Actuarial risk assessment of sexual offenders: The psychometric properties of the Sex Offender Risk Appraisal Guide (SORAG).

The Sex Offender Risk Appraisal Guide (SORAG) is one of the most commonly used actuarial risk assessment instruments for sexual offenders. The aims of the present field study were to examine the predictive validity of the German version of the SORAG and its individual items for different offender subgroups and recidivism criteria in sexual offenders released from the Austrian Prison System (N = 1,104; average follow-up period M = 6.48 years) within a prospective-longitudinal research design. For the prediction of violent recidivism the German version of the SORAG yielded an effect size of AUC = .74 (p < .001, 95% CI = .70-.78). 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. However, the differences were significant only for general recidivism (z = 2.48, p = .001). 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. (PsycINFO Database Record

Progress in violence risk assessment and communication: hypothesis versus evidence.

We draw a distinction between hypothesis and evidence with respect to the assessment and communication of the risk of violent recidivism. We suggest that some authorities in the field have proposed quite valid and reasonable hypotheses with respect to several issues. Among these are the following: that accuracy will be improved by the adjustment or moderation of numerical scores based on clinical opinions about rare risk factors or other considerations pertaining to the applicability to the case at hand; that there is something fundamentally distinct about protective factors so that they are not merely the obverse of risk factors, such that optimal accuracy cannot be achieved without consideration of such protective factors; and that assessment of dynamic factors is required for optimal accuracy and furthermore interventions aimed at such dynamic factors can be expected to cause reductions in violence risk. We suggest here that, while these are generally reasonable hypotheses, they have been inappropriately presented to practitioners as empirically supported facts, and that practitioners' assessment and communication about violence risk run beyond that supported by the available evidence as a result. We further suggest that this represents harm, especially in impeding scientific progress. Nothing here justifies stasis or simply surrendering to authoritarian custody with somatic treatment. Theoretically motivated and clearly articulated assessment and intervention should be provided for offenders, but in a manner that moves the field more firmly from hypotheses to evidence.

Bayes and base rates: what is an informative prior for actuarial violence risk assessment?

Bayes' theorem describes an axiomatic relationship among marginal and conditional proportions within a single "experiment." In many ways, it has been fruitful to greatly extend this idea to the task of drawing inferences from data much more generally. Commonly, what matters is how all prior knowledge is revised (or not) by new findings resulting in posterior (sometimes "subjective") probabilities. And, to address many important problems, it is sensible to conceive of probability in such subjective terms. However, some commentators in the domain of violence risk assessment have assumed an analogous axiomatic relationship among marginals (i.e., priors in the form of base rates) observed in one study and conditionals (i.e., posteriors in the form of revised rates) expected in a separate study or assessment context. We present examples from our own research to suggest this assumption is generally unwarranted and ultimately an unaddressed empirical matter.

Weapon use increases the severity of domestic violence but neither weapon use nor firearm access increases the risk or severity of recidivism.

Use of weapons is a risk factor for domestic violence severity, especially lethality. It is not clear, however, whether access to firearms itself increases assault severity, or whether it is characteristic of a subgroup of offenders who are more likely to commit severe and repeated domestic assault. This reanalysis of 1,421 police reports of domestic violence by men found that 6% used a weapon during the assault and 8% had access to firearms. We expected that firearm use would be rare compared to other weapons and that actual weapon use rather than firearm access would increase the severity of domestic assaults. Firearm access was associated with assault severity, but this was mostly attributable to use of nonfirearm weapons. Weapon use was associated with older age, lower education, and relationship history as well as to assault severity. Victims were most concerned about future assaults following threats and actual injuries. Although firearm access and weapon use were related to actuarial risk of domestic violence recidivism, neither predicted the occurrence or severity of recidivism. We conclude that, consistent with previous research in the United States and Canada, firearm use in domestic violence is uncommon even among offenders with known firearm access. Weapon use is characteristic of a subgroup of offenders who commit more severe domestic violence, and seizure of weapons may be an effective intervention.

How nonrecidivism affects predictive accuracy: evidence from a cross-validation of the Ontario Domestic Assault Risk Assessment (ODARA).

Prediction effect sizes such as ROC area are important for demonstrating a risk assessment's generalizability and utility. How a study defines recidivism might affect predictive accuracy. Nonrecidivism is problematic when predicting specialized violence (e.g., domestic violence). The present study cross-validates the ability of the Ontario Domestic Assault Risk Assessment (ODARA) to distinguish subsequent recidivists and nonrecidivists among 391 new cases with less extensive criminal records than previous cross-validation samples, base rate=27%, ROC area=.67. Excluding ambiguous nonrecidivists increases the base rate to 33%, ROC area=.74. Random samples of 50 recidivists and 50 unambiguous nonrecidivists yield ROC areas from .71 to .80. Published norms significantly underestimate official recidivism. Ambiguous nonrecidivism is prevalent and leads to underestimating base rates and predictive accuracy.

Actuarial assessment of violence risk in hospital-based partner assault clinics.

Hospital-based partner assault clinics are a relatively recent addition to the community response to partner violence. In this study, 66% of 111 women attending hospital clinics for partner assault were physically injured and 43% reported death threats. Few concurrently used other services (shelters or police) and most relied on female friends and relatives for help. Many participants who currently lived with the perpetrator were contemplating leaving but only a third had made plans to do so. Participants faced an unusually high risk of future assault, according to both victim interview using the ODARA actuarial risk assessment and their own perceptions. Findings imply an important role for partner assault clinics and the feasibility of the victim service sector's using the same actuarial risk assessments as the criminal justice system.

An indepth actuarial assessment for wife assault recidivism: the Domestic Violence Risk Appraisal Guide.

An actuarial tool, the Ontario Domestic Assault Risk Assessment (ODARA), predicts recidivism using only variables readily obtained by frontline police officers. Correctional settings permit more comprehensive assessments. In a subset of ODARA construction and cross-validation cases, 303 men with a police record for wife assault and a correctional system file, the VRAG, SARA, Danger Assessment, and DVSI also predicted recidivism, but the Hare Psychopathy Checklist (PCL-R) best improved prediction of recidivism, occurrence, frequency, severity, injury, and charges. In 346 new cases, ODARA and PCL-R independently predicted recidivism. An algorithm was derived for a combined instrument, the Domestic Violence Risk Appraisal Guide (DVRAG), and an experience table is presented (N=649). Results indicated the importance of antisociality in wife assault.

A brief actuarial assessment for the prediction of wife assault recidivism: the Ontario domestic assault risk assessment.

An actuarial assessment to predict male-to-female marital violence was constructed from a pool of potential predictors in a sample of 589 offenders identified in police records and followed up for an average of almost 5 years. Archival information in several domains (offender characteristics, domestic violence history, nondomestic criminal history, relationship characteristics, victim characteristics, index offense) and recidivism were subjected to setwise and stepwise logistic regression. The resulting 13-item scale, the Ontario Domestic Assault Risk Assessment (ODARA), showed a large effect size in predicting new assaults against legal or common-law wives or ex-wives (Cohen's d = 1.1, relative operating characteristic area =.77) and was associated with number and severity of new assaults and time until recidivism. Cross-validation and comparisons with other instruments are also reported.

Applying a forensic actuarial assessment (the Violence Risk Appraisal Guide) to nonforensic patients.

The actuarial Violence Risk Appraisal Guide (VRAG) was developed for male offenders where it has shown excellent replicability in many new forensic samples using officially recorded outcomes. Clinicians also make decisions, however, about the risk of interpersonal violence posed by nonforensic psychiatric patients of both sexes. Could an actuarial risk assessment developed for male forensic populations be used for a broader clientele? We modified the VRAG to permit evaluation using data from the MacArthur Violence Risk Assessment Study that included nonforensic male and female patients and primarily self-reported violence. The modified VRAG yielded a large effect size in the prediction of dichotomous postdischarge severe violence over 20 and 50 weeks. Accuracy of VRAG predictions was unrelated to sex. The results provide evidence about the robustness of comprehensive actuarial risk assessments and the generality of the personal factors that underlie violent behavior.

Actuarial assessment of risk among sex offenders.

The appraisal of risk among sex offenders has seen recent advances through the advent of actuarial assessments. Statistics derived from Relative Operating Characteristics (ROCs) permit the comparison of predictive accuracies achieved by different instruments even among samples that exhibit different base rates of recidivism. Such statistics cannot, however, solve problems introduced when items from actuarial tools are omitted, when reliability is low, or when there is high between-subject variability in the duration of the follow-up. We present empirical evidence suggesting that when comprehensive actuarial tools (VRAG and SORAG) are scored with high reliability, without missing items, and when samples of offenders have fixed and equal opportunity for recidivism, predictive accuracies are maximized near ROC areas of 0.90. Although the term "dynamic" has not been consistently defined, such accuracies leave little room for further improvement in long-term prediction by dynamic risk factors. We address the mistaken idea that long-term, static risk levels have little relevance for clinical intervention with sex offenders. We conclude that highly accurate prediction of violent criminal recidivism can be achieved by means of highly reliable and thorough scoring of comprehensive multi-item actuarial tools using historical items (at least until potent therapies are identified). The role of current moods, attitudes, insights, and physiological states in causing contemporaneous behavior notwithstanding, accurate prediction about which sex offenders will commit at least one subsequent violent offense can be accomplished using complete information about past conduct.