A STUDY WAS CONSTRUCTED TO DETERMINE IF THE RELATIONSHIP OF INDEPENDENT VARIABLES TO THE DEPENDENT VARIABLE OF SENTENCING DIFFERS ACCORDING TO SENTENCING DEFINITION AND STATISTICAL ANALYSIS APPROPRIATE FOR EACH SENTENCING DEFINITION. WHEN SENTENCING IS VIEWED BY TYPE (INCARCERATION, PROBATION, SUSPENSION, ETC.), DISCRIMINANT ANALYSIS WOULD BE THE APPROPRIATE MULTIVARIATE TOOL; IF DEFINED BY LENGTH (AMOUNT IN THE CASE OF FINES), REGRESSION ANALYSIS WOULD BE APPLIED. DATA WERE DRAWN FROM THE PRESENTENCE REPORTS ON FILE IN THE CONNECTICUT DEPARTMENT OF ADULT PROBATION FOR ALL THE CASES FROM THE SUPERIOR COURT'S SENTENCING DOCKET FOR FEBRUARY AND MARCH, 1975, IN HARTFORD, NEW HAVEN, FAIRFIELD, AND NEW LONDON (N=245). THE INDEPENDENT VARIABLES WERE CATEGORIZED UNDER (1) LEGAL (SEVERITY OF OFFENSE, VIOLENCE IN OFFENSE, HISTORY OF VIOLENCE, CRIMINAL RECORD, PROBATION OR PAROLE REVOCATION, AGE AT FIRST COMMITMENT OR CONVICTION); (2) TREATMENT (FAMILY CRIMINAL RECORD, FAMILY STABILITY, MENTAL HEALTH TREATMENT, ALCOHOL TREATMENT, DRUG TREATMENT, HISTORY OF EMPLOYMENT, EMPLOYMENT AT ARREST, EVALUATION OF PSYCHIATRIST); AND (3) ORGANIZATIONAL (PLEA, BAIL, DEFENSE ATTORNEY, VICTIM'S ATTITUDE, PROBATION OFFICER'S RECOMMENDATION). AFTER ALTERNATE DEFINITIONS OF THE SENTENCING DECISIONS WERE ESTABLISHED, SEPARATE ANALYSES OF REGRESSION AND DISCRIMINANT METHODS WERE APPLIED TO THE SAME SAMPLE WITH THE SAME SET OF INDEPENDENT VARIABLES TO SEE IF RESULTS DIFFERED BETWEEN MODELS. WHEN SENTENCE WAS DEFINED BY TYPE (INCARCERATION, PROBATION, ETC.), LEGAL CRITERIA WERE NOT SIGNIFICANTLY RELATED TO THE SENTENCE RENDERED; HOWEVER, WHEN SENTENCE WAS DEFINED BY LENGTH OF SANCTION, LEGAL VARIABLES WERE SIGNIFICANT, NOTABLY THE SEVERITY OF THE OFFENSE. THIS WOULD INDICATE THAT DIFFERENT DEFINITIONS OF SENTENCING CALLING FOR DIFFERING TYPES OF STATISTICAL ANALYSES DO YIELD DIFFERENT RESEARCH RESULTS. TABULAR DATA AND REFERENCES ARE PROVIDED. (RCB)