Focusing on the problems of the federal role in education, the educational "competency" of federal programs, and the usefulness of research-based knowledge for educational practice, this report evaluates the Experimental Schools Program (ESP), a small federal program in the early 1970s that aimed to help poor and minority students by encouraging "comprehensive change" in school districts' educational practices. Data for the evaluation came from ESP documents and interviews with ESP staffers, local school officials, and ESP researchers. After a general discussion of federal-local relations in education, the authors look at the differences between federal and local approaches to change, the variety of local responses to ESP, and the problem of defining comprehensive change. They next examine the conflicts arising from ESP's misunderstanding of school districts' decentralized structure and from its insistence on both local control and detailed federal monitoring. The following section describes ESP's holistic evaluations of its programs, the difficulties in applying holistic methods to the study of ill-defined changes, and the effects of evaluation on local administrators and teachers. The concluding section summarizes the authors' assessment of ESP. (RW)