IT IS ASSUMED THAT, IN MOST CASES, SOCIAL SCIENCE IS BETTER THAN EXECUTIVE GUESS WORK IN DETERMINING THE FEASIBILITY, EFFECTIVENESS, EFFICIENCY, AND EQUITY OF SOCIAL PROGRAMS INTENDED TO DEAL WITH SUCH SOCIAL PROBLEMS AS UNEMPLOYMENT, ILL HEALTH, CRIME, AND UNDEREDUCATION. IT IS ALSO CONTENDED THAT SOCIAL SCIENCE RESEARCH MAKES IMPORTANT CONTRIBUTIONS TO SOCIAL POLICIES AND PROGRAMS DESIGNED TO ALLEVIATE SOCIAL PROBLEMS AND THAT MANY SOCIAL POLICIES AND PROGRAMS DO INDEED MINIMIZE SOCIAL PROBLEMS. BECAUSE IMMEDIATE LARGE-SCALE IMPLEMENTATION OF POLICIES AND PROGRAMS IS RISKY DUE TO THE FACT THAT PREDICTIVE THEORY IS INADEQUATE, SOCIAL EXPERIMENTS ARE ESSENTIAL FOR PROGRESS TO OCCUR WITH MINIMUM RISK. SOCIAL EXPERIMENTS MUST USUALLY BE LARGE BECAUSE LARGE NUMBERS OF SUBJECTS ARE NEEDED TO ASSURE THAT MEASURED RESULTS ARE RELIABLE, COMPLEX CONTEXTS AND CONSTRAINTS IMPOSE SIZE AND DURATION REQUIREMENTS, AND LARGE SOCIAL PROBLEMS ARE INTRINSICALLY INTERDISCIPLINARY. LARGE SOCIAL EXPERIMENTS REQUIRE SUBSTANTIAL FUNDING, AND TREATMENT IN MOST EXPERIMENTS USUALLY AMOUNTS TO AT LEAST $1,000 PER YEAR PER PARTICIPANT. THERE ARE GOOD ARGUMENTS FOR SMALL SOCIAL SCIENCE EXPERIMENTS, PARTICULARLY DISCIPLINARILY-CONCENTRATED RESEARCH PROJECTS THAT DEVELOP INDEPTH SKILLS NEEDED FOR LARGE INTERDISCIPLINARY PROJECTS. SMALL STUDIES IDENTIFY HYPOTHESES AND ALTERNATIVES FOR LARGER EXPERIMENTS TO TEST. THEY CAN BE EXECUTED BY A LARGER AND MORE DIVERSE POPULATION OF RESEARCH ORGANIZATIONS, THUS REDUCING THE RISK OF MONOLITHIC THEORIES PROMULGATED BY A FEW LARGE RESEARCH INSTITUTIONS. LARGE SOCIAL EXPERIMENTS REQUIRE SERIOUS EVALUATION SIMPLY FOR PUBLIC ACCOUNTABILITY REASONS AND ARE LIKELY TO CREATE, DOCUMENT, AND DISSEMINATE NEW KNOWLEDGE. MANY SMALL PROGRAMS ARE INCONCLUSIVE SINCE EVALUATION OFTEN COMES AS AN AFTERTHOUGHT, WITH CONSEQUENT LIMITATIONS ON VALIDITY. THE FEASIBILITY OF A LARGE SOCIAL SCIENCE PROGRAM IS DISCUSSED, AND ARGUMENTS AGAINST INTEGRATED LARGE AND SMALL SOCIAL SCIENCE RESEARCH AND DEVELOPMENT EFFORTS ARE OUTLINED. (DEP)