SOCIAL SCIENCE RESEARCH CAN BE NEITHER AS PRECISE NOR AS STATIC AS LABORATORY RESEARCH. IF A PROGRAM TO AID PEOPLE IS TO BE EFFECTIVE, IT MUST BE FLEXIBLE ENOUGH TO CHANGE ITS FOCUS TO MEET THE NEEDS OF THE PEOPLE IT SERVES. WARD EDWARD'S MULTIATTRIBUTE UTILITIES METHOD FOCUSES ON TWO QUESTIONS: (1) WHAT IS AT STAKE? AND (2) WHAT ARE THE ODDS? THIS APPROACH REQUIRES THAT PROGRAM GOALS BE SPECIFIED AT THE BEGINNING OF THE PROGRAM. THE UTILITIES TO BE MAXIMIZED AND THE ENTITIES TO BE EVALUATED ARE IDENTIFIED. BOTH SUBJECTIVE AND OBJECTIVE FACTORS ARE INCORPORATED INTO THE MEASUREMENT OF THESE UTILITIES. THE FORMULA IS PRESENTED FOR COMPUTING A WEIGHTED VALUE FOR EACH UTILITY IN RELATION TO COST. THE MULTIATTRIBUTE UTILITIES METHOD WAS APPLIED IN THE PLANNING PROCESS AT THE OFFICE OF CHILD DEVELOPENT (OCD). THREE UTILITIES WERE MEASURED FOR ALL OCD PROGRAMS: BENEFITS TO CHILDREN AND FAMILIES; BENEFITS TO OCD AS AN AGENCY IN HEW; AND COSTS. A LIST OF 14 VALUES TO BENEFIT CHILDREN AND FAMILIES WAS DEVELOPED. INCLUDED IN THE LIST WERE SERVICE CONTINUITY, DEVELOPING LEADERSHIP IN COMMUNITIES SERVED, AND ENHANCING PARENTAL INVOLVEMENT IN PROGRAM PLANNING AND MANAGEMENT. BY DETAILING SUCH GOALS IN ADVANCE, THE PROBLEM OF FOCUSING ON TOO NARROW AN ASPECT IS AVOIDED. AGENCY GOALS INCLUDED COORDINATING PROGRAMS FOR CHILDREN AT FEDERAL, STATE, AND LOCAL LEVELS AND INFLUENCING POLITICAL LEADERS TO BE MORE SENSITIVE TO THE NEEDS OF CHILDREN. EVALUATION RESEARCH, THEREFORE, MUST BE CLOSELY LINKED TO THE VALUES AND FINAL DECISIONS OF A PROGRAM. THE OUTCOMES OF THE MULTIATTRIBUTE UTILITIES PLANNING PROCESS PROVIDE THE EVALUATION RESEARCHER WITH THE HYPOTHESIS THAT HE WILL INVESTIGATE, AND THIS APPROACH FORCES HIM OUT OF A NARROW OPERATIONALISM. OLD RESEARCH METHODS MAY NOT WORK. THE DEVELOPMENT OF NEW TECHNIQUES MAY BE CALLED FOR TO MEET SPECIAL NEEDS. (GLR)