THE EVALUATION USED AN EXPERIMENTAL DESIGN, IN WHICH PAIRS OF SCHOOLS WERE RANDOMLY ASSIGNED TO TREATMENT (THOSE THAT RECEIVED FEDERAL FUNDS) AND NONTREATMENT (THOSE THAT DID NOT) GROUPS FOR COMPARISON PURPOSES. THE TREATMENT SCHOOLS IMPLEMENTED MORE HUMAN RELATIONS PROGRAMS, MORE IN-SERVICE PROGRAMS FOR TEACHERS, AND MORE CURRICULUM CHANGES THAN NONTREATMENT SCHOOLS. BOTH TEACHERS AND STUDENTS AT THE TREATMENT SCHOOLS GAVE EVIDENCE OF IMPROVEMENTS IN RACE RELATIONS. THE ACHIEVEMENT TEST SCORES FOR BLACK MALE 10TH GRADERS IN THE TREATMENT SCHOOLS WERE HIGHER THAN THOSE OF STUDENTS AT NONTREATMENT SCHOOLS. THE ACHIEVEMENT TEST RESULTS WERE CONCRETE EVIDENCE THAT THE FUNDING HAD A POSITIVE IMPACT, AND THE EVALUATORS ATTRIBUTED THAT IMPACT IN PART TO THE HUMAN RELATIONS EMPHASIS OF THE PROGRAM. THIS ATTRIBUTION CAME AT A TIME WHEN POLICYMAKERS WERE IN THE PROCESS OF SHIFTING THE PROGRAM'S EMPHASIS AWAY FROM HUMAN RELATIONS TOWARD REMEDIAL EDUCATION PROGRAMS FOR BLACK STUDENTS. FOR SEVERAL REASONS, INCLUDING ASPECTS OF THE EVALUATION FEEDBACK MECHANISM IN THE OFFICE OF EDUCATION, AND THE LIMITED NATURE OF POLICY RECOMMENDATIONS LIKELY TO FLOW FROM ANY EFFECTIVENESS EVALUATION, THE EVALUATORS' RECOMMENDATIONS HAD LIMITED IMPACT ON POLICY. EVALUATORS NEED TO PARTICIPATE IN PLANNING ACTIVITIES WITH PROGRAM MANGERS. THE PLANNING EFFORT SHOULD MAKE IT POSSIBLE TO DETERMINE WHETHER A PROGRAM'S OBJECTIVES ARE WELL ENOUGH ARTICULATED TO JUSTIFY A FULL-SCALE EFFECTIVENESS EVALUATION. IF THEY ARE NOT, THEN THE EVALUATION EFFORT CAN BE DESIGNED TO HELP PROGRAM MANAGERS CLARIFY THEIR OBJECTIVES AND PERHAPS REDIRECT THEIR EFFORTS. (LKM)