Evaluations can be designed to analyze the planning of a project, the structure of a program, the process by which services are delivered, the success of the program and the cost benefit of the program. Each of these questions is answered by a different source of data in a range of research approaches, all using standardized, replicable procedures. Evaluation is always done in a political context; outside evaluators can serve to guard against avoidable bias; and there is considerable value in the evaluator and service provider working together to develop both the program and the research. The chapter provides details on: use of an outside evaluator; formulative evaluations; the classical experiment; need for comparison groups; matching; measuring change; use of pretest data; identification of service provided; defining program objectives; program evaluation; service outcome linkage; what should be measured; testing the achievement of outcomes; use of statistical tests; and writing the final report. Charts, references