Identifying Predictors of Recidivism in a Large Sample of United Kingdom Sexual Offenders: A Prognostic Model

This study uses prognostic modeling to identify the best static and dynamic predictors of sexual and violent recidivism for a sample of 3,773 sexual offenders who received treatment in a correctional establishment in the United Kingdom between 1996 and 2006. The use of individual static items was found to predict recidivism better than a modified version of a risk score produced from the static risk assessment Risk Matrix 2000/sexual dimension (RM2000/s) (Thornton et al., 2003). The best predictors of recidivism were age at release, number of sexual appearances, and number of criminal appearances. Pre- and post-psychological measures did not remain in the model in the presence of these three static variables. Further exploratory analyses found that pretreatment scores on measures related to the socioaffective domain were the most predictive of the dynamic risk domains, but did not add to the predictive power of the static variables. An overall score for deviance was calculated and this score did remain in a model with individual static items. The potential explanations for these findings are discussed along with implications for the assessment of risk in this population.