Sex Offender Risk Appraisal Guide (SORAG)

The Sex Offender Risk Appraisal Guide (SORAG) is a 14-item actuarial scale designed to predict violent, including hands-on, sexual recidivism among men who have committed at least one previous hands-on sexual offense.

The items on the scale are the following:

  1. Lived with both biological parents until age 16
  2. Elementary school maladjustment
  3. History of alcohol problems
  4. Never been married at time of index offense
  5. Criminal history score for nonviolent offenses
  6. Criminal history score for violent offenses
  7. Number of convictions for previous sexual offenses
  8. History of sexual offenses only against girls below 14 years of age (negatively scored)
  9. Failure on prior conditional release
  10. Age at index offense (negatively scored)
  11. Diagnosis of any personality disorder
  12. Diagnosis of schizophrenia (negatively scored)
  13. Phallometric test results indicating deviant sexual interests
  14. Psychopathy Checklist (Revised) score

Each item is scored and then assigned a weight based on the relationship of that item to violent recidivism in the construction sample; the weights are then summed to obtain a total score. The score yields the percentile rank of the offender as compared with the construction sample and the estimated probability of violent recidivism based on a 7- and a 10-year period of opportunity to re-offend. The items can be scored from complete institutional files or from files and interviews (but note that external corroboration of offender self-report is important).

The initial construction sample used by Rice and Harris in 1997 comprised released child molesters and rapists who had been briefly assessed in a Canadian maximum security psychiatric hospital before transfer to corrections and offenders treated in the psychiatric institution. Outcome data were obtained from a variety of official sources, and criminal charges for violent or hands-on sexual crimes were the primary predicted outcome. Outcome data were collected by individuals blind to offenders’ SORAG scores.

The SORAG is closely related to the Violence Risk Appraisal Guide (VRAG), an instrument designed to predict violent recidivism among serious offenders (not exclusively sex offenders), and it contains many of the same items. Accuracy of predicting recidivism is similar in the two instruments among sex offender samples, but the estimated probability of recidivism is higher for the SORAG (reflecting the fact that sex offender samples have higher violent and sexual recidivism rates than unselected samples).

Follow-up studies of sex offenders using the SORAG have examined variations in sample (correctional releasees, forensic psychiatric hospital releasees, offenders supervised in the community), jurisdiction (Canada, United States, Europe), and variations in average follow-up time. The SORAG has been robust to these variations, and SORAG scores are consistently linearly related to an offender’s probability of violent recidivism. In replication studies, the accuracy of prediction as indexed by the area under the curve of the receiver operating characteristic or, equivalently, the common language effect size is about .75, meaning that 75% of randomly selected violent recidivists have higher SORAG scores than randomly chosen men who are not violent recidivists. Accuracy is degraded when recidivism is measured poorly, when data are missing for SORAG variables, and when the length of the follow-up period varies widely across offenders. In 2003, Harris and colleagues found that, under optimum conditions, an accuracy of about .85 can be achieved. Scores on the SORAG correlate with both the speed and severity of recidivistic offenses.

The SORAG has also been used to estimate the likelihood that an offender will recidivate with a specifically sexual offense, although these offenses are a subset of the violent offenses that the SORAG was designed to predict. In 2006, Rice, Harris, Lang, and Cormier found, using actual behaviors involved in offending, that sexually motivated offenses are frequently labeled as violent (nonsexual) offenses in official police “rap sheet” records: For example, sexually motivated homicides were recorded as homicides. It was argued on the basis of these data that a tabulation of violent offenses corresponds more closely to the number of sexually motivated offenses than a tabulation of officially recorded sexual offenses. The prediction of violent offenses rather than sexual offenses appears on the basis of this work to better capture the intent of legislation dealing with the risk sex offenders present to society.


  1. Harris, G. T., Rice, M. E., Quinsey, V. L., Lalumiere, M. L., Boer, D., & Lang, C. (2003). A multi-site comparison of actuarial risk instruments for sex offenders. Psychological Assessment, 15, 413—125.
  2. Quinsey, V. L., Harris, G. T., Rice, M. E., & Cormier, C. (2006). Violent offenders: Appraising and managing risk (2nd ed.). Washington, DC: American Psychological Association.
  3. Rice, M. E., & Harris, G. T. (1997). Cross validation and extension of the Violence Risk Appraisal Guide for child molesters and rapists. Law and Human Behavior, 21, 231-241.
  4. Rice, M. E., Harris, G. T., Lang, C., & Cormier, C. (2006). Violent sex offenses: How are they best measured from official records? Law and Human Behavior, 30, 525-541.

See also: