Predicting re-offending among sexual offenders

Predicting re-offending among sexual offenders PDF Version (10KB)

Research summary
Vol. 9 No. 2
March 2004


Which sexual offenders are most likely to re-offend?


New offences by known sexual offenders invoke considerable public concern. Most sexual offenders are never reconvicted for another sexual offence, but some are much more likely to recidivate than others. Previous research has identified a number of static, historical factors associated with recidivism risk (e.g., prior offences, age). Much less is known about dynamic (potentially changeable) factors – the factors needed for effective treatment and community supervision. As well, experts have disagreed about how to combine risk factors into an overall evaluation. Some experts recommend the actuarial approach in which a mechanical method of combining the risk factors is specified in advance; other evaluators prefer to use their experience and skill to produce unique judgements for each case.


The results of 95 different recidivism studies were summarized. These studies, produced between 1943 and 2003, included more than 31,000 sexual offenders and close to 2,000 recidivism predictions. Two independent raters coded each study. Results were considered valid if consistent findings were observed in at least three different studies.


The sexual offenders most likely to sexually reoffend had deviant sexual interests and antisocial orientations (history of rule violation, lifestyle instability, and antisocial personality). Some of the variables identified in the study have the potential of being useful targets for intervention, such as sexual preoccupations, conflicts in intimate relationships, hostility, and emotional identification with children. For the prediction of violent non-sexual recidivism and general (any) recidivism, the most important factor was antisocial orientation. General psychological problems (e.g., anxiety, depression) and clinical presentation (e.g., denial, motivation for treatment) had little or no relationship with sexual or general recidivism.

Actuarial risk instruments were consistently more accurate than unguided professional opinion for predicting sexual, violent non-sexual and general recidivism. For the prediction of sexual recidivism, there were no significant differences between the commonly used actuarial measures.

Policy implications

  1. Not all sexual offenders should be treated the same. Given the identifiable differences in sexual offenders' recidivism risk, policies applied equally to all sexual offenders will waste resources on low risk offenders while not directing enough attention to high-risk offenders.
  2. Structured, actuarial instruments should be routinely used. Although additional information will be required in many assessment contexts, actuarial instruments have sufficient accuracy that they should be an expected part of sexual offender assessments.
  3. Current risk assessment practises should be reviewed to determine if they are addressing appropriate factors. Evaluators are most likely to make accurate predictions when they focus on risk factors that have been supported by research.


For further information

R. Karl Hanson, Ph.D.
Corrections Research
Public Safety Canada
340 Laurier Avenue West
Ottawa, Ontario K1A 0P8
Phone: 613-991-2840
Fax: 613-990-8295

Date modified: