THE DEVELOPMENT AND VALIDATION OF THE VIOLENCE RISK SCALE: SEXUAL OFFENDER VERSION (VRS:SO) AND ITS RELATIONSHIP TO PSYCHOPATHY AND TREATMENT ATTRITION
The sexual abuse of women and children is a widespread problem. As such, it has become an important priority to advance risk assessment technology to identify high risk offenders to be targeted for treatment. The body of research involves the initial development, revision, and subsequent psychometric evaluation of an instrument designed to appraise risk for sexual offense recidivism - the Violence Risk Scale: Sexual Offender version (VRS:SO). Study 1 describes the initial development of the VRS:SO, which is comprised of Hanson and Thornton's (1999/2000) Static 99 and 19 dynamic risk factors. The instrument was rated on 321 sex offenders treated through the Clearwater Program on the basis of information gathered from their institutional files. Criminal history and recidivism data were gathered from official criminal records. A series of psychometric analyses were conducted. The instrument showed acceptable internal consistency reliability and the results of an exploratory factor analysis on the dynamic component of the instrument yielded an orthogonal three-factor model. Finally, the Static 99, dynamic factor component, total aggregate scale, and derived factors were all found to predict and predict multiple offense related criteria. Study 2 describes the revision of the VRS:SO. This process entailed the replacement of the Static 99 by an 8-item actuarial scale that was constructed on a randomly selected half of the sample, and cross-validated on the remaining half. The sex offender sample was followed an additional two years for an overall average follow-up time of 10.0 years. Psychometric analyses were conducted on the revised instrument. The static factors, dynamic factors, and scale total predicted multiple sexual recidivism criteria, and to a lesser extent, non-sexual violent recidivism. Moreover, ROC analyses conducted over varying follow-up intervals revealed changes in the VRS:SO's predictive accuracy with respect to sexual recidivism as the follow-up time increased. Study 3 investigates the relationship of psychopathy and sexual deviance to recidivism. The PCL-R was rated on a stratified subsample of 113 offenders randomly drawn from the original pool of 321 offenders, and the VRS:SO and recidivism data were incorporated into the analyses. The PCL-R provided a small to moderate prediction of sexual recidivism and a considerably stronger prediction of non-sexual recidivism. The instrument also converged with the VRS:SO in theoretically meaningful ways. Finally, the relationship of psychopathy and sexual deviance to recidivism were examined through survival analyses. Psychopathic sex offenders who were rated high on sexual deviance (as measured by the VRS:SO) evidenced a higher and faster rate of sexual reconviction than non-deviant psychopathic and non-psychopathic sex offenders. Finally, Study 4 examines attrition from sex offender treatment in a sample of 318 offenders from the Clearwater program. Several demographic, psychiatric, and criminal history variables were coded and VRS:SO, PCL-R and recidivism data were incorporated into the analyses. Compared to successful program completers, dropouts had lower education levels, more sporadic employment history, were less likely to have ever been married, spent less time in treatment, and were more likely to receive a diagnosis of Antisocial Personality Disorder or to meet the PCL-R criteria for psychopathy. Although dropout was correlated with scale dimensions of the VRS:SO (indicating that dropouts were higher risk for sexual recidivism), the relationship of the VRS:SO to attrition was diminished considerably after controlling for PCL-R score. Finally, program non-completers were more likely to sexually recidivate (39%) than successful program completers (29%), although this difference was not statistically significant.
Doctor of Philosophy (Ph.D.)
Department of Psychology