COMPARATIVE EVALUATION'S ABILITY TO ASSESS INNOVATIVE COMPONENTS OF PROGRAMS IS TOUTED. THE COMPARATIVE FRAMEWORK FOR COMPARISON ANALYSIS INVOLVES CONTROLLED DATA COLLECTION. THE MATCHED PANEL TECHNIQUE IS VIEWED AS SATISFYING THIS CONDITION FOR CONTROL. THE MATCHED PANEL SYSTEM INVOLVES THE SELECTION OF PERSONS FROM THE GENERAL PRISON POPULATION WHO CONFORM TO THE PERSONAL AND CRIME HISTORY CHARACTERISTICS PROFILE OF PERSONS FROM THE PRISON POPULATION SELECTED TO PASS THROUGH THE CSC PROGRAM. MATCHING WAS DONE USING SEVEN VARIABLES: (1) OFFENSES, (2) LENGTH OF SENTENCE, (3) AGE, (4) CRIMINAL HISTORY, (5) GEOGRAPHIC DISTRIBUTION, (6) RACE, AND (7) QUARTER IN WHICH PAROLED. IN EVALUATING THE PENNSYLVANIA COMMUNITY SERVICE CENTERS, DATA WERE OBTAINED FROM THE PENNSYLVANIA BOARD OF PROBATION AND PAROLE. THE STUDY SAMPLE WAS COMPOSED OF 231 CSC INDIVIDUALS AND 245 NON-CSC INDIVIDUALS. IT WAS FOUND THAT THE NON-CSC OFFENDERS ENCOUNTERED ALMOST THREE TIMES AS MANY POSTINCARCERATION PROBLEMS THAN DID THOSE OFFENDERS WHO HAD GONE THROUGH CSC PROGRAMS. IN ADDITION, IT WAS FOUND THAT THE SEVERITY OF CRIMES COMMITTED BY CSC OFFENDERS WHILE ON PAROLE WAS LESS THAN THAT OF OFFENDERS IN THE MATCHED PANEL. FURTHERMORE, PAROLEES FROM CSC PROGRAMS HAD A HIGHER 'CLEAN' PERCENTAGE OF TOTAL MONTHS AT RISK THAN DID THEIR MATCHED COUNTERPARTS. A COST-BENEFIT ANALYSIS OF CSC PROGRAMS IS DESIRED, BUT IS NOT FEASIBLE USING THE DATA SOURCES AVAILABLE FOR THIS STUDY. HOWEVER, REDUCED RECIDIVISM AND IMPROVED PAROLE PERFORMANCE ARE VIEWED AS OBVIOUS COST REDUCTIONS TO THE CRIMINAL JUSTICE SYSTEM. TABLES ARE INCLUDED IN THE TEXT. (MLC)