Standardized risk assessments are essential to evidence-based responses to criminal behaviour, including sexual offending. Since the 1990s, many actuarial and structured risk assessment instruments have been developed that are now routinely used in sentencing, treatment, and post-release risk management. The majority of these scales, however, were developed using undifferentiated groups, raising questions about their utility within meaningful subgroups, such as persons with low intellectual ability. This study presents meta-analytic findings of the predictive accuracy (discrimination) of risk tools for crime and violence when used with men with low intellectual ability and a history of sexual offending. We also examined age as a recidivism risk predictor. Database searches identified 15 distinct samples (N = 923). Age, as a single variable, showed moderate to large effects for sexual, violent, and general (any) recidivism. Overall, the predictive accuracies of the instruments were moderate and similar to those observed for other samples. Amongst the measures investigated, a measure specifically for persons with low intellectual ability (ARMIDILO-S; Boer et al., 2013 ) demonstrated the highest predictive accuracy. Larger effects were found when recidivism was measured by staff reports than by official records (e.g., charges, convictions). Our findings support the use of existing risk tools with men with low intellectual ability. Researchers should privilege staff reports over official records of recidivism for studies concerned with sexual recidivism.
Hanson et al. (Mon,) studied this question.