994-96 household weight.
Finally, these penultimate combined weights were calibrated one
final time to the March 1994-96 CPS totals along the dimensions
specified above. Unlike the calibration of the
CSFII 1998-only household sample, this time the population
totals represented all U.S. households. The same cells listed
in section 5.5.4 were used.
Table 5-7. Compositing factors for CSFII households with
children 9 or younger
1994-96 1998 Total
Income HH sample sample sample
group comp. size# size size
1994-96
compositing
factor
(a)
1998
compositing
factor
(1 - a)
>= 130%
poverty
Total
Children
7-9 only
Others
< 130% Children
poverty 7-9 only
Others
299
1, 71
99
50
349
2,813 4,609
12 111
787 1,422 2,209
2,981 4,297 7,278
0 .88
0 .51
0 . 90
0.48
0 .12
0.49
0 .10
0 .52
# Households with at least one SP who completed the day 1 Intake.
5-21
Attachment 1-127
-------
5.6 Variance Estimation
5.6.1 Variance estimation fields
As described in Section 3, "Sample Design," Westat's 62
primary sampling unit (PSU) master sample was employed for both
the CSFII/DHKS 1994-96 and the CSFII 1998. This sample of PSUs
contains 24 PSUs selected with certainty. The remaining 38 PSUs
were selected with probability proportional to size from 38 strata,
1 PSU per stratum. Area segments were then selected from each of
the 62 PSUs also with probability proportional to size. The area
segments were randomly allocated to the annual samples, across
quarters of the year, so that the 62 PSUs were fielded at all times
throughout each year. The following approach was used to create a
framework of 2 sampling units per stratum to facilitate variance
estimation procedures. First, 19 variance estimation strata were
formed from the 38 noncertainty PSUs by pairing adjacent PSUs in
the sampling frame. Each PSU within a variance estimation stratum
defines what is referred to as a variance estimation unit. Next,
within each of the 24 certainty PSUs, one-half of the segments were
assigned to one variance estimation unit and the remaining one-half
to another. Because each certainty PSU is considered to be a
separate variance estimation stratum, a total of 43 variance
estimation strata (each containing 2 variance estimation units)
was formed by this process. See section 7.4.2, "Sampling weights
and variance estimation fields," for details of identifying these
variance estimation fields in the data set. This framework
applies to all weighted samples, annual or combined, of the
CSFII 1994-96 and CSFII 1998.
5.6.2 Estimation of Sampling Errors
Linearization method
Estimation of sampling errors may be conducted with a Taylor
series linearization method using the final sampling weights
described in the above sections along with the variance
estimation strata and variance estimation units described in
section 5.6.1. Software packages such as SUDAAN and Stata can
be used to obtain estimates using the linearization method.
5-22
Attachment 1-128
-------
Jackknife replicate method
Alternatively, sampling errors may be estimated using the
jackknife technique described here. The construction of
jackknife replicate weights makes use of the variance
estimation stratum/variance estimation unit structure described
above in section 5.5.1. As an illustration of how a jackknife
variance estimator can be calculated, let y denote a weighted
survey estimate (for example, total fat intake) calculated using
the full-sample weights. Let y(j) be the corresponding weighted
estimate calculated using the j-th set of replicate weights
(j = 1, 2, ..., 43). The estimated variance of y is then given
by the formula
Var(y) = SUM (y(j) - y)**2,
where the summation extends over the 43 sets of jackknife
replicate weights. Forty-three replicates were created by
applying this process to each of the 43 variance estimation
strata.
A jackknife replicate is created by dropping out one of the two
variance estimation units from a variance estimation stratum and
doubling the initial probability weights of the individuals in
the other variance estimation unit in that stratum. The entire
weighting process as described in the previous sections was
repeated for each replicate. Individuals who were not in the
current replicate subsample were assigned a corresponding
replicate weight of zero. In this way, series of replicate
weights were generated for each sample person or household.
Together with the final, full-sample weights, these replicate
weights were designed for the calculation of sampling errors.
Using a replication method to calculate sampling errors of
survey-based estimates makes complicated variance estimation
formulas unnecessary. The jackknife replication method used
here is also designed to reflect the stratification and
clustering used in the CSFII/DHKS sample design and to capture
the effects of the raking ratio adjustments mentioned in the
sections above.
5-23
Attachment 1-129
-------
Replicate weights are provided for use with each of the sets of
sampling weights listed in Table 5-2. There are seven files
altogether, found in the \jacknife directory on disk 2:
j kw4yrcs.dat
j kwanncs.dat
j kw3yrcs.dat
j kw4yrhh.dat
j kw3yrhh.dat
j kwanndh.dat
j kw3yrdh.dat
Day 1 and two-day weights for the combined
CSFII 1994-96, 1998 (4-year) sample
Day 1 and two-day weights for annual samples
(1994, 1995, 1996, 1998)
Day 1 and two-day weights for the CSFII 1994-96
combined (3-year) sample
Household weights for the combined
CSFII 1994-96, 1998 (4-year) sample
Household weights for the CSFII 1994-96
combined (3-year) sample
DHKS and two-day DHKS weights for the annual
samples (1994, 1995, 1996)
DHKS and two-day DHKS weights for the
DHKS 1994-96 combined (3-year) sample
Corresponding file formats are provided in section 9.4 and SAS
programs for reading the data files are provided in section 10.4.
The annual and 4-year person-level files each contain one record
per CSFII respondent (21,662 records total, 5,559 from CSFII 1998).
The 3-year person-level file contains one record per CSFII
respondent from 1994-96 (16,103 records). The 2-day weight fields
are blank for respondents providing only one day of intake.
The DHKS files each contain one record per DHKS respondent in
1994-96 (5,765 records). The 2-day DHKS weight fields are blank
for DHKS respondents who did not provide a second day of intake.
The 4-year household-level file contains one record per CSFII
household (12,364 records total, 4,297 from the CSFII 1998). The
3-year household-level file contains one record per CSFII household
from 1994-96 (8,067 records). A field indicating the survey
year, the full-sample sampling weights, and the variance-estimation
stratum and unit are included in each file.
The replicate weighting process described above was designed and
implemented by Westat, Inc., who have also created a variance
estimation program, WesVarPC, which runs on computers using the
Windows operating system and is available to the public at no
charge. A commercial version, WesVar Complex Samples, is also
available from SPSS. Information about both programs may be
found at Westat's home page at . Note that
in WesVarPC terminology, the JK2 method was used in constructing
these replicate weights.
5-24
Attachment 1-130
-------
5.7 CSFII 1994-96 (3-Year) Household Sampling Weights:
Original Documentation
These weights permit household level estimates using the
fields that are present on household record type 15. The
data contained in the record type 15 fields include household
participation in programs such as WIC and Food Stamps, income
and food-related expenditures, and food sufficiency. The
3-year weights, calibrated to 3-year averages of population
characteristics, are intended to be used with the 3-year
CSFII data set. They may be used with the annual subsets,
however, as long as it is understood that the annual subsets
were not calibrated to annual population characteristics.
If annual totals are being estimated, the weights should each
be multiplied by 3 to scale the weights appropriately. Such
scaling is not necessary for the estimation of means or
percentages.
5.7.1 How the 3-year household sampling weights were constructed
In general, the analysis of data from surveys having complex
designs requires the use of sample weights to compensate for
variable probabilities of selection, differential nonresponse
rates, and possible deficiencies in the sampling frame. For
the 1994-96 CSFII/DHKS, the overall probabilities of
selecting sample persons were designed to vary by sex, age,
and income level to meet precision goals specified by ARS.
For this reason, the probability of selecting a household
into the sample is directly related to the composition of the
household at the time of screening. The construction of
household sample weights was performed by ARS using the
design developed by Westat, Inc.
Weighting design
The 3-year CSFII households were weighted in the following
steps. First, a base weight equal to the reciprocal of the
probability of selection was assigned to each household. The
base weights were then adjusted for nonresponse at two levels
within weighting classes defined by variables that were
determined to be correlated with response rates. The first
was a screener-level adjustment using 57 classes defined by
combinations of region, quarter, MSA status, and minority
status of the segment. The second was a household level
adjustment using 8 classes defined by combinations of region,
MSA status, minority status of segment and household income
as a percentage of poverty. Finally, to compensate for
random variation in the observed sample counts and possible
undercoverage of some groups, the nonresponse-adjusted
weights were ratio adjusted to the average population
estimates from the March Current Population Surveys for
1994, 1995, and 1996.
5-25
Attachment 1-131
-------
Base weights
The base weight associated with a household is the
reciprocal of the overall probability of including that
household in the survey. For each year of the CSFII/DHKS,
sample households were selected through a complex multistage
sample design involving the selection of primary sampling
units (PSUs), area segments within PSUs, and households
within segments. The eligibility of households for the CSFII
was determined by household income level and the sex and age
of its members at the time of screening. The product of
steps 1, 2, and 3 below determines the probability of
selection for eligible households. Since segments were
allocated for selection over the 3 years of the survey, a
factor of 3 is included in probability of selecting area
segments. As with the individual weights, the reciprocal of
this probability is the household base weight.
1. The probability of selecting the PSU.
2. The probability of selecting the segment within
the PSU.
3. The probability of selecting the household within
the segment.
CSFII nonresponse adjustments
Not all households completed the household interview but
all households had a member to provide a Day-1 intake.
Those households that did not provide a household
interview are included in the nonresponse adjustment as
participating households. This was done because household
eligibility and participation were determined by the
presence and participation of a specific household member.
There were 41 households where a Day-1 intake was
completed but the household questionnaire was not. In
these cases most of the household information is missing
or was imputed on record type 15. Otherwise, to
compensate for nonresponse, the following procedures were
used to adjust the household base weights.
5-26
Attachment 1-132
-------
First the weight was adjusted for screening nonresponse.
This adjustment was made within classes created by
grouping households by census region (see 1994-96 CD-ROM
documentation "Region" in section 14, "Glossary"), MSA
status (see 1994-96 CD-ROM documentation "Metropolitan
Statistical Area" in section 14, "Glossary"), minority
status of area segment (high or low minority) and
quarter of field operations. Within each class, the
base weight of each eligible household was increased by
a factor equal to the inverse of the screening rate
within the class. This adjustment is the same screener
adjustment made in constructing the individual
sample weights.
The screener nonresponse-adjusted weight was then
adjusted again to account for household nonresponse.
A different set of weighting classes was used for this
adjustment. A CHAID analysis was performed by ARS to
determine the groupings for the household level
nonresponse adjustments. The new classes were
defined by income level, census region, MSA status,
and minority status of the segment. Only those
households which had eligible sample persons but did
not complete any day 1 intakes were considered
nonresponding. As in the screener nonresponse
adjusted weight, this adjustment is equal to the
inverse of the household response rate within the
classes. The result of this step was a set of
nonresponse-adjusted base weights for responding
households. The nonresponse-adjusted base weight
(WT_H_ADJ) is included in the weight file.
Post-stratification and population adjustments
Finally, the nonresponse-adjusted weights were
calibrated using an iterative process called "raking
ratio weighting" to produce final weights that sum
to the average of population totals over the 3-year
period of the CSFII/DHKS. The totals are from the
March (1994, 1995, and 1996) Current Population
Surveys (CPS). The cells used to define the totals
were generally the same as those used for the
individual weight reflecting household totals. Day
and quarter of intake were replaced by day and
quarter of the household interview. Household size
was added. Race and ethnic origin variables are
based on characteristics of the female head of
household when present; otherwise, the male head of
household.
5-27
Attachment 1-133
-------
1. Home ownership and age of the head of household
2. Season of household interview (winter, spring,
summer, fall)
3. Day of week of household interview
4. Census region
5. MSA status (metropolitan/nonmetropolitan)
6. Household income as percentage of poverty level
(using the appropriate poverty thresholds)
7. Household received food stamps in past 12 months
8. Presence in household of persons 18 and older
9. Presence in household of children under 6 years
10. Presence in household of children 6 to 17 years
11. Presence of female head of household 40 years or
younger and no one in the household under 18 years
12. Employment status of the head of household
13. Race (black or nonblack) of the head of household
14. Ethnic origin (Hispanic or non-Hispanic) of the
head of household
15. Household size
To illustrate the adjustments, table 5-8 shows, by
weighting variable, the 1994-96 CSFII unweighted sample
sizes, the weighted percentage distributions following
nonresponse adjustments (but before calibration to
population targets), and the population targets for all
responding households.
5-28
Attachment 1-134
-------
Table 5-8. Unweighted household sample sizes, weighted
percentage distributions following nonresponse
adjustments, and population targets, CSFII 1994-96
Variable
Sample
size
Nonresponse
adj ustment
Population
targets*
Total
Home ownership/age
Home owned
20-39
40-59
60 and older
Home not owned
20-39
40-59
60 and older
Number
8, 067
1, 813
1,909
1, 598
1, 623
660
464
Percent- --
100.0 100.0
22 . 7
24 . 0
19.8
19 . 9
8 . 0
5.6
19.3
25 . 9
19.6
20.7
8 . 9
5.6
Season of interview
Winter 1,943
Spring 2,122
Summer 1,988
Fall 2,014
Day of week of interview
Sunday 952
Monday 1,348
Tuesday 1,246
Wednesday 1,226
Thursday 968
Friday 919
Saturday 1,408
Census region
Northeast 1,499
Midwest 1,958
South 2,866
West 1,744
MSA status
MSA (metropolitan) 6,092
Non-MSA 1,975
24 .3
26.2
24 . 7
24 . 8
11 . 8
16 .6
15 . 5
15 .3
12 . 1
11.2
17 . 5
19.3
24 . 1
34 . 7
21.9
76 .2
23 .8
25 . 0
25 . 0
25 . 0
25 . 0
14 .3
14 .3
14 .3
14 .3
14 .3
14 .3
14 .3
19 . 9
23 . 9
35 .1
21.1
78 .8
21.2
¦- continued
5-29
Attachment 1-135
-------
Table 5-S
Continued.
Variable
Sample
size
Nonresponse
adj ustment
Population
targets*
Number
- Percent -
Household income as
percentage of poverty
level
0-75% 888
76-130% 1,156
131-300% 2,665
Over 300% 3,358
Household received food
stamps in past 12 months
Yes 1,011
No 7,056
Presence in household
of persons 18 and older
Exactly 1 2,019
Exactly 2 4,832
Other than 1 or 2 1,216
Presence in household
of children under 6
and 6-17
Children under 6
Children 6-17 1,128
No children 6-17 1,329
No children under 6
Children 6-17 1,380
No children 6-17 4,230
10.3
13 .2
33 .8
42 . 8
11. 7
88.3
24 . 5
60.2
15 .2
13 . 9
16 .6
17 . 0
52 . 5
8.4
10 . 9
31.8
48 . 9
9 . 0
91. 0
31.3
54 .2
14 . 5
8.5
9.5
19 . 9
62 .1
Presence of female head
of household 40 or
younger and no one in
household under 18
Yes 449
No 7,618
Employment status
Have job 4,3 55
Do not have job 3,712
5 .7
94 .3
54 . 6
45.4
9.7
90.3
57 . 7
42 .3
¦- continued
5-30
Attachment 1-136
-------
Table 5-8. Continued.
Sample
Nonresponse
Population
Variable
size
adj ustment
targets*
Number
Percent-
Race
Black
993
12 .1
11.6
Non-black
7, 074
87 .1
88.4
Ethnic origin
Hispanic
755
9 . 0
8 . 0
Non-Hispanic
7,312
91. 0
92 . 0
Household size
1 Member
1,464
17 . 9
24 . 8
2 Members
2,429
30.3
32.3
3 or more members
4 , 174
51.8
42 . 9
* Calculated using 1994-96 Current Population Survey data
except for the variables "season of interview" and "day of
week of household interview." Since the goal of the CSFII
was to estimate behavior on an average day, each day of the
week received an equal value of 14.3 percent, and each
season received a value of 25 percent.
5-31
Attachment 1-137
-------
5.7.2 Use of the household sampling weights
The household sample in the 1994-96 CSFII consists of all
households where at least one sample person was selected
and provided a Day-1 intake. This is true regardless of
whether a household questionnaire was completed. The use
of the weights should be restricted to household
information only (record type 15). No connections to
sample persons or their intakes should be assumed in
using the household weights.
Summary of final household weights
Table 5-9 summarizes the set of final household weights.
The table shows the sample size; the sum of the weights;
the coefficient of variation of the weights (CV), defined
as the ratio of the standard deviation of the weights to
the mean of the weights expressed as a percentage; and
the variance inflation factor (VIF), defined as
1 + (CV/100)**2. This last statistic, which is
equivalent to the ratio of the mean of the squared weights
to the square of the mean of the weights, represents the
anticipated proportional increase in the variance of
survey estimates resulting from the variation in the
weights. For example, it is anticipated that the variance
of a household estimate will be 1.2 times what it would
have been had all the weights been equal. The VIF can be
used in the role of the "broadly calculated average design
effect" specified in reporting guidelines adopted by ARS
(FASEB/LSRO 1995).
Table 5-9. Summary of final household sample weights
Sample
Sum of
CV
VIF =
size
weights
1+(CV/100)**2
8, 067
98,574,761
45.88%
1.21
5-32
Attachment 1-138
-------
Variance Estimation Fields
As described in CSFII/DHKS 1994-96 documentation
(USDA 1998) section 3.2.1, "Sample design,"
Westat's 62 primary sampling unit (PSU) master sample
was employed for CSFII/DHKS 1994-96. This sample of
PSUs contains 24 PSUs selected with certainty. The
remaining 38 PSUs were selected with probability
proportional to size from 38 strata, 1 PSU per stratum.
Thirty-six area segments were then selected from each
of the 62 PSUs also with probability proportional to
size. The thirty-six area segments were randomly
allocated to the annual samples, twelve per year and
three per quarter, so that the 62 PSUs were fielded at
all times throughout the three years.
The following approach was used to create a framework
of 2 sampling units per stratum to facilitate variance
estimation procedures. First, 19 variance estimation
strata were formed from the 38 noncertainty PSUs by
pairing adjacent PSUs in the sampling frame. Each PSU
within a variance estimation stratum defines what is
referred to as a variance estimation unit. Next,
within each of the 24 certainty PSUs, one-half of the
segments were assigned to one variance estimation unit
and the remaining one-half to another. Because each
certainty PSU is considered to be a separate variance
estimation stratum, a total of 43 variance estimation
strata (each containing 2 variance estimation units)
was formed by this process. See CSFII/DHKS 1994-96
documentation (USDA 1998) section 7.4.2, "Sampling
weights and variance estimation fields," for details
on identifying these variance estimation fields in the
data set.
Estimation of Sampling Errors - Linearization method
Estimation of sampling errors may be conducted with a
Taylor series linearization method using the final
sample weights described in CSFII/DHKS 1994-96
documentation (USDA 1998) sections 5.1.2 and 5.1.3 along
with the variance estimation strata and variance
estimation units described in section 5.1.4. Software
packages such as SUDAAN and Stata can be used to
obtain estimates using the linearization method.
5-33
Attachment 1-139
-------
Estimation of Sampling Errors - Jackknife replicate method
Alternatively, sampling errors may be estimated using the
jackknife technique described here. The construction of
jackknife replicate weights makes use of the variance
estimation stratum/variance estimation unit structure
described in CSFII/DHKS 1994-96 documentation (USDA 1998)
section 5.1.4. To illustrate how a jackknife variance
estimator can be calculated, let y denote a weighted
survey estimate (for example, total fat intake) calculated
using the full-sample weights. Let y(j) be the corresponding
weighted estimate calculated using the j-th set of replicate
weights (j = 1, 2, .,43). The estimated variance of y is
then given by the formula
Var(y) = SUM (y(j) - y)**2,
where the summation extends over the 43 sets of jackknife
replicate weights. Forty-three replicates were created by
applying this process to each of the 43 variance estimation
strata.
A jackknife replicate is created by dropping out one of
the two variance estimation units from a variance
estimation stratum and doubling the initial probability
weights of the households in the other variance estimation
unit in that stratum. The entire weighting process as
described in the previous sections of this document was
repeated for each replicate. Households not in the current
replicate subsample were assigned a corresponding replicate
weight of zero. In this way, a series of replicate weights
was generated for each household. Together with the final,
full-sample weights, these replicate weights were designed
for the calculation of sampling errors.
Using a replication method to calculate sampling errors
of survey-based estimates makes complicated variance
estimation formulas unnecessary. The jackknife
replication method used here is also designed to reflect
the stratification and clustering used in the CSFII/DHKS
sample design and to capture the effects of the raking
ratio adjustments mentioned in CSFII/DHKS 1994-96
documentation section 5.1.2.5.
The replicate weighting process described above was
designed and implemented by Westat, Inc., who has also
created a variance estimation program, WesVarPC, which
runs on computers using the Windows operating system
and is available to the public at no charge. The
software can be downloaded from Westat's home page at
. In WesVarPC terminology, the
JK2 method was used in constructing these replicate weights.
5-34
Attachment 1-140
-------
Programs and examples of output
Note: The following programs were written to accompany the
release of the CSFII 1994-96 household sampling weights,
not this 1994-96, 1998 release. The main purpose of two of
the programs was to merge the final household sampling weights
into an existing file derived from household record type 15.
That merging process is not necessary with this release
because both the 3-year and 4-year final household sampling
weights have been included on record type 15. Also note that
the 3-year household jackknife replicate weight file,
jkw3yrhh.dat, has a different format than the file read in
by the following program. Appropriate input programs for both
the survey data files and the replicate weight files from this
release may be found in section 10. Nevertheless, these input
programs and programming examples from the original
documentation may be useful to users of household level data.
The following are three SAS programs used to prepare data
files and three examples of using the household data and
sampling weights. Program 1 is a SAS program that reads
the ASCII household weight file (hhwgt.dat) and creates a
SAS system file. Program 2 adds the household sampling
weight to an existing CSFII household-level SAS system
file. Program 3 is a modified version of Program 2 that
prepares a data file used as input by two of the examples.
Example 1 is a simple SAS program that produces weighted
percentages of selected household variables. Examples 2
and 3 demonstrate the use of SUDAAN and WesVarPC for the
estimation of standard errors of percentages. Example 2
is a SUDAAN program; Example 3 describes the preparation
procedure necessary for using WesVarPC with CSFII data.
The latter two examples examine household income as a
percentage of poverty level (POVCAT) and the adequacy of
the food supply of the household (FOODDESC). Both
variables are modified by program 3 to create
two-category variables. Levels 1 and 2 of FOODDESC have
been combined to identify those households where "enough
food eaten" was reported. Levels 3 and 4 have been
combined to identify households where "not enough food
eaten" was reported. Households with other values of
FOODDESC are not represented in the analysis. Levels
2 and 3 of POVCAT have been combined for those households
that have income over 13 0% of poverty. Level 1
represents those households that are below 131% of
poverty (see documentation section 3.5, "Glossary").
5-35
Attachment 1-141
-------
* programl.sas
*
* This SAS program reads the entire household weight
* file and creates the SAS system file, HHWGT,
* containing the same data.
*
* These programs assume that the directory \data9496
* holds all CSFII SAS files as well as the downloaded
* ascii file containing the household sampling
* weights. The LIBNAME and FILENAME statements
* should be modified as appropriate.
*
•kieieieieieieieieieieieieieieieieieieieieieieieieieieieieieieieieieieieieieieieieieieieieieieieieieieieie
options Is = 78 ps = 55;
libname dir9496 'C:\data94 96';
filename hhwgt 'C:\data9496\hhwgt.dat';
data dir9496.hhwgt (compress = 'yes');
infile hhwgt lrecl = 386;
input hhid 1-5 wt3_hh 8-15 wt_h_adj 16-23
(r3_hh_01-r3_hh_43) (43 * 8.)
hh_bwt 368-375 wt_h_scr 376-383 varstrat 384-
varunit 386;
hhid
=
"Household
ID"
wt3
_hh
"Full-sample household weight"
wt
_h_adj =
"Non-response adjusted base
weight
r3_
hh
01 =
"Replicate
household
weight
- 1"
r3_
"hh
02 =
"Replicate
household
weight
- 2"
r3_
"hh
03 =
"Replicate
household
weight
- 3"
r3_
"hh
04 =
"Replicate
household
weight
- 4"
r3_
"hh
05 =
"Replicate
household
weight
- 5"
r3_
"hh
06 =
"Replicate
household
weight
- 6"
r3_
"hh
07 =
"Replicate
household
weight
- 7"
r3_
"hh
08 =
"Replicate
household
weight
- 8"
r3_
"hh
09 =
"Replicate
household
weight
- 9"
r3_
"hh
10 =
"Replicate
household
weight
- 10"
r3_
"hh
11 =
"Replicate
household
weight
- 11"
r3_
"hh
12 =
"Replicate
household
weight
- 12"
r3_
"hh
13 =
"Replicate
household
weight
- 13"
r3_
"hh
14 =
"Replicate
household
weight
- 14"
r3_
"hh
15 =
"Replicate
household
weight
- 15"
r3_
"hh
16 =
"Replicate
household
weight
- 16"
r3_
"hh
17 =
"Replicate
household
weight
- 17"
r3_
"hh
18 =
"Replicate
household
weight
- 18"
r3_
"hh
19 =
"Replicate
household
weight
- 19"
r3_
"hh
20 =
"Replicate
household
weight
- 20"
r3_
"hh
21 =
"Replicate
household
weight
- 21"
r3_
"hh
22 =
"Replicate
household
weight
- 22"
5-36
Attachment 1-142
-------
r3
hh
23 =
"Replicate
household
weight
23"
r3
hh
"24 =
"Replicate
household
weight
24"
r3
hh
"25 =
"Replicate
household
weight
25"
r3
hh
"26 =
"Replicate
household
weight
26"
r3
hh
"2 7 =
"Replicate
household
weight
27"
r3
hh
"28 =
"Replicate
household
weight
28"
r3
hh
"2 9 =
"Replicate
household
weight
29"
r3
hh
"3 0 =
"Replicate
household
weight
30"
r3
hh
"31 =
"Replicate
household
weight
31"
r3
hh
"32 =
"Replicate
household
weight
32"
r3
hh
"33 =
"Replicate
household
weight
33"
r3
hh
"34 =
"Replicate
household
weight
34"
r3
hh
"3 5 =
"Replicate
household
weight
35"
r3
hh
"3 6 =
"Replicate
household
weight
36"
r3
hh
"3 7 =
"Replicate
household
weight
37"
r3
hh
"3 8 =
"Replicate
household
weight
38"
r3
hh
"3 9 =
"Replicate
household
weight
3 9"
r3
hh
"4 0 =
"Replicate
household
weight
40"
r3
hh
"41 =
"Replicate
household
weight
41"
r3
hh
"42 =
"Replicate
household
weight
42"
r3
_hh_
"43 =
"Replicate
household
weight
43"
hh
bwt
"Household
base weight"
wt
h scr =
"Screener
adjusted household
base weight
varstrat =
"Variance
strata"
varunit
"Variance
estimation
unit"
proc means;
run;
************* End Of PITOCJITclTTl 1 ************************ ~
5-37
Attachment 1-143
-------
* program2.sas
*
* This SAS program adds the household sampling
* weight, WT3_HH, to an existing household-level SAS
* file such as the file created from record type 15
* by the READ15.SAS program on the 1994-96 CD-ROM.
* The file created by PR0GRAM1.SAS supplies the
* sampling weight.
*
* These programs assume that the directory \data9496
* holds all CSFII SAS files. The LIBNAME statement
* should be modified as appropriate.
*
•kieieieieieieieieieieieieieieieieieieieieieieieieieieieieieieieieieieieieieieieieieieieieieieieieieieie
options Is = 78 ps = 55;
libname dir9496 'C:\data94 96';
•kieieieieieieieieieieieieieieieieieieieieieieieieieieieieieieieieieieieieieieieieieieieieieie
* *
* Delete or modify the KEEP option in the *
* following statement to add the replicate *
* sampling weights to the RT15 file. The *
* replicate weights are required if software *
* such as WESVAR, utilizing a replication *
* method, is used for variance estimation *
* (see example 3). *
* *
•kieieieieieieieieieieieieieieieieieieieieieieieieieieieieieieieieieieieieieieieieieieieieieie .
f
data dir9496.rtl5 (compress = 'yes');
merge dir9496.rtl5
dir9496.hhwgt (keep = hhid wt3_hh);
by hhid;
proc means;
run;
•kieieieieieieieieieieieie End Of P IT OCJ1T 3. TTl 2
5-38
Attachment 1-144
-------
* program3.sas
*
* This SAS program prepares an input file for the two
* variance estimation programs, examples 2 and 3. It is
* used for three purposes. The first purpose is to create
* a PC SAS file in an older than current version, version
* 6.04, that both PC-based SUDAAN and Wesvar can read
* directly. Secondly, the replicate weights are collected
* from the file created by PROGRAM1.SAS. Thirdly, the
* variables used for analysis in example programs are
* created. Only the variables needed for the examples are
* retained.
*
***********************************************************
options Is = 78 ps = 55;
libname dir9496 'C:\data94 96';
libname dir2 v604 'c:\data94 96';
data dir2.pgm3 (keep = hhid wt3_hh varstrat varunit
r3_hh_01-r3_hh_43 underl31 enough);
merge dir9496.rtl5 (keep = hhid povcat fooddesc wt3_hh
varstrat varunit)
dir9496.hhwgt (keep = hhid r3_hh_01- r3_hh_43);
by hhid;
if fooddesc in(l, 2) then
enough = 1;
else if fooddesc in (3, 4) then
enough = 2;
if (povcat eq 1)
under131 = 1;
else
under131 = 2;
label underl31 =
enough
proc means;
run;
************* End Of PITOCJ1T3.TTl 3 ************************ ~
then
'Income status'
'Enough to eat'
5-39
Attachment 1-145
-------
* examplel.sas
*
* This SAS program produces weighted frequencies of
* several household level variables. The input file
* is the file created by PR0GRAM2.SAS
*
* These programs assume that the directory \data9496
* holds all CSFII SAS files. The LIBNAME statement
* should be modified as appropriate.
*
•kieieieieieieieieieieieieieieieieieieieieieieieieieieieieieieieieieieieieieieieieieieieieieieieieieieieie
options Is = 78 ps = 60;
options nodate nonumber nocenter;
libname dir9496 'C:\data94 96';
proc freq data = dir9496,rtl5;
tables povcat fooddesc fs_rcvl2 urb region;
weight wt3_hh;
format povcat povcat. fooddesc fooddesc.
fs_rcvl2 yn789f. urb urb. region region.;
title 'Example 1: Weighted frequencies of household
'level data, 1994-96 CSFII';
run;
************* End of Example 1 program **************
5-40
Attachment 1-146
-------
Example 1: Weighted frequencies of household level data,
1994-96 CSFII
Annual income: % of poverty category
POVCAT
Frequency
Percent
Cumulative Cumulative
Frequency Percent
0 - 130%
131 - 350%
Over 3 50%
19520256
39468942
39585563
19.8
40 . 0
40.2
19520256
58989198
98574761
19.8
59.8
100 . 0
Description of food eaten in HH
FOODDESC
Frequency
Percent
Cumulative
Frequency
Cumulative
Percent
Enough - 1 75210037 76.3 75210037
Enough - 2 20344690 20.6 95554727
Sometimes not enough 1803550 1.8 97358277
Often not enough 336251 0.3 97694528
Not ascertained 880233 0.9 98574761
76 .3
96 . 9
98 .8
99 .1
100 . 0
Food stamps: in last 12 months
Cumulative
Cumulative
FS_RCV12
Frequency
Percent
Frequency
Percent
Yes
8693044
8.8
8693044
8.8
No
88663791
89.9
97356835
98.8
Refused
237994
0.2
97594829
99 . 0
Don't know
102999
0.1
97697828
99 .1
Not ascertained
876933
0.9
98574761
100 . 0
Urbanization
URB
Frequency
Percent
Cumulative Cumulative
Frequency Percent
MSA, central city 31977978 32.4 31977978 32.4
MSA, not central city 45717307 46.4 77695285 78.8
Non-MSA 20879476 21.2 98574761 100.0
Region
REGION
Frequency
Percent
Cumulative Cumulative
Frequency Percent
Northeast
19586188
19.9 19586188
19 . 9
Midwest
23591612
23.9 43177800
43 .8
South
34604507
35.1 77782307
78 . 9
West
20792454
21.1 98574761
100 . 0
•kieieieieieieieieieieieie End Of
Example 1 output listing
•kieieieieieieieie .
t
5-41
/•kieieieieieieieieieieieieieieieieieieieieieieieieieieieieieieieieieieieieieieieieieieieieieieieieieieieieieieieieieieieieieieie
*
¦k
Example2.pre
* This SUDAAN program provides an example of computing the
* standard error of estimates from the CSFII 1994- 96. SUDAAN
* is a program containing procedures designed to be used to
* analyze data from complex sample surveys such as the CSFII.
*
* This program was written to be used by the stand- alone
* version of SUDAAN. The input file is the SAS system file
Attachment 1-147
-------
* created by PR0GRAM2.SAS which created a version 6.04 PC SAS
* system file. This program provides the basic statements
* needed to inform SUDAAN of the CSFII sample design
* information needed for the estimates.
*
* The SUDAAN procedure used here is PROC CROSSTAB. The
* procedure call specifies a "with replacement" design
* (design = wr). A NEST statement is used to define the
* required design parameters, VARSTRAT, the variance-estimation
* stratum, and VARUNIT, the variance-estimation unit which is
* used as a primary sampling unit or PSU.
*
* Notes: The data directory must be set to the directory
* containing the input file. Also, a LEVEL.DBS may be
* placed in that directory to supply variable formats.
*
•kieieieieieieieieieieieieieieieieieieieieieieieieieieieieieieieieieieieieieieieieieieieieieieieieieieieieieieieieieieieieieieie
proc crosstab data = pgm3 filetype = sas design = wr;
nest varstrat varunit;
weight wt3_hh;
subgroup underl31 enough;
levels 2 2;
tables underl31 * enough;
print nsum rowper colper serow secol /style = nchs ;
************* Enci 0f Example 2 program **************** ;
5-42
Attachment 1-148
-------
Example 2 output listing
Research Triangle Institute
The CROSSTAB Procedure
by: Income status, Enough to eat.
Income status
Enough to eat Sample Row Col SE Row SE Col
Size Percent Percent Percent Percent
Total
Total
8007
100.00
100.00
0 .00
O
O
o
Enough
7791
97.81
100.00
0 .19
0 .00
Not enough
216
2 .19
100.00
0 .19
0 .00
Below 131% of
poverty
Total
2083
100.00
19 . 90
0 .00
0 . 96
Enough
1908
91.80
18 .68
0 .71
0 . 93
Not enough
175
8.20
74 . 54
0 .71
3 .19
131% +
Total
5924
100.00
80 .10
0 .00
0 . 96
Enough
5883
99.30
81.32
0 .11
0 . 93
Not enough
41
0 .70
25.46
0 .11
3 .19
•kieieieieieieieieieieieie End
of Example
2 output
listing
•kieieieieieieieie .
t
5-43
Attachment 1-149
-------
Example3
Example 3 used the WesVarPC software to estimate percentages
and their standard errors. The file created by PR0GRAM2.SAS
provided the input. During the preparation step, the SAS
file PGM3.SSD was imported, the analysis variables, full
sample weight and replicate weights identified, and the
replication method JK2 selected.
The output shown below was produced by a table request of
underl31 * enough, asking for percentages of the sum of the
weights, and with other specifications as shown below.
***************************************************************
Example 3 output listing
PC WESVAR VERSION NUMBER:
TIME THE JOB EXECUTED:
INPUT DATASET NAME:
OUTPUT LISTING:
2 . 12
08:19:23 02/19/99
C:\data9496\Pgm3.var
C:\data9496\pgm3.LST
OPTION NOSUMMARY IS:
OPTION FUNCTION LOG IS:
OPTION ALIGNMENT IS:
OPTION EXPORT IS:
VARIANCE ESTIMATION METHOD:
FINITE POPULATION CORRECTION FACTOR:
VALUE OF ALPHA (CONFIDENCE INTERVAL !
DEGREES OF FREEDOM:
t VALUE:
OFF
OFF
OFF
OFF
JK2
1.00000
0.05000 (95.00000
INFINITE
1.960
OPTION COMPLETE IS:
FULL SAMPLE WEIGHT:
REPLICATE WEIGHTS:
ANALYSIS VARIABLES:
COMPUTE STATISTIC:
TABLE REQUESTS:
ON
WT3_HH
R3_HH_01...R3_HH_43
None Specified.
None Specified.
UNDER13l*ENOUGH
FACTOR(S) :
1.00
NUMBER OF REPLICATES:
NUMBER OF OBSERVATIONS READ:
WEIGHTED NUMBER OF OBSERVATIONS READ:
43
8067
98574761.000
5-44
Attachment 1-150
-------
TABLE
REQUEST : UNDER131 * ENOUGH
UNDER131
ENOUGH
EST_TYPE
ESTIMATE
STDERROR
N
Below 131%
Enough
PERCENT
18 .27
0 . 16
1908
Below 131%
Not enough
PERCENT
1.63
0 . 12
175
Below 131%
MARGINAL
PERCENT
19 . 90
0 . 11
2083
131% +
Enough
PERCENT
79 . 54
0 . 13
5883
131% +
Not enough
PERCENT
0 .56
0 .10
41
131% +
MARGINAL
PERCENT
80 .10
0 .11
5924
MARGINAL
Enough
PERCENT
97 . 81
0 .16
7791
MARGINAL
Not enough
PERCENT
2 .19
0 .16
216
MARGINAL
MARGINAL
PERCENT
100 . 00
0 .00
8007
Below 131%
Enough
COLPCT
18 .68
0 .14
1908
Below 131%
Not enough
COLPCT
74 . 54
3.39
175
Below 131%
MARGINAL
COLPCT
19 . 90
0 .11
2083
131% +
Enough
COLPCT
81.32
0 .14
5883
131% +
Not enough
COLPCT
25 .46
3.39
41
131% +
MARGINAL
COLPCT
80 .10
0 .11
5924
MARGINAL
Enough
COLPCT
100 .00
0 .00
7791
MARGINAL
Not enough
COLPCT
100 . 00
0 .00
216
MARGINAL
MARGINAL
COLPCT
100 .00
0 .00
8007
Below 131%
Enough
ROWPCT
91.80
0.58
1908
Below 131%
Not enough
ROWPCT
8.20
0.58
175
Below 131%
MARGINAL
ROWPCT
100 . 00
0 .00
2083
131% +
Enough
ROWPCT
99 .30
0 .12
5883
131% +
Not enough
ROWPCT
0 .70
0 .12
41
131% +
MARGINAL
ROWPCT
100 .00
0 .00
5924
MARGINAL
Enough
ROWPCT
97 . 81
0 .16
7791
MARGINAL
Not enough
ROWPCT
2 .19
0 .16
216
MARGINAL
MARGINAL
ROWPCT
100 .00
0 .00
8007
************* End of Example 3 output listing *********;
5-45
Attachment 1-151
-------
5.8 References
FASEB/LSRO (Federation of American Societies for Experimental
Biology, Life Sciences Research Office). 1995. Joint policy
on variance estimation and statistical standards on NHANES III
and CSFII reports ... (Appendix III). In: Third Report on
Nutrition Monitoring in the United States. Prepared for the
Interagency Board for Nutrition Monitoring and Related Research.
USDA Publication.
SAS Institute, Inc. 1990. SAS language: Reference,
version 6 first edition. SAS Institute, Inc., Cary, NC.
Shah, BV, Barnwell, BG, and Bieler, GS. 1997.
SUDAAN User's Manual, Release 7.5. Research Triangle
Park, NC: Research Triangle Institute.
USDA (U.S. Department of Agriculture, Agricultural
Research Service). 1998. 1994-96 Continuing Survey
of Food Intakes by Individuals and 1994-96 Diet and
Health Knowledge Survey. CD-ROM. Available from
National Technical Information Service,
Springfield, VA. (NTIS Accession No. PB98-500457)
USDC/BOC (U.S. Department of Commerce, Bureau of the Census).
1994. Current Population Survey, March 1994. Machine-readable
data file.
USDC/BOC (U.S. Department of Commerce, Bureau of the Census).
1995. Current Population Survey, March 1995. Machine-readable
data file.
USDC/BOC (U.S. Department of Commerce, Bureau of the Census).
1996. Current Population Survey, March 1996. Machine-readable
data file.
USDC/BOC (U.S. Department of Commerce, Bureau of the Census).
1998. Current Population Survey, March 1998. Machine-readable
data file.
5-46
Attachment 1-152
-------
6. USING THE CSFII 1994-96, 1998 DATA
6.3 Statistical Notes
6.3.1 Statistical software
Because of the complex sample design of the CSFII, ARS
recommends that data users calculate standard errors and
coefficients of variation for descriptive and related
statistics using software that takes the sample design and
weighting into account. The PSUs used in the design can be
paired as in a stratified sample where the Taylor Series
expansion method can be used. This will allow software such
as SUDAAN or Stata to be used when studying population
subgroups. The fields VARSTRAT and VARUNIT are located in
positions 11-12 and 13 in the data file, respectively. These
fields represent the nesting fields STRATUM and PSU used for
Taylor Series expansion estimation of standard errors.
Also, see section 7.4.2, "Sampling weights and variance
estimation fields." Replicate weights, as provided by the
jackknife replication method, can also be used as described
in section 5.6.2, "Estimation of Sampling Errors." See
section 5, "SAMPLING WEIGHTS," for more information on
weighting procedures.
6.3.2 Guidelines for the use of sampling weights
Weights should always be used when calculating descriptive
statistics. This is because descriptive statistics are meant
to provide summary information about the entire population
under study, not just the sample. Included under the heading
of descriptive statistics are measures of central tendency,
such as means and medians, as well as measures of variability,
such as variances.
Most statistical software packages allow the user to compute
weighted descriptive statistics although they may not estimate
variances properly. If in doubt, the analyst is advised
to consult a survey statistician.
Attachment 1-153
-------
ATTACHMENT 2
The EPA Food Commodity Vocabulary
(Source: CSFII Documentation)
Attachment 2-1
-------
U.S. Environmental Protection Agency
THE EPA FOOD COMMODITY VOCABULARY
EPA's list of food commodities, that is, agricultural food items from either plant or animal sources in raw or processed forms, includes items based on the following rationale.
A food commodity is included by EPA if it is cited in the Residue Chemistry Guidelines, Table 1, entitled: Raw Agricultural and Processed Commodities and Feedstuffs
Derived from Crops (Guideline prepared by the Office of Prevention, Pesticides, and Toxic Substances; available electronically from the EPA Public Access webpage. To
access the guideline, go to: http: / /www. epa .gov/OPPTS_Harmonized/8 6 0_Residue_Chemistry_Test_Guidelines/Series . Then, select
860 .1000) .
Additionally, a food commodity is included in the following list if it currently has a tolerance. Other food
commodities are included if foods known to contain them have been reported to be consumed in the 1994-96 CSFII, or in
prior food consumption surveys, or if they are known to be ingredients in commercially available baby food. Some few
other food commodities may exist in the list below as they were revealed to be consumed by humans subsequent to
research undertaken by EPA staff in the preparation of numerous internal reports on use of agricultural animals or
crops.
THE EPA FOOD COMMODITY VOCABULARY
Master List June 15, 2000
EPA Food
Commodity (FC)
Code
Food Commodity (FC)
Weight Basis of Food
Commodity (FC):
amount=edible portion
as consumed, except
where otherwise noted
(in gm/kg body
weight/day)
Comments
95000010
Acerola
weight of fruit;
excludinq seed
18000020
Alfalfa, seed
weight of dry seed
Alfalfa sprouts are the
human food item.
14000030
Almond
weiqht of nutmeat
14000031
Almond- babvfood
same as almond
Attachment 2-2
-------
EPA Food
Commodity (FC)
Code
Food Commodity (FC)
Weight Basis of Food
Commodity (FC) :
amount=edible portion
as consumed, except
where otherwise noted
(in gm/kg body
weight/day)
Comments
14000040
Almond, oil
weiqht of oil
14000041
Almond, oil- babyfood
weight of oil
04010050
Amaranth, leafy
weight of leaf
Includes Tampala, Chinese
spinach, lambsquarter,
and pokeweed (pokeberry).
95000060
Amaranth, grain
dry weight of grain;
include Amaranth flour
11000070
Apple, fruit with peel
weight of apple;
including peel,
excluding core and
stem
11000080
Apple, peeled fruit
weight of apple;
excluding peel, core
and stem
11000081
Apple, peeled fruit -
babyfood
weight of apple;
excluding peel, core
and stem
11000090
Apple, dried
dry weight; excluding
peel, core, stem
11000091
Apple, dried - babyfood
dry weight; excluding
peel, core, stem
Attachment 2-3
-------
EPA Food
Commodity (FC)
Code
Food Commodity (FC)
Weight Basis of Food
Commodity (FC) :
amount=edible portion
as consumed, except
where otherwise noted
(in gm/kg body
weight/day)
Comments
11000100
Apple, juice
weight of juice at
single strength (or
standard dilution)
11000101
Apple, juice - babyfood
weight of juice at
single strength (or
standard dilution)
11000110
Apple, sauce
weight of applesauce
11000111
Apple, sauce - babyfood
weight of applesauce
12000120
Apricot
weight of pulp, with
or without peel;
excluding pit
12000121
Apricot- babyfood
weight of pulp, with
or without peel;
excluding pit
12000130
Apricot, dried
dry weight of pulp,
with or without peel;
excluding pit
12000140
Apricot, juice
weight of juice at
single strength (or
standard dilution)
Attachment 2-4
-------
EPA Food
Commodity (FC)
Code
Food Commodity (FC)
Weight Basis of Food
Commodity (FC):
amount=edible portion
as consumed, except
where otherwise noted
(in gm/kg body
weight/day)
Comments
12000141
Apricot, juice- babyfood
weight of juice at
single strength (or
standard dilution)
01030150
Arrowroot, flour
dry weight of flour
01030151
Arrowroot, flour- babyfood
dry weight of flour
95000160
Artichoke, globe
edible portion of
flowerhead
01030170
Artichoke, Jerusalem
edible portion of
tuber
04010180
Aruqula
weight of leaves
95000190
Asparagus
weight of edible
portion of
spears/stems
95000200
Avocado
weight of pulp;
excluding skin and pit
09020210
Balsam pear
weight of whole fruit
Includes Balsam apple,
Chinese cucumber, and
Bittermelon.
95000220
Bamboo, shoots
weight of shoots
Bamboo shoots are the
human food item.
Attachment 2-5
-------
EPA Food
Commodity (FC)
Code
Food Commodity (FC)
Weight Basis of Food
Commodity (FC) :
amount=edible portion
as consumed, except
where otherwise noted
(in gm/kg body
weight/day)
Comments
95000230
Banana
weight of pulp;
excluding peel; juice
95000231
Banana- babyfood
weight of pulp;
excluding peel; juice
95000240
Banana, dried
dry weight of dried
pulp; excluding peel
(include weight of
fruit from chips)
95000241
Banana, dried- babyfood
dry weight of dried
pulp; excluding peel
(include weight of
fruit from chips)
15000250
Barley, pearled barley
dry weight of barley
15000251
Barley, pearled barley -
babyfood
dry weight of barley
15000260
Barley, flour
dry weight of flour
Includes malt and whole
barley.
15000261
Barley, flour- babyfood
dry weight of flour
Includes malt and whole
barley.
15000270
Barlev, bran
dry weiqht of bran
Attachment 2-6
-------
EPA Food
Commodity (FC)
Code
Food Commodity (FC)
Weight Basis of Food
Commodity (FC):
amount=edible portion
as consumed, except
where otherwise noted
(in gm/kg body
weight/day)
Comments
19010280
Basil, fresh leaves
weight of leaves and
stems
19010281
Basil, fresh leaves -
babyfood
weight of leaves and
stems
19010290
Basil, dried leaves
dry weight of leaves
and stems
19010291
Basil, dried leaves-
baby food
dry weight of leaves
and stems
06030300
Bean, black, seed
dry weight of bean
Includes black turtle
bean, Bayo, and brown
bean.
06020310
Bean, broad, succulent
weight of bean and pod
Also called fava bean.
06030320
Bean, broad, seed
dry weight of bean
Also called fava bean.
06020330
Bean, cowpea, succulent
weight of bean;
excluding pod
Includes cowpea, crowder
pea, blackeye pea, and
southern pea.
06030340
Bean, cowpea, seed
dry weight of bean
Includes cowpea, crowder
pea, blackeye pea, and
southern pea.
06030350
Bean, great northern,
seed
dry weight of bean
Attachment 2-7
-------
EPA Food
Commodity (FC)
Code
Food Commodity (FC)
Weight Basis of Food
Commodity (FC):
amount=edible portion
as consumed, except
where otherwise noted
(in gm/kg body
weight/day)
Comments
06030360
Bean, kidney, seed
dry weiqht of bean
06020370
Bean, lima, succulent
weight of bean;
excludinq pod
06030380
Bean, lima, seed
dry weiqht of bean
06030390
Bean, mung, seed
dry weight of bean
Bean sprouts are the
human food item.
06030400
Bean, navy, seed
dry weiqht of bean
Includes pea bean.
06030410
Bean, pink, seed
dry weiqht of bean
06030420
Bean, pinto, seed
dry weight of bean
Include calico and red
Mexican bean.
06010430
Bean, snap, succulent
weight of bean and pod
Includes green bean,
runner bean, and wax
bean.
06010431
Bean, snap, succulent-
baby food
weight of bean and pod
Includes green bean,
runner bean, and wax
bean.
21000440
Beef, meat
weight of meat;
excluding the weight
of bone, and all
nutrient fat
Consider veal as
equivalent to beef.
Include bison and
buffalo.
Attachment 2-8
-------
EPA Food
Commodity (FC)
Code
Food Commodity (FC)
Weight Basis of Food
Commodity (FC):
amount=edible portion
as consumed, except
where otherwise noted
(in gm/kg body
weight/day)
Comments
21000441
Beef, meat- babyfood
weight of meat;
excluding the weight
of bone, and all
nutrient fat
21000450
Beef, meat, dried
weight of dried meat;
excluding bones, and
trimmable fat
21000460
Beef, meat byproducts
weight of meat;
excluding bone; may
contain some fat
Includes brain, heart,
lung, sweetbread, tail,
tripe, tongue, and head.
21000461
Beef, meat byproducts-
baby food
weight of meat;
excluding bone; may
contain some fat
Includes brain, heart,
lung, sweetbread, tail,
tripe and tongue.
21000470
Beef, fat
weight of nutrient fat
only; includes
nutrient fat from beef
meat
21000471
Beef,fat- babyfood
weight of nutrient fat
only; includes
nutrient fat from beef
meat
21000480
Beef, kidney
weight of organ
including nutrient fat
Attachment 2-9
-------
EPA Food
Commodity (FC)
Code
Food Commodity (FC)
Weight Basis of Food
Commodity (FC):
amount=edible portion
as consumed, except
where otherwise noted
(in gm/kg body
weight/day)
Comments
21000490
Beef, liver
weight of organ
including nutrient fat
21000491
Beef, liver- babyfood
weight of organ
including nutrient fat
01010500
Beet, qarden, roots
weight of roots; juice
01010501
Beet, garden, roots -
babyfood
weight of roots
02000510
Beet, qarden, tops
weight of leaves
Include pumpkin leaves.
01010520
Beet, sugar
dry weight of sugar
(sucrose)
Sucrose is a disaccharide
obtained from sugar cane
and sugar beet.
01010521
Beet, sugar- babyfood
dry weight of sugar
(sucrose)
Sucrose is a disaccharide
obtained from sugar cane
and sugar beet.
01010530
Beet, suqar, molasses
weight of molasses
01010531
Beet, sugar, molasses -
babyfood
weight of molasses
95000540
Belgium endive
weight of leaves
Also called Witloof
chicory.
Attachment 2-10
-------
EPA Food
Commodity (FC)
Code
Food Commodity (FC)
Weight Basis of Food
Commodity (FC):
amount=edible portion
as consumed, except
where otherwise noted
(in gm/kg body
weight/day)
Comments
13010550
Blackberry
weight of berry
Includes Marionberry,
Olallieberry, and
Youngberry.
13010560
Blackberry, juice
weight of juice at
single strength (or
standard dilution)
Includes Marionberry,
Olallieberry, and
Youngberry.
13010561
Blackberry, juice -
babyfood
weight of juice at
single strength (or
standard dilution)
Includes Marionberry,
Olallieberry, and
Youngberry.
13020570
Blueberry
weight of berry
13020571
Blueberry- babyfood
weight of berry
13010580
Boysenberry
weight of berry
14000590
Brazil nut
weight of nutmeat
95000600
Breadfruit
weight of pulp;
excluding peel
05010610
Broccoli
weight of flower heads
and adjoining stems
05010611
Broccoli- babyfood
weight of flower heads
and adjoininq stems
Attachment 2-11
-------
EPA Food
Commodity (FC)
Code
Food Commodity (FC)
Weight Basis of Food
Commodity (FC):
amount=edible portion
as consumed, except
where otherwise noted
(in gm/kg body
weight/day)
Comments
05010620
Broccoli, Chinese
weight of flower buds,
adjoining stems and
leaves
05020630
Broccoli raab
weight of flower buds,
adjoining stems and
leaves
05010640
Brussels sprouts
weiqht of leaf sprouts
15000650
Buckwheat
dry weight of groats;
whole qroat flour
15000660
Buckwheat, flour
dry weiqht of flour
01010670
Burdock
weiqht of roots
14000680
Butternut
weiqht of nutmeat
05010690
Cabbaqe
weiqht of leaves
05020700
Cabbage, Chinese, bok choy
weight of leaves and
stems
05010710
Cabbaqe, Chinese, napa
weiqht of leaves
05010720
Cabbaqe, Chinese, mustard
weiqht of leaves
95000730
Cactus
weight of pulp;
excluding peel; juice
Includes Prickly pear,
Cactus pads and flowers,
Nopales, and Aloe vera.
Attachment 2-12
-------
EPA Food
Commodity (FC)
Code
Food Commodity (FC)
Weight Basis of Food
Commodity (FC) :
amount=edible portion
as consumed, except
where otherwise noted
(in gm/kg body
weight/day)
Comments
95000740
Canistel
weight of pulp;
excluding peel
Also called Eggfruit.
09010750
Cantaloupe
weight of pulp;
excluding seeds and
outer rind
Includes wintermelon.
04020760
Cardoon
weight of leaf stalks
95000770
Carob
dry weight of bean;
flour
01010780
Carrot
weight of roots, with
or without peel,
excluding tops
01010781
Carrot- babyfood
weight of roots, with
or without peel,
excluding tops
01010790
Carrot, juice
weight of juice at
single strength (or
standard dilution)
09010800
Casaba
weight of pulp,
excluding seeds and
rind
14000810
Cashew
weiqht of nutmeat
Attachment 2-13
-------
EPA Food
Commodity (FC)
Code
Food Commodity (FC)
Weight Basis of Food
Commodity (FC) :
amount=edible portion
as consumed, except
where otherwise noted
(in gm/kg body
weight/day)
Comments
01030820
Cassava
weight of root; dry
tapioca
Includes tapioca.
01030821
Cassava- babyfood
weight of root; dry
tapioca
Includes tapioca.
05010830
Cauliflower
weight of flower heads
and adjoining stems
Includes Broccoflower and
green cauliflower.
01010840
Celeriac
weight of tuberous
root
04020850
Celery
weight of leaf stalk
04020851
Celery- babyfood
weight of leaf stalk
04020860
Celery, juice
weight of juice as
single strength (or
standard dilution)
04020870
Celtuce
weight of stalks and
leaves
09020880
Chayote, fruit
weight of fruit
Also called Christophine,
and Mirliton.
95000890
Cherimoya
weight of fruit;
excluding peel
Also called custard
apple.
Attachment 2-14
-------
EPA Food
Commodity (FC)
Code
Food Commodity (FC)
Weight Basis of Food
Commodity (FC):
amount=edible portion
as consumed, except
where otherwise noted
(in gm/kg body
weight/day)
Comments
12000900
Cherry
weight of fruit;
including skin;
excluding pit and stem
Includes sweet cherry and
sour or tart cherry.
12000901
Cherry- babyfood
weight of fruit;
including skin;
excluding pit and stem
Includes sweet cherry and
sour or tart cherry.
12000910
Cherry, juice
weight of juice at
single strength (or
standard dilution)
Includes sweet cherry and
sour or tart cherry.
12000911
Cherry, juice- babyfood
weight of juice at
single strength (or
standard dilution)
Includes Sweet cherry and
Sour or tart cherry.
14000920
Chestnut
weight of nutmeat
40000930
Chicken, meat
weight of flesh;
excluding the weight
of bone, total
nutrient fat, and skin
40000931
Chicken, meat- babyfood
weight of flesh;
excluding the weight
of bone, total
nutrient fat, and skin
40000940
Chicken, liver
weight of organ;
including nutrient fat
Attachment 2-15
-------
EPA Food
Commodity (FC)
Code
Food Commodity (FC)
Weight Basis of Food
Commodity (FC):
amount=edible portion
as consumed, except
where otherwise noted
(in gm/kg body
weight/day)
Comments
40000950
Chicken, meat byproducts
weight of meat;
excluding bone; may
include some trimmable
fat and/or skin
Includes giblets (heart),
necks, gizzard, feet, and
tail.
40000951
Chicken, meat byproducts-
baby food
weight of meat;
excluding bone; may
include some trimmable
fat and/or skin
Includes giblets (heart),
necks, gizzard, feet, and
tail.
40000960
Chicken, fat
weight of nutrient fat
only; includes weight
of nutrient fat from
chicken meat and skin
40000961
Chicken, fat- babyfood
weight of nutrient fat
only; includes weight
of nutrient fat from
chicken meat and skin
40000970
Chicken, skin
weight of skin only
(0 grams nutrient fat)
40000971
Chicken, skin- babyfood
weight of skin only
(0 grams nutrient fat)
06030980
Chickpea, seed
dry weight of bean
Also called garbanzo
bean.
Attachment 2-16
-------
EPA Food
Commodity (FC)
Code
Food Commodity (FC)
Weight Basis of Food
Commodity (FC):
amount=edible portion
as consumed, except
where otherwise noted
(in gm/kg body
weight/day)
Comments
06030981
Chickpea, seed - babyfood
dry weight of bean
Also called garbanzo
bean.
06030990
Chickpea, flour
dry weight of flour
Also called garbanzo
bean.
01011000
Chicory, roots
weight of roots
02001010
Chicory, tops
weight of leaves
09021020
Chinese waxgourd
weight of flesh;
including or excluding
peel
Includes Togan and
wintermelon.
19011030
Chive
weight of leaves
04011040
Chrysanthemum, garland
edible leaves and
stems
Includes Chrysanthemum,
edible leaved.
19021050
Cinnamon
dry weight of spice:
stick or ground powder
19021051
Cinnamon- babyfood
dry weight of spice:
stick or ground powder
10001060
Citrus citron
weight of pulp;
excluding peel
10001070
Citrus hybrids
weight of pulp;
excluding seeds and
peel
Includes tangelo, Tangor,
Chironja, and Calamondin.
Attachment 2-17
-------
EPA Food
Commodity (FC)
Code
Food Commodity (FC)
Weight Basis of Food
Commodity (FC):
amount=edible portion
as consumed, except
where otherwise noted
(in gm/kg body
weight/day)
Comments
10001080
Citrus, oil
weiqht of oil
95001090
Cocoa bean, chocolate
weight of chocolate;
cocoa butter
95001100
Cocoa bean, powder
weiqht of powder
95001110
Coconut, meat
weight of meat;
excluding milk and
shell
95001111
Coconut- meat, babyfood
weight of meat;
excluding milk and
shell
95001120
Coconut, dried
dry weight of meat;
excluding milk and
shell
95001130
Coconut, milk
weight of milk only;
excluding meat and
shell
95001140
Coconut, oil
weiqht of oil
95001141
Coconut, oil- babyfood
weiqht of oil
95001150
Coffee, roasted bean
dry weiqht of bean
95001160
Coffee, instant
dry weight of powder
or qranules
Attachment 2-18
-------
EPA Food
Commodity (FC)
Code
Food Commodity (FC)
Weight Basis of Food
Commodity (FC):
amount=edible portion
as consumed, except
where otherwise noted
(in gm/kg body
weight/day)
Comments
05021170
Collards
weiqht of leaves
19021180
Coriander, leaves
weight of leaves
Includes Chinese parsley
and cilantro leaf.
19021181
Coriander, leaves -
babyfood
weight of leaves
Includes Chinese parsley
and
cilantro leaf.
19021190
Coriander, seed
weight of seed
Includes Chinese parsley
and cilantro seed.
19021191
Coriander, seed - babyfood
weight of seed
Includes Chinese parsley
and
cilantro seed.
15001200
Corn, field, flour
dry weight of whole
grain flour; masa
harina
15001201
Corn, field, flour-
baby food
dry weight of whole
grain flour; masa
harina
Attachment 2-19
-------
EPA Food
Commodity (FC)
Code
Food Commodity (FC)
Weight Basis of Food
Commodity (FC):
amount=edible portion
as consumed, except
where otherwise noted
(in gm/kg body
weight/day)
Comments
15001210
Corn, field, meal
dry weight of whole
grain or degermed
meal; dry weight of
corn ingredient from
corn or cornmeal based
chips or snacks; dry
weight of hominy
Includes hominy.
15001211
Corn, field, meal-
baby food
dry weight of whole
grain or degermed
meal; dry weight of
corn ingredient from
corn or cornmeal based
chips or snacks; dry
weight of hominy.
Includes hominy.
15001220
Corn, field, bran
dry weight of bran
15001230
Corn, field, starch
dry weight of corn
starch
15001231
Corn, field, starch-
baby food
dry weight of corn
starch
15001240
Corn, field, syrup
weight of syrup
15001241
Corn, field, syrup-
baby food
weight of syrup
15001250
Corn, field, oil
weiqht of oil
Attachment 2-20
-------
EPA Food
Commodity (FC)
Code
Food Commodity (FC)
Weight Basis of Food
Commodity (FC) :
amount=edible portion
as consumed, except
where otherwise noted
(in gm/kg body
weight/day)
Comments
15001251
Corn, field, oil- babyfood
weiqht of oil
15001260
Corn, pop
weight of kernels;
excludinq cob and husk
15001270
Corn, sweet
weight of kernels;
excludinq cob and husk
15001271
Corn, sweet- babyfood
weight of kernels;
excludinq cob and husk
95001280
Cottonseed, oil
weiqht of oil
95001281
Cottonseed, oil - babyfood
weiqht of oil
11001290
Crabapple
weight of pulp;
excluding core and
stem; including or
excludinq peel
95001300
Cranberry
weiqht of berry
95001301
Cranberry- babyfood
weiqht of berry
95001310
Cranberry, dried
dry weiqht of berry
95001320
Cranberry, juice
weight of juice at
single strength (or
standard dilution)
Attachment 2-21
-------
EPA Food
Commodity (FC)
Code
Food Commodity (FC)
Weight Basis of Food
Commodity (FC):
amount=edible portion
as consumed, except
where otherwise noted
(in gm/kg body
weight/day)
Comments
95001321
Cranberry, juice- babyfood
weight of juice at
single strength (or
standard dilution)
04011330
Cress, qarden
weight of leaves
04011340
Cress, upland
weight of leaves
Includes Yellow rocket
and Winter cress.
09021350
Cucumber
weight of flesh and
seeds; including or
excluding peel
13021360
Currant
weight of berry
13021370
Currant, dried
dry weight of berry
04011380
Dandelion, leaves
weight of leaves
01031390
Dasheen, corm
weight of the corm
Includes taro.
02001400
Dasheen, leaves
weight of the leaves
Includes taro.
95001410
Date
weight of fruit,
excluding pit
13011420
Dewberry
weight of berry
19021430
Dill, seed
dry weight of seed
19011440
Dill
weiqht of leaves
Also called dillweed.
Attachment 2-22
-------
EPA Food
Commodity (FC)
Code
Food Commodity (FC)
Weight Basis of Food
Commodity (FC):
amount=edible portion
as consumed, except
where otherwise noted
(in gm/kg body
weight/day)
Comments
70001450
Egg, whole
weight of white and
yolk; excluding shell
70001451
Egg, whole - babyfood
weight of white and
yolk;
excluding shell
70001460
Egg, white
weight of egg white
70001461
Egg, white (solids)-
babyfood
rehydrated weight of
egg white solids
70001470
Egg, yolk
weight of egg yolk
70001471
Egg, yolk- babyfood
weight of egg yolk
08001480
Eggplant
weight of whole
vegetable; including
seeds, with or without
peel
13021490
Elderberry
weight of berry
04011500
Endive
weight of leaves
Includes escarole.
95001510
Feij oa
weight of pulp,
excluding peel
Also called pineapple
guava.
04021520
Fennel, Florence
weight of leaves
Includes Italian and
sweet fennel.
Attachment 2-23
-------
EPA Food
Commodity (FC)
Code
Food Commodity (FC)
Weight Basis of Food
Commodity (FC):
amount=edible portion
as consumed, except
where otherwise noted
(in gm/kg body
weight/day)
Comments
95001530
Fiq
weiqht of fruit
95001540
Fiq, dried
dry weiqht of fruit
14001550
Filbert
weiqht of nutmeat
Also called hazelnut.
14001560
Filbert, oil
weiqht of oil
Also called hazelnut.
80001570
Fish- freshwater finfish
weiqht of edible
portion; excludinq
head, tail, scales,
fins, viscera,
inedible bones and
skin
See Appendix A ("The Fish
List") for categorization
of various fish species
into appropriate FC, ie,
for species that are
specifically fresh water
fish, salt water fish,
crustacean-shellfish, and
mollusc-shellfish.
80001580
Fish- freshwater finfish,
farm raised
weiqht of edible
portion; excludinq
head, tail, scales,
fins, viscera,
inedible bones and
skin
See Appendix A.
Attachment 2-24
-------
EPA Food
Commodity (FC)
Code
Food Commodity (FC)
Weight Basis of Food
Commodity (FC):
amount=edible portion
as consumed, except
where otherwise noted
(in gm/kg body
weight/day)
Comments
80001590
Fish- saltwater finfish,
tuna
weight of edible
portion; excluding
head, tail, scales,
fins, viscera,
inedible bones and
skin
See Appendix A.
80001600
Fish- saltwater finfish,
other
weight of edible
portion; excluding
head, tail, scales,
fins, viscera,
inedible bones and
skin
See Appendix A.
80001610
Fish- shellfish,
crustacean
weight of edible
portion; excluding
shell, gills, and
viscera
See Appendix A.
80001620
Fish- shellfish, mollusc
weight of edible
portion; excluding
shells
See Appendix A.
95001630
Flaxseed, oil
weight of oil;
nutrient fat from flax
seeds
Also called Solin.
03001640
Garlic
weight of bulb;
excluding skin (outer
scales)
Attachment 2-25
-------
EPA Food
Commodity (FC)
Code
Food Commodity (FC)
Weight Basis of Food
Commodity (FC):
amount=edible portion
as consumed, except
where otherwise noted
(in gm/kg body
weight/day)
Comments
03001650
Garlic, dried
dry weight of bulb;
excluding skin (outer
scales)
03001651
Garlic, dried- babyfood
dry weight of bulb;
excluding skin (outer
scales)
01031660
Ginger
weight of roots;
excluding peel
01031661
Ginger - babyfood
weight of roots,
excluding peel
01031670
Ginger, dried
dry weight of roots;
excluding peel
01031680
Ginseng, dried
dry weight of roots
23001690
Goat, meat
weight of meat;
excluding weight of
bone, and all nutrient
fat and skin
23001700
Goat, meat byproducts
weight of meat;
excluding bone; may
include some trimmable
fat and/or skin
Includes brain, heart,
and head.
Attachment 2-26
-------
EPA Food
Commodity (FC)
Code
Food Commodity (FC)
Weight Basis of Food
Commodity (FC):
amount=edible portion
as consumed, except
where otherwise noted
(in gm/kg body
weight/day)
Comments
23001710
Goat, fat
weight of nutrient fat
only; include nutrient
fat from qoat meat
23001720
Goat, kidney
weight of organ
including nutrient fat
23001730
Goat, liver
weight of organ
including nutrient fat
13021740
Gooseberry
weiqht of berry
95001750
Grape
weight of grape, with
skin, and with or
without seeds
Includes Muscadine.
95001760
Grape, juice
weight of juice at
single strength (or
standard dilution)
95001761
Grape, juice- babyfood
weight of juice at
single strength (or
standard dilution)
95001770
Grape, leaves
weiqht of leaves
95001780
Grape, raisin
dry weiqht of raisin
Includes Zante currant.
95001790
Grape, wine and sherry
weight of wine or
sherry
Attachment 2-27
-------
EPA Food
Commodity (FC)
Code
Food Commodity (FC)
Weight Basis of Food
Commodity (FC) :
amount=edible portion
as consumed, except
where otherwise noted
(in gm/kg body
weight/day)
Comments
10001800
Grapefruit
weight of pulp;
excluding seeds and
rind
10001810
Grapefruit, juice
weight of juice at
single strength (or
standard dilution)
06031820
Guar, seed
weight of bean
06031821
Guar, seed - babyfood
weight of bean
95001830
Guava
weight of pulp;
excluding peel; juice
95001831
Guava- babyfood
weight of pulp;
excluding peel; juice
Attachment 2-28
-------
EPA Food
Commodity (FC)
Code
Food Commodity (FC)
Weight Basis of Food
Commodity (FC) :
amount=edible portion
as consumed, except
where otherwise noted
(in gm/kg body
weight/day)
Comments
19011840
Herbs, other
weight of leaves and
stems
See Appendix B ("The Herb
List"). Note that some
herbs are not included in
this list as they are
unique FCs and are cited
in this vocabulary
separately. These latter
include: basil, chive,
coriander (cilantro),
leaf, dill (dillweed),
fennel, Florence (Italian
and sweet), lemongrass,
marjoram (oregano),
parsley and savory
(winter and summer).
Attachment 2-29
-------
EPA Food
Commodity (FC)
Code
Food Commodity (FC)
Weight Basis of Food
Commodity (FC):
amount=edible portion
as consumed, except
where otherwise noted
(in gm/kg body
weight/day)
Comments
19011841
Herbs, other- babyfood
weight of leaves and
stems
See Appendix B ("The Herb
List"). Note that some
herbs are not included in
this list as they are
unique FCs and are cited
in this vocabulary
separately. These latter
include: basil, chive,
coriander (cilantro),
leaf, dill (dillweed),
fennel, Florence (Italian
and sweet), lemongrass,
marjoram (oregano),
parsley, and savory
(summer and winter).
14001850
Hickory nut
weiqht of nutmeat
95001860
Honey
weiqht of honey
95001861
Honey- babyfood
weiqht of honey
09011870
Honeydew melon
weight of pulp;
excluding seeds and
rind
95001880
Hod
weiqht of dried hops
Attachment 2-30
-------
EPA Food
Commodity (FC)
Code
Food Commodity (FC)
Weight Basis of Food
Commodity (FC):
amount=edible portion
as consumed, except
where otherwise noted
(in gm/kg body
weight/day)
Comments
24001890
Horse, meat
weight of meat;
excluding bone,
trimmable fat or skin
01011900
Horseradish
weight of root, ground
13021910
Huckleberry
weight of berry
95001920
Jaboticaba
weight of whole fruit
95001930
Jackfruit
weight of pulp;
excluding peel
05021940
Kale
weight of leaves
Includes mizuna.
95001950
Kiwifruit
weight of pulp;
excluding peel
05011960
Kohlrabi
weight of leaves and
stems
10001970
Kumquat
weight of fruit;
including peel;
excluding seeds
03001980
Leek
weight of whole plant;
including leaves and
bulb
Attachment 2-31
-------
EPA Food
Commodity (FC)
Code
Food Commodity (FC)
Weight Basis of Food
Commodity (FC):
amount=edible portion
as consumed, except
where otherwise noted
(in gm/kg body
weight/day)
Comments
10001990
Lemon
weight of pulp;
excluding seeds and
peel
10002000
Lemon, juice
weight of juice at
single strength (or
standard dilution)
10002001
Lemon, juice- babyfood
weight of juice at
single strength (or
standard dilution)
10002010
Lemon, peel
weiqht of peel only
19012020
Lemonqrass
weiqht of leaves
06032030
Lentil
dry weight of edible
seed
04012040
Lettuce, head
weight of leaves;
juice
04012050
Lettuce, leaf
weiqht of leaves
Includes romaine.
10002060
Lime
weight of pulp;
excluding seeds and
peel
10002070
Lime, juice
weight of juice at
single strength (or
standard dilution)
Attachment 2-32
-------
EPA Food
Commodity (FC)
Code
Food Commodity (FC)
Weight Basis of Food
Commodity (FC):
amount=edible portion
as consumed, except
where otherwise noted
(in gm/kg body
weight/day)
Comments
10002071
Lime, juice- babyfood
weight of juice at
single strength (or
standard dilution)
13012080
Loganberry
weight of berry
95002090
Longan
weight of pulp;
excluding seeds and
rind
11002100
Loquat
weight of pulp;
excluding seeds and
skin
95002110
Lychee
weight of pulp;
excluding peel
95002120
Lychee, dried
weight of dried pulp
14002130
Macadamia nut
weight of nutmeat
95002140
Mamey apple
weight of pulp;
excluding peel, core
and stem
Also called Mamey.
95002150
Mango
weight of pulp;
excluding peel
95002151
Mango- babyfood
weight of pulp;
excluding peel
Attachment 2-33
-------
EPA Food
Commodity (FC)
Code
Food Commodity (FC)
Weight Basis of Food
Commodity (FC) :
amount=edible portion
as consumed, except
where otherwise noted
(in gm/kg body
weight/day)
Comments
95002160
Manqo, dried
weiqht of dried pulp
95002170
Mango, juice
weight of juice at
single strength (or
standard dilution)
95002171
Mango, juice - babyfood
weight of juice at
single strength
(or standard dilution)
95002180
Maple, suqar
dry weiqht of suqar
95002190
Maple syrup
weiqht of syrup
19012200
Marj oram
weight of leaves and
stems
Includes oregano.
19012201
Marjoram - babyfood
weight of leaves and
stems
Includes oregano.
28002210
Meat, game
weight of meat;
excluding bone,
trimmable fat and skin
Includes armadillo, bear,
beaver, caribou, deer,
elk, frog, groundhog,
moose, snake, opossum,
raccoon, squirrel,
turtle.
27002220
Milk, fat
weight of nutrient fat
only
Attachment 2-34
-------
EPA Food
Commodity (FC)
Code
Food Commodity (FC)
Weight Basis of Food
Commodity (FC):
amount=edible portion
as consumed, except
where otherwise noted
(in gm/kg body
weight/day)
Comments
27002221
Milk, fat - baby
food/infant formula
weight of nutrient fat
only
27012230
Milk, nonfat solids
Remaining weight of
milk after subtracting
milk fat and moisture
content of milk
Includes casein; also,
includes lactose added to
items that are neither
commercial baby foods nor
infant formulas.
27012231
Milk, nonfat solids- baby
food/infant formula
Remaining weight of
milk after subtracting
milk fat and moisture
content of milk
Includes casein; also,
note that lactose added
to commercial baby foods
or infant formulas is not
included here, but is
represented under EPA
Food Commodity Code
#27032251 (Milk, sugar
[lactose]-babyfood/infant
formula).
27022240
Milk, water
Moisture content of
milk
27022241
Milk, water-
babyfood/infant formula
Moisture content of
milk
27032251
Milk, sugar (lactose)-
baby food/infant formula
dry weight of lactose-
report only when
consumed from baby
food/infant formula
Includes lactose added to
items that are either
commercial baby foods or
infant formulas.
Attachment 2-35
-------
EPA Food
Commodity (FC)
Code
Food Commodity (FC)
Weight Basis of Food
Commodity (FC):
amount=edible portion
as consumed, except
where otherwise noted
(in gm/kg body
weight/day)
Comments
15002260
Millet, qrain
dry weiqht of qrain
95002270
Mulberry
weight of fruit
95002280
Mushroom
weight of caps or caps
and stems
05022290
Mustard greens
weight of leaves and
stems
12002300
Nectarine
weight of pulp;
including peel;
excluding pit and stem
15002310
Oat, bran
dry weight of bran
15002320
Oat, flour
dry weight of flour
15002321
Oat, flour- babyfood
dry weight of flour
15002330
Oat, qroats/rolled oats
dry weight
15002331
Oat, groats/rolled oats-
baby food
dry weight
95002340
Okra
weight of pods;
including seeds
95002350
Olive
weight of fruit;
excluding pit
95002360
Olive, oil
weiqht of oil
Attachment 2-36
-------
EPA Food
Commodity (FC)
Code
Food Commodity (FC)
Weight Basis of Food
Commodity (FC):
amount=edible portion
as consumed, except
where otherwise noted
(in gm/kg body
weight/day)
Comments
03002370
Onion, dry bulb
weight of bulb;
excluding outer skin
03002371
Onion, dry bulb- babyfood
weight of bulb;
excluding outer skin
03002380
Onion, dry bulb, dried
dry weight of bulb;
excluding outer skin
03002381
Onion, dry bulb, dried-
baby food
dry weight of bulb;
excluding outer skin
03002390
Onion, green
weight of bulb or bulb
and leaves
10002400
Orange
weight of pulp;
excluding seeds and
peel
10002410
Orange, juice
weight of juice at
single strength (or
standard dilution)
10002411
Orange, juice- babyfood
weight of juice at
single strength (or
standard dilution)
10002420
Orange, peel
weight of peel only
95002430
Palm heart, leaves
weight of stem and
leaves
Attachment 2-37
-------
EPA Food
Commodity (FC)
Code
95002440
95002441
95002450
95002451
95002460
95002470
04012480
19012490
19012491
01012500
01012510
Food Commodity (FC)
Palm, oil
Palm, oil - babyfood
Papaya
Papaya- babyfood
Papaya, dried
Papaya, juice
Parsley, leaves
Parsley, dried leaves
Parsley, dried leaves
babyfood
Parsley, turnip rooted
Parsnip
Weight Basis of Food
Commodity (FC):
amount=edible portion
as consumed, except
where otherwise noted
(in gm/kg body
weight/day)
Comments
weiqht of oil
weiqht of oil
weight of pulp;
excluding peel and
seeds
as above
weiqht of dried pulp
weight of juice at
single strength (or
standard dilution)
weight of leaves and
stems
dried weight of leaves
and stems
dried weight of leaves
and stems
weight of roots and
leaves
weight of roots with
or without peel
Attachment 2-38
-------
EPA Food
Commodity (FC)
Code
Food Commodity (FC)
Weight Basis of Food
Commodity (FC):
amount=edible portion
as consumed, except
where otherwise noted
(in gm/kg body
weight/day)
Comments
01012511
Parsnip - babyfood
weight of roots with
or without
peel
95002520
Passionfruit
weight of pulp;
excluding seeds and
peel
95002521
Passionfruit- babyfood
weight of pulp;
excluding seeds and
peel
95002530
Passionfruit, juice
weight of juice at
single strength (or
standard dilution)
95002531
Passionfruit, juice-
baby food
weight of juice at
single strength (or
standard dilution)
95002540
Pawpaw
weight of pulp;
excluding peel
06022550
Pea, succulent
weight of peas
06022551
Pea, succulent- babyfood
weight of peas
06032560
Pea, dry
dry weight of pea
06032561
Pea, dry- babyfood
dry weiqht of pea
Attachment 2-39
-------
EPA Food
Commodity (FC)
Code
Food Commodity (FC)
Weight Basis of Food
Commodity (FC) :
amount=edible portion
as consumed, except
where otherwise noted
(in gm/kg body
weight/day)
Comments
06012570
Pea, edible podded
weiqht of pea and pod
06032580
Pea, piqeon, seed
dry weiqht of pea
06022590
Pea, pigeon, succulent
weight of pea
In U.S. presently,
usually found as canned
piqeon peas.
12002600
Peach
weiqht of pulp, with
or without peel;
excludinq pit
12002601
Peach- babyfood
weiqht of pulp, with
or without peel;
excludinq pit
12002610
Peach, dried
weiqht of dried pulp,
with or without peel;
excludinq pit
12002611
Peach, dried- babyfood
weiqht of dried pulp,
with or without peel;
excludinq pit
12002620
Peach, juice
weiqht of juice at
sinqle strenqth (or
standard dilution)
12002621
Peach, juice- babyfood
weiqht of juice at
sinqle strenqth (or
standard dilution)
Attachment 2-40
-------
EPA Food
Commodity (FC)
Code
Food Commodity (FC)
Weight Basis of Food
Commodity (FC):
amount=edible portion
as consumed, except
where otherwise noted
(in gm/kg body
weight/day)
Comments
95002630
Peanut
weight of nutmeat;
excluding shell
95002640
Peanut, butter
weight of ground
peanuts; excluding
shell
95002650
Peanut, oil
weight of oil
12002660
Pear
weight of pulp, with
or without peel;
excluding core and
stem
Include Oriental pear.
12002661
Pear- babyfood
weight of pulp, with
or without peel;
excluding core and
stem
12002670
Pear, dried
weight of dried pulp,
with or without peel
12002680
Pear, juice
weight of juice at
single strength (or
standard dilution)
12002681
Pear, juice- babyfood
weight of juice at
single strength (or
standard dilution)
14002690
Pecan
weiqht of nutmeat
Attachment 2-41
-------
EPA Food
Commodity (FC)
Code
Food Commodity (FC)
Weight Basis of Food
Commodity (FC):
amount=edible portion
as consumed, except
where otherwise noted
(in gm/kg body
weight/day)
Comments
08002700
Pepper, bell
weight of flesh;
excluding seeds and
stem
Includes sweet pepper,
cooking pepper, pimento,
and banana pepper.
08002701
Pepper, bell- babyfood
weight of flesh;
excluding seeds and
stem
Includes sweet pepper,
cooking pepper, pimento,
and banana pepper.
08002710
Pepper, bell, dried
dry weight of flesh
only
Includes sweet pepper,
cooking pepper, pimento,
and banana pepper.
08002711
Pepper, bell, dried-
baby food
dry weight of flesh
only
Includes sweet pepper,
cooking pepper, pimento,
and banana pepper.
08002720
Pepper, non-bell
weight of flesh, with
or without seeds;
excluding stem
08002721
Pepper, non-bell, -
babyfood
weight of flesh, with
or without
seeds; excluding stem
08002730
Pepper, non-bell, dried
dry weight of flesh or
flesh and seeds
19022740
Pepper, black and white
dry weight of pepper
19022741
Pepper, black and white-
babvfood
dry weight of pepper
Attachment 2-42
-------
EPA Food
Commodity (FC)
Code
Food Commodity (FC)
Weight Basis of Food
Commodity (FC):
amount=edible portion
as consumed, except
where otherwise noted
(in gm/kg body
weight/day)
Comments
95002750
Peppermint
weight of leaves and
stems
95002760
Peppermint, oil
weiqht of oil
95002770
Persimmon
weight of entire
fruit, pulp and peel
95002780
Pine nut
weiqht of nutmeat
Also called piqnolia.
95002790
Pineapple
weight of pulp;
excluding leaves and
outer peel
95002791
Pineapple- babyfood
weight of pulp;
excluding leaves and
outer peel
95002800
Pineapple, dried
weight of dried pulp
only
95002810
Pineapple, juice
weight of juice at
single strength (or
standard dilution)
95002811
Pineapple, juice- babyfood
weight of juice at
single strength (or
standard dilution)
14002820
Pistachio
weiqht of nutmeat
Attachment 2-43
-------
EPA Food
Commodity (FC)
Code
95002830
95002840
12002850
12002851
12002860
12002861
12002870
12002871
12002880
Food Commodity (FC)
Plantain
Plantain, dried
Plum
Plum- babyfood
Plum, prune, fresh
Plum, prune, fresh-
babyfood
Plum, prune, dried
Plum, prune, dried-
baby food
Plum, prune, juice
Weight Basis of Food
Commodity (FC):
amount=edible portion
as consumed, except
where otherwise noted
(in gm/kg body
weight/day)
Comments
weight of pulp;
excluding skin
weight of dried pulp
only
weight of pulp with
peel; excluding pit
weight of pulp with
peel; excluding pit
weight of plum, with
peel; excluding pit
weight of plum, with
peel; excluding pit
weight of dried flesh,
with or without peel;
excluding pit
weight of dried flesh,
with or without peel;
excluding pit
weight of juice at
single strength (or
standard dilution)
Attachment 2-44
-------
EPA Food
Commodity (FC)
Code
Food Commodity (FC)
Weight Basis of Food
Commodity (FC):
amount=edible portion
as consumed, except
where otherwise noted
(in gm/kg body
weight/day)
Comments
12002881
Plum, prune, juice-
baby food
weight of juice at
single strength (or
standard dilution)
95002890
Pomegranate
weight of pulp;
excluding peel and
seeds
Seeds are usually not
consumed.
25002900
Pork, meat
weight of meat;
excluding weight of
bone, and all nutrient
fat and skin
25002901
Pork, meat- babyfood
weight of meat;
excluding weight of
bone, and all nutrient
fat and skin
25002910
Pork, skin
dry weight of skin
including nutrient fat
e.g., pork rind snacks.
25002920
Pork, meat byproducts
weight of meat;
excluding bone; may
include some trimmable
fat and skin
Includes ears, jowl,
chitterlings, stomach
(maw), fatback, and feet.
25002921
Pork, meat byproducts-
baby food
weight of meat;
excluding bone; may
include some trimmable
fat and skin
Includes ears, jowl,
chitterlings, stomach
(maw), fatback, and feet.
Attachment 2-45
-------
EPA Food
Commodity (FC)
Code
Food Commodity (FC)
Weight Basis of Food
Commodity (FC):
amount=edible portion
as consumed, except
where otherwise noted
(in gm/kg body
weight/day)
Comments
25002930
Pork, fat
weight of nutrient
fat only; includes
weight of nutrient fat
from meat
25002931
Pork, fat- babyfood
weight of nutrient
fat only; includes
weight of nutrient fat
from meat
25002940
Pork, kidney
weight of organ
including nutrient fat
25002950
Pork, liver
weight of organ
including nutrient fat
01012960
Potato, chips
weight of potato from
chip or stick, with or
without peel
01012970
Potato, dry (granules/
flakes)
dry weight of granules
or flakes
01012971
Potato, dry (granules/
flakes)- babyfood
dry weight of granules
or flakes
01012980
Potato, flour
dry weight of flour or
potato starch
01012981
Potato, flour - babyfood
dry weight of flour or
potato starch
Attachment 2-46
-------
EPA Food
Commodity (FC)
Code
Food Commodity (FC)
Weight Basis of Food
Commodity (FC):
amount=edible portion
as consumed, except
where otherwise noted
(in gm/kg body
weight/day)
Comments
01012990
Potato, tuber, w/peel
weight of tuber;
includinq peel
01012991
Potato, tuber, w/peel-
babyfood
weight of tuber;
including peel
01013000
Potato, tuber, w/o peel
weight of tuber;
excluding peel
01013001
Potato, tuber, w/o peel-
baby food
weight of tuber;
excluding peel
60003010
Poultry, other, meat
weight of meat;
excluding weight of
bone, and all nutrient
fat, and skin
Includes dove, duck, emu,
goose, guinea hen,
ostrich, partridge,
pheasant, pigeon, quail,
squab, wild duck.
60003020
Poultry, other, liver
weight of organ,
including nutrient fat
Includes dove, duck, emu,
goose, guinea hen,
ostrich, partridge,
pheasant, pigeon, quail,
scruab, wild duck.
Attachment 2-47
-------
EPA Food
Commodity (FC)
Code
Food Commodity (FC)
Weight Basis of Food
Commodity (FC) :
amount=edible portion
as consumed, except
where otherwise noted
(in gm/kg body
weight/day)
Comments
60003030
Poultry, other, meat
byproducts
weight of meat;
excluding bone; may
contain some trimmable
fat and/or skin
Includes dove, duck, emu,
goose, guinea hen,
ostrich, partridge,
pheasant, pigeon, quail,
squab, wild duck.
Byproducts include (as
for chicken) giblets,
gizzard, neck, feet, and
tail, as applicable to
species of poultry.
60003040
Poultry, other, fat
weight of nutrient
fat only; includes
weight of nutrient fat
from meat and skin
Includes dove, duck, emu,
goose, guinea hen,
ostrich, partridge,
pheasant, pigeon, quail,
squab, wild duck.
60003050
Poultry, other, skin
weight of nutrient
fat only; includes
weight of nutrient fat
from meat and skin
Includes dove, duck, emu,
goose, guinea hen,
ostrich, partridge,
pheasant, pigeon, quail,
squab, wild duck.
95003060
Psyllium, seed
dry weight of psyllium
husks
10003070
Pummelo
weight of pulp;
excluding seeds and
peel
Attachment 2-48
-------
EPA Food
Commodity (FC)
Code
Food Commodity (FC)
Weight Basis of Food
Commodity (FC):
amount=edible portion
as consumed, except
where otherwise noted
(in gm/kg body
weight/day)
Comments
09023080
Pumpkin
weight of pulp;
excluding seeds and
rind
09023090
Pumpkin, seed
weight of dried seed
only
11003100
Quince
weight of pulp;
excluding seeds and
peel
95003110
Quinoa, grain
weight of grain or
Quinoa flour
29003120
Rabbit, meat
weight of meat;
excluding bone,
trimmable fat and skin
04013130
Radicchio
weight of leaves
01013140
Radish, roots
weight of roots
02003150
Radish, tops
weight of leaves
01013160
Radish, Oriental, roots
weight of roots
02003170
Radish, Oriental, tops
weight of leaves
05023180
Rape greens
weight of leaves and
stems
95003190
Rapeseed, oil
weiqht of oil
Includes canola oil.
Attachment 2-49
-------
EPA Food
Commodity (FC)
Code
Food Commodity (FC)
Weight Basis of Food
Commodity (FC) :
amount=edible portion
as consumed, except
where otherwise noted
(in gm/kg body
weight/day)
Comments
95003191
Rapeseed, oil - babyfood
weiqht of oil
Includes canola oil.
14013200
Raspberry
weiqht of berry
14013201
Raspberry- babyfood
weiqht of berry
14013210
Raspberry, juice
weiqht of juice at
sinqle strenqth (or
standard dilution)
14013211
Raspberry, juice
babyfood
weiqht of juice at
sinqle strenqth (or
standard dilution)
04023220
Rhubarb
weiqht of stalks;
excludinq leaves
15003230
Rice, white
dry weiqht of qrain
15003231
Rice, white- babyfood
dry weiqht of qrain
15003240
Rice, brown
dry weiqht of qrain or
brown rice flour
15003241
Rice, brown- babyfood
dry weiqht of qrain or
brown rice flour
15003250
Rice, flour
dry weiqht of flour
15003251
Rice, flour- babyfood
dry weiqht of flour
15003260
Rice, bran
dry weiqht of bran
Attachment 2-50
-------
EPA Food
Commodity (FC)
Code
Food Commodity (FC)
Weight Basis of Food
Commodity (FC) :
amount=edible portion
as consumed, except
where otherwise noted
(in gm/kg body
weight/day)
Comments
15003261
Rice, bran- babyfood
dry weight of bran
01013270
Rutabaga
weight of roots;
excluding tops
15003280
Rye, grain
dry weight of grain;
weight of whole grain
flour
15003290
Rye, flour
dry weight of flour
95003300
Safflower, oil
weight of oil
95003301
Safflower, oil - babyfood
weight of oil
01013310
Salsify, roots
weight of roots
Also called oyster plant.
02003320
Salsify, tops
weight of leaves
Also called oyster plant.
95003330
Sapote, Mamey
weight of pulp;
excluding peel
Includes black sapote,
Mamey sapote, and white
sapote.
19013340
Savory
weight of leaves and
flower buds
Includes summer and
winter savory.
95003350
Seaweed
weight of vegetation
(wet and dry)
Includes algae, Irish
moss, kelp, spirulina,
aqar, laver, and wakame.
Attachment 2-51
-------
EPA Food
Commodity (FC)
Code
Food Commodity (FC)
Weight Basis of Food
Commodity (FC):
amount=edible portion
as consumed, except
where otherwise noted
(in gm/kg body
weight/day)
Comments
95003351
Seaweed - babyfood
weight of vegetation
(wet and dry)
Includes algae, Irish
moss, kelp, spirulina,
agar, laver, and wakame.
95003360
Sesame, seed
dry weight of seed
95003361
Sesame, seed- babyfood
dry weight of seed
95003370
Sesame, oil
weight of oil
95003371
Sesame, oil- babyfood
weight of oil
03003380
Shallot
weight of bulb;
excluding skin
26003390
Sheep, meat
weight of meat;
excluding weight of
bone, total nutrient
fat and skin
26003391
Sheep, meat- babyfood
weight of meat;
excluding weight of
bone, total nutrient
fat and skin
26003400
Sheep, meat byproducts
weight of meat;
excluding bones; may
include some trimmable
fat and/or skin
Includes brain and
tongue.
Attachment 2-52
-------
EPA Food
Commodity (FC)
Code
Food Commodity (FC)
Weight Basis of Food
Commodity (FC):
amount=edible portion
as consumed, except
where otherwise noted
(in gm/kg body
weight/day)
Comments
26003410
Sheep, fat
weight of nutrient fat
only; includes
nutrient fat from meat
26003411
Sheep, fat- babyfood
weight of nutrient fat
only; includes
nutrient fat from meat
26003420
Sheep, kidney
weight of organ,
including nutrient fat
26003430
Sheep, liver
weight of organ,
including nutrient fat
15003440
Sorqhum, qrain
dry weiqht of qrain
15003450
Sorqhum, syrup
weiqht of syrup
95003460
Soursop
weight of pulp;
excludinq peel
06003470
Soybean, seed
dry weight of seed or
bean
06003480
Soybean, flour
dry weight of flour,
soy meal, soy protein
concentrate and
isolate
Attachment 2-53
-------
EPA Food
Commodity (FC)
Code
Food Commodity (FC)
Weight Basis of Food
Commodity (FC):
amount=edible portion
as consumed, except
where otherwise noted
(in gm/kg body
weight/day)
Comments
06003481
Soybean, flour- babyfood
dry weight of flour,
soy meal,
soy protein
concentrate and
isolate
06003490
Soybean, soy milk
total weight of milk
06003491
Soybean, soy milk-
babyfood or infant formula
total weight of milk
06003500
Soybean, oil
weiqht of oil
Includes lecithin.
06003501
Soybean, oil- babyfood
weiqht of oil
Includes lecithin.
95003510
Spanish lime
weight of pulp;
excluding seeds and
pulp
Includes Genip.
95003520
Spearmint
weight of leaves and
stems
95003530
Spearmint, oil
weiqht of oil
Attachment 2-54
-------
EPA Food
Commodity (FC)
Code
Food Commodity (FC)
Weight Basis of Food
Commodity (FC):
amount=edible portion
as consumed, except
where otherwise noted
(in gm/kg body
weight/day)
Comments
19023540
Spices, other
dry weight
See Appendix C ("The
Spice List"). Note that
this list does not
include some spices that
are unique FCs and are
listed separately in this
Commodity Vocabulary.
These latter spices
include: cinnamon, dill
(seed), coriander
(cilantro)-seed, and
black and white pepper.
19023541
Spices, other- babyfood
dry weiqht
See Appendix C.
04013550
Spinach
weight of leaves;
juice
04013551
Spinach- babyfood
weight of leaves;
juice
09023560
Squash, summer
weight of flesh, seeds
and peel
Includes crookneck
squash, kampyo, scallop
squash, straightneck
squash, vegetable marrow,
and zucchini.
Attachment 2-55
-------
EPA Food
Commodity (FC)
Code
Food Commodity (FC)
Weight Basis of Food
Commodity (FC):
amount=edible portion
as consumed, except
where otherwise noted
(in gm/kg body
weight/day)
Comments
09023561
Squash, summer- babyfood
weight of flesh, seeds
and peel
Includes crookneck
squash, kampyo, scallop
squash, straightneck
squash, vegetable marrow,
and zucchini.
09023570
Squash, winter
weight of flesh;
excluding seeds and
peel
Includes butternut
squash, hubbard squash,
calabaza, acorn squash,
and spaghetti squash.
09023571
Squash, winter- babyfood
weight of flesh;
excluding seeds and
peel
95003580
Starfruit
weight of fruit,
including seeds and
peel
Also called Carambola.
95003590
Strawberry
weight of berry;
excluding leaf cap
95003591
Strawberry- babyfood
weight of berry,
excluding leaf cap
95003600
Strawberry, juice
weight of juice at
single strength (or
standard dilution)
Attachment 2-56
-------
EPA Food
Commodity (FC)
Code
Food Commodity (FC)
Weight Basis of Food
Commodity (FC):
amount=edible portion
as consumed, except
where otherwise noted
(in gm/kg body
weight/day)
Comments
95003601
Strawberry, juice -
babyfood
weight of juice at
single strength (or
standard dilution)
95003610
Sugar apple
weight of pulp;
excluding peel
Include atemoya.
95003620
Sugarcane, sugar
dry weight of cane
sugar (sucrose)
Sucrose is a disaccharide
obtained from sugar cane
and sugar beet.
95003621
Sugarcane, sugar- babyfood
dry weight of cane
sugar (sucrose)
Sucrose is a disaccharide
obtained from sugar cane
and sugar beet.
95003630
Sugarcane, molasses
weight of molasses
Includes blackstrap
molasses.
95003631
Sugarcane, molasses -
babyfood
weight of molasses
95003640
Sunflower, seed
dry weight of seeds
95003650
Sunflower, oil
weight of oil
95003651
Sunflower, oil - babyfood
weight of oil
01033660
Sweet potato
weight of roots, with
or without peel; juice
01033661
Sweet potato- babvfood
weiqht of root; juice
Likelv without peel.
Attachment 2-57
-------
EPA Food
Commodity (FC)
Code
Food Commodity (FC)
Weight Basis of Food
Commodity (FC):
amount=edible portion
as consumed, except
where otherwise noted
(in gm/kg body
weight/day)
Comments
04023670
Swiss chard
weight of leaves and
stalks
95003680
Tamarind
weight of pulp,
including seeds;
excluding peel; juice
10003690
Tangerine
weight of pulp;
excluding seeds and
peel
Include mandarin.
10003700
Tangerine, juice
weight of juice at
single strength (or
standard dilution)
Include mandarin.
01013710
Tanier, corm
weight of corm
Also called Cocoyam.
95003720
Tea, dried
dry weight of tea
leaves
95003730
Tea, instant
dry weight of powder
08003740
Tomatillo
weight of fruit;
excluding outer husks
08003750
Tomato
weight of pulp, seeds
and skin; tomatoes
without skin but with
seed (example canned
whole tomatoes)
Attachment 2-58
-------
EPA Food
Commodity (FC)
Code
Food Commodity (FC)
Weight Basis of Food
Commodity (FC):
amount=edible portion
as consumed, except
where otherwise noted
(in gm/kg body
weight/day)
Comments
08003751
Tomato- babyfood
weight of pulp, seeds
and skin; tomatoes
without skin but with
seed (example canned
whole tomatoes)
08003760
Tomato, paste
weight of concentrated
tomato pulp from food
described as paste;
excluding seeds and
skin
08003761
Tomato, paste- babyfood
weight of concentrated
tomato pulp (from food
described as paste);
excluding seeds and
skin
08003770
Tomato, puree
weight of concentrated
tomato pulp (from food
described as puree or
sauce); excluding
seeds and skin
08003771
Tomato, puree- babyfood
weight of concentrated
tomato pulp (from food
described as puree or
sauce); excluding
seeds and skin
Attachment 2-59
-------
EPA Food
Commodity (FC)
Code
Food Commodity (FC)
Weight Basis of Food
Commodity (FC) :
amount=edible portion
as consumed, except
where otherwise noted
(in gm/kg body
weight/day)
Comments
08003780
Tomato, dried
weight of dried tomato
(may include skin
and/or seeds)
08003781
Tomato, dried - babyfood
weight of dried tomato
(may include skin
and/or seeds)
08003790
Tomato, juice
weight of juice at
single strength (or
standard dilution)
95003800
Tomato, Tree
weight of the tree
tomato, excluding peel
Also called Tamarillo.
15003810
Triticale, flour
dry weight of flour
15003811
Triticale, flour- babyfood
dry weight of flour
50003820
Turkey, meat
weight of meat;
excluding bone, all
nutrient fat, and skin
50003821
Turkey, meat- babyfood
weight of meat;
excluding bone, all
nutrient fat, and skin
50003830
Turkey, liver
weight of organ,
including nutrient fat
Attachment 2-60
-------
EPA Food
Commodity (FC)
Code
Food Commodity (FC)
Weight Basis of Food
Commodity (FC) :
amount=edible portion
as consumed, except
where otherwise noted
(in gm/kg body
weight/day)
Comments
50003831
Turkey, liver- babyfood
weight of organ,
including nutrient fat
50003840
Turkey, meat byproducts
weight of meat;
excluding bone (may
contain some fat
and/or skin)
Includes gizzard, heart,
neck, and tail.
50003841
Turkey, meat byproducts-
baby food
weight of meat;
excluding bone (may
contain some fat
and/or skin)
Includes gizzard, heart,
neck, and tail.
50003850
Turkey, fat
weight of nutrient
fat only; includes
nutrient fat from meat
and skin
50003851
Turkey, fat- babyfood
weight of nutrient
fat only; includes
nutrient fat from meat
ans skin
50003860
Turkey, skin
weight of skin only
(0 grams nutrient fat)
50003861
Turkey, skin - babyfood
weight of skin only
(0 grams nutrient fat)
19023870
Turmeric
weiqht of roots
Attachment 2-61
-------
EPA Food
Commodity (FC)
Code
Food Commodity (FC)
Weight Basis of Food
Commodity (FC):
amount=edible portion
as consumed, except
where otherwise noted
(in gm/kg body
weight/day)
Comments
01013880
Turnip, roots
weight of roots
02003890
Turnip, tops
weight of leaves and
stems
95003900
Vinegar
Weight of vinegar made
from apple, grape or
rice juice in recipe
14003910
Walnut
weight of nutmeat
86003920
Water, dilution, source NS
water to dilute or
reconstitute a juice,
beverage, or soup
This FC is of interest to
EPA, but will be acquired
by a separate mechanism.
86003930
Water, tapwater- direct
(drinking)
This FC is of interest to
EPA, but will be acquired
by a separate mechanism.
86003940
Water- indirect (cooking)
This FC is of interest to
EPA, but will be acquired
by a separate mechanism.
86003950
Water, bottled water
This FC is of interest to
EPA, but will be acquired
by a separate mechanism.
86003960
Water, commercial beverage
This FC is of interest to
EPA, but will be acquired
by a separate mechanism.
Attachment 2-62
-------
EPA Food
Commodity (FC)
Code
Food Commodity (FC)
Weight Basis of Food
Commodity (FC):
amount=edible portion
as consumed, except
where otherwise noted
(in gm/kg body
weight/day)
Comments
95003970
Water chestnut
weight of tuber
95003980
Watercress
weight of leaves and
stems; juice
09013990
Watermelon
weight of pulp and
rind; excluding seeds
To include weight of
pickled watermelon rind.
09014000
Watermelon, juice
weight of juice at
single strength (or
standard dilution)
15004010
Wheat, grain
dry weight of grain
Include whole kernel,
cracked wheat, whole
grain flour, bulgur,
couscous.
15004011
Wheat, grain - babyfood
dry weight of grain
Include whole kernel,
cracked wheat, whole
grain flour, bulgur,
couscous
15004020
Wheat, flour
dry weight of flour
15004021
Wheat, flour- babyfood
dry weight of flour
15004030
Wheat, germ
dry weight of germ
15004040
Wheat, bran
dry weight of bran
15004050
Wild rice
dry weiqht of qrain
Attachment 2-63
-------
EPA Food
Commodity (FC)
Code
Food Commodity (FC)
Weight Basis of Food
Commodity (FC):
amount=edible portion
as consumed, except
where otherwise noted
(in gm/kg body
weight/day)
Comments
01034060
Yam, true
weiqht of roots
01034070
Yam bean
weiqht of roots
Also called jicama.
Attachment 2-64
-------
APPENDIX A
THE FISH LIST
FISH-SALTWATER Finfish. Tuna
Tuna
FISH-SALTWATER Finfish. Other
Alewife
Menhaden
Spot
Anchovy
Monkfish
Swordfish
Barracuda
Mullet
Tilefish
Bass
Perch
Tomcod
Bluefish
Pollock
Whitefish
Bonita
Pompano
Whiting
Butter fish
Porgy
Cisco
Rockfish
Eel
Cod
Roe (herring, sea urchin)
Croaker
Roughy
Dolphinfish
Sablefish
Drum
Sardine
Flatfish
Sea Trout
Flounder
Scup
Grouper
Shad
Haddock
Shark
Halibut
Skate
Herring
Smelt
Kingfish
Snapper
Mackerel
Sole
Attachment 2-65
-------
APPENDIX A
THE FISH LIST, Continued
FISH-SHELLFISH. Crustacean
Crab
Crayfish and Crawfish
Cuttlefish
Lobster
Shrimp and Prawns
FISH-FRESHWATER. Finfish
Carp
Catfish (not farm raised)
Caviar (Sturgeon)
Pike
Salmon (Chinook, Chum, Coho, Sockeye)(not farm raised)
Tilapia (not farm raised)
Sturgeon
Trout (not farm raised)
FTSH-SHELLFISH. Mollusc
Abalone
Clam
Conch
Mussel
Octopus
Oyster
Scallop
Snail
Squid
FISH-FRESHWATER. Finfish-
Farm Raised
Bluegill
Catfish, Channel
Trout (Rainbow, Brook, Lake)
Tilapia
Salmon (Chinook, Chum,
Coho, Sockeye)
Attachment 2-66
-------
APPENDIX B
l lll HERB LIST
Angelica
Balm
Woodruff
Wormwood
Borage
Burnet
Camomile
Catnip
Chervil, dried leaves
Chinese chive
Clary
Costmary
Curry, dried leaves
Horehound
Hyssop
Lavender
Lovage, leaves
Marigold
Nasturtium
Pennyroyal
Rosemary
Rue
Sage
Sweet bay
Tansy
Tarragon
Thyme
Winter green
Note that some herbs are not included in this list as they are unique FCs and are cited in
this vocabulary separately. These latter include: basil, chives, coriander (cilantro), leaf,
dill (dillweed), fennel, Florence (Italian and sweet), lemongrass, marjoram (oregano),
parsley, and savory (summer and winter).
Attachment 2-67
-------
APPENDIX C
THE SPICE LIST
Allspice
Anise, seed
Anise, star
Annatto, seed
Caper, buds
Caraway
Caraway, black
Cardamom
Cassia, buds
Celery, seed
Clove, buds
Cumin
Fennel, common
Fennel, Florence, seed
Fenugreek, seed
Grains of Paradise
Juniper Berry
Lovage, seed
Mace
Mustard, seed
Nutmeg
Poppy, seed
Saffron
Vanilla
Note that this list does not include some spices that are unique FCs and are listed
separately in this Commodity Vocabulary. These latter spices include: cinnamon;
seed; coriander (cilantro), seed; black and white pepper.
Attachment 2-68
-------
APPENDIX D
SELECTED REFERENCES FOR THE EPA FOOD COMMODITY VOCABULARY
Arthey, D. and P.R. Ashust. 1996. Fruit processing. Blackie Academic and professional. NY. 248 pp.
Bebee, C.N. 1990. The Protection of Tropical and Subtropical Fruits. 1979-April 1990. USDA National Agricultural
Library. BLA No. 97.
Chaisson, C., B. Petersen, S. White, and A. Clayton. 1984. Tolerance Assessment System: Crops to Food Map. US
EPA. HED Toxicology branch. 58 pp.
Codex Alimentarius. 1993. Pesticides Residues in Food. Section 2. Codex Classification of Foods and Animal Feeds.
FAO/WHO. Rome, Italy. Vol. 2. 218 pp.
Considine, D.M. and G.D. Considine. 1982. Foods and Food Production Encyclopedia. Van Nostrand Reinhold, Co.
NY. 2305 pp.
Ensminger, M., J. Oldfield, and W. W. Heineman. 1990. Feed and Nutrition Digest. Second Ed. Regu Publication.
Clovis, CA. 795 pp.
Farrell, K.T. 1990. Spices, Condiments, and Seasonings. Second Ed. AVI Book. VanNostrand Reinhold Co. NY.
244 pp.
Heaton, D.D. 1997. A Produce Reference Guide to Fruits and Vegetables From Around the World. Nature's Harvest.
Food Product Press. NY. 415 pp.
Hoseney, R.C. 1994. Principles of Cereal Science and Technology. Second Ed. American Association of Cereal
Chemists. St. Paul, MN. 378 pp.
Janick, J. Editor. 1996. Progress in New Crops. American Society of Horticulture Science Press. Alexandria, VA.
600 pp.
Larkcom, J. 1991. Oriental Vegetables. Kodansha, Int. NY. 232 pp.
Magness, J.R., G. M. Markle, and C.C. Compton. 1971. Food and Feed Crops of the United States. USDA IR-4.
Rutgers, NJ. IR Bulletin No. 1. Bulletin 828.
Markle, G.M., J.J. Baron, and B.A. Schneider. 1998. Food and Feed Crops of the United States. Second Ed. Meister
Publications. Willoughby, OH. 517 pp.
Morton, J.F. 1987. Fruits of Warm Climates. First Ed. Media Inc., Greensboro, NC. 505 pp.
Pennington, J.A.T. 1997. Bowes and Church's Food Values of Portions Commonly Used. Seventeenth Ed.
Lippincot-Raven Publishers. Philadelphia, PA. 481 pp.
Attachment 2-69
-------
APPENDIX D (continued)
Randolph, S. and M. Synder. 1993. The Seafood List. US FDA . Washington, DC 68 pp. .
Raymond, E. 1999. The Packer 1999 Produce and Availability Guide. Vance Publishing Co. Lenexa, KS. 552 pp..
Schneider, Bernard A. 1992. Comparison of USD A Human Nutrition Information Service (HNIS) Legume
Vegetables Vocabulary With OPP Commodity Vocabularies. US EPA. September 4.
Schneider, B.A. 1993. Response to Commodities, Subcommodities, and Commodity Groups Not Yet Included In
The New DRES Vocabulary. Part I: A - CN, Entries 1 - 206. US EPA.
Schneider, B.A. 1993. Response to Commodities, Subcommodities, and Commodity Groups Not Yet Included In
The New DRES Vocabulary. Part II: CO - E, Entries 207 - 283. US EPA.
Schneider, B.A. 1993. Response to Commodities, Subcommodities, and Commodity Groups Not Yet Included
In The New DRES Vocabulary. Part III: F - G, Entries 284 - 379. US EPA.
Schneider, B.A. 1993. Response to Commodities, Subcommodities, and Commodity Groups Not Yet Included In
The New DRES Vocabulary. Part IV: H - K, Entries 380 - 419. US EPA.
Schneider, B.A. 1993. Response to Commodities, Subcommodities, and Commodity Groups Not Yet Included In
The New DRES Vocabulary. Part V: L - M, Entries 420 - 508. US EPA.
Schneider, B.A. 1993. Response to Commodities, Subcommodities, and Commodity Groups Not Yet Included In
The New DRES Vocabulary. Part VIA: O - Pears, Entries 509 - 554. US EPA.
Schneider, B.A. 1993. Response to Commodities, Subcommodities, and Commodity Groups Not Yet Included In
The New DRES Vocabulary. Part VIB: Peas - Pumpkins, Entries 555 - 621. US EPA.
Schneider, B.A. 1993. Response to Commodities, Subcommodities, and Commodity Groups Not Yet Included In
The New DRES Vocabulary. Part VII: Q - R, Entries 622 - 663. US EPA.
Schneider, B.A. 1994. Response to Commodities, Subcommodities, and Commodity Groups Not Yet Included In
The New DRES Vocabulary. Part VIII: S, Entries 664 - 784. US EPA.
Schneider, B.A. 1994. Response to Commodities, Subcommodities, and Commodity Groups Not Yet Included In
The New DRES Vocabulary. Part IX: T, Entries 785 - 833. US EPA.
Schneider, B.A. 1994. Response to Commodities, Subcommodities, and Commodity Groups Not Yet Included In
The New DRES Vocabulary. Part X: U - Z, Entries 834 - 920. US EPA.
Schneider, B.A. 1994. Response to Public Comments on Pesticide Tolerance Crop Grouping Regulation: Proposed
Rule (40CFR § 180). Part II: Evaluation of Supporting Data Concerning Commodity Common Names and Proposed
Additions to Several Crop Groups. CB# 12719. DP Barcode D196080. US EPA. OPPTS, HED, CBTS. June 22. 15
pp.
Schneider, B.A. 1996. Uses of Brans of Wheat, Oats, Barley, and Rye As Foods. D220744. US EPA. OPP HED Feb.
14. 32 pp.
Schneider, B.A., Yuen-Shaung Ng, H. Jamerson, C. Colledge, C. Lang, I. Hornstein, R. Cromwell. 1999. EPA Food
and Feed Tolerance Vocabulary. US EPA. OPP. HED.
Attachment 2-70
-------
APPENDIX D (continued)
Schneider, Elizabeth. 1985. Uncommon Fruits and Vegetables: A Common Sense Guide. Harper and Row
Publishers, Inc. NY. 546 pp.
Smith, D.S., J.N. Cash, W.K. Nip, and Y.H. Hui. 1997. Processing Vegetables Science and Technology. Technomics
Publ. Co. 434 pp.
U.S. EPA. 1993. Pesticide Tolerances: Portion of Food Commodities to Be Analyzed for Pesticide Residues.
Proposed Rule. OPP-3000243. Federal Register Vol. 58. No. 187. Pp. 50888-50893. September 23.
U.S. EPA. 1993. Table 1. Raw Agricultural and Processed Commodities and Feedstuffs Derived from Crops.
Subdivision G: Residue Chemistry Guidelines. OPPTS860.1000
Attachment 2-71
------- |