Cross-Validation of the Discrimination and Calibration Properties of the VRAG-R in a Treated Sexual Offender Sample

Abstract

The present study featured an examination of the predictive properties of the Violence Risk Appraisal Guide–Revised (VRAG-R) in a treated sample of sexual offenders, using modern risk metrics. The Sex Offender Risk Appraisal Guide (SORAG) and the original Violence Risk Appraisal Guide (VRAG) were examined for comparison purposes. The three measures were rated archivally on 296 treated sexual offenders followed up 17.6 years. VRAG-R scores demonstrated good discrimination of recidivists from nonrecidivists for sexual (area under the curve [AUC] = .60-.67) and violent (AUC = .70-.78) recidivism, and were incremental in the prediction of violent, but not sexual, recidivism after controlling for baseline sexual violence risk and treatment change. The VRAG-R bin structure demonstrated good calibration, although the present sample generated lower 5-year estimates of general violence compared with the normative sample. Application of the VRAG-R in the assessment and management of violence risk, via integration with dynamic risk assessment information, is discussed.