Predicting parole decision in Kansas via discriminant analysis

Abstract

Discriminant analysis was used in this case study to assess differences among variables considered to be important in the parole decision-making process in Kansas. A subjective rating scale completed by parole board members and socio-demographic data were examined. The analysis indicated that eight variables in the rating scale, including parole plan and recidivism risk, were important variables in deciding whether or not to parole an inmate. Socio-demographic variables especially significant were age of inmate at hearing, class of felony, and history of alcohol usage. Suggestions are made about how the parole authority in Kansas could refine parole guidelines using the rating scale (as a measure of parole risk) combined with a measure of offense seriousness (utilizing objective data from the computerized information system available in Kansas). Consequent political, ethical, and legal concerns are acknowledged.