Rasch Analyses of Very Low Food Security among Households and Children in the Three City Study.

The longitudinal Three City Study of low-income families with children measures food hardships using fewer questions and some different questions from the standard U.S. instrument for measuring food security, the Household Food Security Survey Module (HFSSM) in the Current Population Survey (CPS). We utilize a Rasch measurement model to identify thresholds of very low food security among households and very low food security among children in the Three City Study that are comparable to thresholds from the HFSSM. We also use the Three City Study to empirically investigate the determinants of food insecurity and of these specific food insecurity outcomes, estimating a multivariate behavioral Rasch model that is adapted to address longitudinal data. The estimation results indicate that participation in the Supplemental Nutrition Assistance Program and the Temporary Assistance for Needy Families program reduce food insecurity, while poverty and disability among caregivers increase it. Besides its longitudinal structure, the Three City Study measures many more characteristics about households than the CPS. Our estimates reveal that financial assistance through social networks and a household's own financial assets reduce food insecurity, while its outstanding loans increase insecurity.