THE PURPOSE OF THIS STUDY WAS TO DETERMINE THE REHABILITATIVE EFFICACY OF THE PROGRAM AS MEASURED BY REDUCED CRIMINAL ACTIVITY AND GENERAL SOCIAL ADJUSTMENT, AND TO DEMONSTRATE THE UTILITY FOR EVALUATION RESEARCH OF DISCRIMINANT ANALYSIS AND BASE-EXPECTANCY ANALYSIS. DISCRIMINANT ANALYSIS PERMITS DUPLICATION OF THE SELECTION PROCESS WITH ANOTHER SAMPLE OF ELIGIBLE INDIVIDUALS FROM THE SAME POPULATION, AND BASE-EXPECTANCY ANALYSIS PERMITS THE RESEARCHER TO SEPARATE OUT THE EFFECTS OF TREATMENT AND SELECTION. THE SAMPLE OF ELIGIBLE MEN CHOSEN FOR THE PROGRAM INCLUDED ALL PRISON RESIDENTS WITH A HIGH SCHOOL DIPLOMA OR ITS EQUIVALENT. DISCRIMINANT ANALYSIS WAS USED TO SEPARATE SELECTION FROM TREATMENT EFFECTS, AND BASE EXPECTANCY PROCEDURES WERE USED TO MEASURE POSTRELEASE ADJUSTMENT AND TO CONTROL FOR SELECTION PROCESSES. THE COLLEGE PROGRAM HAD NO GENERAL BENEFICIAL EFFECTS TO COUNTERACT SERIOUS RECIDIVISM; HOWEVER, WITHIN DIFFERENT RISK LEVELS, SOME BENEFITS APPEARED FOR THOSE INMATES WHO WERE LEAST LIKELY TO SUCCEED WITHOUT THE TREATMENT REPRESENTED BY THE COLLEGE PROGRAM. ALTHOUGH NEITHER TECHNIQUE CAN COMPLETELY COMPENSATE FOR LACK OF RANDOMIZATION, BOTH CAN SIGNIFICANTLY REDUCE SELECTION BIAS AND, WHEN USED TOGETHER, CAN PROVIDE MUTUALLY SUPPORTING CONFIRMATION OF THE MAGNITUDE OF THE SELECTION EFFECT. STATISTICAL FIGURES AND TABLES ARE INCLUDED. (AOP)