Abstract
This study explores the relationships between program practices and program impacts on earnings and welfare payments in California's welfare-to-work program, known as GAIN. Practices and impacts (based on a random assignment experiment) are compared across six counties and (on some measures) 20 local offices. The findings challenge some popular theories of "what works best" in welfare-to-work programs and illustrate how comparisons of a small number of sites within multisite studies can help researchers "get inside the black box, " a common problem in evaluation research.
Get full access to this article
View all access options for this article.
