Criminology Research Council grant ; (6/00-01)
This study assessed the productive utility of static (unchangeable factors) and dynamic (both criminogenic and non-criminogenic needs) risk factors to predict violent and sexual offending behaviour using retrospective data of 1,259 Indigenous males from Western Australia. The utility was examined using three independent samples and this culminated in a highly accurate model for sexual offenders (3-Predictor model). A model that was accurate in predicting non-sexual violent behaviour could not be developed using the available data.
The risk items in the 3-Predictor Model for sexual offenders were unrealistic long-term goals, unfeasible release plans and poor coping skills. The predictive accuracy of recidivism (sensitivity) was 92.3 per cent, while the predictive accuracy of desisting (specificity) was 94.3 per cent. This model outperformed several other risk instruments and achieved comparable results when applied to a sample of 96 non-Indigenous sexual offenders. The finding that the 3-Predictor Model was also accurate in predicting sexual reoffending for non-Indigenous offenders was unexpected, as was the finding that all three predictors in the 3-Predictor Model were dynamic. The prominence of dynamic predictors demonstrates that intervention is likely to make a difference and has implications for both correctional and community services concerned with decreasing the likelihood of reoffending behaviour. Moreover, it appears important to proceed with the development and refinement of a risk of sexual offending tool for male Indigenous offenders given the relative accuracy of the 3-Predictor Model viz a viz the other instruments it was compared with.