Validation of the Risk and Needs Assessment Used in the Classification for Parole and Probation of Virginia\u27s Adult Criminal Offenders

This study analyzes the effectiveness of risk-needs assessment used in the classification of probation and parole for adult criminal offenders in the Commonwealth of Virginia. The use of risk-needs classification instruments dramatically increased in the 1980\u27s and 1990\u27s, but some research has indicated that these instruments do not have universal applicability and must be validated to the population to which they are applied. The research selected a useable cohort (n = 13,011) of probationers/parolees in the Commonwealth of Virginia from the 20, 647 cases opened in 1994 and tracked them for three years, using Virginia Department of Corrections records, to measure recidivism based upon return to the state correctional system (from another crime or a technical violation) or absconsion. A ten percent sample (n = 1,299) was drawn and assessment instruments of individuals in this sample were analyzed. Statistical analysis was conducted on the 1,013 usable responses to validate the major subgroups of gender, race, age, and parole/probation. Additional surveys were conducted to assess inter-rater reliability and the face-validity of the instrument by users. Results indicate that the current risk-needs assessment classifies offenders in Virginia reasonably well. Using actual data from Virginia, a new discriminant function was calculated and validated, and the new function yielded no remarkable improvement over the Wisconsin instrument now in use. Subgroup analysis indicated that there are some appreciable differences in classifying for success or failure. The variables with the most predictive value (consistent using both discriminant stepwise and multiple regression stepwise analysis) were the questions dealing with prior parole/probation revocations, the percentage of time employed during the last twelve months, and drug usage. Face validity surveys indicated that respondents were often not aware of the superiority of actuarial over clinical methods, and the reliability survey supported instrument reliability. Recommendations include automation of data for continued analysis, the use of interval data, future revalidation to verify cutoff scores, and educating field personnel of the instrument\u27s validity and superiority of actuarial over clinical methods