Rice and Grant T. The predictive accuracy for general and violent recidivism was slightly higher than for general sexual and sexual hands-on recidivism. The effect sizes were found to be higher for the child molester sample than for rapists. Further analyses exhibited the SORAG to have incremental predictive validity beyond the VRAG and the PCL-R, and to remain the only significant predictor for violent recidivism once all 3 instruments were forced into a combined regression model.
Twelve out of the 14 SORAG items were found to have a significant positive relationship with violent recidivism. The comparison of the relative and absolute risk indices between the Austrian and the Canadian samples showed that the normative data distribution yielded more absolute risk indices or less relative risk indices meaningful differences between the 2 countries.
SORAG, actuarial risk assessment, sexual offender, recidivism, validity, reliability Supplemental materials: In modern forensic psychology, there basically exist two different standardized methodological approaches to risk assessment e.
Fundamental charac- teristics of an actuarial risk assessment instrument ARAI are, first, that all risk factors included are empirically validated and combined into a total score based on an a priori fixed algorithm for combining the individual risk factors. Second, the total score can be linked to different kinds of empirically derived risk communi- cation procedures: The current research project was conducted in accordance with the legal and ethical demands of the Austrian Department of Justice and in accordance with the national Data Protection Act.
The views expressed are those of the authors and not necessarily those of the Austrian Prison System. Martin Rettenberger and Reinhard Eher would like to express their appreciation and admiration of the tremendous scientific contribution made by Marnie E. Harris, who passed away during the publication process of the present manuscript. Our deep condolences to their families, friends, and colleagues. This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
In the last few years a couple of meta-analyses about the predictive accuracy of risk assessment instruments were published basically confirming that standardized and structured risk assessment usually yields moderate to large effect sizes in the prediction of recidivism in different offender samples Singh et al.
The SORAG is conceptualized for sexual offenders to assess violent recidivism risk that includes sexual offenses involving physical contact with the victim sexual hands-on recidivism. First, violent recidi- vism was operationalized by using criminal charges rather than new convictions.
They further argue that even though every official recidivism source inevitably underestimates the true recidivism rate because of unknown numbers of unre- ported and unrecorded offenses, it can be assumed that charges are a more sensitive measure of recidivism than convictions.
One reason for the fact that the majority of actuarial risk assessment instruments measure the risk of sexual rather than general violent recidivism is that sexually violent predator civil commitment statutes in the United States generally require an assessment of the risk of a sexually motivated reoffense Doren, The re- sults indicated that a substantial number of actually sexually mo- tivated violent offenses were not recorded as sexual on police rapsheets.
Another, but not less important, issue is the question of whether individual risk factors included in a particular risk assessment instrument are actually all related to the outcome and whether these associations might be replicated in independent studies with different samples Coid et al.
For example, previous studies reported that the predictive validity of the SORAG was higher for extrafamilial child molesters compared to other sexual offender subgroups, and that the predictive accuracy for intrafamilial child molesters was relatively low in general Bartosh et al. The aim of the present study was to contribute to answering these unresolved research questions concerning the SORAG by examining its predictive validity in a large sample of prison- released sexual offenders within a prospective-longitudinal field study research design by using different statistical analysis meth- ods.
In the next step of our data analyses we investigated the predictive accuracy for different subgroups of sexual offenders child molesters vs.
Then we calculated absolute and relative risk indices and compared the results with the absolute and relative risk indices reported by the Canadian researcher group who had developed the SORAG Quin- sey et al. This document is copyrighted by the American Psychological Association or one of its allied publishers.
All those independent variables were assigned to either one of the following predictor catego- ries: Variables from the predictor pool without a significant bivariate relationship with the outcome were not considered for further analyses.
Also, those variables of highly collinear pairs were removed that had the lower correlation with the outcome. Multiple regression analyses identified 12 VRAG items showing an independent and incre- mental contribution to the prediction of violent recidivism: Because the use of individual item weights led to a small but significant improve- ment in predictive accuracy, every single risk factor was weighted in accordance to its deviation from the general base rate using a method described by Nuffield Until now a number of empirical studies on the psychometric properties of the VRAG have been conducted for current over- views see, e.
The authors slightly mod- ified the VRAG by eliminating two items female index victim and index victim injury and by adding four items prior history of violent offending, prior convictions for sexual offenses, adult female or male child victims, and sexual deviance , leading to the item version of the SORAG.
Table 1 gives an overview of the items, the individual item score ranges, and the bivariate correlations between the items and violent recidivism in the developmental sample Quin- sey et al.
By means of these risk categories, it is possible to infer empirically calculated probabilities of violent including sexual recidivism after 7 and 10 years at risk absolute risk indices. Up to four missing items can be prorated or substituted Quinsey et al.