Do Standard Risk Prediction Instruments Apply to Female Parolees?

Abstract

The research on female parolees has been rather scant compared to the voluminous research on males. Among other reasons, a primary problem has been the difficulty of drawing a sufficient number of female cases for multivariate analysis. This manuscript reports on an analysis of 546 females in a randomly-selected sample of 4,047 parolees. The primary question is whether a classification system created on the dominantly male population will work equally well for females. Logistic regression analyses indicate that the classification variables and the subsequently-developed classification system are suitable for use with females. Alternative variables, likely to affect female parolees, were tested with the classification model and failed to offer better predictive results for females. While jurisdictions using other classification models may want to be cautious, the commonality of prediction variables among various standard instruments suggests that these findings may be generally applicable.