ABSTRACT
These tools generally involve prediction of reoccurrence of the target behavior (sexual re-offense or violent recidivism) within a specified time frame (e.g., five years, seven years). Some of these tools contain solely static variables (i.e., those which are fixed such as number of prior offenses) and others identify both static and dynamic variables (i.e., those which are subject to intervention, such as participation in treatment), which are designed to assist clinicians in designing treatment or management plans. Current actuarial tools are based upon archival review of the files of violent and sex offenders, mostly from Canadian and British samples (Hanson 1997; Quinsey et al. 1998; Hanson and Thornton 1999). Variables commonly used to predict general criminal recidivism include having a history of juvenile and adult criminal behavior, being young, being unmarried, having
criminal associates, and demonstrating antisocial attitudes and values, and psychopathy (Quinsey et al. 1998). For sexual recidivism, an additional variable includes having a history of prior sex offenses. The follow-up period refers to time in the community after release from custody and prior to re-offense. The actuarial tools generally use a five-to ten-year follow-up study period. These tools refer to the most recent offense as the ‘index’ offense, and subsequent offenses as prior offenses. A statistical procedure is used to determine which combination of the variables best predicts violence or sex offender recidivism. Survival analysis as a statistical method is also employed to calculate risk of re-offense at each year postrelease. In this instance, ‘survival’ refers to the number of individuals remaining in the community (i.e., not arrested and/or incarcerated) at each year post-release. There is not a general consensus about calculating the weights of individual variables on the actuarial tools. The percentage of individuals in each rating level (e.g., 1, 2, 3, etc.) who have committed acts of sexual or general violence are calculated and used to assign a label of low, moderate or high risk. High risk may be defined as over 50 per cent of the individuals in that rating number who have re-offended violently or sexually when released to the community. The tools are under constant revision informed by research, but the general methodology as described above remains the same, and there are others under development (e.g., MNSOST-R; Epperson et al. 1998).
