(fiCT 0&1897 OolH'do> /l^f 2 tfF'/^/w,
USDA Measuring Food Security in the United States
United States
Department ol
Agriculture
Food and
Consumer
Service
Office©!
Analytitand
Evaluation
Household Food Security
in the United States in 1995
Summary Report
of the Food Security
Measurement Project
\*
<p.00<O%?>
USDA
United ttatM
Dapartniant of
Agrtcuttura
Food and
Comumor
Sorvteo
Offteaof
Analysis an
Evaluation
Household Food
Security in the
United States
in 1995
Summary Report
of the Food Security
Measurement Project
September 1997
Prepared for:
Gary W. Bickel, Project Officer
U.S. Department of Agriculture
Food and Consumer Service
3101 Park Center Drive
Alexandria, VA 22302
under contract no. 53-3198-5-028
Prepared by:
William L. Hamilton, Project Director*
John T. Cook, Principal Investigator1*
William W. Thompson'
Lawrence F. Buron*
Edward A. Frongillo, Jr. ■
Christine M. Olson0
Cheryl A. Wehlei"
* Abt Associates, Inc.
b Tufts University Center on Hunger,
Poverty, and Nutrition Policy
' Cornell University Division of
Nutritional Sciences
d C.A.W. and Associates
TABLE OF CONTENTS
Chapter One
Chapter Two
Chapter Three
Chapter Four
EXECUTIVE SUMMARY i
FOOD SECURITY AND HUNGER MEASUREMENT IN THE
UNITED STATES 1
Introduction 1
Food Insecurity and Hunger Measurement—Background 2
Food Insecurity and Hunger Measurement—Conceptual Basis 4
The Hypothesis of Hunger as Severe Food Insecurity 7
The Role of Food Security Measurement 9
Conclusion 11
THE FOOD SECURITY SUPPLEMENT TO THE CURRENT
POPULATION SURVEY 13
The Food Security Measurement and Research Conference 13
Questions in the Food Security Survey Supplement 15
The Current Population Survey Sample 17
The CPS Food Security Supplement Sample 18
THE FOOD SECURITY MEASUREMENT SCALE 21
Scale Development 21
12-Month and 30-Day Scales 23
Questions Used in the Scale 24
Relative Severity of Questions in the Scale 27
Household Values on the Scale 29
Overview of Scale Development Results 31
THE FOOD SECURITY STATUS MEASURE 33
Defining Ranges on the Food Security Scale 36
Food Insecure 37
Food Insecure with Moderate Hunger 38
Food Insecure with Severe Hunger 41
Response Profile of Households in the Four Categories 42
Relationship of Food Security Status to Other Measures 44
Prepared by Abt Associates Inc. n
Table of Contents
Chapter Five
Chapter Six
PREVALENCE OF FOOD INSECURITY AND HUNGER 47
Prevalence Estimates 47
Participation in Food Assistance Programs 54
State-Level Food Security Prevalence Estimates 56
Thirty-Day Prevalence Estimates 56
Estimating the Number of Persons in Food-Insecure Households .... 57
Comparing Food Security Prevalence Estimates from
Various Sources 58
Conclusion 60
STRENGTHS AND LIMITATIONS OF THE FOOD SECURITY
MEASURE 61
Results of the Scaling Analysis 61
Summary of Prevalence Estimates 62
Notes on Validity and Accuracy 62
Future Directions 65
REFERENCES 67
Appendix A SUPPLEMENT TO THE APRIL 1995 CURRENT POPULATION
SURVEY
Appendix B QUESTIONS TESTED FOR THE FOOD SECURITY SCALES:
UNWEIGHTED RESPONSE FREQUENCIES
Appendix C PREVALENCE ESTIMATES AND STANDARD ERRORS BY STATE
Appendix D ESTIMATED STANDARD ERRORS FOR PREVALENCE TABLES IN
CHAPTER FIVE
Appendix E DISTRIBUTION OF PERSONS IN HOUSEHOLDS BY FOOD
SECURITY STATUS CLASSIFICATION
Appendix F PARTICIPANTS IN FEDERAL INTERAGENCY WORKING GROUP
FOR FOOD SECURITY MEASUREMENT
Prepared by Abt Associates Inc.
LIST OF EXHIBITS
Exhibit 2-1 Summary of Food Security Survey Items by Subject Area IS
Exhibit 2-2 Description of the Final Food Security Supplement Sample by
Type of Household 20
Exhibit 3-1 Questions Included in the Food Security Scale 25
Exhibit 3-2 Severity Ranking of Questions in Food Security Scale 28
Exhibit 3-3 Population Distribution by Selected Household Scale Values 30
Exhibit 4-1 Severity Ranges on the Food Security Scale 39
Exhibit 4-2 Response Profile by Category 43
Exhibit 4-3 Relationship of the Food Security Status Measure to
Other Variables 45
Exhibit 5-1 Prevalence of Household Food Security Status by Selected
Characteristics of Households 48
Exhibit 5-2 Percent of Households Receiving Food Assistance in the Past 30
Days, by Food Security Status 55
Exhibit B-l Responses to Questions Tested for the Food Security Scales
Exhibit C-l Estimated Food Security Prevalences by State: Twelve Months
Preceding the Survey
Exhibit C-2 Standard Errors for State Prevalence Estimates: Twelve Months
Preceding the Survey
Exhibit D-l Standard Errors: Prevalence of Household Food Security Status
by Selected Characteristics of Households: 12-Month Scale
Exhibit El Estimated Distribution of Persons With Selected Characteristics Living
in Households with Each Food Security Status: 12-Month Scale
Prepared by Abt Associates Inc. K\
ACKNOWLEDGEMENTS
The analyses presented here build on the work of a great many people, carried out over
several years. In particular, the members of the Federal Interagency Working Group for Food
Security Measurement were the architects of the Food Security Supplement to the Current
Population Survey, the foundation on which the present analysis rests. The participants in that
interagency group are listed in Appendix G of the report.
For the analysis project itself, the project team benefitted from the contribution of staff
in several federal agencies, as well as members of the project's expert review panel. Gary
Bickel, Margaret Andrews, and Steven Carlson (USDA, Food and Consumer Service) provided
guidance, support, spirited debate, and detailed critiques throughout the project. Thoughtful
review and commentary came also from Mark Nord, Donald Rose, and Victor Oliveira (USDA,
Economic Research Service), Ronette Briefel (DHHS, Centers for Disease Control and
Prevention, National Center for Health Statistics), and Richard Bavier (Office of Management
and Budget). Finally, we are especially grateful for the time and effort devoted by the members
of the expert review panel listed below, most of whom participated in a group review meeting
in Washington and many of whom also provided insightful written reviews.
Expert Review Panel Members
Dr. George H. Beaton
Department of Nutritional Sciences
University of Toronto
Dr. Janice Dodds
School of Public Health
Department of Nutrition
University of North Carolina-Chapel Hill
Dr. Jean-Pierre Habicht
James Jamison Professor of Nutritional Epide-miology
Division of Nutritional Sciences
Cornell University
Dr. Helen Jensen
Professor of Economics, Head of Food and
Nutrition Policy Research
Center for Agricultural and Rural Develop-ment
(CARD)
Iowa State University
Dr. Stanley R. Johnson
Vice Provost for Extension
Director, Center for Agricultural and Rural
Development
Iowa State University
Prepared by Abt Associates Inc. v
rrr'A
Acknowledgements
Ms. Anna Kondratas
Co-Director, Assessing the New Federalism
Project
The Urban Institute
Dr. Shiriki Kumanyika
Chair, National Nutrition Monitoring Advisory
Council
College of Associated Health Professions
Department of Human Nutrition and Dietetics
University of Chicago
Ms. Kathryn Porter
Research Director
Center on Budget and Policy Priorities
Dr. Robert Reischauer
The Brookings Institution
Dr. Victor Sidel
Distinguished Professor of Social Medicine
Montefiore Medical Center and Albert Einstein
College of Medicine
Dr. Valerie Tarasuk
Department of Nutritional Sciences
University of Toronto
Dr. Benjamin Wright
Director, Cognitive Research
MESA Psychometric Laboratory and Press
Department of Education
University of Chicago
Prepared by Abt Associates Inc. V\
EXECUTIVE SUMMARY
In April 1995, the U.S. Bureau of the Census conducted the first Food Security
Supplement to its regular Current Population Survey (CPS). With about 45,000 household
interviews, the Food Security Supplement provides the basis for the first comprehensive
measurement of food insecurity and hunger in a nationally-representative sample of U.S.
households. This survey is the cornerstone of the food security measurement project begun in
1992 to carry out a key task assigned by the Ten-Year Comprehensive Plan for the National
Nutrition Monitoring and Related Research Program (NNMRRP). The task is to develop a
standard measure of food insecurity and hunger for the United States, for use at national, state,
and local levels.
This project has been a cooperative undertaking by the responsible federal government
agencies under the leadership of the Food and Consumer Service (FCS) of the U.S. Department
of Agriculture jointly with the National Center for Health Statistics/Centers for Disease Control
and Prevention (NCHS) of the Department of Health and Human Services. Academic and other
private-sector research experts in the field of food security and hunger measurement have aided
the project from its beginning, achieving a substantial public/private partnership in the effort to
develop a state-of-the-art food security survey questionnaire, statistical measurement method, and
food insecurity and hunger measures and prevalence estimates for the nation.
The present study reports the first of these national prevalence estimates for food
insecurity and hunger for the 12-month period ending in April 1995, based on the CPS data and
applying a sophisticated statistical measurement method that creates a detailed scale for
measuring the underlying level of severity of food insecurity and hunger experienced in U.S.
households. Based on this food security scale, a simpler measure is constructed that classifies
households into several broad ranges or levels of severity, defining four categories of food
security status for U.S. households:
• food secure,
• food insecure without hunger,
• food insecure with moderate hunger, and
• food insecure with severe hunger.
Prepared by Abt Associates Inc. V\\
Executive Summary
The categorical measure allows one to estimate the number of American households that
experience food insecurity and hunger within each of the broad levels specified. The measure
is designed to be useful primarily for monitoring changes in prevalence over time, and
comparing prevalence across groups within the population, on a sustained, consistent basis.
Background and Definitions
Food security has been defined briefly as "assured access to enough food for an active,
healthy life." The household should have access to enough food, the food should be nutritionally
adequate, it should be safe, and the household should be able to obtain it through normal
channels. Although all of these dimensions of food security are important, the measure
presented here focuses on whether the household has "enough" food, as perceived and reported
by adult members of the household. When food insecurity on this central dimension reaches
severe levels, actual hunger for household members is the result.
Hunger is defined briefly as "the uneasy or painful sensation caused by a lack of food."
The CPS Food Security Supplement aims to measure only that hunger which results from the
financial resource constraint of the household—from being unable to afford enough food. The
survey does not measure hunger that results from being too busy to eat, from voluntary fasting,
from illness, or from any other cause except lack of financial resources. Thus, food insecurity
and hunger measured here are clearly related to general income poverty. They focus, however,
on only one area of household circumstances, rather than on the general problem of whether
resources are adequate to cover all areas of need.
Interest in measuring food insecurity and hunger springs from two sources. First, food
security is an important dimension of basic individual and family well-being, analogous to health
or housing. Food insecurity and hunger are undesirable in their own right, and possible
precursors to more serious health and developmental problems. Monitoring food security is
important for understanding one fundamental component of the well-being of the American
population and for identifying geographic or other subgroups with particularly undesirable and
high-risk conditions.
Second, numerous public and private food assistance programs attempt to ameliorate
food insecurity and hunger. Accurate measurement of food insecurity and hunger are important
for program planners and policy makers to assess adequately die effectiveness of these programs
Prepared by Abt Associates Inc. H
Executive Summary
in meeting their intended objectives. This need for concrete indicators of program outcomes
takes on new importance for federal agencies under the mandate of the 1993 Government
Performance and Results Act (GPRA), which requires agencies to give increased, explicit
attention to such indicators.
The government's food security measurement effort was built upon extensive private-sector
research in the late 1980s that expanded and sharpened the understanding of food security,
food insecurity, and hunger. This work led to the development by an expert working group of
the American Institute of Nutrition of the following conceptual definitions, which were published
by the Life Sciences Research Office (LSRO) of the Federation of American Societies for
Experimental Biology (Anderson/AIN/LSRO, 1990):
• Food security - "Access by all people at all times to enough food for an active,
healthy life. Food security includes at a minimum: (1) the ready availability of
nutritionally adequate and safe foods, and (2) an assured ability to acquire
acceptable foods in socially acceptable ways (e.g., without resorting to emergency
food supplies, scavenging, stealing, or other coping strategies)."
• Food insecurity — "Limited or uncertain availability of nutritionally adequate and
safe foods or limited or uncertain ability to acquire acceptable foods in socially
acceptable ways."
• Hunger — "The uneasy or painful sensation caused by a lack of food. The
recurrent and involuntary lack of access to food. Hunger may produce malnutrition
over time. . . . Hunger ... is a potential, although not necessary, consequence
of food insecurity."
These definitions underlie the CPS Food Security Supplement and the new measurement
scale discussed below, with the one additional qualification, already described, that only
resource-constrained or poverty-linked food insecurity and hunger are intended to be captured
by the measure.
The Food Security Scale
The Food Security Supplement contains a large battery of questions asking respondents
about various aspects of food sufficiency in their households. Taken individually, none of these
questions can provide a measure of the severity and extent of food insecurity or hunger. Taken
together, a systematic set of 18 of the CPS questions (those with strong statistical properties
Prepared by Abt Associates Inc. iii
/X
Executive Summary
identified by the measurement method) do provide such a measure. The CPS questions ask
about five general types of household food conditions, events, or behaviors:
• Anxiety that the household food budget or food supply may be insufficient to meet
basic needs;
• Perceptions that the food eaten by household members was inadequate in quality
or quantity;
• Reported instances of reduced food intake, or consequences of reduced food intake
(such as the physical sensation of hunger or reported weight loss) for adults in the
household;
• Reported instances of reduced food intake or its consequences for children in the
household; and
• Coping actions taken by the household to augment their food budget or food supply
(such as borrowing from friend,-, or family or getting food from emergency food
pantries).
All of the CPS food security questions explicitly condition the event or behavior
identified as being due to financial limitation (such it "... because we couldn't afford enough
food" or "because there wasn't enough money to buy food.") Each question addresses an
explicit time frame, either the past 12 months or the past 30 days. Several key items include
follow-up questions on how often the event or condition occurred within the past 12 months or
the past 30 days.
Two separate measurement scales were developed, one for the severity of food
insecurity within the 12-month period, the other for the 30~day period. The 12-month scale
covers a broader range of severity levels of food insecurity and hunger, because fewer questions
were asked in the 30-day time frame. The more comprehensive 12-month measure is expected
to be the more useful, both for research and policy purposes, and is the focus of discussion in
this report.
The scaling methodology began with exploratory linear and non-linear factor analyses
to determine the number of distinct factors that should be represented. Scales were estimated
using a Rasch measurement model, a form of non-linear factor analysis in the family of Item
Prepared by Abt Associates Inc. iv
Executive Summary
Response Theory models.1 Most food insecurity and hunger questions met the statistical
criteria for inclusion in the models, although the resource augmentation questions did not. The
final 12-month food security scale is based on answers to 18 questions, including some from
each of the first four types of questions identified above.
Key findings during the scaling analysis were as follows:
• The results are consistent with previous research characterizing food insecurity as
a "managed process" through several stages or levels of severity (Radimer et al.,
1992). In this process, households first note serious inadequacy in their food
supply, feel anxiety about the sufficiency of their food to meet basic needs, and
make adjustments to their food budget and food served. As the situation becomes
more severe, adults experience reduced food intake and hunger, but they spare the
children this experience. In the third stage, children also suffer reduced food
intake and hunger and adults' reductions in food intake are more dramatic.
The severity ranking of questions in the measurement scale proceeds generally in
this order. At the same time, it shows that all three stages fit well in a single
scale, which means that the level of severity of food insecurity can be measured
as an essentially unidimensional aspect of the food insecurity/hunger phenomenon.
• The measurement models were tested with three different population groups:
households with children; those without children but with one or more elderly
members (age 60 or older); and those with neither children nor elderly members.
Tests showed that a single scale can be used with all three populations.
• An extensive series of tests found the food security scale to have good reliability,
including good internal (or content) validity and good external (or construct)
validity.
Defining Levels of Severity of Food Insecurity and Hunger
Four categories of food security status are defined, based on the distinct behavioral
stages associated with the managed process of food insecurity and hunger:
• Food secure — Households show no or minimal evidence of food insecurity.
1IRT models are a form of statistical measurement model developed in educational testing, where test
Hems vary systematically in difficulty and the overall score measures the level of difficulty that the tested
individual has mastered. In the present application, the severity of food insecurity that the household has
experienced is analogous to the level of test difficulty that an individual has mastered.
Prepared by Abt Associates Inc.
tfl
Executive Summary
Food insecure without hunger — Food insecurity is evident in households'
concerns and in adjustments to household food management, including reduced
quality of diets. Little or no reduction in household members' food intake is
reported.
• Food insecure with moderate hunger — Food intake for adults in the household
has been reduced to an extent that it implies that adults have repeatedly experienced
the physical sensation of hunger. Such reductions are not observed at this stage for
children in the household.
• Food insecure with severe hunger — Households with children have reduced the
children's food intake to an extent that it implies that 'he children have experienced
the physical sensation of hunger. Adults in households with and without children
have repeatedly experienced more extensive .eductions in food intake at this stage.
Each household is classified into one of the four food security status categories on the
basis of its value on the food security scale; Exhibit ES-1 illustrates the process. Households
with zero scale score are those reporting no indications at all of food insufficiency or insecurity.
Households with low scale values are those reporting very slight experiences of food insecurity.
Both these groups are classified as food secure. At the other extreme, households with high
scale values are those who report experiencing all or nearly all of the conditions covered by the
scale, and are classified as food insecure with severe hunger. A household classified into a
particular category must normally have experienced all of the conditions associated with the less-severe
categories, plus at least two or three of the conditions associated with the assigned
category.
The Prevalence of Food Insecurity and Hunger in the United States
The large majority of American households were food secure in the year ending April
1995. About 88.1 percent of the approximately 100 million households in the United States are
classified as food secure over that period, as illustrated in Exhibit ES--2. About 11.9 million
households, however, experienced food insecurity at some level during that year.
Most of the food insecure households are classified as food insecure without hunger (7.8
percent, or 7.8 million households). About 4.1 percent, however, are classified as food insecure
with hunger. Thus, one or more adult members of some 4.2 million American households are
estimated to have experienced reduced food intake and hunger as a result of financial constraints
in the year ending in April 1995.
Prepared by Abt Associates Inc. vi • Yd
Executive Summary
Exhibit ES-1
THE FOOD SECURITY STATUS CATEGORIES
keU's
Vmbum
Feed
Seam*,
Seek
The Severity Ranking efQuestions Reflects
the Managed Process »f Feed Insecurity
Feed Security Status Depends en
the CemmpUH Set efCemmmiem
Experienced by tke HmuuheU
No Food
Insecurity
orHufor
CoadMoas
Reported
W
PnpandbyAbtAieockites Inc. vn yiii
Executive Summary
Exhibit ES-2
PREVALENCE OF FOOD SECURITY AND HUNGER, 1995
JBBE3KB&BBB5&
fvoitnmamHtaiHmgmTm
Among the households experiencing some level of hunger, about 800,000 (0.8 percent)
are classified as food insecure with severe hunger. In these households, children as well as
adults experienced reduced food intakes and hunger. Adults in these households had very
substantial reductions in food intake, such as not eating for a whole day because of lack of
money.
Food insecurity is clearly related to income and poverty, but the relationship is not
exact. Not all poor households are food insecure, and only a small percentage of households
with below-poverty incomes experience actual hunger (13.1 percent). The percent of households
estimated to experience food insecurity is somewhat less than the poverty rate for individuals in
the same period (12 percent vs. 15 percent). More than a third of poor households are classified
as food insecure, whereas only 8 percent of households with above-poverty incomes are food
insecure, and moat of those have near-poverty incomes. Public and private food assistance
programs may account for the fact that so many poor households are food secure, but this
hypothesis has not yet been analyzed.
Even though food insecurity does not exactly follow income lines, food insecurity tends
to be concentrated in population groups that have comparatively high poverty rates. For
example, food insecurity rates are higher than average in female-headed households, in
Prepared by Abt Associates Inc. viii W
Executive Summary
households with children (especially young children), in Black and Hispanic households, and in
central city areas.
Next Steps
The present analysis represents an important step in the measurement of food security,
food insecurity, and hunger, but much more lies ahead for the food security measurement
project. A task for the immediate future is to identify subsets of the questions in the CPS Food
Security Supplement, and appropriate scaling procedures, so that smaller survey efforts can
approximate the scale presented here with reasonable reliability. Another ongoing effort is to
refine and strengthen the Food Security Supplement itself, so that the annual surveys planned
for the future will yield comparable and increasingly reliable information. In the longer term,
FCS and the larger research community will be undertaking several lines of data collection and
analysis to understand better the phenomenon of food insecurity and to apply that understanding
in the design and implementation of nutrition policies and food assistance programs.
Prepared by Abt Associates Inc. IX
CHAPTER ONE
FOOD SECURITY AND HUNGER MEASUREMENT
IN THE UNITED STATES
Introduction
One of the basic aims of U.S. public policy in the latter half of the 20th century has
been to assure that all Americans have enough to eat. The President's Task Force on Food
Assistance stated the theme in its 1984 report:
// has long been an article of faith among the American people that no one in
a land so blessed with plenty should go hungry. . . . Hunger is simply not
acceptable in our society. — Task Force Report, p. 2
The commitment to reduce and ultimately eliminate poverty-linked hunger in the United
States has been expressed in the allocation of public resources to major public programs of food
assistance targeted to families and persons in need. Beginning in the 1960s, food assistance
programs grew to be an important part of the general social safety net of government-aided
programs aimed at reducing poverty. By 1996, some $35.6 billion of federal funds were
devoted to food assistance to American families and single persons. Nevertheless, despite the
amount of these resource transfers, either as direct emergency food aid or financial means to
obtain food through normal channels of trade, food insecurity and hunger due to lack of adequate
financial resources continues to be a problem for some Americans.
In order for the policies and programs aimed at reducing food insecurity and hunger to
be directed effectively, it is important to be able to measure with some degree of confidence the
conditions that the policies and programs are intended to affect. Lack of reliable measures with
which to gauge their impact may hamper the effectiveness and appropriateness of the policies
and programs themselves; at the least, lack of such measures leaves policymakers and the public
in doubt as to die actual effect of food-assistance programs. The 1984 Task Force Report noted
die lack, up to that time, of any authoritative measure of die number of people in the U.S.
Prepared by Abt Associates Inc.
Chapter One: Food Security and Hunger Measurement in the United States
experiencing poverty-related hunger, and the problem for policy making caused by this lack of
a reliable hunger measure.l
Food Insecurity and Hunger Measurement—Background
In 1977, the federal government began collecting information on food sufficiency in
American households through a single question included in the periodic national food
consumption surveys conducted by the U.S. Department of Agriculture (USDA). In the 1980s,
additional questions on food insecurity and hunger were included in the Third National Health
and Nutrition Examination Survey (NHANES III) conducted by the Department of Health and
Human Services, National Center for Health Statistics/Centers for Disease Control and
Prevention.
The challenge implicit in the 1984 President's Task Force Report—to develop a valid
and reliable measure of the severity and extent of hunger in the U.S.—was taken up most
actively by scientists and researchers in the private sector, both in academia and under
sponsorship by concerned social-policy and policy-research organizations. These private-sector
efforts to develop and implement technically competent, scientifically grounded measurement of
the conditions of food insecurity and hunger in the U.S. in the latter 1980s, demonstrated the
feasibility of developing such measures.2 This body of research and field survey experience
produced methodologically sophisticated, empirically grounded measurement scales for food
1 "While we have found evidence of hunger in the sense that some people have difficulty obtaining
adequate access to food, we have also found that it is at present impossible to estimate the extent of that
hunger. We cannot report on any indicator nut will tell us by bow much hunger has gone up in recent yean.
.... Since general claims of widespread hunger can neither be positively refuted nor definitively proved, it
seems likely that the issue of hunger will remain on our national policy agenda for an indefinite future."
(Task Force Report, Chapter S: How Much Hunger is There in America?—Conclusion, p. 39.)
2 Two major sustained research efforts in particular during this period provided die Hchatal basis for the
direct household-level measurement of food insecurity and hunger under working definitions relevant to the
U.S. context. One is the work of Wehler and colleagues, beginning with the 1983 Massachusetts Nutrition
Survey and continuing with the 1983 New Haven Risk Factor Study, the initial pilot study of the Community
Childhood Hunger Identification Project, or CCHIP (Wehler, 1986; Wehler, Scott and Anderson, 1991,1992,
1995a,b). The other is die work of Radimer and colleagues in the Cornell University Division of Nutritional
Sciences, including Radimer's 1990 doctoral dissertation and subsequent work at Cornell to develop and
extend this approach (Radimer, 1990; Radimer, Olson and Campbell, 1990; Campbell, 1991; Radimer et al.,
1992; Kendall, Olson and Frongillo, 1993; Olson, Frongillo and Kendall, 1993). A third important
contribution to this body of research, focusing on food insecurity and hunger as experienced by elderly
persons, is Burt, 1993, and Cohen, Bun and Schulte, 1993.
Prepared by Abt Associates Inc.
Chapter One: Food Security and Hunger Measurement in the United States
insecurity and hunger in households lacking resources to obtain sufficient food, and demonstrated
the practical means of creating such measures from reasonably obtainable social survey data.
During the same period, a consensus was emerging within the nutrition community over
the appropriate conceptual basis for identifying and measuring U.S. hunger, viewing it as an
element or consequence of a broader condition of food inadequacy associated with poverty and
identified as "food insecurity." An important step in this direction had been taken by the
President's Task Force in recognizing the distinction between clinical or medical definitions of
hunger, on the one hand, and "hunger as commonly defined," on the other. Simply put, the
medical definitions associate hunger closely with malnutrition, "a weakened, disordered
condition brought about by prolonged lack of food" (Report, p. 34), identifiable from clinical
indicators such as weight loss in adults and serious underweight or stunting of growth in
children. By the time hunger shows up in these clinical measures, however, the condition has
persisted over a long period of time. The clinical definition and measures of hunger thus do not
provide sensitive indicators of food insufficiency and hunger as these are primarily experienced
in the U.S. context. Nor do they respond to the policy concern to address hunger and the risk
factors for hunger—especially for children—as soon as these appear, rather than only after they
have persisted for extended periods at substantial levels of severity.3
3 In addressing "hunger at commonly defined," the 1984 President's Task Force helped clarify the shift
from an exclusively medical definition of hunger to an alternative social definition more relevant to actual
U.S. conditions. The discussion also anticipates die later attention to food insecurity, recognizing that a
broader condition of food problem than hunger, as such, provides the context within which resource-constrained
hunger is experienced:
To many people hunger means not just symptoms that can be diagnosed by a physician,
it bespeaks the existence of a social, not a medical, problem: a situation in which
someone cannot obtain an adequate amount offood, even if the shortage is not prolonged
enough to cause health problems. It ism* experience of being unsatisfied, of not getting
enough to eat. This, of course, is the sense in which people ordinarily use the word. It
is also the sense in which the witnesses before us and many of the reports and documents
we haw studied have spoken ofhunger.. .. And in mis sense, we cannot doubt mat there
is hunger in America. This is the sad truth. It is easy to think of examples of this kind
of hunger: children who sometimes are sent to bed hungry because their parents find it
impossible to providefor them; parents, especially mothers, who sometimes forgofood so
that their families may eat; the homeless who must depend on the largess of charity or
who areforced to scavengeforfood or beg; and people who do not eat property in order
mat they save money to pay rent, utilities, and other bills. (Report, p. 36)
Prepared by Abt Associates Inc.
Chapter One: Food Security and Hunger Measurement in the United States
Two events in 1990 mark the emergence of consensus on the appropriate concepts of
food insecurity and hunger relevant for the U.S. First was sponsorship and publication by the
American Institute of Nutrition (AIN) of a major report prepared by the Life Sciences Research
Office (LSRO) of the Federation of American Societies for Experimental Biology, Core
Indicators ofNutritional Statefor Difficult-to-Sample Populations (Anderson/LSRO, 1990). The
AIN/LSRO report provides authoritative definitions of food security, food insecurity, and hunger
as key areas for further development and measurement. These LSRO definitions provide the
basic conceptual underpinnings for the present measurement project and guided the development
of its measurement objectives.
The second event noting a coming of age of food security measurement was the passage
by the U.S. Congress of the National Nutrition Monitoring and Related Research Act of 1990,
mandating creation of a joint plan of action by USDA and DHHS for comprehensive nutritional
monitoring of the U.S. population. Subsequently, the Ten-Year Comprehensive Plan for the
National Nutrition Monitoring and Related Research Program (NNMRRP) included the task
assignment to:
Recommend a standardized mechanism and instrument^) for defining and
obtaining data on the prevalence of "food insecurity" or "food insufficiency"
in the U.S. and methodologies that can be used across the NNMRRP and at
State and local levels.
Responsibility for carrying out the development of standardized measures of food
insecurity and insufficiency for the U.S. is assigned under the Ten-Year Plan jointly to the Food
and Consumer Service (FCS) of USDA and the National Center for Health Statistics/Centers for
Disease Control and Prevention (NCHS) of DHHS. Beginning in 1992, FCS and NCHS
established a federal interagency working group to carry out the assigned task, initiating the
present food security measurement project. The present report represents the first major product
resulting from this continuing development effort.
Food Insecurity and Hunger Measurement—Conceptual Basis
As noted, the 1990 AIN/LSRO report presents the nutrition community's understanding,
gained from die research on food insecurity and hunger up to that time, and provides the
conceptual basis for the present measurement project. The report defines food insecurity and
Prepared by Abt Associates Inc. 4
Chapter One: Food Security and Hunger Measurement in the United States
hunger in a way that clarifies the meaning of hunger, as directly experienced, spells out the
relationship between food insecurity and hunger, and makes it possible to measure them both
across the full range of severity of these conditions as they are experienced. Thus, the LSRO
definitions of food insecurity and hunger are critical in helping define the measurement
objectives of the present project. The conceptual definitions provided by the AIN/LSRO report
are referred to herein as the LSRO definitions (Anderson/LSRO, 1990, p. 1598). They are:
Food security — Access by all people at all times to enoughfoodfor an active, healthy
life. Food security includes at a minimum: (1) the ready availability of nutritionally
adequate and safe foods, and (2) an assured ability to acquire acceptable foods in
socially acceptable ways (e.g., without resorting to emergency food supplies,
scavenging, stealing, or other coping strategies).
Food insecurity — Limited or uncertain availability of nutritionally adequate and safe
foods or limited or uncertain ability to acquire acceptable foods in socially acceptable
ways.
Hunger — The uneasy or painful sensation caused by a lack offood. The recurrent
and involuntary lack of access to food. Hunger may produce malnutrition over time.
Hunger, as the recurrent and involuntary lack of access to food which may produce
malnutrition over time, is discussed asfood insecurity in this report.
Hunger, in its meaning of the uneasy or painful sensation caused by a lack offood, is
in this definition a potential, although not necessary, consequence offood insecurity.
Malnutrition is also a potential, although not necessary, consequence of food insecurity
(Ibid., p. 1576).
These conceptual definitions are consistent with the sequence of household food
conditions and behaviors revealed in the earlier research on hunger measurement. The
understanding of the phenomenon of food insecurity and hunger that they present recognizes the
distinction between the medical and social definitions of hunger described in the President's Task
Force Report, and clarifies the relationship of hunger to less severe conditions of food
insufficiency. The LSRO definitions also reflect efforts to make scientific research findings from
hunger and nutrition studies more relevant and useful in the public policy arena (Habicht and
Meyers, 1991) and to reduce confusion arising from multiple definitions and interpretations of
the term hunger.
Prepared by Abt Associates Inc.
Chapter One: Food Security and Hunger Measurement in the United States
In this perspective, hunger has the common meaning of a physical sensation that is
familiar to everyone through direct personal experience. Poverty-linked hunger, the potential
object of public policy concern, is distinguished from other hunger by its primary proximate
cause. The hunger identified by this definition occurs as a consequence of food insecurity,
nested within that broader poverty-linked concept.4
The LSRO definitions clarify the relationship between the concepts of food insecurity
and hunger. Hunger is a "potential although not necessary consequence of food insecurity."
Recognizing this relationship opened the possibility of measuring hunger and food insecurity
together, by means of a single measurement scale. In such a scale, hunger would lie in the more
severe part of the range. The less severe part of the range would capture more limited food
insufficiency and some of the household coping behaviors that represent responses to food
insufficiency.
This idea of a coherent underlying phenomenon, varying through distinct levels of
severity and revealing an orderly sequence of characteristic conditions and behaviors, provides
the basis for the current measurement effort. If a comprehensive set of indicators for the various
aspects of food insufficiency and associated household coping responses are found to fall in a
regular, orderly sequence from recognizably less severe to more severe conditions of
inadequacy, then a scaled measure is both feasible and appropriate for gauging the severity and
extent of the phenomenon.
Food insecurity and hunger may also be seen as one potential facet of poverty, as
manifested in this particular area of bask need. Observing the distinct conditions and behaviors
that characterize food insecurity reveals the kind of economizing and coping efforts that
households make in trying to manage their available resources when these are insufficient to
fully cover basic needs. The identification and measurement of food insecurity and hunger may
thus help provide better understanding of poverty in general. Moreover, if the observed
4 The terminology and concept of food security and food insecurity, which originally referred to issues
of community-wide food supply in lower-income countries, were found useful in describing issues of
househokl-levc: food sufficiency and access in the U.S. as well. During the same period, the international
literature on food security in low-income countries was also beginning to apply the concept to die household
level, as a feasible and sensitive early warning indicator for potential or approaching food sufficiency problems
in the general population or population subgroups. (See, e.g., Daniel G. Maxwell, Measuring Food
Insecurity: The Frequency and Severity of "Coping Strategies," Washington, DC: International Food Policy
'search Institute, Discussion Paper #8, December 1995.)
Prepared by Abt Associates Inc.
Chapter One: Food Security and Hunger Measurement in the United States
indicators of food insecurity are surveyed and measured consistently over time, then the regular
national measurement of food insecurity and hunger can provide an important supplement to the
established measures of income poverty as a tool for monitoring changes over time in the well-being
of the population, and differences across population subgroups.
The Hypothesis of Hunger as Severe Food Insecurity
The idea of an orderly, normal sequence of behaviors as households strive to cope with
increasingly insufficient food resources, represents a central working hypothesis of the present
hunger measurement project. In summary, the hypothesis is that hunger may be seen as a
consequence of persistent or worsening food insecurity, appearing when the condition reaches
sufficiently severe levels. Hunger is viewed as nested within the broader concept and experience
of household food insecurity, and food insecurity results from an immediate lack of household
income or other financial resources.
When they experience food insecurity due to limited or reduced resources, household
food managers, usually mothers or female heads of household, may attempt to deal with an
insufficient household food supply through a variety of coping behaviors and strategies. This
management of the conditions of food insecurity may include attempts to augment household
food from irregular or emergency sources, and may involve reductions in the quality and/or
quantity of food available within the household. Moreover, this managed process occurs within
the context of tightly constrained economic choices, likely to involve uncertainty as to future
availability of adequate food. Consequently, it is identifiable in part by characteristic affective
states, such as anxiety or worry about whether food or money will last, or whether more can
be obtained before food supplies run out.
Under this hypothesis, if household food sufficiency declines further, efforts to manage
the process eventually require reductions in food intake among one or more household members.
Reduced food intake is likely to occur initially via reduced serving sizes, reduced overall meal
sizes, or skipped meals. These behaviors will provide the first indication that actual hunger,
"the uneasy or painful sensation caused by a lack of food," is being experienced by household
members as part of the effort to manage an insufficient household food supply. In households
with children, reduction of food intake is expected to occur first among adults, as they attempt
to spare the children from food intake reduction.
Prepared by Abt Associates Inc. 7
Chapter One: Food Security and Hunger Measurement in the United States
If efforts to cope with an intensifying degree of food insecurity are not successful,
reductions in food intake and hunger will also occur among children in the household. When
children's hunger occurs it may be viewed as indicating a more severe condition, partly because
the consequences of hunger are likely to be more damaging for children than adults, and partly
because adults in the household normally will have experienced hunger for some period of time
prior to the children. If household hunger persists or recurs often enough, observable signs of
malnutrition will appear among either the adults or children in the household, or both (Radimer,
Olson and Campbell, 1990; Radimer et al., 1992; Wehler, Scott and Anderson, 1992). Before
such clinical signs of malnutrition become evident, however, the quantity and nutritional quality
of diets in food-insecure households will necessarily have been deficient for some extended
period of time.
The central hypothesis that food insecurity and hunger represent a coherent range of
conditions and experience, that these are different and distinct from the nutritional quality of
diets, and that they can be directly observed and measured, is put to the test by the attempt to
develop a measure that is based on the hypothesis. Other implications of the hypothesis,
however, are not addressed by the measurement itself. For example, the expected relationship
between food insecurity and hunger as measured in this study and the nutritional quality of diets
as measured by nutritionists can be tested only after the food security measure is available for
comparison with established nutritional measures. If the food insecurity and hunger measure is
found to be closely correlated with dietary quality, food insecurity and hunger measures may
prove useful as simple indirect indicators of the nutritional adequacy of diets. It will be
important for future research to explore the exact nature of the interrelationships among poverty-linked
food insecurity, hunger, and malnutrition.5
The Role of Food Security Meat urement
Reliable measures of food insecurity and hunger and consistent estimates of their
prevalence in the population can meet the needs of policymakers in designing and directing
effective policies and programs to address these conditions. Although considerable progress
5 Several research studies have demonstrated the link between food insufficiency as experienced and
nutritional inadequacy of diets. A recent example is D. Rose and V. Oliveira, "Nutrient Intakes of Individuals
from Food Insufficient Households in the United States," American Journal of Public Health (forthcoming).
Prepared by Abt Associates Inc.
Chapter One: Food Security and Hunger Measurement in the United States
occurred during the past decade in developing technically sound, scientifically-grounded methods
to measure food insecurity and hunger, accurate national measures from which consistent
prevalence estimates could be derived have not been available. Inclusion of the goal to develop
such measures in the ten-year comprehensive plan for the NNMRRP reflected the widely-held
view within both the social policy and scientific communities of the importance of the food
security of the nation's population. As a result, designing a survey instrument for collecting
national data on food security and applying state-of-the-art measurement methods to create
reliable and consistent national benchmark measures was identified as an explicit objective of
national policy.
Accurate measurement of these conditions on a consistent basis from year to year is
expected to provide a valuable tool for administrators and policymakers at several levels, state
and local as well as national. Such measures can help identify those segments of the population
most in need, assess the impacts of changing economic conditions and public programs on this
basic element of well-being, and monitor the success of efforts to reduce poverty-linked hunger
over time. For these uses, the most important aspect of the measures is their degree of
reliability and consistency: the measures should provide the ability to track year-to-year changes
in food insecurity and hunger at several specified, well-defined levels of severity, and provide
a reliable set of standard national benchmark measures for consistent application and comparison
with equivalent state and local measures.
From the standpoint of sound measurement method, the foremost concern is that the
measures of food insecurity and hunger that are developed yield valid and reliable descriptive
statements about the existence and extent of the phenomenon. The food security measurement
project cannot determine the causes of food insecurity, nor whether its existence is a serious
social problem requiring a policy response. Those judgments will be made by policymakers,
advocates, and the general public. Results from the consistent and reliable measurement of food
security can, however, be expected to help inform and strengthen those judgements. In the
remainder of this chapter, we summarize some of the considerations contributing to incorpora-tion
of food security measurement into the national policy agenda.
Child and Adult Health Considerations. Economically, a well-prepared work force
is essential to America's success in the rapidly changing global economy. Sound physical and
mental health are key factors in providing the skilled, well-educated workers demanded by
Prepared by Abt Associates Inc. 9
Chapter One: Food Security and Hunger Measurement in the United States
increasingly technical service-oriented domestic labor markets. Good nutrition throughout the
life cycle, but especially during childhood, is a necessary prerequisite for successful
physiological and cognitive development and maintenance of sound health (Munro, Suter and
Russell, 1987; Duncan, Brooks-Gunn and Klebanov, 1994; Pollitt, 1994; Frazao, 1995;
Kretchmer, Beard and Carlson, 1996).
Evidence from recent research in child development indicates that school performance,
cumulative educational achievement, and mastery of skills are affected both by physiological
factors related to adequate nutrient intake and by factors related to food security and sufficiency
of food intake (Pollitt, 1994). In addition to detrimental effects on physical growth and
cognitive development resulting from chronic or severe undernutrition, serious cumulative
deficits also accompany chronic lack of access to adequate food (Pollitt, Leibel and Greenfield,
1981; Meyers et al., 1989). Simple hunger—"the uneasy or painful sensation caused by a lack
of food"—can interfere with a variety of behaviors necessary to successful learning—e.g.,
concentration, ability to maintain the focus of one's attention, achievement motivation, and
inclination toward physical activity.
Young children especially need frequent intake of nutritionally adequate food to
maintain food energy stores needed for effective activity. A child's small liver size relative to
total body mass limits its capacity to store sufficient glycogen for ready conversion to energy
over extended time periods. Therefore, children need to eat more frequently and regularly than
adults to maintain needed levels of available energy. Moreover, most nutrient requirements
increase dramatically during periods of rapid growth, further amplifying the importance of
adequate nutritious food for overall healthy growth and development.
The concept of "sentinel groups," as applied to disease and nutrition surveillance
systems, is prominent in public health. Sentinel groups can be predictive of future events or
conditions, and are often selected for monitoring as a result. Such groups have characteristics
that make them likely to be the first in the population to contract a disease or suffer from
malnutrition (Anderson/LSRO, 1990, pp. 1574-1575). Knowledge of changes in conditions
among sentinel groups can often enable policymakers and health officials to implement responses
that help avoid widespread occurrence of more costly diseases or conditions. Food-insecure
households may comprise a sentinel group in which hung* T, undernutrition, and poor health are
Prepared by Abt Associates Inc. 10
Chapter One: Food Security and Hunger Measurement in the United States
more likely to occur (Munro, Suter and Russell, 1987; Wehler, Scon and Anderson, 1992;
Pollitt, 1994; FrazSo, 1995; Wehler, Scott and Anderson, 1995a,b).
Several surveys included in the NNMRRP provide information on food intake and
undernutrition. The emergence of nutrient deficiencies over time, however, implies that
households in which affected individuals reside are likely to have experienced what is now
understood to be a progression through worsening levels of food insecurity. In this view, hunger
and undernutrition are understood to occur at the more severe levels of food insecurity, whereas
serious nutrient deficiencies are likely to occur only after chronic food insecurity with hunger
has been experienced. The progressive and nested nature of hunger and undernutrition within
food insecurity thus make measures that identify the entire range of food insecurity valuable as
sensitive leading indicators for more serious health consequences. Thus, accurate and reliable
measures of food insecurity at its various levels of severity will provide valuable information for
informing and guiding national and state policies.
Conclusion
The new CPS food security data, and the standard measurement method for severity and
extent of food insecurity based on the data, are expected to provide useful resources for research
into the causes and consequences of food insecurity and hunger. The subject area of food
security poses challenges and opportunities for researchers, particularly because of the overlap
between public health and nutrition concerns, on the one hand, and concerns of general poverty
policy, on the other. The food security measures provide new information relevant to both these
fields, and to the relationship between them. The utility of the new data and measurement for
research, however, is only a secondary reason for obtaining them. The primary purpose is to
provide a broad new assessment and monitoring tool for policymakers and administrators of
government food assistance programs at all levels.
Chapters Two through Four of this report describe the operational measurement
concepts, survey questionnaire design, food security data collected, and the analytic procedures
used in developing a measurement scale. Chapter Five presents the initial prevalence estimates
for food insecurity and hunger in the United States resulting from the new measure. Chapter
Six discusses the reliability and limitations of the measure.
Prepared by Abt Associates Inc. \ \
Chapter One: Food Security and Hunger Measurement in the United States
More detailed explanation and documentation of the methods used in developing the
food security measure are presented in the companion volume to the present report (Hamilton
etal., 1997).
Prepared by Abt Associates Inc. 12
CHAPTER Two
THE FOOD SECURITY SUPPLEMENT TO
THE CURRENT POPULATION SURVEY
This chapter briefly describes the development of the Food Security Supplement
questionnaire and the subsequent data collection effort undertaken for USDA by the U.S. Bureau
of the Census as a part of the Current Population Survey (CPS) for April 1995. The chapter
includes a short description of the 1994 Food Security Measurement and Research Conference
that preceded development of the national-level food security questionnaire for use in the CPS.
The final survey instrument that emerged from this conference and the subsequent development
process is described, as are the basic CPS sample and the Food Security Supplement subsample.
The Food Security Measurement and Research Conference
The Food Security Supplement instrument is based upon a synthesis of tested material
reported from earlier research. Initial consensus on the content of the instrument for national
use was attained during the January 1994 Food Security Measurement and Research Conference
convened jointly by FCS and the National Center for Health Statistics (NCHS) of the Centers
for Disease Control and Prevention. The conference was attended by nearly 100 professionals
with direct experience in areas related to nutrition, health, economics of food consumption, food
security policy, and hunger measurement. This working conference included presentations by
the authors of the primary research related to food insecurity and hunger measurement over the
previous decade.1 The second half of the conference was devoted to identifying a consensus
(with the aid of professional facilitators) among participants regarding the optimal content and
form of a food security survey instrument for application at the national level.
Several key issues that had been insufficiently addressed by earlier work needed to be
clarified before the LSRO conceptual definitions could be adapted for national data collection.
The resolution of these issues by FCS, conference participants, and a federal interagency
1 Transcript* of the presentations and discussion from this conference, with background papers and
participant list, were published in a volume entitled "Food Security Measurement and Research Conference:
Papers and Proceedings," USDA FCS, Office of Analysis and Evaluation, June 1995).
Prepared by Abt Associates Inc. 13
Chapter Two: The Food Security Supplement to the Current Population Survey
working group on food security measurement, led to the measurement approach implemented
in the current study.2 The key issues were:
• How to treat aspects of food insecurity that are not necessarily caused by a lack of
adequate household income, but are relevant for households at all income levels
(e.g., food safety concerns). The decision was to limit the current measure to
clearly poverty-linked or resource-constrained food insecurity and hunger.
• Whether to limit operational definitions to only those aspects of food security that
can be captured in household-level surveys. It was agreed that the FCS effort
should limit its measurement approach to the household. It was noted that agencies
involved in collecting individual-level data might develop complementary
approaches for measuring food insecurity at the specific individual level, whereas
issues ofcommunity food security would require a different data collection strategy
and orientation, outside the scope of the present effort.
• Whether indicators of nutritional adequacy would be incorporated into the
operational definition and measurement of food security. The decision was to
focus on the behavioral and experiential dimensions of food insecurity and hunger,
which were seen as the major gap in existing information and an essential
component for policymakers.
• How to estimate the prevalence of food insecurity and hunger from the resulting
data. Participants agreed on the desirability of scaling items into a single measure
across all observed levels of severity of the phenomenon being measured, if
feasible, and to develop a standard set of prevalence estimates at several designated
levels of severity for consistent application and comparison across data sets from
year to year.
During the year following the Food Security Measurement and Research Conference,
the national survey questionnaire underwent extensive further development, testing, and
refinement. Participants in the conference working sessions and the federal interagency working
group continued their contributions to this work, along with survey method specialists from the
Census Bureau's Center for Survey Methods Research (CSMR). The revised survey instrument
resulting from this development process was field-tested by the Census Bureau in August 1994
with approximately 600 regular CPS sample households. These field test results were analyzed
2 The measurement approach and its background in the research literature are described in Bickei,
Andrews and Klein, 1996. Participants in die Federal Interagency Working Group on Food Security
Measurement are listed in Appendix G.
Prepared by Abt Associates Inc. 14
Chapter Two: The Food Security Supplement to the Current Population Survey
by CSMR, and the instrument was further refined to incorporate a number of subsequent CSMR
recommendations.
The final version of the food security questionnaire was administered by the Census
Bureau as a supplement to its regular April 1995 CPS. In its final form, the questionnaire
contained 58 items intended to identify three levels of severity of food insecurity, including two
levels involving hunger on both a 12-month and a 30-day basis.
Questions in the Food Security Supplement
The questions in the food security questionnaire can be grouped into eight subject areas,
as summarized in Exhibit 2-1. The items in Part I are included primarily for the purpose of
helping validate the food security and hunger measures developed. Items in Part II are included
for assessing program impacts on food insecurity and hunger. Questions in Parts III-VIII were
designed to reflect the full observed range of severity of U.S. household food insecurity and
hunger, and to provide potential indicator items for inclusion in scale development analyses.
Two additional dimensions are embodied in the candidate scale items in Parts III-VIII
of Exhibit 2-1. All questions are asked of an adult respondent, usually the household member
with greatest knowledge of the household's food shopping and consumption, and relate generally
to the household unit as a whole. Some questions ask specifically about conditions or circum-stances
of the respondent, others ask about the adults generally in the household, and some ask
about the children generally (in households where children are present). Thus, items can be
classified as "Household," "Adult," or "Child" items.
In a second important distinction, all questions relate to one of two separate time
frames: the past 12 months or the past 30 days. Some 12-month items are followed by
subsequent items asking how often, or in how many months, a condition occurred during the
past 12 months. Similarly, several 30-day questions have follow-up items asking how often, or
in bow many days, a condition occurred during the previous 30 days. A few questions form
four-item sequences following the pattern: (i) "did it ever occur within the past 12 months?,"
(ii) "if so, in how many months did it occur?," (iii) "did it occur within the past 30 days?," and
(iv) "if so, on how many days did it occur?"
Both the adult-child and time dimensions of the items are conceptually related to aspects
of the hypothesized managed process of household efforts to cope with food insufficiency, as
Prepared by Abt Associates Inc. 15
Chapter Two: The Food Security Supplement to the Current Population Survey
Exhibit 2-1
SUMMARY OF FOOD SECURITY SURVEY ITEMS BY SUBJECT AREA
Description of Surrey Item Part Items in Each Part
Parti:
Weekly household food expenditures by place of purchase
(Eight items)
Q1-Q8
Part II:
Food assistance program participation by type of program
(Eight items)
Q9. Q9A - Q9G
Part III:
USDA and NHANES-III food sufficiency items, and follow-up
(Four items)
QUA, Q11.Q12. Q13
Part TV:
Existence of conditions requiring food-insufficiency coping behaviors
(Three items)
Q15-Q17
PartV:
Household food-supply-augmentation coping behaviors
(Six items)
Q18 - Q23
Part VI:
Adult food intake reduction items
(15 items)
Q24 - Q39
Part VII:
Child food intake reduction items
(13 items)
Q40 - Q52. Q57
Part VIII:
Radimer-scale food sufficiency items (adult and child food quality and
quantity concerns)
(Six items)
Q53-Q58
described in Chapter One. Research evidence had shown that when hunger emerges in food-insecure
households it usually appears first among adult members, affecting children only at
more severe levels.3 Thus, items addressing aspects of food-intake sufficiency for adults and
for children separately can provide a basis for measuring household food insecurity across
differing levels of severity.
Several types of periodicity have been observed in studies of household food
insufficiency and hunger.4 Food insecurity at the less severe levels is expected to be more
chronic in nature and less subject to this periodicity. For example, concerns about the adequacy
of household food supplies may persist for some time after a household experiences inadequate
3 Radimer, 1990; Radimer et at., 1992; Wehler, Scott and Anderson, 1992; Olson, Frongillo and Kendall,
1995.
4 See, for example, transcriptions of presentations by John Cook, Vaierie Tarasuk, and Janet Fitchen
included in "Food Security Measurement and Research Conference: Papers and Proceedings," USDA FCS,
June 1993.
Prepared by Abt Associates Inc. 16
Chapter Two: The Food Security Supplement to the Current Population Survey
food supplies. Hunger is a more acute condition, however, and in the U.S. context is more
likely to occur only periodically within households. For example, hunger may occur at the end
of month, when household food resources are depleted, but then subside after paychecks, food
stamps, or transfer payments are received. The two time periods addressed by survey items (12
months and 30 days) and follow-up items regarding frequency of occurrence are designed to
capture some part of this periodic aspect of food insecurity.
The Current Population Survey Sample
The Food Security Supplement was first fielded as a part of the April 1995 CPS. The
CPS is a nationally-representative monthly survey conducted by the Bureau of the Census in
approximately 58,000 households throughout the U.S. The CPS is a probability sample based
on a stratified sampling design. The overall sample is selected from lists of housing unit
addresses obtained from the most recent decennial census, and updated for new construction.5
The CPS sample is a state-based design, with primary sampling units (PSUs), consisting
of counties or groups of counties, selected in an initial sampling stage. The PSUs are grouped
into strata, with all strata defined within state boundaries. The sample is allocated among the
states to produce both state and national estimates with the required reliability, while keeping
total sample size to a minimum.6 Each stratum consists of one or more PSUs, with one PSU
chosen for the sample from within each stratum with probability proportional to its population
as of the most recent dfrmnial census.
In a second step, a sample of addresses is obtained within each sample PSU. Most
addresses are selected from census lists in a single stage within the selected PSU, although for
a relatively small proportion a second stage of selection is necessary. This two-step process is
roughly equivalent to a simple sampling plan of dividing each state into ultimate sampling units
5 This brief summary of the CPS sample is bated on documentation provided to users of the CPS public
use data tapes. For more detail, tee the U.S. Department of Labor, Bureau of Labor Statistics, "Redesign
of the Sample for the Current Population Survey," Employment and Earnings 41(5): 7-10, May 1994.
6 The ntft"4*1"1" of standard errors of state-level estimates produced from CPS data are related to the size
of each state's population. Therefore, estimates for states with large populations will be more reliable and
stable over time than those for stales with smaller populations. In general, state-level estimates for the ten
to twelve stales with largest populations are fairly stable, whereas those for the other states may vary
considerably from year to year due to greater sampling error and larger standard errors.
Prepared by Abt Associates Inc. 17
Chapter Two: The Food Security Supplement to the Current Population Survey
(USUs), each containing about four neighboring housing units, and selecting cluster samples of
these USUs for the interview.
The variables used for stratification within each state derive from the principal uses of
the CPS in providing reliable data for estimating labor force participation and characteristics.
The same stratification variables are used in all states, and include employment and
unemployment statistics by male, female, and total population; employment by occupation;
change in population; racial and ethnic composition of population; and other variables.
Each CPS sample is divided into eight approximately equal rotation groups, with each
group interviewed for four consecutive months, dropped out for eight months, then brought back
in for four more consecutive months before being permanently retired. This "four months fr-eight
months out—four months in" rotation leads to improved reliability of estimates of month-to-
month and year-to-year changes.
The weights for all interviewed households in the CPS sample are adjusted to account
for occupied households for which no information could be obtained. Some reasons for non-interview
include absence, impassable roads, refusals, or unavailability for other reasons. If a
respondent is reluctant to participate in the CPS, the interviewer informs the regional office
staff, and a follow-up letter is sent to the household with a fuller explanation of the CPS. If this
procedure fails to achieve participation, a supervisory field representative recontacts the
household and attempts to obtain participation through efforts to accommodate the respondent's
concerns. The CPS non-interview rates range around 5-6 percent monthly.
The CPS Food Security Supplement Sample
Approximately 53,700 households completed the April 1995 basic CPS questionnaire,
and were invited to answer the Food Security Supplement. Of these, 44,730 households
completed ine supplement, implying a non-interview rate of 16.7 percent below the basic CPS
sample. The respondents completing the supplement included households at all income levels,
both above and below the federal poverty thresholds. Special weights were computed to adjust
the final supplement sample for the demographic characteristics of supplement non-interviews.
The Food Security Screener. The complete Food Security Supplement instrument was
administered to all households with incomes at or below 185 percent of the federal poverty level
for the 12 months prior to their entry into the CP sample. This is the income-poverty threshold
Prepared by Abt Associates Inc. 18
Chapter Two: The Food Security Supplement to the Current Population Survey
used in determining eligibility for some federal assistance programs (e.g., WIC and reduced-price
school lunch and breakfast programs). All households with incomes below this level
received all parts of the questionnaire.
Preliminary analyses of NHANES-III data had indicated that some households with
annual incomes above 185 percent of poverty may have experienced food insufficiency sometime
during the period covered by that survey (1988-94), based on their responses to food sufficiency
items included in the NHANES questionnaire. To reduce the risk of screening out any currently
food-insecure households with prior-year annual incomes above 185 percent of poverty, three
additional routes for passing through the screener were included for higher-income households.
These were: (1) reporting sometimes or often not having enough to eat on either of the two
versions of the food sufficiency question (QUA, or Qll and Q12); (2) a combined answer
pattern indicating the possibility of (low-severity) food insecurity (Q15 - "did you ever run short
of money and try to make your food or food money go further?" plus reporting "enough but not
the kinds of food wanted" in QUA or Q12); and (3) an affirmative answer to Q16 - "did you
ever run out of the foods that you needed to make a meal and didn't have money to get more?"
Of the 44,730 households that completed the Food Security Supplement, a total of
18,453 households passed this screener and were asked the full battery of food security and
hunger questions. This group comprised the preliminary analysis sample for developing the food
security scale. These included 15,662 households with incomes below 185 percent of poverty
and 2,791 households with higher incomes. Initial analyses determined that an additional 83
households lacked responses on some important items, and these were dropped from the sample.
This created a final analysis sample of 18,370 households used in the development of the
measurement scales for food insecurity and hunger.
To allow assessment of reliability of the measurement scales and their invariance across
different household types, the analysis sample was randomly subdivided into four subsamples.
Initial scale development analyses and modeling were implemented using one of these
subsamples, with the remaining three preserved for use in reliability and invariance testing. The
Food Security Supplement sample is shown by household type in Exhibit 2-2.
Prepared by Abt Associates Inc. 19
Chapter Two: The Food Security Supplement to the Current Population Survey
Exhibit 2-2
DESCRIPTION OF THE FINAL FOOD SECURITY SUPPLEMENT SAMPLE
BY TYPE OF HOUSEHOLD
Household Type
Households
with Children
Households with
Elderly and No
Children
Households
with No Elderly
or Children
Number of households in the population 38,232,774 27,851,187 34,354,945
Proportion of households in the population 38.1% 27.7% 34.2%
Number of households in the sample (total:
44.730)1 16,954 12.503 15,273
37.9% 28.0% 34.1%
Number of sample households passing die
screener (total: 18,453)
7,998 5,731 4,724
Proportion of sample households passing
the screener
43.3% 31.1% 25.6%
* Households completing (he survey. Of those respondents completing the Supplement, 83 provided incomplete information
on food security hems and were dropped from the final analysis sample.
Prepared by Abt Associates Inc. 20
CHAPTER THREE
THE FOOD SECURITY MEASUREMENT SCALE
The questions included in the CPS Food Security Supplement were designed to represent
the full range of severity of food insecurity and hunger as experienced in U.S. households, in
order to allow the development of a comprehensive food security measurement scale. The
purpose of such a scale is to combine a household's answers to many survey questions into a
single measure of the severity of food insecurity and hunger, where the household's score on
the measurement scale indicates the level of severity of food insecurity it has experienced. This
chapter describes the two scales that have been developed. One measures the full range of food
insecurity and hunger on a 12-month basis; the other focuses on only the more severe conditions
of reduced food intake and hunger measured on a 30-day basis.
Scale Development
The process of developing, refining, and testing the scales occupied nearly a year, from
the autumn of 1995 through the summer of 1996. The methods used and results obtained are
summarized briefly below and described more fully in the study's technical report (Hamilton et
al., 1997).
Each of the questions considered as candidates for the food security scale refers
explicitly to either the 12-month or the 30-day time frame. After early descriptive and
exploratory analyses, these two groups of questions were separated, and distinct models were
estimated for the 12-month and 30-day periods. The procedures described below apply generally
to both the 12-month and the 30-day models, although the 12-month scale will be the main focus
of the following discussion.1
Linear Analysis. Exploratory analyses were fust conducted using linear factor analysis
methods. This analysis phase was principally devoted to replicating analyses reported in the
existing literature to determine whether the findings of prior research were applicable to the
national population-level CPS data. These analyses focused on households with children, which
1 The companion Technical Report volume provides a description of die 30-day scale and presents
of the prevalence of hunger within the 30-day period.
Prepared by Abt Associates Inc. 21
Chapter Three: The Food Security Measurement Scale
were asked all questions in the Food Security Supplement. Results showed general conformity
with previous research. Analyses suggested that either a one- or two-factor model would best
fit the data in linear models.
Exploratory Non-linear Analysis. Because most questions in the Food Security
Supplement are asked in dichotomous or categorical form, a non-linear factor analysis model was
considered best suited to the structure of the data. Exploratory analyses were conducted, fitting
a series of alternative models to determine whether a single- or multi-factor model would best
fit the data. Results indicated the unidimensional model to be most appropriate. Thus, the
results support the hypothesis that the severity of food insecurity and hunger can be validly
viewed as a single continuous dimension, along which various aspects of household food
sufficiency and food management behaviors are arrayed.
Preliminary Model Estimation. The statistical approach chosen was the Rasch model,
a concise one-factor non-linear Item Response Theory (IRT) model that was fit to the CPS data
using a specialized software package.2 Using a one-fourth random subset of the CPS data, a
preliminary model was fit for the subpopulation of households with children within that one-quarter
sample. The model was refined iteratively. Fit statistics were examined for each
question in the candidate set, items that failed to meet threshold criteria were discarded, and the
model was re-estimated with the new candidate list.
Tests for Invariance. The model estimated for households with children was then
estimated separately for two other groups: households without children but with one or more
elderly members, and households with neither children nor elderly members. A high level of
correspondence was found among the models fit to the three separate household types, indicating
that food security and hunger could be measured for all three populations using the same scale.
A single model was therefore estimated for the full sample population, combining all three
household types.
Tests for Robustness. The preliminary model estimated with the one-fourth subsample
was then fit to the remaining three one-fourth partitions of the sample. Essentially identical
2 IRT describes a general type of measurement model developed by the educational testing industry for
use in developing and scaling tests such as aptitude tests. IRT models provide a way to measure the overall
ability level of an individual being tested, based on widely varying difficulty of particular questions, and on
the individual's overall pattern of response to the entire set of questions.
Prepared by Abt Associates Inc. 22
Chapter Three: The Food Security Measurement Scale
results were found for all subsamples, which indicates that the model should be stable across
repeated samples of households. The model was therefore re-estimated from the entire CPS
sample.
Tests for Reliability. A variety of statistical tests for reliability were performed,
including tests specific to the Rasch model and several tests commonly used for scales developed
through linear analyses. Tests indicated quite good reliability for the 12-month scale and
moderate reliability for the 30-day scale (Hamilton et ai, 1997).
12-Month and 30-Day Scales
Although food security measurement scales were developed for both the 12-month and
30-day time frames, this report gives primary emphasis to the 12-month scale, which is
considered the more broadly useful measure.
The difference between questions asked in the 30-day time frame and the parallel 12-
month questions is solely a matter of calendar timing. The 12-month questions ask whether the
household experienced a particular condition at any time during the year ending in April 1995,
whereas the 30-day questions ask whethe [hose conditions that were experienced during the year
also were experienced during the 30-day period prior to the survey.3 The questions do not differ
in the severity of the condition they measure, but because of this difference in time periods, one
would expect to find more positive responses to the 12-month questions than to their parallel 30-
day versions, and the data bear out this expectation.
The questions included in die 12-month scale differ substantively from those in the 30-
day scale in one important respect. A number of questions about less severe food insecurity
conditions (for example, whether the respondent worried that the household would run out of
food before getting money to buy more) were asked in the 12-month time frame but not the 30-
day frame. The 12-month scale is therefore able to describe a broader range of food insecurity
conditions. This makes the 12-month scale better suited to a number of policy and research
purposes, and also gives it stronger statistical properties.
3 More precisely, the qaeitiont refer to the time period ending on the day of the interview, which
occurred during the period April 16-22. 1995.
Prepared by Abt Associates lie. 23
Chapter Three: The Food Security Measurement Scale
The remainder of this volume accordingly focuses mainly on the 12-month scale. As
noted above, details on the 30-day scale are presented in this study's technical report.
Questions Used in the Scale
All 12-month questions in the Food Security Supplement were tested for possible
inclusion in the scale.4 Most candidate questions met the statistical criteria for inclusion in the
final version of the model. Exhibit 3-1 lists the questions that are included, showing them in
the order in which they appear in the questionnaire.
The questions included in the scale capture four kinds of situations or events. All are
related to the general definition of food insecurity presented earlier, which includes a
psychological dimension as well as qualitative and quantitative aspects of food supply and food
intake. The four kinds of situation are:
• Anxiety or perception that the household food budget or food supply was
inadequate (Q53, Q54)
• Perceptions that the food eaten by adults or children was inadequate in quality or
quantity (Q32, Q55, Q56, Q57, Q58)
• Reported instances of reduced food intake, or consequences of reduced intake (such
as feelings of hunger or reported weight loss) for adults in the household (Q24,
Q28, Q35, Q38)
• Reported instances of reduced food intake, or its consequences, for children (Q40,
Q43, Q47, Q50)
A number of the questions in the CPS Supplement did not fit the 12-month model, and
are therefore not included in the measurement scale. Three of the excluded questions indicate
relatively less severe conditions of food insecurity, such as concerns about the adequacy of the
4 Specifically, those considered were the 12-month questions in die series from Q1S through Q58 (see
Appendix A). The response frequencies for the 12-month and 30-day questions in this sequence are shown
m Appendix B.
Prqmnd by Abt Associates Inc. 24
Chapter Three: The Food Security Measurement Scale
Exhibit 3-1
QUESTIONS INCLUDED IN THE FOOD SECURITY SCALE
Question
Number Question
24,25 In the last 12 months, did you or other adults in your household ever cut the size of your
meals or skip meals because there wasn't enough money for food?
How often did this happen—almost every month, some months but not every month, or in
only 1 or 2 months?
28,29 In the last 12 months, did you or other adults in your household ever not eat for a whole
day because there wasn't enough money for food?
How often did this happen—almost every month, some months but not every month, oi in
only 1 or 2 months?
32 In the last 12 months, did you ever eat less than you felt you should because there wasn't
enough money to buy food?
35 In the last 12 months, were you ever hungry but didn't eat because you couldn't afford
enough food?
38 Sometimes people lose weight because they don't have enough to eat. In the last 12
months, did you lose weight because there wasn't enough food?
40" In the last 12 months, did you ever cut the size of any of the children's meals because
there wasn't enough money for food?
43', 44» In the last 12 months, did any of the children ever skip a meal because there wasn't
enough money for food?
How often did this happen—almost every month, some months but not every month, or in
only 1 or 2 months?
47» In the last 12 months, were die children ever hungry but you just couldn't afford more
food?
50" In die last 12 months, did any of the children ever not eat for a whole day because there
wasn't enough money for food?
53 "I worried whether our food would run out before we got money to buy more." Was that
often, sometimes, or never true for you in the last 12 months?
54 "The food that we bought Just didn't last, and we didn't have money to get more." Was
mat often, sometimes, or never true for you in the last 12 months?
55 "We couldn't afford to eat balanced meals. * Was that often, sometimes, or never true for
you in the last 12 months?
56" "We couldn't feed the children a balanced meal because we couldn't afford that." Was
that often, sometimes, or never true for you in the last 12 months?
57" "The children were not eating enough because we just couldn't afford enough food." Was
that often, sometimes, or never true for you in the last 12 months?
58« "We relied on only a few kinds of low-cost food to feed the children because we were
running out of money to buy food." Was that often, sometimes, or never true for you in
the last 12 months?
* Question asked only of households wkh children.
Prepared by Abt Associates Inc. 25
Chapter Three: The Food Security Measurement Scale
food budget or food supply, or adjustments to me type of food served.3 Although these
particular questions were excluded, this range of comparatively less severe food insecurity
experience is captured by other questions that did meet the statistical goodness-of-fit criteria for
inclusion in the model.
The other five excluded questions ask about actions that a household might take to cope
with food insecurity by seeking external food resources (examples are borrowing food or money
for food from friends or relatives, or getting meals at soup kitchens).6 These "resource
augmentation" questions have a peculiar relationship to food insecurity. On the one hand, they
represent household responses to a situation of food insecurity, and thus provide conceptually
valid indicators of the existence of the condition: households that are food secure are not
expected to take such actions. On the other hand, a household that successfully augments its
food resources may thereby become less food insecure, so these indicators do not fit well in
scales measuring the severity of the condition. Probably because of the complicated nature of
their relationship to food insecurity, the resource augmentation or coping questions did not meet
the statistical criteria for inclusion in the food security measurement model.
All questions are entered in the models in dichotomous "yes/no" form. Three follow-up
questions in the 12-month series ask whether a situation occurred "almost every month, some
months but not every month, or in only 1 or 2 months." These question: were receded to
combine the first two response categories into "three or more months."7 Questions S3-S8 ask
the respondent whether the condition was "often, sometimes, or never" true in the past 12
months. The first two of these response categories are combined into "sometimes or often."
5 These were Q1S ("Did you ever run short of money and try to make your food or your food money go
further?"); Q16 ("Did you ever run out of the foods that you needed to make a meal and didn't have money
to get more?"); and Q20 ("Did you ever serve only a few kinds of low-cost foods—like rice, beans, macaroni
products, bread or potatoes—for several days in a row because you couldn't afford anything else?").
6 The specific questions are Q18 (get or borrow food from friends or relatives); Q19 (children eat at home
of friends or relatives); Q21 (put off paying bills); Q22 (get food from church or food pantry); and Q23 (get
meals at soup kitchen).
7 The other category, "less than three months," is coded to include respondents that answered negatively
to the base question.
Prepared by Abt Associates Inc. 26
Chapter Three: The Food Security Measurement Scale
Relative Severity of Questions in the Scale
The analytic software that estimates the measurement scale computes an "item
calibration" value for each question included in the scale. The item calibration score indicates
the relative severity of the food insecurity or hunger condition represented by each question.8
Exhibit 3-2 illustrates the pattern of relative question severity. Questions representing less
severe levels of food insecurity and hunger are located at the bottom of the chart, and those
measuring more severe levels are at the top. Questions that are grouped closely together can
be considered to represent approximately the same level of severity of food insecurity and
hunger.
The pattern of question severity reflects the three progressive stages of food insecurity
that previous research has found for households with children. The first stage involves
adjustments to the overall household food budget and to patterns of food acquisition and use.
In the second stage, adults reduce food intake while generally protecting the children's intake
levels. The third stage involves reductions in food intake for children as well as more
pronounced reductions for adults.
The item calibration scores generally correspond to this pattrrn. The least severe
questions reflect concerns about the food supply and adjustments to the kind and quality (but not
the amount) of food eaten. The most severe questions indicate reduction of children's food
intake or drastic reductions in adult intake (not eating for a whole day). In between lie the
questions indicating reductions in adult food intake.
Some overlap in item calibration scores, or severity, exists in the groups of questions.
For example, the least severe child-oriented questions in the scale (QS8 and QS6) occur quite
early in the severity sequence, whereas the most severe adult-oriented question (Q29) occurs at
quite an advanced point in the sequence. This suggests that the movement between stages does
not occur as abrupt or uniform behavior shifts, but as graduated adjustments that are likely to
differ from household to household.
1 In educational testing, the item calibration is considered to describe the relative difficulty of questions
in a test. The item calibration score for a particular question depends on the overall response pattern by all
the persons initially taking die test (i.e., when the "test is calibrated") and is based on die probability that
houseJolds with a given value (overall test score) on the scale answer the question affirmatively. Households
with scale value equal to the item calibration are predicted to have a SO percent chance of answering die
question affirmatively.
Prepared by Abt Associates Inc. 27
Chapter Three: The Food Security Measurement Scale
Exhibit 1-2
SEVERITY RANKING OF QUESTIONS IN FOOD SECURITY SCALE
10
i OB© CUM MM Ml lorNh* My
7-
8 H Q44 am iHfiiMi. **
< 043 CMMSUMM
* QMAMtMt
Q47CMU
I Q40ChMMMl
•■lfocwhoteMy.3*
OSfl f#f WfMMS MrY
e-o
g ■ OMWimiHirtMiHfyHMlMK
= 8H QSTCMMMlMiMI
£ (lQMAMiB>Mib,l»i
4-
QM
024AM
3-
- QHRMt
2- QMPtH
QttWinMtalMHHMHl
tkmntuti m*> OJ(bMtMTM)k>n
Prepared by Abt Associates frtc. 28
Chapter Three: The Food Security Measurement Scale
Household Values on the Scale
The scaling model also assigns to each household a value on the scale. The household
value is based on the number of questions the respondent answers affirmatively, adjusted for the
number and relative severity of the questions the respondent answers.9 Among households that
answer the same set of questions, those that give more affirmatives have higher values on the
scale.
The analysis indicates that household response patterns are largely ordered. That is,
a household that answers a particular question affirmatively tends to affirm all less severe
questions as well. It is appropriate to characterize households with n affirmative responses as
having affirmed the n least severe questions, because that is the single most common, or modal,
pattern.10
The vast majority of households have the lowest possible value on the scale (a scale
score of zero), indicating that they did not respond affirmatively to any food insecurity or hunger
questions. Many of these households were screened out because their incomes were above 185
percent of the poverty level and they gave no indication of food insecurity in die preliminary
screener questions; such households amount to about 40 percent of all respondents. Another
large group of households passed the screen and were asked all food insecurity questions, but
responded negatively to all of them.
Thus, 82 percent of all households surveyed had the lowest possible value on the scale.
The other 18 percent answered at least one question affirmatively and therefore have values
above the minimum. The proportion of households at each successively greater level of severity
declines rapidly, as indicated in Exhibit 3-3. The exhibit shows the percentage of the sample
with household values at or above selected levels. The selected levels are the modal household
values associated with each non-child question in the scale—that is, each question that is
9 The adjustment is necessary because all respondents do not answer all questions. Eight of the 18
questions in the 12-month scale are asked only to households with children. In addition, a few respondents
simply fail to respond to some of the questions they are asked.
10 Most households follow the modal pattern in their responses to the scaled questions, but not all
households do. For example, a household with n affirmative responses may answer negatively to one of die
less severe questions (i.e., less severe than the nth question}, but answer affirmatively to one of the more
severe questions. Such a household would have the same value on the scale as a household following die
modal pattern. Households without children whose responses exactly follow the modal pattern amount to 82
percent of all households without children in die sample.
Prepared by AIM Associates Inc.
Chapter Three: The Food Security Measurement Scale
Exhibit 3-3
POPULATION DISTRIBUTION BY SELECTED HOUSEHOLD SCALE VALUES
069 Wonted food would run out (0.9)
Q54 Food bought (Mnltatt (1.9)
066 Adult not MI balanotd msals (2.7)
024 Adult cut or skipped ma* (3.5)
032 AduN Ml taw than should (42)
025 AduR out/ridp moo*. 3+ montw (4.9)
036 AduK hungry but dU not Mt (5.7)
038 Adult lost vmkjht (6.5)
Q28 Adult not oMwhotedoy (7.4)
029 Aduk not oat whoto day, 3* month* (5.1) -
fModol houtohold vakjoo In uiiait—M1
Oft 6ft 10ft 16ft
Porasnt At or Abovo Modal Household VaJuo
20ft
applicable to all households with or without children. For each question in sequence, the modal
household value is the scale value assigned to all those households that said "yes" to that
specified question and to all less severe questions, but that said "no" to all the more severe
questions. The chart can thus be read as indicating the percent of households that answered
affirmatively to the specified question and to all less severe questions.
About 18 percent of the sample households have scale values above the minimum, as
mrtiratrd in the top bar on die chart. Only 8 percent have scale values at or above die level
corresponding to die first direct indicator of reduced food intake (Q24). Just 0.1 percent have
scale values in die highest range shown on die chart, associated with adults repeatedly not eating
for a whole day (Q29).
Prepared by Abt Associates Inc. 30
Chapter Three: The Food Security Measurement Scale
Overview of Scale Development Results
In summary, the food security scale has three key properties. First, it captures multiple
facets of food insecurity within the single dimension of relative severity, an important exception
being resource augmentation actions that households take to address food insecurity.11
Resource augmentation aside, the analysis indicates that food insecurity and hunger can be
viewed as a unidimensional phenomenon that increases in severity from essentially no food
insecurity up through the most severe level measured in the U.S. context.12
Second, the relative severity of the questions that make up the scale conforms well with
past research. The severity ranking is quite consistent with the concept of food insecurity and
hunger as a managed process, going through distinct behavioral stages that first involve the
household budget and food use patterns, then the reduction of adult food intake, and finally
reductions in the food intake of children in the household.
Finally, a relatively small proportion of survey respondents is seen to have any
measurable level of food insecurity or hunger. This is the expected result in a survey
representing the entire U.S. population. The proportion amiinishes rapidly at higher, more
severe levels of the scale.
11 other H—JMI of food insecurity included in the LSRO conceptual definition, such as nutritional
inadequacy of diets, problematic food safety, and food access problems apart from the household's own
resource limitations, are not intended to be captured in the present measurement of food insecurity.
12 AMM—scale developed for use in economically less-developed countries would be expected
to include conditions more severe than those incorporated here, such as severe malnutrition and starvation.
Prepared by Abt Associates Inc. 31
41
CHAPTER FOUR
THE FOOD SECURITY STATUS MEASURE
The analysis reported in Chapter Three supports the hypothesis that food insecurity and
hunger can be viewed as an ordered, sequential phenomenon. Households are distributed along
a range that runs from no indication of food insecurity at all, through increasing levels of
severity, up to the most severe measured level of food insecurity and hunger. Households in
the United States are very largely food secure, falling outside the measured range of food
insecurity and hunger. Among the minority of U.S. households that arc measurably food
insecure, most are concentrated at the less severe end of the continuum, with only a small
fraction at the most severe end.
It is useful for policy purposes to divide the population into groups that can be identified
as experiencing different designated levels of severity of food insecurity. This entails
subdividing the food security scale into separate ranges, so that a household's value on the scale
classifies the household as falling in a particular severity level or category of food insecurity.
A number of research efforts have categorized their study populations into two or three groups,
such as food secure vs. food insecure (Burt, 1993), or nothmigiy.atriskofhuiiger, andhimgry
(Wehkr, Scott and Anderson, 1992).
Four categories are defined in the present analysis, based on the differing behavioral
patterns that characterize different broad ranges of severity of food insecurity and hunger.
Previous research as weU as the analysis reported here suggests that food insecurity manifests
at the household level as a managed process of efforts to cope with inadequate supplies of food
and resources to obtain food (Radimer, Olson, and CampbeU, 1990; Basiotis, 1992; Radimer
et of., 1992; Wehler, Scott and Anderson, 1992; Burt, 1993; Cohen, Burt, and Schulte, 1993).
This managed process moves through an observable set of stages as food insecurity increases.
In the first stage, household members exrwieiice food msufficieiicy and aii^
situation, and adjust their bud^ aiid food inaiiageniem rotterns. For example, they may worry
that their food will not last until they have nioiiey to biiy more, they inay substn^
cheaper foods in their diet, and they may eat the same few low-cost foods several days in a row.
In the second stage, adults reduce their food htake, but m households wim chiWren they ra^
Prepared by Abt Associates Inc. 33
Chapter Four: The Food Security Status Measure
food to avoid reducing the children's food intake. Adults may be hungry, but normally they try
to protect their children from being hungry. In the third stage, the children also experience a
reduction in food intake and hunger, and adults' food intake is more sharply reduced (e.g., going
an entire day with no food). The transition from one stage or broad range of food insecurity
to the next may be a gradual alteration of conditions and behaviors, or may be more sharply
demarcated, but in either case, it is likely to occur differently in different households.
Nonetheless, it appears that, overall, distinct behavior patterns exist at different levels of food
insecurity and hunger.
These observed behavioral patterns provide the foundation for defining the four
categories of food insecurity used in the present measurement project.1 In order of increasing
severity of food insecurity, they are as follows:
• Food secure — Households show no or minimal evidence of food insecurity.2
• Food insecure without hunger — Food insecurity is evident in households'
concerns and in adjustments to household food management, including reductions
in diet quality, but with no or limited reductions in quantity of food intake. In
terms of the LSRO definitions, households cannot predictably obtain access to an
adequate quantity and/or quality of acceptable food, but household members are not
evidently experiencing hunger due to resource scarcity.
• Food insecure with moderate hunger — Food intake for adults in the household
is reduced to an extent that implies that adults are experiencing hunger due to lack
of resources.
1 This four-way categorization is applied only to the 12-month food security scale. Because die 30-day
■cale does not measure die less severe condition of food insecurity short of actual hunger, a three-way
categorization is used for that scale: (1) no hunger evident; (2) food insecure with moderate hunger; and (3)
food insecure with severe hunger. The conceptual and operational definitions of the latter two categories are
parallel to those for the two most severe categories on the 12-month seals.
2 Most U.S. households show no signs of food insecurity, based on consistendy negative responses to
several broad screener SjBSsfMI in the CPS questionnaire. Households in this category, and with prior-year
annual incomes over J85 percent of poverty, were screened out of the remainder of the Supplement at that
point and direcdy clarified as food-secure without further analysis. A much smaller proportion of higher-income
households passed through the food security screener, by virtue of positive answers to at least one of
the screener questions. Most of mete households were also subsequently classified as food-secure, bated on
further analysis. Some of this latter group of food-secure households show one. or at most two, fHMftHial
positive indirariom of food insecurity in the CPS data. To be classified as food-insecure, however, a
household had to show at least three positive indicators of food uisecurity from the set of food security items
beyond the srtttutt questions.
Prepared by Abt Associates Inc.
Chapter Four: The Food Security Status Measure
• Food insecure with severe hunger — Households with children reduce the
children's food intake to an extent that implies that the children experience hunger
as a result of inadequate household resources. Adults in households with or
without children experience extensive reductions in food intake (e.g., going whole
days without food).
It is important to emphasize that these behavioral classifications do not imply policy
judgments. Policymakers, advocates, and the public at large must decide whether and at what
level food insecurity and hunger may constitute a social problem that merits public concern and
government action.
Readers should also note the necessary role of judgment involved in giving exact
operational definition to these categories. The particular categories specified for classifying
households according to level of severity of food insecurity are grounded in observable patterns
of behavior, and the placement of boundary lines between categories attempts to reveal the
nature of these patterns as clearly as possible, given the content of the available data. The exact
placement of the classification boundaries, however, necessarily involves interpretive judgment.
Judgment is involved in determining how well each indicator item in the data corresponds to one
or another of the designated ranges of severity described above. In addition, judgment is
required in identifying or selecting the particular indicator that best represents the dividing line
or transition from one <Vfjgn»t<«i severity range to the next. Reasonable people can disagree
about whether the dividing lines between the several designated severity ranges should be located
somewhat differently.
The most important uses of the food insecurity and hunger measures, however, will be
in examining chai^ii^ severity and extent of needs over time, or differing needs across
population groups. Comparing the prevalence of food insecurity from one year to the next and
across population groups on a consistent basis can help identify changing levels and location of
need, and help inform decisions as to whether re-targeting of assistance may be needed. In such
analyses, die key requirement is mat the dividing lines be robust, defined in an operationally
clear and consistent way over time and across subgroups. The emphasis of the analysis has
therefore been on establishing a clear and replicable logic for defining categories, rather than
seeking universal agreement on the appropriateness of each dividing line between categories.
Prepared by Abt Associates Inc. 35
Chapter Four: The Food Security Status Measure
Defining Ranges on the Food Security Scale
To classify households into the categories described above, the food-security
measurement scale is subdivided into corresponding ranges. The general procedure for defining
these ranges is summarized below. Subsequent sections review the logic for each of the specific
ranges.
The definition of ranges takes advantage of the scaling model's estimates of the relative
severity of the questions that make up the food security scale. Because the model indicates that
responses are sufficiently well ordered, it is acceptable to assume that a household answering
a particular question affirmatively also answers all less severe questions affirmatively. This is
the predominant actual pattern, or pattern of the "modal households." This allows behavioral
ranges on the food security scale to be identified by considering the substantive content of sets
of adjacent questions. Thus, if all the questions prior to question n in level of severity are
judged to reflect conditions of food insecurity but not hunger, whereas question n is deemed to
be an indicator of actual hunger, the boundary between food insecurity without hunger and food
insecurity with hunger can be set between question n-1 and question n in the severity ranking.
Question n itself would then be considered a threshold or boundary indicator for the more severe
category of food insecurity with hunger.3
Although the discussion below focuses on the individual questions that border the
boundaries between ranges of the scale, it is important to bear in mind that households are
classified on the basis of their overall pattern of responses to the entire sequence of questions
making up the measurement scale. No single question, no single condition is used to classify
households. Rather, classification depends on the accumulated evidence, from the entire set of
questions, that the household has (or has not) experienced a series of successively more severe
conditions and behaviors.
3 Technically, the boundary is established at a particular value on the teak. Once a boundary question
U chosen, the boundary is act at the acale score of those modal households that answer all less-severe
questions and the selected boundary question affirmatively, while answering all more-severe questions
negatively. In the simplest case, if question n in the severity ranking is the boundary question and all
households respond to the same total number of questions, the boundary is established at die scale value of
households that answer exactly n questions affirmatively.
Prepared by Abt Associates Inc. 36
Chapter Four: The Food Security Status Measure
Food Insecure
Food security — Access by allpeople at all times to enoughfoodfor an active,
healthy life.... Food insecurity exists whenever the availability ofnutritionally
adequate and safe foods or the ability to acquire acceptablefoods in socially
acceptable ways is limited or uncertain. — Anderson/LSRO, 1990, pp. 1575-
1576
The above definitions suggest that households are food insecure if they do not have, or
cannot be reasonably sure of having enough food, of acceptable quality, to meet basic needs.
Most questions in the Food Security Supplement, including all of the items retained in
constructing the measurement scale, are pertinent to this concept of food insecurity.4 A few
questions measure either the respondent's level of uncertainty about the future adequacy of the
household's food supply, or the retrospective assessment of the food supply.3 A larger number
of questions ask about events or conditions that can result from an inadequate food supply, such
as not eating balanced meals, cutting or skipping meals, or losing weight because of not having
enough food. All questions explicitly mention resource constraint as the immediate cause of
food insufficiency through phrases such as "because you didn't have enough money."
Food insecure households are defined operationally as those which, at a minimum,
express concerns about the adequacy of the household food supply and report some adjustments
to dietary intake. Exhibit 4-1 illustrates this criterion. The two least severe questions in die 12-
month scale concern the households' food supply, asking whether household members "worried
that our food would run out" or whether the "food that we bought just didn't last" (Q53 and
Q54). The third question in the severity ranking asks about the failure to eat halairrri meals,
an adjustment to nutritional (and conventional) quality of household members' diets (Q55). A
respondent who answers all three of these questions affirmatively is deemed to show sufficient
evidence of food insecurity to have met unambiguously die operational criterion for that
4 The definition alto refers to the household's need for access to food through "socially acceptable ways."
This dimension of the definition may be captured in the CPS datt with questioM abotf copmg actrvities sudi
as getting emergency food from food pantries or eating meals at soup kitchens. As explained above (p. 26),
however, these items are not included in the measurement scale for severity of food insecurity. (See the
technical report for further discussion of the food-augmenting coping-behavior questions.)
9 For example, Q53 asks whether respondents "worried our food would run out before we got more."
Q54 asks whether "the food that we bought just didn't last, and we didn't have money to get more."
Prepared by Abt Associates Inc. 37
Chapter Four: The Food Security Status Measure
classification.6 The boundary between "food secure" and "food insecure without hunger" is
therefore drawn between Q54 and Q55 in the scale sequence.
Exhibit 4-1 illustrates the application of this definition. Households that answer no
questions affirmatively, or that affirm only the one or two least severe questions, are classified
as food secure. Households that give affirmative answers to three or more questions are placed
into one of the three categories of food insecurity. The least severe condition classified as food
insecure is that represented by affirmative answers to the three least severe questions in the scale
sequence.
Food Insecure with Moderate Hunger
Hunger — The uneasy or painful sensation caused by lack offood. The
recurrent and involuntary lack of access to food... a potential, although not
necessary, consequence of food insecurity.1 — Anderson/LSRO, 1990, p.
1598, 1576
As the severity of food insecurity increases, the household reaches a point at which
further economizing on food costs requires reduction in food intakes such that household
members experience hunger as a consequence of the household's financial resource limitation.
Hunger, as the term is used here, is a physical sensation caused by a lack of food, where that
lack of food results from scarce or limited household financial resources. The exact level at
which the lack of food is certain to produce hunger varies substantially among individuals. The
physiological literature indicates that virtually any noticeable reduction from an individual's
This minimum requirement ofthree affirmative responses is more stringent than most previous literature,
which typically has classified households as food insecure if they respond iffirmatively to any one or more
food insecurity indicator questions. Research has shown that households answering as few as one of the
questions positively have significantly reduced household food supplies, and that women in these households
have reduced intakes of fruits and vegetables and increased body man indices (Kendall, Olson and Fronjillo,
1995).
The approach to range definition used here sets each boundary at a level disc requires two or three
affirmative responses to questions measuring the rendition of interest. This strategy reduces the likelihood
that a household will be placed in a too-severe category of food insecurity because of an eiTooeous affirmative
response (a "false positive" classification). The trade-off is an increased likelihood that a household will be
placed into a less severe category than actually merited (a 'false negative" idfnrifkation).
7 The present project makes explicit the condition that the measurement objective is limited to hunger
resulting from inadequate resources.
Prepared by Abt Associates htc.
Chapter Four: The Food Security Status Measure
Exhibit 4-1
SEVERITY RANGES ON THE FOOD SECURITY SCALE
Questions Aaw>dated with Each Food Security Statin Lord
Modal houael»ld« arc tno«eil»wto«ar*7fe<^ ordered
to»w*rili1 amend iiffinnntively. k abo amwen all ton leverc questions affirmatively. Thto to dw predominant
roaponea paaan among survey households.
^^U^^mmdkmmmtmmm,mwmmBmmomamtm.m^m^»imii»imam. some of these
I luth M| miiBian ■ -fthr wpjajr ■*—-—-—-*- - -~*~i»i»-ir.uv >~. n^aaa chnMaaloa mie
would be identified u food insecure.
To be classified in a given food aacaiky category, nodal households mutt respond affirmatively to all question.
anoctottd wkli ton severe categories, pha oaa or -ore of the on«o^ anoctond wkt t» cangory kao wkfck Ike
household is cbtstifted. Onw h—hjMi(I I . nitlTnXlT* —' ■f-^-""*"^"''-*"
of affirmative M
Prepared by Abt Associates Inc. 39
Chapter Four: The Food Security Status Measure
normal level of intake can produce the physical sensation of hunger, although the sensation
apparently occurs differently for different people, and different people describe it differently.8
The definition of the range of food insecurity with hunger therefore focuses mainly on behavioral
questions that ask about reductions in food intake, initially for adults and subsequently for
children.
Several questions used in the scale explicitly measure situations in which adults in the
household experience reduced food intake as a result of inadequate resources. Two others ask
about potential consequences of reduced food intake, one question referring to the sensation of
hunger and another asking about weight loss. In order of increasing severity, the questions are:
• Q24 cut or skip meals because there wasn't enough money for food
• Q32 eat less than you felt you should because there wasn't enough money to
buy food
• Q25 cut or skip meals because there wasn't enough money for food, in three
or more months
• Q35 hungry but didn't eat because you couldn't afford enough food
• Q38 lost weight because there wasn't enough food
• Q28 not eat for a whole day because there wasn't enough money for food
• Q29 not eat for a whole day because there wasn't enough money for food, in
three or more months
The questions that ask about cutting or skipping meals and not eating for a whole day are asked
with reference to "you or other adults in your household." The other questions are asked only
about "you," die adult respondent.
The questions pertinent to reduced food intake by adults are generally concentrated in
die middle of the overall severity ranking for food insecurity, but overlap with both the less
severe food insecurity questions and the more severe questions indicating reduced food intake
by children. The threshold question used in drawing the boundary between food insecure
1 Thin literature is summarized in the companion Technical Report volume, Appendix A. See especially,
the lefcitncca cited therein, Lappalainen et at., 1990, Mattes and Friedman, 1993; and Read, French
1994.
Pnpartd by Abt Associates Inc. 40
Chapter Four: The Food Security Status Measure
without hunger and food insecure with moderate hunger is Q25, which identifies a recurrent
pattern of cutting or skipping of meals by adults in the household. Because this is the third
question in the sequence asking about reductions in adults' food intake, and because it indicates
multiple instances of reduced intake, it is deemed that households that reach this level on the
scale have at least one, and potentially more, adult members who have experienced resource-constrained
hunger. Thus, households that report repeated reductions in adults' food intake, in
combination with affirmative responses to all less severe questions, are classified as food
insecure with moderate hunger (see Exhibit 4-1).
Food Insecure with Severe Hunger
The most severe range of food insecurity measured by the scale is characterized by
reduced food intake and consequent hunger for children. Most of the questions that pertain to
reduced food intake and hunger among children are similar to questions asked about adults.
They are listed below in order of increasing severity.
• Q57 children were not eating enough because couldn't afford enough food
• Q40 cut the size of children's meals because there wasn't enough money for
food
• Q47 children were hungry but couldn't afford more food
• Q43 children skipped meals because there wasn't enough money for food
• Q44 children skipped meals, in three or more months
• Q50 children did not eat for a whole day because there wasn't enough money
for food
The least severe of these questions (Q57) falls roughly in the middle of the range of questions
shown earlier for adults. The remaining questions are all more severe than any adult-oriented
item except the one measuring adults not eating for a whole day. Adults not eating for a whole
day (Q28) has nearly the same severity as children being hungry (Q47).9
9 Q28 and Q47 have item calibrations of 6.4 and 6.5, respectively. See Exhibit 3-2.
Prepared by Abt Associates Inc. 41
Chapter Four: The Food Security Status Measure
The intent in defining the most severe category of food insecurity is to focus on the
condition of children. Specifically, the selected boundary item in the children's series is the
question that asks whether the "children were ever hungry but you just couldn't afford more
food" (Q47).
The classification procedure, however, must apply equally to households with and
without children. A straightforward way to accomplish this objective is to select as the threshold
item an adult-oriented question that is similar in severity to Q47. Accordingly, the question
about adults not eating for a whole day (Q28), which has almost the same item calibration score
as the question about children being hungry, is used as the boundary question. Households, both
those with and without children, that report that one or more adults did not eat for a whole day,
and that respond affirmatively to all less severe questions, are classified as food insecure with
severe hunger.
Response Profile of Households in the Four Categories
If ail survey responses were perfectly ordered, all households would fit the modal
pattern that, although predominant in the actual data, is not universal. With perfectly-ordered
data, we would expect to see very clear-cut differences between the response patterns of
households classified into the different food security categories. For instance, 100 percent of
the households categorized as food insecure without hunger would answer affirmatively either
the three, four, five, six, or seven least-severe questions. No one in that category would give
affirmative answers to any of the more severe questions in the scale (the eleven questions from
Q25 onwards), because those questions lie beyond the boundary for the next more severe
category, food insecurity with moderate hunger. For the five questions within the severity-range
category of food insecure without hunger (Q55-Q32), the more severe questions would have
systematically fewer positive responses than the less severe questions.
All of these patterns can be seen as general tendencies in Exhibit 4-2, although the
divisions are not absolute because not all the survey responses are perfectly ordered. For
example, among households classified as food insecure without hunger, more than 70 percent
responded positively to all of the three least severe questions, whereas fewer than 20 percent
responded positively to any one of the eleven most severe items, and less than 5 percent
responded positively, on average, to these eleven severe items. The percent of positive
Prepared by Abt Associates Inc. 42
Chapter Four: The Food Security Status Measure
Exhibit 4-2
RESPONSE PROFILE BY CATEGORY
(Percentage of Household! in Each Food Security Category
Answering Each Question Affirmatrreiy)
■mcMon 01 RXM i
I Of fDOd WttBKwft M» J,n«»n MM — ** MM .nnan« HUM »|-a li ■ J *- —^— --* » * h■■■ ■ ■ M! M1 naicHori of reaounx-coawTuBca nungcr evneu for oouaeaota
cvideat fcr adalt household nwiatiwri.
evident for duklrca in houKboid *nd/or iodkalon of severe adult
iothefourfroup.it: 14.192.1.934.
Prepared by Abt Associates Inc. 43
Chapter Four: The Food Security Status Measure
responses to each item consistently increases from left to right across the table, reflecting
increasing levels of food insecurity; within each food insecurity category, the percent of positive
responses declines from top to bottom, as the severity level of the questions increases.
Relationship of Food Security Status to Other Measures
Food insecurity is, by definition, a result of constrained financial resources. One would
therefore expect income to be related to food security status. At the same time, one would not
expect the correlation to be perfect for several reasons. In particular, in-kind food assistance
programs, which are designed to ameliorate food insecurity, are specifically targeted to
households with low income. Thus, food insecurity should depend on income in combination
with other factors that ameliorate the effect of low income, especially program participation.
In fact, food security is clearly related to income, as shown in Exhibit 4-3.10 Among
households whose income is less than half of the federal poverty level, 41 percent are classified
as having experienced some kind of food insecurity in the past 12 months,11 and 5 percent fall
into the most severe category of food insecurity. In contrast, 96 percent of the households with
annual incomes above 185 percent of the poverty level are classified as food secure.12
Food security status is also related to the level of household expenditures for food.
Households reporting that they usually spend less than $20 per household member per week are
much more likely to be classified as food insecure than those spending $40 per week or more
(21 percent vs. 6 percent).
Finally, the food insecurity categories defined here show close links to the food
sufficiency measure mat has been used in much previous research. Of the respondents who say
in the food sufficiency measure that they have "enough of the kinds of food we want to eat,"
10 Income is measured in this analysis as cash income, exclusive of in-kind food assistance. Including
the cash value of such assistance might lead to a stronger relationship between income and food security.
11 This includes all households classified into any of the three food insecurity categories—i.e., those in
the three right-hand columns of Exhibit 4-3.
12 Because annual income and food security status are not measured for precisely the same period, it is
possible for a household with apparently substantial income to be accurately identified as food insecure. In
fact, it would be possible even if the two constructs were measured for exactly the same 12-month period.
For example, a head of household could have substantial earnings for the first nine months of the year and
then lose his or her job. Such a household might well be food insecure in the last months of the year.
Prepared by Abt Associates Inc. 44
Chapter Four: The Food Security Status Measure
Exhibit 4-3
RELATIONSHIP OF THE FOOD SECURITY STATUS
MEASURE TO OTHER VARIABLES
Households
in Sample
Food Security Status*
Food
Secure
Food
Insecure
without
Hunger
Food Insecure
with
Moderate
Hunger
Food
Insecure with
Severe
Hunger
Income Relative to Poverty Lime*
<50% 2.219 59.5% 24.2% 11.4% 4.9%
50-100% 4,431 69.6 20.1 8.2 1.9
101-183% 8,944 82.6 1