THE ROLE OF THE SOCIAL SCIENTIST IS CRUCIAL IN THE DEVELOPMENT OF PROGRAMS THAT WILL EFFECT LARGE-SCALE SOCIAL CHANGE. THIS ROLE IS EVIDENCED IN THE DESIGN OF SOCIAL PROGRAMS WHOSE IMPACT CAN BE PROPERLY MEASURED. THE DEVELOPMENT AND IMPLEMENTATION OF THE EVALUATION PROCESS ITSELF IS EMPHASIZED AND THE BUILDING OF A FUND OF KNOWLEDGE DERIVED FROM EVALUATION RESEARCH THAT CAN BE USED IN PROGRAM DEVELOPMENT. WITHIN THE RESEARCH ENVIRONMENT, MASSIVE SOCIAL PROGRAMS ALMOST ALWAYS ARE FUNDED BY POLITICAL BODIES OR PRIVATE FOUNDATIONS. THEY FREQUENTLY EXIST IN AN ENVIRONMENT OF ARBITRARY AND CHANGING DECISIONMAKING BY FUNDING AUTHORITIES, SO THAT THE CLASSICAL INDEPENDENT-DEPENDENT VARIABLE MODEL FOR EVALUATION IS NOT FEASIBLE. IT IS RECOMMENDED, THEREFORE, THAT ACTION-RESEARCH BE DEVELOPED IN TERMS OF A SERIES OF STAGED INPUTS AND OUTPUTS. THIS KIND OF DESIGN IS ILLUSTRATED IN A JUVENILE DELINQUENCY PREVENTION PROJECT IN BOSTON, MASS. IN ORDER TO INFLUENCE SOCIAL POLICY, FINDINGS FROM SOCIAL-ACTION EXPERIMENTS MUST PROVIDE A BASIS FOR THE EFFICIENT ALLOCATION OF FINANCIAL AND HUMAN RESOURCES IN THE SOLUTION OF SOCIAL PROBLEMS. IN ORDER TO ACCOMPLISH THIS, EVALUATION RESEARCH MUST FULFILL THE FOLLOWING REQUIREMENTS: (1) THE DETERMINATION OF EFFICACY WHICH INVOLVES THE APPROXIMATION OF THE EXPERIMENTAL MODEL AS CLOSELY AS POSSIBLE IN EVALUATION; (2) ACCOUNTABILITY WHICH INVOLVES THE DETERMINATION THAT THE NUMBER OF CLIENTS SERVED AND THE SIZE OF THE POTENTIAL AGGREGATE MAKES THE PROGRAM WORTHWHILE; AND (3) THE DETERMINATON OF EFFICIENCY WHICH INVOLVES FINDING THE PER UNIT CHANGE IMPACT OF A PROJECT. PROBLEMS INVOLVE THE MEASUREMENT OF THE RELATIONSHIP BETWEEN SETS OF CHANGE, INTERPRETING THE MEANING OF CHANGE IN THE DEPENDENT VARIABLE, AND SUBJECT INVOLVEMENT IN MORE THAN ONE PROGRAM. (RCB)