Predicting Recidivism in Sex Offenders Using the SVR-20: The Contribution of Age-at-release

Abstract

Sex offenders (N = 468) were released from custody and recidivism outcome was recorded. The Sexual Violence Risk-20 (SVR-20) was scored for each offender and the relationship between age-at-release and SVR-20 item and total scores was examined. SVR-20 total scores were not correlated with age-at-release (r = .-057). SVR-20 scores were combined with a score representing the age of the offender at their release from custody. On the basis of ROC analysis, predictive accuracy was significantly enhanced when age-atrelease was included in the risk score. We suggest that the SVR-20, and perhaps other similar risk instruments, could be improved by including age-at-release information. We discuss the possibility that the advantage obtained by empirical actuarial instruments may be due in part to their close relation with age-at-release.