NEP IS DESIGNED FOR STEP-BY-STEP IMPLEMENTATION, SO THAT AFTER PARTIAL IMPLEMENTATION, A PROJECT MAY BE HALTED AND MONEY SAVED IF EVALUATION IS JUDGED TO BE INFEASIBLE. THE FOLLOWING STEPS ARE DESCRIBED: (1) DISCOVERING PREVAILING EXPECTATIONS FOR THE PROGRAM BY EXAMINING EXISTING LEGISLATION, GRANT APPLICATIONS, EVALUATION REPORTS, AND OTHER DOCUMENTATION; (2) GETTING THE BIG PICTURE BY CONDUCTING ONSITE VISITS AND INTERACTING WITH APPROPRIATE PERSONNEL; (3) DEVELOPING A LOGIC MODEL BY LINKING INPUTS TO PROCESSES AND PROCESSES TO OUTCOMES AND IMPACTS; (4) DEVELOPING MEASURES FOR ALL VARIABLES IN THE LOGIC MODEL; (5) PREPARING FOR HYPOTHESIS TESTING; (6) DEVELOPING A SIMPLE PROGRAM MONITORING AND EVALUATION DESIGN; (7) IDENTIFYING PROBLEMS IN FILLING GAPS IN THE KNOWLEDGE ABOUT THE PROGRAM; (8) DEVELOPING ALTERNATE EVALUATION DESIGNS FOR FILLING KNOWLEDGE GAPS; AND (9) PRETESTING THE EVALUATION DESIGNS. STEPS ONE THROUGH FIVE WERE APPLIED TO ADULT INSTITUTIONS FOR SENTENCED FELONS UNDER A SINGLE INSTITUTIONAL ADMINISTRATION HAVING AT LEAST ONE PROGRAM OR AREA WHERE MALE AND FEMALE INMATES INTERACT. APPLYING STEP ONE, MANY EXPECTATIONS WERE FOUND; SOME WERE UNRELATED TO THE PRESENCE OF BOTH SEXES IN AN INSTITUTION; E.G., PROVIDING PROGRAM ACCESS TO FEMALE INMATES EQUAL TO THAT OF MALES IN RESPONSE TO LEGAL PRESSURE AND EXPANDING FEMALE CORRECTIONAL STAFF CAREER LADDERS. EXPECTATIONS CENTERED ON SUCH GENERAL BENEFITS AS A MORE NORMAL, LESS INSTITUTIONAL ATMOSPHERE AND ON ADVERSE CONSEQUENCES, PARTICULARLY FOR WOMEN. CHANGES IN SECURITY PRACTICES AND COMMUNITY RELATIONSHIPS WERE ALSO ADDRESSED. IT WAS OBSERVED THAT DIFFERENT CORRECTIONAL PHILOSOPHIES ENGENDERED DIFFERENT EXPECTATIONS. STEP TWO--SITE VISITS TO 10 COED INSTITUTIONS--SHOWED GREAT VARIATION IN COED PRISONS. STEP THREE UNCOVERED THREE PROGRAMMATIC MODELS BASED ON ASSUMING THAT COCORRECTIONS FILL A POSITIVE TREATMENT FUNCTION: THE REINTEGRATION MODEL, INSTITUTIONAL CONTROL MODEL, AND THE THERAPY MODEL. TWO NONPROGRAMMATIC MODELS FOCUS ON THE PRESENCE OF MALE-FEMALE INTERACTION AND TRY TO MINIMIZE ITS EFFECT ON INSTITUIONAL LIFE: THE SURVEILLANCE AND SANCTION MODEL AND THE ALTERNATE CHOICE MODEL. ALL OF THESE WERE PRESENT TO SOME EXTENT IN EACH PRISON. STEP FOUR TRANSLATED THE LOGIC MODEL CONCEPTS INTO MEASURABLE OUTCOMES, AND STEP FIVE WAS NOT ACCOMPLISHED BECAUSE OF INSUFFICIENT EVALUATION DATA. SUPPORTING DATA OR REFERENCES ARE NOT INCLUDED. (MLC)