A POLITICAL CLIMATE THAT ENCOURAGES SOCIAL REFORMERS TO PREJUDGE AND COMMIT THEIR POLITICAL PRESTIGE TO THE SUCCESS OF A PARTICULAR APPROACH TOWARD A COMPLEX SOCIAL PROBLEM TENDS TO DISCOURAGE DESIGN FOR AND THE EXERCISE OF OBJECTIVE EVALUATION OF SOCIAL PROGRAMS. THE PUBLIC SHOULD BE EDUCATED TO HAVE A REALISTIC VIEW OF THE TRIAL AND ERROR, EXPERIMENTAL CHARACTER OF ATTEMPTS TO DEAL WITH DIFFICULT SOCIAL PROBLEMS. THE DESIGNERS AND ADMINISTRATORS OF PROGRAMS THEN WILL BE MORE INCLINED TO PLAN AND IMPLEMENT PROJECTS ACCORDING TO EXPERIMENTAL DESIGNS THAT WILL PERMIT EFFECTIVE EVALUATION OF A GIVEN APPROACH. TRUE EXPERIMENTATION WHICH PROVIDES FOR RANDOMLY ASSIGNED EXPERIMENTAL AND CONTROL GROUPS IS RECOMMENDED WHERE POSSIBLE. THE ALLOCATION OF SCARCE RESOURCES BY LOTTERY, A STAGGERED PHASING IN OF A PROGRAM FOR VARIOUS RECIPIENTS, AND THE USE OF THE PILOT PROJECT CONCEPT ARE BELIEVED TO PROVIDE ETHICAL AND WORKABLE PROCEDURES FOR PROVIDING CONTROL GROUPS TO BE EVALUATED ON THE BASIS OF THEIR NOT RECEIVING THE PROGRAM SERVICES THAT THE EXPERIMENTAL GROUP IS RECEIVING. IF A REFORM PROGRAM MUST BE INTRODUCED ACROSS THE BOARD SO THAT THE TOTAL UNIT RECEIVES EXPERIMENTAL TREATMENT AND A CONTROL GROUP IS NOT POSSIBLE, THE INTERRUPTED TIME-SERIES DESIGN IS SUGGESTED TO TEST THE GROUP BEFORE AND AFTER THE SERVICES OF THE PROGRAM. IF SIMILAR UNITS EXIST UNDER INDEPENDENT ADMINISTRATION, A CONTROL SERIES DESIGN IS CONSIDERED APPROPRIATE WHERE ONE GROUP RECEIVES THE TREATMENT AND THE OTHER DOES NOT. INTERNAL AND EXTERNAL PROBLEMS AFFECTING THE VALIDITY OF TRUE AND QUASI-EXPERIMENTAL DESIGNS ARE DISCUSSED AND ILLUSTRATED USING ACTUAL PROGRAMS AS EXAMPLES. (RCB)