The five essays that compose this report criticize, from federal, state, and local perspectives, current methods of evaluating government-sponsored education programs. A major recurring theme is that experimental design methods, most commonly used by the Department of Education, do not provide adequate information for policy-makers' needs. The essays recommend other methods that address policy-makers' immediate concerns, including such issues as resource use and distribution of funds, fidelity of implementation, and needs of target groups, in addition to the traditional focus on student outcomes. Underlying this theme is the view that the proper relationship between evaluation and policy-making is important, but ill understood, with the consequence that contemporary evaluation methods are frequently of little value in formulating education policy. Taken together, these essays offer a generally consistent set of views on the current state of program evaluation in the federal education system, and on developing new approaches to making evaluation more useful for policy. (Author/MLF)