Home Research Corrections Research summaries The content of sexual offender risk scales

The content of sexual offender risk scales

Research summary
Vol. 8  No. 2
March 2003

Question

How can we improve risk scales for sexual offenders?

Background

The effective management of sexual offenders requires accurate information about their risk for recidivism. High risk sex offenders may serve more time in prison, benefit from more intensive treatments, and require closer community supervision than low risk sexual offenders.

Evaluating recidivism risk is difficult. Decisions-makers often have limited information and few resources to complete their evaluations. As well, the accuracy of unguided professional judgements is only slightly above chance levels. Consequently, evaluators are increasingly relying on structured, actuarial risk scales. These scales combine a small number of static, historical variables (e.g., previous offences) into an overall risk score. These actuarial risk scores are more accurate than unguided professional judgement, but none of the existing scales measure all relevant risk factors. Consequently, evaluators often wonder how to interpret the combined results of an actuarial risk score and other known risk factors.

When external factors overlap with items already in a scale, then these factors should not change the overall risk level indicated by the actuarial score. It is often difficult to know what is being assessed, however, because many of the current scales selected their items on a purely statistical basis.

Method

A review of previous risk scales identified 22 variables that could be score based on easily available information derived from the offender’s age, official criminal history, and victim characteristics. These variables were organized into conceptual categories guided by theory, how much they were correlated with each other, and how much they correlated with recidivism. The data was drawn from 10 different samples of sexual offenders from diverse jurisdictions (Canada, US, England and Wales). The recidivism rate was 15.7% for sexual recidivism and 27.7% for any violent recidivism after an average follow-up period of 7 years (total sample of the 4,596).

Answer

Five content areas each contributed to the prediction of sexual and violent recidivism: a) young age at release; b) persistence of sexual offending (number and rate of prior offences); c) deviant sexual interests (boy victims, multiple pre-pubescent victims); d) relationship to victims (non-related, strangers); and e) general criminality (prior violent and non-violent offences, violation of conditional release). The predictive accuracy of the combined items was similar to that found for other scales focusing on static, historical risk variables.

The identified domains were sufficiently clear that evaluators can anticipate how these subscales overlap with external measures. For example, a diagnosis of antisocial personality disorder would not be an aggravating factor for offenders with a high score on general criminality because such highly criminal offenders would be expected to fit the diagnosis of antisocial personality. There is no clear explanation, however, for why having stranger victims or unrelated victims should be associated with increased risk. An initial hypothesis that high risk offenders should have a wide range of potential victims was not supported; most measures of diverse victim types (e.g., total age range of victims, victims outside and inside the family) contributed little to risk prediction.

Policy implications

1. The most accurate evaluations of recidivism risk consider a variety of risk factors. No single factor is sufficient to determine whether an offender will or will not sexually reoffend.
   
2. It is possible to obtain moderate levels of predictive accuracy using easily obtained information from official criminal histories. Such simple, actuarial scales can be a cost effective option for decision-makers faced with limited information and limited resources.
   
3. In order to improve the effectiveness of interventions, research should shift from pure prediction to understanding the reasons for recidivism among sexual offenders.

Source

Hanson, R. K., & Thornton, D. (2003). Notes on the development of Static-2002. User Report 2003-01. Ottawa: Department of the Solicitor General of Canada.

For further information

R. Karl Hanson, Ph.D.
Public Safety and Emergency Preparedness Canada
340 Laurier Avenue West
Ottawa, Ontario
K1A 0P8
Tel (613) 991-2840
Fax (613) 990-8295
e-mail Karl.hanson@psepc.gc.ca