Food Insecurity or Poverty? Measuring Need-Related Dietary Adequacy

Year: 2001

Research Center: Institute for Research on Poverty, University of Wisconsin-Madison

Investigator: Currie, Janet, Jayanta Bhattacharya, and Steven Haider

Institution: Department of Economics, University of California, Los Angeles

Project Contact:
Janet Currie, Professor
UCLA Department of Economics
405 Hilgard Avenue
Los Angeles, CA 90095-1477


The U.S. Department of Agriculture monitors the food security of U.S. households through an annual survey that contains questions about behaviors that signal an inability to meet food needs because of financial constraints. The survey has been conducted annually since 1995 as part of the Current Population Survey (CPS) and has been adopted, at least in part, by many other surveys. Numerous researchers have used these questions to analyze a variety of topics, with reports published in medical and public health journals. An advantage of the questions is that they are relatively inexpensive to administer compared with biomedical measures or dietary recall.

Several recent studies examined the validity of the food insecurity questions. These studies examined how the questions are correlated among themselves (that is, their internal validity) and how the questions are correlated with demographic characteristics, household characteristics, and dietary outcomes (that is, their external validity). Generally, these studies found the food insecurity questions to be correlated in expected ways with both internal and external factors. For example, using the Continuing Survey of Food Intake by Individuals, one author found that in households reporting insufficient food, most household members had a significantly lower intake of most vitamins and minerals than members of other households. One exception was that preschoolers in food-insecure households did not suffer from low consumption.

The authors assessed the empirical content of the food insecurity questions, advancing the literature in several directions. First, rather than simply examine whether the food insecurity questions were correlated with other factors, they focused on how well they were correlated. The standard poverty measure serves as a useful benchmark for these purposes because it has been used extensively and can be computed from many different data sets. Second, this report used a unique dataset, the National Health and Nutrition Examination Survey III (NHANES III).1 In addition to the dietary recall information collected in other data sets, NHANES III collects and analyzes blood from its participants. Thus, the authors could examine measures of diet adequacy from individuals of all ages without recall or proxy bias. Third, the authors examined how the correlations between the responses to the food security questions and dietary outcomes varied by age. This last point is particularly valuable, given that standard food insecurity questions make distinctions by age. For example, one distinction between the CPS’s two most severe categories of food insecurity rests on whether children are skipping meals.

The study found that the responses to the food security questions are correlated with the diets of older household members but are not consistently correlated with the diets of children. In contrast, poverty is consistently related to the diets of preschoolers. Among adults, poverty and food insecurity questions are good predictors of diet. However, poverty may be a better overall predictor of diet quality, since it is more consistently related to a range of dietary outcomes than the food insecurity questions.

Although the focus of this research was related to measurement, it is important to note two substantive aspects of the study’s findings. First, individuals in poverty tend to have different dietary outcomes even at the basic level of vitamin deficiencies and anemia. This finding is true for most age groups in the population, including the youngest and oldest, the two particularly vulnerable age groups. Second, the study reveals several underlying behavioral issues. For example, it found much variation by age in the relationship between poverty and dietary outcomes. Adult dietary outcomes are more correlated with poverty than are child outcomes, and dietary outcomes of younger children are more correlated with poverty than are the dietary outcomes of older children. It is likely that parents protect their children from the effects of poverty to the extent that they can and that older children have more opportunities to supplement their consumption outside the home. It would be useful to have a better understanding of these protective family behaviors.