Analysis of Food Stamp Participation Using Qualitative Choice Models

Abstract Food stamp program participation at the household level was analyzed using a nationwide sample of households as well as refined measures of residential location, source of income, and household age distribution. Additionally, an empirical comparison of probit analysis vis‐à‐vis logit analysis was provided. The criteria for the comparison consisted of the use of estimated parameter values, signs, magnitudes, and test statistics, goodness‐of‐fit to sample data, predictive ability to independent data samples, and non‐nested testing procedures. On the basis of this application, the two alternatives appeared to perform equally well.