THE BASIC HYPOTHESIS WAS THAT BY MAKING USE OF STATISTICAL TECHNIQUES WHICH ALLOW FOR A RELAXATION OF BOTH THE LINEARITY AND HOMOSCEDASTICITY ASSUMPTIONS, IT SHOULD BE POSSIBLE TO OBTAIN RESULTS THAT WILL ALLOW FOR MORE ACCURATE AND CONFIDENT INFERENTIAL STATEMENTS REGARDING THE CHARACTERISTICS OF GIVEN CRITERION POPULATIONS, IN THIS CASE PAROLE SUCCESSES AND PAROLE FAILURES. ELEVEN PERSONALITY VARIABLES WERE SELECTED, ON A PRIORI GROUNDS, FOR EVALUATION IN TERMS OF THEIR DESCRIPTIVE POTENTIAL. WAS COLLECTED BY MEANS OF THE JESNESS PSYCHOLOGICAL INVENTORY. SUBJECTS WERE SAMPLED FROM THE POPULATION OF 336 MALE, CALIFORNIA CENTRAL VALLEY (SACRAMENTO, STOCKTON, MODESTO) COMMUNITY TREATMENT PROJECT YOUTHS WHO PARTICIPATED AS EXPERIMENTAL CASES BETWEEN THE YEARS 1961 AND 1969. FOR EACH CASE WITHIN THE POPULATION, THE NUMBER OF PAROLE SUSPENSIONS RECEIVED DURING A PERIOD OF UP TO 24 PAROLE MONTHS WAS COUNTED FOR EACH OF THREE VIOLATION SEVERITY CATEGORIES. DATA WAS ANALYZED AND GROUPED BY THREE MAIN CATEGORIES: PAROLE SUCCESSES, MODERATE SUCCESSES, AND PAROLE FAILURES. THE ANALYSES SHOWED THAT DIFFERENT SETS OF TRAITS WERE FOUND TO BE OF RELEVANCE TO PAROLE SUCCESSES AND PAROLE FAILURES. ALSO, IT WAS FOUND THAT A LARGE PROPORTION OF THE VARIABLES THAT HAD BEEN EXAMINED WERE RELATED TO THE CRITERION OF PAROLE ADJUSTMENT IN WAYS WHICH DEPARTED FROM THE CONDITIONS OF LINEARITY AND HOMOSCEDASTICITY, WHICH ARE USUALLY ASSUMED IN MOST TRADITIONAL STATISTICAL ANALYSES. (AUTHOR ABSTRACT MODIFIED)