The United States is viewed by the world as a country with plenty of food, yet not all households in America are food secure, meaning access at all times to enough food for an active, healthy life. A proportion of the population experiences food insecurity at some time in a given year because of food deprivation and lack of access to food due to economic resource constraints. Still, food insecurity in the United States is not of the same intensity as in some developing countries.
Since 1995 the U.S. Department of Agriculture (USDA) has annually published statistics on the extent of food insecurity and food insecurity with hunger in U.S. households. These estimates are based on a survey measure developed by the U.S. Food Security Measurement Project, an ongoing collaboration among federal agencies, academic researchers, and private organizations. It is an experiential measure based on reported behaviors, experiences, and conditions in response to questions in a household survey. The measure was developed over the course of several years in response to the National Nutrition Monitoring and Related Research Act of 1990 (NNMRR). The legislation specifically called for development of a standardized mechanism and instrument(s) for defining and obtaining data on the prevalence of food insecurity in the United States and methodologies that can be used across the NNMRR programs and at the state and local levels.
The USDA estimates of food insecurity are based on data collected annually in the Food Security Supplement (FSS) to the Current Population Survey (CPS). On the basis of the number of food-insecure conditions reported, households are classified into one of three categories for purposes
of monitoring and statistical analysis of the food security of the U.S. population: (1) food secure, (2) food insecure without hunger, and (3) food insecure with hunger.
The USDA estimates, published in a series of annual reports, are widely used by government agencies, the media, and advocacy groups to report the extent of food insecurity and hunger in the United States, to monitor progress toward national objectives, to evaluate the impact of particular public policies and programs, as a standard by which the performance of USDA programs is measured, and as a basis for a diverse body of research relating to food assistance programs.
In addition, USDA has a program of research for improving the measurement and understanding of food security. Despite these efforts, some major questions continue to be raised regarding the underlying concepts, the estimation methods, and the design and clarity of the questions used to construct the food insecurity scale.
USDA requested the Committee on National Statistics of the National Academies to convene a panel of experts to undertake a two-year study in two phases to review at this 10-year mark the concepts and methodology for measuring food insecurity and hunger and the uses of the measures. The specific tasks to be addressed in Phase 1 of the study were:
the appropriateness of a household survey as a vehicle for monitoring on a regular basis the prevalence of food insecurity among the general population and within broad population subgroups, including measuring frequency and duration;
the appropriateness of identifying hunger as a severe range of food insecurity in such a survey-based measurement method;
the appropriateness, in principle and in application, of item response theory and the Rasch model as a statistical basis for measuring food insecurity;
the appropriateness of the threshold scores that demarcate food insecurity categories—particularly the categories “food insecure with hunger” and “food insecure with hunger among children”—and the labeling and interpretation of each category;
the applicability of the current measure of the prevalence of food insecurity with hunger for assessing the effectiveness of USDA food assistance programs, in connection with the Government Performance and Results Act performance goals for the Food and Nutrition Service; and
future directions to consider for strengthening measures of hunger prevalence for monitoring, evaluation, and related research purposes.
In Phase 2 of the study the panel was to consider in more depth the issues raised in Phase 1 relating to the concepts and methods used to measure food security and make recommendations as appropriate. In addition, the panel was asked to address and make recommendations on:
the content of the 18 items and the set of food security scales based on them currently used by USDA to measure food insecurity;
how best to incorporate and represent information about food security of both adults and children at the household level;
how best to incorporate information on food insecurity in prevalence measures;
needs and priorities for developing separate, tailored food security scales for population subgroups, for example, households versus individuals, all individuals versus children, and the general population versus homeless persons; and
future directions to consider for strengthening measures of food insecurity prevalence for monitoring, evaluation, and related research purposes throughout the national nutrition monitoring system.
The Committee on National Statistics appointed a panel of 10 experts to examine the above issues. In order to provide timely guidance to USDA, the panel issued an interim Phase 1 report, Measuring Food Insecurity and Hunger: Phase 1 Report. That report presented the panel’s preliminary assessments of the food security concepts and definitions; the appropriateness of identifying hunger as a severe range of food insecurity in such a survey-based measurement method; questions for measuring these concepts; and the appropriateness of a household survey for regularly monitoring food security in the U.S. population. It provided interim guidance for the continued production of the food security estimates. This final report primarily focuses on the Phase 2 charge. The major findings and conclusions based on the panel’s review and deliberations are summarized below, followed by the text of all of the recommendations.
FINDINGS AND CONCLUSIONS
Concepts and Definitions
The broad conceptual definitions of food security and insecurity developed by the expert panel convened in 1989 by the Life Sciences Research Office (LSRO) of the Federation of American Societies for Experimental
Biology have served as the basis for the standardized operational definitions used for estimating food security in the United States. Food security according to the LSRO definition means access at all times to enough food for an active, healthy life. Food insecurity exists whenever the availability of nutritionally adequate and safe foods or the ability to acquire acceptable foods in socially acceptable ways is limited or uncertain. Food insecurity as measured in the United States refers to the social and economic problem of lack of food due to resource or other constraints, not voluntary fasting or dieting or because of illness or for other reasons. Although lack of economic resources is the most common constraint, food insecurity can also be experienced when food is available and accessible but cannot be used because of physical or other constraints, such as limited physical functioning by elderly people or those with disabilities.
Food insecurity is measured as a household-level concept that refers to uncertain, insufficient, or unacceptable availability, access, or utilization of food. It is therefore households that are classified as food secure or food insecure. It means that one can measure and report the number of people who are in food-insecure households (even though not everyone in the household need be food insecure themselves). When a household contains one or more food-insecure persons, the household is considered food insecure.
A full understanding of food insecurity requires the incorporation of its frequency and duration because more frequent or longer duration of periods of food insecurity indicate a more serious problem. Frequency and duration are therefore important elements for USDA to consider in the concept, operational definition, and measurement of household food insecurity and individual hunger.
The LSRO conceptual definition of hunger adopted by the interagency group on food security measurement is: “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” (Anderson, 1990, pp. 1575, 1576). This language does not provide a clear conceptual basis for what hunger should mean as part of the measurement of food insecurity. The first phrase “the uneasy or painful sensation caused by a lack of food” refers to a possible consequence of food insecurity. The second phrase “the recurrent and involuntary lack of access to food” refers to the whole problem of food insecurity, the social and economic problem of lack of food as defined above.
Unlike food insecurity, which is a household-level concept, hunger is an individual-level concept. The Household Food Security Survey Module (HFSSM) in the Food Security Supplement to the CPS measures food insecurity at the household level; it does not measure the condition of hunger at
the individual level. The HFSSM does include items that are related to being hungry. Some or all of these items are probably appropriate in the food insecurity scale, but they contribute to the measurement of food insecurity and not the measurement of hunger.
The panel therefore concludes that hunger is a concept distinct from food insecurity, which is an indicator of and possible consequence of food insecurity, that can be useful in characterizing severity of food insecurity. Hunger itself is an important concept that should be measured at the individual level distinct from, but in the context of, food insecurity.
The broad conceptual definition of household food insecurity includes more elements than are included in the current USDA measure of food insecurity. Not all elements of the consensus conceptual definition of food insecurity have been incorporated into the USDA measurement of food insecurity in the United States. It was a decision of the Food Security Measurement Project to limit the operational definition and measurement approach to only those aspects of food insecurity that can be captured in a household-level survey. The other conceptually separable aspects of food insecurity are potentially distinct empirical dimensions. For example, the measurement does not include the supply of food or its safety or nutritional quality; these additional aspects would require developing measures and fielding separate surveys to measure them. Moreover, the food supply in the United States is generally regarded as safe, and nutritional adequacy is already assessed by other elements of the nutrition monitoring system, in particular the continuing National Health and Nutrition Examination Survey. The panel therefore concludes that it is neither required nor necessarily appropriate for USDA to attempt to measure all the elements of the broad conceptual definition of food insecurity as part of the HFSSM.
The labeling used to categorize food insecurity is at the heart of the criticism of the current measurement system. In particular, the category “food insecure with hunger” has come under scrutiny because of disagreement over whether hunger is actually measured. The rationale for including hunger in the label for the classification is understandable. Hunger is a politically sensitive and evocative concept that conjures images of severe deprivation, and the HFSSM does include some items that are specifically related to hunger. However, the measurement of food insecurity rather than hunger is the primary focus of the HFSSM. As an indication of the severity of food insecurity, the HFSSM asks the household respondent if in the past 12 months she or he has experienced being hungry because of lack of food due to resource constraints. This is not the same as evaluating individual members of the household in a survey as to whether or not they have experienced hunger. The panel urges USDA to consider alternate labels to convey the severity of food insecurity without the problems inherent in the current labels.
Survey Measurement of Food Insecurity and Hunger
The panel reviewed the current questions used to measure food insecurity and hunger, considered the relationship among the three major aspects of food insecurity and hunger embodied in the questions (whether the household experienced uncertainty, the perception of insufficiency in quality of diet, and reduced food intake or the feeling of hunger), and identified several design issues in the HFSSM that should be addressed.
USDA’s food security scale measures the severity of food insecurity in surveyed households and classifies their food security status during the previous year. The frequency of food insecurity and the duration of spells of insecurity are not assessed directly in the HFSSM questions that are used to classify households by food security status. Although some of the response options do offer the choice of “often, sometimes, or never,” these response options are not sufficient measures of frequency, and they are not included in the construction of the scale. In addition to the items in the HFSSM, the full supplement includes questions that focus on duration. However, these questions are not part of the 18-item HFSSM, although they have been used in research to estimate the percentage of the population that is food insecure on a given day in a given month. A recent study undertaken by USDA researchers examined the extent to which food insecurity and hunger are occasional, recurring, or frequent in the U.S. households that experience them. The panel recommends further research on the frequency and duration of food insecurity.
The panel reviewed the 18 items that constitute the food insecurity scales as well as the entire questionnaire module in which these are embedded and found many issues of questionnaire design. Consistent terminology, clustering questions so as to focus on a specific reference person or reference group (e.g., the respondent, all adults in the household, all children in the household) and on a specific reference period (e.g., 12 months versus 30 days), and developing response options that most closely map to the respondent’s representation of the behavior or attitude are all means by which questions can be designed to reduce cognitive burden and thereby improve the validity and reliability of the measures. Inevitably, questionnaire design requires balancing multiple intents and principles, and there is no perfect questionnaire design. Nevertheless, the panel concludes that the questions in the HFSSM in particular and the FSS in general can be improved by attending to these design principles to the extent possible.
Item Response Theory
USDA uses the Rasch model, a specific type of item response theory (IRT) model, to estimate the food insecurity of households. Several issues
have been raised about the use of IRT models in the measurement of food insecurity and, in particular, the use of the Rasch model.
The panel reviewed IRT and related statistical models and discussed their use and applicability to the development of such classifications as food insecurity. The panel recommends modifications of the current IRT methodology used by USDA to increase the amount of information that is used and to make the methodology more appropriate to the types of data that are currently collected using the Food Security Supplement to the Current Population Survey.
The panel reviewed how the latent variable models are estimated and issues of identifiability of these models and how IRT models are used by USDA in the measurement of food insecurity. On the basis of this review, the panel suggests how the models might be used in better ways to accomplish this measurement and recommends a simple way to modify the existing models currently used by USDA to take into account the polytomous nature of the data collected.
Survey Vehicles to Measure Food Insecurity and Hunger
USDA bases its annual report and estimates of the prevalence of food insecurity on data collected from the Food Security Supplement to the Current Population Survey. The Household Food Security Survey Module, or a modification of it, is or has been used in several surveys. One of the main objectives of the annual food insecurity measure is to monitor the estimated prevalence of food insecurity, as well as changes in its prevalence over time, at the national and state levels to assess both program policies and the possible need for program development.
After reviewing the key features of selected national surveys—the Current Population Survey, the National Health Interview Survey, the National Health and Nutrition Examination Survey, and the Survey of Income and Program Participation—the panel compared the relative merits of each, for either carrying the Food Security Supplement or conducting research to supplement the information obtained from it. The panel recommends research and testing to understand better the strengths and weaknesses of each survey in relation to the Current Population Survey, leading to the selection of a specific survey vehicle for the Food Security Supplement, or for supplementing that information for research purposes.
Food Insecurity Estimates as a Measure of Program Performance Assessment
Currently, the Food and Nutrition Service in USDA uses trends in the prevalence of food insecurity with hunger based on the HFSSM as a mea-
sure of its annual performance to implement the Government Performance and Results Act of 1993 (GPRA). That law requires government agencies to account for progress toward intended results of their activities. It requires that specific performance goals be established and that annual measurement of these output goals be undertaken to determine the success or failure of the program. The panel was asked to comment on the applicability of these data for this purpose.
The panel concludes that an overall national estimate of food insecurity is not appropriate as a measure for meeting the requirements of the GPRA. Even an appropriate measure of food insecurity or hunger using appropriate samples would not be a useful performance indicator of food assistance programs, because their performance is only one of many factors that result in food insecurity or hunger. Consequently, changes in food insecurity and hunger could be due to many factors other than the performance of the food safety net.
The panel concludes that relying exclusively on trends in prevalence estimates of food insecurity as an indicator of program results is inappropriate. To assess program results, a better understanding is needed of the transitions into and out of poverty made by low-income households and the kind of unexpected changes that frequently bring about alterations—for good or bad—in households participating in food assistance programs.
The panel is impressed with the extensive research thus far undertaken, and with the continuing research carried out by USDA. The panel urges that the research program be continued and makes several recommendations for its direction in the future.
The panel concludes that the measurement both of food insecurity and of hunger is important. The recommendations in the report are intended to improve these measurements, so that policy makers and the public can be better informed. Toward this end, the panel has recommended research efforts that should lead to improved concepts, definitions, and measurement of food insecurity and hunger. The panel has provided a detailed discussion of the analytical methods used by USDA and made recommendations for further research to improve the accuracy of the food insecurity scale and on survey alternatives. The panel recognizes that such research will take time.
On the basis of its findings and conclusions, the panel presents recommendations in five areas: concepts and definitions, labeling of food insecu-
rity data outcomes, survey measurement, item response theory and food insecurity, and survey vehicles to measure food insecurity and hunger. The text of the recommendations, grouped according to these areas, follows, keyed to the chapter in which they appear in the body of the report.
Concepts and Definitions
Recommendation 3-1: USDA should continue to measure and monitor food insecurity regularly in a household survey. Given that hunger is a separate concept from food insecurity, USDA should undertake a program to measure hunger, which is an important potential consequence of food insecurity.
Recommendation 3-2: To measure hunger, which is an individual and not a household construct, USDA should develop measures for individuals on the basis of a structured research program, and develop and implement a modified or new data gathering mechanism. The first step should be to develop an operationally feasible concept and definition of hunger.
Recommendation 3-3: USDA should examine in its research program ways to measure other potential, closely linked, consequences of food insecurity, in addition to hunger, such as feelings of deprivation and alienation, distress, and adverse family and social interaction.
Recommendation 3-4: USDA should examine alternate labels to convey the severity of food insecurity without the problems inherent in the current labels. Furthermore, USDA should explicitly state in its annual reports that the data presented in the report are estimates of prevalence of household food insecurity and not prevalence of hunger among individuals.
Recommendation 4-1: USDA should determine the best way to measure frequency and duration of household food insecurity. Any revised or additional measures should be appropriately tested before implementing them in the Household Food Security Survey Module.
Recommendation 4-2: USDA should revise the wording and ordering of the questions in the Household Food Security Survey Module. Examples of possible revisions that should be considered include improvements in the consistent treatment of reference periods, reference units, and response options across questions. The revised questions should reflect modern cognitive questionnaire design principles and new data collection technology and should be tested prior to implementation.
Item Response Theory and Food Insecurity
Recommendation 5-1: USDA should consider more flexible alternatives to the dichotomous Rasch model, the latent variable model that underlies the current food insecurity classification scheme. The alternatives should reflect the types of data collected in the Food Security Supplement. Alternative models that should be formally compared include:
Recommendation 5-2: USDA should undertake the following additional analyses in the development of the underlying latent variable model:
Recommendation 5-3: To implement the underlying latent variable model that results from the recommended research, USDA should develop a new classification system that reflects the measurement error inherent in latent variable models. This can be accomplished by classifying households probabilistically along the latent scale, as opposed to the current practice of deterministically using the observed number of affirmations. Furthermore, the new classification system should be more closely tied to the content and location of food insecurity items along the latent scale.
Recommendation 5-4: USDA should study the differences between the current classification system and the new system, possibly leading to a simple approximation to the new classification system for use in surveys and field studies.
Recommendation 5-5: USDA should consider collecting data on the duration of spells of food insecurity in addition to the currently measured intensity and frequency measures. Measures of frequency and duration spells may be used independently of the latent variable measuring food insecurity.
Survey Vehicles to Measure Food Insecurity and Hunger
Recommendation 6-1: USDA should continue to collaborate with the National Center for Health Statistics to use the National Health and Nutrition Examination Survey to conduct research on methods of measuring household food insecurity and individual hunger and the consequences for nutritional intake and other relevant health measures.
Recommendation 6-2: USDA should carefully review the strengths and weakness of the National Health Interview Survey in relation to the Current Population Survey in order to determine the best possible survey vehicle for the Food Security Supplement at a future date. In the meantime, the Food Security Supplement should continue to be conducted in the Current Population Survey.
Recommendation 6-3: USDA should explore the feasibility of funding a one-time panel study, preferably using the Survey of Income and Program Participation, to establish the relationship between household food insecurity and individual hunger and how they co-evolve with income and health.