Use of Dynamic Risk Instruments to Assess Sexual Violence Risk in a Community-Supervised Sample of Men with Sexual Offense Convictions
The present study examined the predictive validity and psychometric properties of several actuarial risk measures developed to estimate likelihood of sexual recidivism, and one protective factor measure developed to assess protective factors related to desistance from sexual offending. Each of Static-99R (Helmus et al., 2012), Violence Risk Scale - Sexual Offense Version (VRS-SO; Wong et al., 2003-2017), STABLE-2007 (Hanson et al., 2007), Sex Offender Treatment Intervention Progress Scale (SOTIPS; McGrath et al., 2012), and Structured Assessment of Protective Factors for Violence Risk- Sexual Offence version (SAPROF-SO; Willis et al., 2017-2020) was rated based on file information of 200 community-supervised men with sexual offense convictions who were court mandated to receive assessments (and often treatment) at an outpatient forensic clinic in Edmonton, Alberta. Recidivism information was available for 172 men; mean follow-up time was 8.6 years. Nine percent of the sample was charged or convicted of a new sexual offense, 18.5% for any new violent (including sexual) charge or conviction, and 33% for any new charge or conviction. Predictive validity for all tools was obtained with respect to sexual, violent (including sexual), and general recidivism. All measures significantly predicted sexual (AUC = .65- .72) recidivism and dynamic measures were sometimes incremental to static measures in the prediction of sexual recidivism, depending on the pairing of predictors. An exploration of the structural properties of the VRS-SO dynamic items revealed a three-factor solution isomorphic to previous research (Olver et al., 2007; Olver & Eher, 2019). Further discrimination and calibration findings will be discussed, including implications for assessment and treatment of men convicted of sexual offenses.
sexual offending, actuarial risk assessment, recidivism
Doctor of Philosophy (Ph.D.)