![[Feedback]](/icons/banner/feedbackACT.gif){kind=icon}
![[For Subscribers]](/icons/banner/subscriberACT.gif){kind=icon}
![[Archive]](/icons/banner/archiveACT.gif){kind=icon}
![[Search]](/icons/banner/searchACT.gif){kind=icon}
![[Contents]](/icons/banner/tocACT.gif){kind=icon}
|
|
|
|
|
||
Division of Nutritional Sciences, Cornell University, Ithaca, NY 14853–6301
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONVALIDATION OF MEASURESVALIDITY OF FOOD SECURITY...SUMMARYREFERENCES |
|---|
KEY WORDS: • hunger • food insecurity • validity • measurement • humans
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONVALIDATION OF MEASURESVALIDITY OF FOOD SECURITY...SUMMARYREFERENCES |
|---|
, Hamilton et al. 1997
, Rose et al. 1995
). The primary measure
developed from this Food Security Supplement was based on measurement
concepts, methods and items from two previously developed measures from
Radimer/Cornell (Radimer et al. 1990
and 1992
) and the Community
Childhood Hunger Identification Project (CCHIP; Wehler et al. 1992
). The purpose of this paper is to present the evidence available that questionnaire-based measures, in particular the national food security measure, provide valid measurement of food insecurity and hunger for population and individual uses. The paper first discusses basic ideas about measurement and criteria for establishing validity of measures. These criteria are then used to structure an examination of the research results available for establishing the validity of food security measures.
|
VALIDATION OF MEASURES |
|---|
|
TOPABSTRACTINTRODUCTIONVALIDATION OF MEASURESVALIDITY OF FOOD SECURITY...SUMMARYREFERENCES |
|---|
).
We are interested in measuring the relative degree or severity of food
insecurity and hunger. The Food Security Supplement makes this
measurement through questionnaire-based items that ask respondents to
report behaviors and experience directly. The approaches that had
previously been commonly used to measure food insecurity relied upon
indirect indicators, often with unknown validity. An evaluation by the
U.S. General Accounting Office (1986)
criticized these approaches for
not measuring food insecurity directly. At about the same time,
research based on the conviction that it was possible to measure food
insecurity directly was undertaken at Cornell University. Because the
phenomenon of food insecurity, as experienced by people, was not well
understood, a rigorous naturalistic paradigm was chosen to understand,
define and measure food insecurity based on the way in which people
actually experience it (Radimer et al.1990
and 1992
).
Validation is the process of determining whether a method is suitable
for providing useful analytical measurement for a given purpose and
context. A method suitable for providing useful analytical measurement
for a given purpose and context is one for which the following are
true: 1) its construction is well-grounded in an
understanding of the phenomenon; 2) its performance is
consistent with that understanding; 3) it is precise within
specified performance standards; 4) it is dependable within
specified performance standards; 5) it is accurate within
specified performance standards; and 6) its accuracy is
attributable to the well-grounded understanding for that purpose and
context. If all of these criteria are fulfilled, then the method is
valid for that purpose and context (Habicht et al. 1979
, Koch 1987
).
Several possible purposes for populations and for individuals exist. Many of these are potentially relevant to a measure of food security. Some purposes for population include the following:
Estimation of prevalence (How many people are affected?)
Determination of causes (Why are people affected?)
Targeting (Who is affected?)
Monitoring (How is the situation changing?)
Evaluation of programs (Who has benefitted and how?)
Some purposes for individuals include the following:
Screening (Is the person at risk?)
Diagnosis (Does the person have the problem?)
Monitoring (Is the person's situation improving?)
The Food Security Supplement was specifically intended to be used to estimate prevalence and to monitor populations, but several other population and individual uses can be envisioned, including the identification of the food security status of households or individuals.
|
VALIDITY OF FOOD SECURITY MEASURES |
|---|
|
TOPABSTRACTINTRODUCTIONVALIDATION OF MEASURESVALIDITY OF FOOD SECURITY...SUMMARYREFERENCES |
|---|
Well-grounded construction.
The most important evidence that the construction of the food security
measures is well-grounded in understanding is the formal, in-depth
interviews of rural women with children conducted by Radimer et al.
(1990
and 1992
). An additional basis is the informal knowledge from
contact with people who are food insecure, for example, through the
CCHIP surveys (Wehler et al. 1992
). The in-depth interviews of Radimer
clarified our understanding of the following facts: 1) food
insecurity is experienced differently at the household, adult and child
levels; 2) food insecurity has four components; and
3) families experience sequenced levels of severity of food
insecurity with hunger as the most extreme consequence of the
progression of food insecurity. Two components of food insecurity,
quantity and quality of food, are related directly to food. Two
components, certainty and acceptability, are psychological and social
in nature. Furthermore, food insecurity is a managed process; in their
management of food insecurity, families aim to protect children from
food insecurity and its consequences. The Radimer/Cornell items were
developed from words taken from the in-depth interviews of the women
experiencing food insecurity, which contributes to their
construct-validity. CCHIP items were developed from the knowledge of
project investigators who were in contact with people experiencing food
insecurity.
Performance consistent with understanding.
The consistent manner in which the food security measures meet the
well-grounded understanding has been demonstrated in four ways. First,
factor analysis has shown that the conceptualized components of food
insecurity and hunger are confirmed in empirical data (Hamilton et al. 1997
, Kendall et al. 1995
, Radimer et al. 1992
). Second, the proportion
of affirmative responses for items has been examined to determine if
the conceptualized sequence of severity is reflected in the sequence of
responses in items. Third, extensive cognitive testing of measured
items has been done (Alaimo 1997
, Hamilton et al. 1997
) to determine
and ensure that items ask respondents a meaningful question that they
can answer and interpret as the developers intended.
Fourth, the consistency of patterns of affirmative responses across
populations has been examined. For example, Figure 1
shows the pattern of affirmative responses for five population surveys
that had eight items in common with the Radimer/Cornell measure.
Although the prevalence of the items differed markedly across the
samples (due in part to differential oversampling of low income
households), the patterns were parallel. The only exceptions are two
items from the Hispanic Health Survey (Perez-Escamilla et al. 1997
).
This uncertainty item may have been answered affirmatively more often
in the Hispanic Health Survey because of the general economic and
social uncertainty in that community at the time of the survey (R.
Perez-Escamilla, personal communication) or perhaps for other reasons.
Similar examinations of the pattern of responses made for U.S.
subpopulations (defined by race/ethnicity, income and household
composition) by the national food security measure showed parallel
curves, indicating very good consistency of responses.
|
Precision is the extent to which repeated measurements yield the same
value. The usual internal consistency method has demonstrated very good
precision (Cronbach's 
0.85) for the Radimer/Cornell (Kendall et al.1995
, Radimer et al. 1992
), CCHIP (Wehler et al. 1992
) and the U.S.
food security measures (Hamilton et al. 1997
). Dependability is the
extent to which differences in a measure consistently reflect
differences in the phenomena. For example, fluctuation in hydration is
a source of undependability when body weight is used as a measure of
nutritional status in infants. Given the relatively long time span of
12 mo relevant for the national food security measure, dependability is
not an issue, but it might be for short time spans if, for example,
transient events in people's lives influence assessment of their food
security status.
Accuracy.
Accuracy is the extent to which a measure provides unbiased assessment of the phenomena. Accuracy is achieved by construction, which rests upon the depth of understanding of the phenomena. Accuracy is assessed by in-depth analysis and by relating the measure to a criterion measure, which may be a more definitive measure, determinant or consequence.
The hierarchy of possible measures is the following: definitive,
reference and routine (Uriano and Cali, 1977
). Definitive measures
achieve high accuracy because they rely on first principles, i.e., they
reflect in a fundamental way the theoretical structure of the phenomena
they purport to represent. Reference measures achieve accuracy because
they directly and closely relate to the phenomena of interest, and
accuracy is demonstrated by comparison to definitive measures. Routine
measures are usually fast and inexpensive; they require relatively
unsophisticated personnel, whereas accuracy is demonstrated by
comparison to reference measures.
The Food Security Supplement was intended to provide a reference
measure. The demonstration of accuracy of a reference measure is best
done by comparison to a definitive measure. To develop a definitive
measure, a consensus method was used with a rich set of information
from a 1993 survey of 193 households with women and children living at
home in a rural county. The aim was to define a criterion measure for
food security to compare with food security items from Radimer/Cornell
and CCHIP (Frongillo et al. 1997
). The criterion measure that was
developed approximated the most definitive measure possible, one that
would be gained from an in-depth understanding of the experience itself
through a personal interview with the respondent. Two researchers with
very different experiences achieved good agreement after working
independently, and excellent consensus after working together, in
categorizing the households. The Radimer/Cornell and CCHIP measures had
good specificity (i.e., percentage of truly food secure correctly
classified, 63–71%) and excellent sensitivity (i.e., percentage of
truly food insecure correctly classified, 84–89%) compared with the
criterion measure. Estimates of the prevalence of household food
insecurity from the criterion, Radimer/Cornell and CCHIP measures were
almost identical. Similar results have been obtained for the
Radimer/Cornell measure by Anne-Marie Hamelin in Québec (personal
communication) and in a study of elderly urban Black and rural White
women and men (Wolfe et al. 1998
).
Additional evidence for the accuracy of questionnaire-based measures of
food security has come from comparisons of the Radimer/Cornell or the
national food security measure with a large number of determinants and
consequences such as income, education, participation in food
assistance programs, having savings, food expenditures and food
consumption (Hamilton et al., 1997
, Kendall et al. 1994
, 1995
and 1996
,
Olson et al. 1994
and 1997
). These studies have demonstrated that the
food security status of groups of households is associated with these
determinants and consequences in the expected manner.
Attribution of accuracy.
It is crucial to establish whether the apparent accuracy of a measure
is actually due to the well-grounded understanding. Otherwise, a
measure that is supposedly validated may well actually be useless. For
example, the apparent accuracy of a measure of lean body mass of small
animals has been shown to be attributable to the measure's relation to
total body weight and not to its relation to actual lean body mass as
had been thought (Bell et al. 1994
). This issue for the assessment of
food security was addressed by asking whether the food security measure
accounted for variability in the definitive criterion measure after
accounting for other socioeconomic factors (Frongillo et al. 1997
).
Multinomial logistic regression was used to examine a sequence of
models for the Radimer/Cornell measure. For example, income alone had a
model fit (-2 log-likelihood) of 233 (6 df). When employment status was
added, the fit improved to 201 (18 df). When the Radimer/Cornell
measure was added to the other two variables in model, the fit was 182
(24 df), a substantial further improvement (P <
0.0005). Nearly identical results were found for the CCHIP measure.
These results demonstrate that the food security scale measures food
insecurity status beyond what income and employment (and other
socioeconomic variables that were examined) can explain.
|
SUMMARY |
|---|
|
TOPABSTRACTINTRODUCTIONVALIDATION OF MEASURESVALIDITY OF FOOD SECURITY...SUMMARYREFERENCES |
|---|
To assess accuracy for uses at the level of individuals, a rich set of information available from survey interviews was used to develop a definitive criterion measure. Both the Radimer/Cornell and CCHIP questionnaire-based measures had good specificity and excellent sensitivity compared with the criterion measure.
These results support the use of the items from the Food Security Supplement measure, which is based on the Radimer/Cornell and CCHIP measures, for the purposes of estimating the prevalence of food insecurity and hunger in the U.S. population. Furthermore, the excellent sensitivity of these measures means that they can be used validly to identify households for food insecurity and hunger, and to target portions of the population for food programs. Further validation research is required for subgroups of the population; this may be different from that in Upstate New York, which has been studied most intensively for validation purposes. Continued validation work should be done with the national food security measure to ensure its accuracy for monitoring population changes in prevalence and changes in households and individuals.
This approach of using questionnaire items to assess people's experience of food insecurity has the direct potential to be important in providing a common means for assessing food security in other countries. Food insecurity is a social as well as biological, nutritional and economic phenomenon. This approach provides a way of capturing the social as well as other aspects of this phenomenon. Even the best economic measures fail to capture important social and contextual variability. Research that uses qualitative and quantitative methods to fully understand how people in a variety of countries experience and think about food insecurity is required. Then, this understanding of the experience of food insecurity should be used to develop systematically robust and contextually sensitive measures that have demonstrated validity, building upon what has been done for the U.S. national food security measure.
|
ACKNOWLEDGMENTS |
|---|
|
FOOTNOTES |
|---|
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONVALIDATION OF MEASURESVALIDITY OF FOOD SECURITY...SUMMARYREFERENCES |
|---|
1. Alaimo, K. (1997) Food Insecurity, Hunger, and Food Insufficiency in the United States: Cognitive Testing of Questionnaire Items and Prevalence Estimates from the Third National Health and Nutrition Examination Survey. Masters thesis, Cornell University, Ithaca, NY..
2. Bell R. C., Lanou A. J., Frongillo E. A., Jr, Levitsky D. A., Campbell T. C.. Accuracy and reliability of TOBEC (Total Body Electrical Conductivity) for determining body composition of rats in experimental studies. Physiol. Behav. 1994;56:767-773.[Medline]
3. Carlson S. J., Andrews M. S., Bickel G. W.. Measuring food insecurity and hunger in the United StatesDevelopment of a national benchmark measure and prevalence estimates. J. Nutr. 1999;129:510S-516S.
4. Frongillo, E. A., Jr. (1994) The use of scaling and indexing to measure the severity of food insecurity. In: Conference on Food Security Measurement and Research. Food and Consumer Service, U.S. Department of Agriculture and National Center for Health Statistics, Centers for Disease Control and Prevention, U.S. Department of Health and Human Services, Washington, DC..
5. General Accounting Office (1986) Hunger counties: methodological review of a report by the Physician Task Force on Hunger. U.S. General Accounting Office, Washington, DC..
6. Habicht J.-P., Yarbrough C., Martorell R.. Anthropometric field methodscriteria for selection. Jelliffe D. B. Jelleffe E.F.P. eds. Human NutritionA Comprehensive Treatise 1979;Vol. 2 Plenum Press New York, NY.. .
7. Hamilton, W. L., Cook, J. T., Thompson, W. W., Buron, L. F., Frongillo, E. A., Jr., Olson, C. M. & Wehler, C. A. (1997) Household Food Security in the United States in 1995: Summary Report of the Food Security Measurement Project. Report prepared for the USDA Food and Consumer Service, Alexandria, VA..
8. Kendall, A., Olson, C. M. & Frongillo, E. A., Jr. (1994) Validation of the Radimer/Cornell Hunger and Food Insecurity Measures. Final Project Report to the New York State Department of Health. Cornell University, Ithaca, NY..
9. Kendall A., Olson C. M., Frongillo E. A., Jr. Validation of the Radimer/Cornell measures of hunger and food insecurity in a general population survey. J. Nutr. 1995;125:2793-2801.
10. Kendall A., Olson C. M., Frongillo E. A., Jr. Relationships of hunger and food insecurity with food availability and consumption. J. Am. Diet. Assoc. 1996;96:1019-1024.[Medline]
11. Koch D. D.. Concepts in the validation of neurochemical methodsthe proper generation and use of statistics. Life Sci 1987;41:853-856.[Medline]
12. Olson, C. M., Frongillo, E. A., Jr. & Kendall, A. (1994) Validation of Measures for Estimating the Prevalence of Hunger and Food Insecurity in the Current Population Survey Module: A Combination of Cornell and CCHIP Items. Final Technical Report to United States Department of Agriculture. Cornell University, Ithaca, NY..
13. Olson C. M., Rauschenbach B. S., Frongillo E. A., Jr, Kendall A.. Factors contributing to household food insecurity in a rural upstate New York county. Fam. Econ. Nutr. Rev. 1997;10:2-17.
14. Perez-Escamilla, R., Himmelgreen, D. A. & Ferris, A. (1997) Community Nutritional Problems Among Latino Children in Hartford, Connecticut. Connecticut Family Nutrition Program Technical Report #1, Storrs and Hartford, CT..
15. Radimer K. L., Olson C. M., Campbell C. C.. Development of indicators to assess hunger. J. Nutr. 1990;120:1544-1548.
16. Radimer K. L., Olson C. M., Greene J. C., Campbell C. C., Habicht J.-P.. Understanding hunger and developing indicators to assess it in women and children. J. Nutr. Educ. 1992;24:36S-44S.
17. Rose D. R., Basiotis P. P., Klein B. S.. Improving federal efforts to assess hunger and food insecurity. Food Rev 1995;18:18-23.
18. Uriano G. A., Cali J. P.. Role of reference materials and reference methods in the measurement process. Gould R. F. eds. Validation of the Measurement Process 1977 American Chemical Society Washington, DC.. .
19. Wehler C. A., Scott R. I., Anderson J. J.. The Community Childhood Hunger Identification Projecta model of domestic hunger—demonstration project in Seattle, Washington. J. Nutr. Educ. 1992;24:29S-35S.
20. Wernimont G.. Statistical control for measurement process. Gould R. F. eds. Validation of the Measurement Process 1977 American Chemical Society Washington, DC.. .
21.
Wolfe W. S., Olson C. M., Kendall A., Frongillo E. A., Jr. Hunger and food insecurity in the elderlyits nature and measurement. J. Aging Health 1998;10(3):327-350.
This article has been cited by other articles:
|
|
 |
|
  W. Gonzalez, A. Jimenez, G. Madrigal, L. M. Munoz, and E. A. Frongillo Development and Validation of Measure of Household Food Insecurity in Urban Costa Rica Confirms Proposed Generic Questionnaire J. Nutr., March 1, 2008; 138(3): 587 - 592. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
W. S. Wolfe, E. A. Frongillo, and P. Valois Understanding the Experience of Food Insecurity by Elders Suggests Ways to Improve Its Measurement J. Nutr., September 1, 2003; 133(9): 2762 - 2769. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 M. C Gulliford, D. Mahabir, and B. Rocke Food insecurity, food choices, and body mass index in adults: nutrition transition in Trinidad and Tobago Int. J. Epidemiol., August 1, 2003; 32(4): 508 - 516. [Abstract] [Full Text] [PDF]
|
|
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
