A cautionary tale: Using propensity scores to estimate the effect of food stamps on food insecurity

This article uses propensity scores to evaluate the effect of food stamps on food insecurity, a measure of inadequate food supply. It relies on data from the Early Childhood Longitudinal Study-Kindergarten Cohort. By balancing treatment and comparison groups on covariates, the propensity score method adjusts for bias caused by observed variables. This method may be preferable to regression because it does not rely on a linear functional form to adjust for potential confounding variables. Results show that food stamps do not decrease the probability of being food insecure, although they lessen the severity of the problem according to some models. However, propensity scores rely on several stringent assumptions, including the need for a common support region (where two compared groups share the same characteristics) and a properly specified model. Propensity scores should therefore be employed with caution.