The book's recommendations are based on several years of seminar discussion through the Stanford University Evaluation Consortium. The analysis is based on the recognition that evaluation should not be only a service to central decisionmakers, but should help everyone in a pluralistic society understand what programs accomplish and why they fall short of their objectives. Evaluation is described as a political institution, but the evaluator is considered primarily an educator. Thus, the evaluator helps members of a policyshaping community recognize their own interests, weigh the consequences of alternative approaches, and discover new opportunities for action. Implications of this perspective for the planning, analysis, and reporting stages of evaluation are described. The book concludes with a description of the existing and recommended evaluation institutions, suggestions of alternative arrangements for contracting and sponsorship, and suggestions regarding the training of evaluators, the criticism of evaluations, and interpretations of research. Numerous case examples from the fields of crime, education, health care, human services, and other areas are cited. Tables, subject index, a name index, and a list of abut 400 references are included. (Author abstract modified)