THE USE OF RANDOMIZED CONTROL GROUP PROCEDURES AS METHODS FOR PROGRAM EVALUATIONS HAS BEEN CRITICIZED AND HAS CAUSED A DIVISION AMONG THOSE EVALUATORS WHO ARE EXPERIMENTALISTS AND THOSE WHO ARE NOT. SUCH A SITUATION IS UNNECESSARY SINCE EVALUATION RESEARCH SHOULD BE ABLE TO ACCOMODATE MULTIPLE METHODOLOGICAL APPROACHES. IN FACT, IT MAY ONLY BE THROUGH THE USE OF MULTIPLE RESEARCH STRATEGIES THAT PROGRAM EVALUATION'S ROLE IN SOCIAL POLICY FORMATION CAN BE REALIZED. THE VARIETY OF METHODS AVAILABLE TO EVALUATION RESEARCHERS--TRUE EXPERIMENTAL, QUASI-EXPERIMENTAL, SURVEY, STATISTICAL--ARE FULLY COMPATIBLE WITH ONE ANOTHER. THE CHOICE OF METHODOLOGY SHOULD DEPEND ON THE CONTEXT OF THE PROGRAM AND THE REQUIRED DECISION NEEDS. IT SHOULD BE POSSIBLE TO USE METHODOLOGIES EITHER SINGLY OR TOGETHER AS PART OF THE EVALUATION OF A PARTICULAR PROGRAM. A CASE STUDY ILLUSTRATES ONE APPLICATION OF A COMBINED MODEL IN AN ATTEMPT TO GAIN THE STRONGEST INFERENCE POSSIBLE ABOUT IMPACT IN A STUDY OF TELEVISION DIAGNOSIS IN RURAL AREAS, WHILE RETAINING THE PRAGMATIC CONVENIENCE OF THE NONEXPERIMENTAL MODELS FOR ANSWERING OTHER QUESTIONS. THE MULTIPLE-METHODOLOGICAL APPROACH ADVOCATED SUGGESTS THAT THE TYPICAL BASIC-TO-APPLIED RESEARCH PROCESS BE MODIFIED SO THAT BASIC RESEARCH STUDIES CAN BE CONDUCTED AS PART OF PROGRAM EVALUATION. RESEARCH IS THUS CONCEPTUALIZED AS A CYCLICAL PROCESS BETWEEN THE LABORATORY AND THE FIELD, AND BETWEEN THEORY AND ACTION. THE ARTICLE CONCLUDES THAT THIS TYPE OF EVALUATION RESEARCH CAN ONLY LEAD TO BETTER UNDERSTANDING OF THE SOCIAL WORLD AND THE PROSPECT OF BETTER SOCIAL POLICY. REFERENCES ARE CITED. (MJW)