TO DATE, MOST PAROLE PREDICTION STUDIES HAVE FOCUSED ON VARIOUS CLUSTERS OF PERSONAL CHARACTERISTICS. FEW STUDIES HAVE FOCUSED ON THE COMMUNITY INTO WHICH THE PAROLED PERSON IS RELEASED, INTERPERSONAL RELATIONSHIPS, OR ATTITUDINAL FACTORS. QUALITY OF SUPERVISION, THE ORIENTATION OF THE PAROLE OFFICER, AND THE HONESTY OF THE PAROLEE IN ANSWERING EVALUATION QUESTIONS ARE ALSO FACTORS. THE BIGGEST PROBLEM, HOWEVER, IN THAT PAROLE STUDIES HAVE ONLY INDICATED THE EXTREMES WHICH EITHER HAVE POOR PROGNOSIS OR EXTREMELY GOOD PROGNOSIS; LITTLE HELP IS GIVEN FOR THE VAST NUMBER OF CASES IN THE MIDDLE. PERHAPS THE EMPHASIS SHOULD SHIFT FROM 'WHO SHOULD BE PAROLED' TO 'WHEN SHOULD SOMEONE BE PAROLED.' STUDIES HAVE SHOWN THAT THE LONGER A PERSON REMAINS IN PRISON, THE MORE EMBITTERED HE BECOMES AND THE MORE SERIOUS THE RISKS PRESENTED ON FINAL RELEASE. PAROLE COULD ALSO BE OFFERED TO MORE DUBIOUS PAROLE RISKS WITH INTENSIFIED SUPERVISION AS A COMPENSATING FACTOR. STUDIES MIGHT ALSO ISOLATE VARIOUS TYPES OF CRIMINALS WITHIN CERTAIN BROAD CATEGORIES. FOR EXAMPLE, ARSONISTS AS A CATEGORY HAVE HIGH RECIDIVISM RATES. HOWEVER, A STUDY OF 138 ARSONISTS CLASSIFIED THEM INTO SIX TYPES (REVENGE FIRESETTERS, EXCITEMENT FIRESETTERS, INSTITUTIONALIZED FIRESETTERS, INSURANCE-CLAIM FIRESETTERS, VANDALISM FIRESETTERS, AND CRIMINAL VINDICATION FIRESETTERS). EACH SPECIFIC TYPE REFLECTED CONTRASTING LEVELS AND PATTERNS OF PAROLE BEHAVIOR. EXPANDED PERSPECTIVE IS ESSENTIAL IF PAROLE IS TO BE SUCCESSFUL AS A PROCESS OF TRANSITION FROM THE STATIC PRISON ENVIRONMENT TO THE DYNAMISM OF THE CONVENTIONAL SOCIAL SYSTEM. RESEARCH MUST HAVE A STRONGER THEORETICAL BASE IF IT IS TO HELP ACHIEVE THIS EXPANDED PERSPECTIVE.