Evaluating the effects of changes in AFDC: Methodological issues and challenges

Analysts are commonly called upon to perform the difficult task of evaluating the effects of specific changes in public policy upon the behavior of individuals, such as a change in the provisions of the Aid to Families with Dependent Children (AFDC) program relating to the benefits of those in the program who find work. When charged with such a task, analysts commonly try to answer the question by tracing the behavior of a fixed panel of individuals, comparing the experience of the group before and after the change in policy. That approach, however, risks major errors; in the case of the AFDC program, for instance, changes in the work benefit provisions affected the decisions of some who might have come into the program, a consequence that would not be picked up by a fixed panel of initial recipients. Cross‐sectional data drawn independently from a general population at points in time before and after a policy change can often provide a more valid measure of the effects of the policy change than can panel data; moreover, cross‐sectional data are usually less expensive and more readily available.