Developing a Conceptual Model for
Risk Management of Sex Offenders

July 6 juillet 16:00 – 17:45
Room NCDH–200

Chair:    Howard E. Barbaree , Centre for Addiction and Mental Health, Toronto

Actuarial assessment of risk for reoffense: A comparison of five current instruments

Michael C. Seto , Calvin M. Langton , Howard E. Barbaree

Centre for Addiction and Mental Health, Toronto

Research has identified a number of static variables that predict risk for recidivism among sex offenders. Reflecting this work, a number of actuarial risk instruments have been introduced in recent years for use with adult sex offenders. Five well-known instruments are the Violence Risk Appraisal Guide (VRAG), Sex Offender Risk Appraisal Guide (SORAG), Rapid Risk Assessment for Sexual Offense Recidivism (RRASOR), Static-99, and the MnSOST-R. The Static-99 combines the four RRASOR items with additional items from Thornton’s Structured Anchored Clinical Judgement, while the SORAG is a modification of the VRAG, to be used specifically with sex offenders (Hanson & Thornton, 1999; Quinsey, Harris, Rice, & Cormier, 1998). This study compared the accuracy of these five risk instruments in predicting violent and sexual recidivism in a sample of 224 sex offenders assessed at the Warkworth Sexual Behaviour Clinic, a treatment program located within a Canadian medium security penitentiary, and subsequently released on parole or at the end of their sentences. The results are discussed in terms of selecting risk assessment instruments and using actuarial predictions in risk management decisions.

Sex offender risk assessment: Integration of treatment performance with actuarial risk

Edward J. Peacock

Correctional Service of Canada

Calvin M. Langton , Howard E. Barbaree

Centre for Addiction and Mental Health, Toronto

Sex offender treatment programs are intended to promote cognitive and behavioral changes that result in decreased recidivism. Consequently, it is often assumed that participants who perform well in such programs have made important changes that decrease their risk for reoffense. This presentation examines the issue of whether information about treatment performance contributes to the prediction of recidivism beyond that provided by actuarial risk scores. Findings are reported for a sample of 153 male sex offenders who participated in treatment while incarcerated, were subsequently released and then followed in the community. Treatment performance, as measured by clinician ratings and treatment-related variables such as attendance, and actuarial risk scores are examined as predictors of recidivism. The findings indicate that integrating treatment performance with actuarial scores did not improve the prediction of recidivism. The implications for clinical practice and subsequent research will be discussed.

Clarifying the role of moderating variables in risk assessment

Calvin M. Langton , Howard E. Barbaree

Centre for Addiction and Mental Health, Toronto

Despite the growing body of research that supports the predictive validity of actuarial assessments of risk for reoffense with sex offenders, concerns remain about the clinical utility of actuarial tools. One primary objection to their use is their reliance on static (historical) factors and the general absence of dynamic factors in their composition, which renders the assessment insensitive to interventions and affords the clinician little guidance about how best to manage an offender in the short term. The provision of treatment and community supervision practices are two dynamic factors that would be expected to lower the probability that an offender will recidivate. An important question for risk assessors is how best to integrate information pertaining to these two factors into an assessment of risk, and this is the subject of the final presentation in this symposium. However, an important first step in addressing the question includes clarification of the conceptual differences between static and dynamic factors. In this presentation, the role of moderating and mediating variables in research is considered in reference to the static-dynamic factor distinction. Theoretical and clinical implications are then discussed.

A matrix model of risk classification and management

Howard E. Barbaree , Calvin M. Langton

Centre for Addiction and Mental Health, Toronto

The presentations in this symposium illustrate two important points: (1) Accurate prediction of reoffense amongst sex offenders is possible using actuarial assessment, and (2) Questions remain about how best to integrate information from interventions in assessments of risk. Some authors have proposed that actuarial assessment establishes a benchmark of static risk which can then be modified according to treatment-related information. As an alternative, we are proposing a matrix model in which actuarial assessment is used to classify offenders into risk categories. Risk classification is then combined with risk management strategy in a two-dimensional matrix. A different set of rules for risk management are applied according to risk category: In low risk categories, management involves less intensive monitoring and supervision; at higher levels of risk, more intensive supervision or more restrictive management options are applied. In this system of risk management, offenders are not moved from one risk category to another based on treatment-related information. Once their risk category is established through actuarial assessment, they remain in that category. However, within each category, offenders who complete treatment and are compliant with other requirements can progress towards community release or other objectives. Once an offender is released to the community, the intensity by which he is supervised is determined by his risk classification.


Return to Schedule | Return to Montreal 2001 Main Page | Return to IALMH Main Page