Using Household Fit Indices to Examine the Psychometric Quality of Food Insecurity Measures.

Abstract

This study focuses on model-data fit with a particular emphasis on household-level fit within the context of measuring household food insecurity. Household fit indices are used to examine the psychometric quality of household-level measures of food insecurity. In the United States, measures of food insecurity are commonly obtained from the U.S. Household Food Security Survey Module (HFSSM, 18 items) of the Current Population Survey Food Security Supplement (CPS-FSS). These measures, in various forms, are used to inform national programs and policies related to food insecurity. Data for low-income households with children from recent administrations of the HFSSM (2012-2014) are used in this study (N = 7,324). The results suggest that there are detectable levels of misfit with Infit mean square error (MSE) statistics ranging from 6.73 % to 21.33% and Outfit MSE statistics ranging from 5.31% to 9.68%. The data suggest for Outfit MSE statistics that (a) male respondents, (b) respondents with lower levels of education, and (c) respondents who did not report participating in SNAP (Supplemental Nutrition Assistance Program, formerly the Food Stamp Program) tend to have more misfit. For Infit MSE statistics, lack of homeownership appears to be a predictor of misfit. The implications of this research for future research, theory, and policy related to the measurement of household food insecurity are discussed.