A Reconsideration of an Oft-Used Method of Program Evaluation

Craig G. Gundersen, U.S. Department of Agriculture (DOA)
Dean Jolliffe, U.S. Department of Agriculture (DOA)
Laura Tiehen, U.S. Department of Agriculture (DOA)

In program evalutations, participants are often compared with eligible non-participants with respect to an ouctome consistent with program goals. If participants are better off, this is a recommendation for the program; if participants are as well-off or worse off, this is not a recommendation. Such conclusions, we argue, may unnecessarily paint some programs in a negative light and limit the opportunities for program improvement. We, instead, propose an alternative interpretation whereby we incorporate the knowledge of policymakers and program administrators. After incorporating this knowledge (and the observed and unobserved characteristics of households) the interpretation of the sign and signficicance of the relevant coefficient will change. After a theroetical treatment of our argument, we then turn to a consideration of the impact of the Food Stamp Program on food insecurity. For this, we use the Current Population Survey and reconsder the usual comparisons between participants and non-participants.