THE STUDY POPULATION WAS SUBDIVIDED THREE TIMES ON THE BASIS OF: TYPE OF COMMITMENT OFFENSE, RACIAL AFFILIATION, AND ADMISSION STATUS. MULTIPLE REGRESSION ANALYSIS WAS APPLIED TO EACH SUBGROUP, AND ADDITIONAL REGRESSION ANALYSIS WAS APPLIED TO EACH SUBGROUP AND ADDITIONAL REGRESSION RUNS WERE UNDERTAKEN TO DETERMINE THE RELATIONSHIP OF ANY ONE GROUP'S PREDICTIVE EQUATION TO THE OTHER SUBGROUPS OF THE DIVISION. RESULTS INDICATED THAT DIFFERENT PREDICTIVE ITEMS MAY BE PREDICTIVE FOR DIFFERENT OFFENDER GROUPS, ALTHOUGH THE PRIMARY INTENT OF THE ANALYSIS, TO INCREASE PREDICTIVE POWER, WAS NOT SUCCESSFUL. APPENDED TO THE TEXT ARE EXTENSIVE TABLES THAT PRESENT VARIABLES USED DURING THE CONTINUATION PERIOD ANALYSIS. (AUTHOR ABSTRACT)...KAP