Validity and reliability of the Violence Risk Scale-Sexual Offense version in a community sexual offense outpatient setting.

Abstract

The present study examined the convergent, structural, and predictive properties of Violence Risk Scale-Sexual Offense version (VRS-SO) scores in a sample of 200 men on community supervision for sexual offenses, attending forensic community outpatient services and followed up an average 8.6 years. The VRS-SO and two additional dynamic sexual recidivism risk measures-STABLE 2007 and Sex Offender Treatment Intervention and Progress Scale (SOTIPS)-were coded archivally from clinic files; Static-99R ratings were extracted. Recidivism data were captured from Royal Canadian Mounted Police records. VRS-SO static, dynamic, and total scores demonstrated expected patterns of convergence with total and subscale scores of the risk measures. Moreover, a confirmatory factor analysis of the VRS-SO dynamic item scores demonstrated acceptable model fit for a correlated three-factor solution consistent with prior confirmatory factor analyses. Discrimination analyses demonstrated that VRS-SO dynamic and total scores and STABLE 2007 scores had large prediction effects for 5-year sexual recidivism (area under the curves [AUCs] = .71-.72) while SOTIPS had a medium effect for this outcome (AUC = .67); the measures yielded medium to large effects for nonsexual recidivism. Cox regression survival analyses demonstrated that VRS-SO dynamic, Sexual Deviance factor, and SOTIPS scores each incrementally predicted sexual recidivism controlling for Static-99R or VRS-SO static factor scores. VRS-SO calibration analyses demonstrated that expected or predicted 5-year sexual recidivism rates showed generally close correspondence to the rates predicted or observed in the present community sample. Results support the psychometric properties of the VRS-SO, a sexual violence risk assessment and treatment planning measure, to a community outpatient sample. (PsycInfo Database Record (c) 2024 APA, all rights reserved).