Logistic Regression Analysis of Factors That Predict Sex Offender Recidivism (From Handbook of Sex Offender Treatment, P 26-1 - 26-16, 2011, Barbara K. Schwartz, ed. - See NCJ-243091)

• NCJ Number: 243117
• Author(s): Lawrence L. Bench, Ph.D.; Terry Allen, Ph.D.
• Date Published: 2011
• Length: 16 pages
• Sale Source: Civic Research Institute, Inc
• Publication Format: Book Chapter
• Publication Type: Research (Applied/Empirical); Literature Review
• Language: English
• Country: United States of America

Annotation

After a brief review of previous studies on sex offender recidivism, this chapter presents the methodology and findings of a study that addressed some of the methodological problems identified in previous research on sex offender recidivism.

Abstract

The overall conclusion of the study is that although prior criminal activity had an important role in predicting the likelihood of sexual recidivism, the history of parole technical violations was the most important predictor of sex offender recidivism. Four different models were constructed and evaluated for various forms of recidivistic activity for a sample of 387 male sex offenders who were under the supervision of the Utah Department of Corrections at some time between 1979 through 2005. Model 1, measured recidivism that did not involve sex offenses, and model 2 examined recidivism for only sex offenses. For these two models, there were no statistically significant variables that predicted non-sex offenses as a group or sex offenses as a group. The dependent variable for model 3 distinguished offenders who returned to prison for any reason (technical violations and more serious crimes) from those who did not return. Stepwise logistic regression analysis determined that only 3 of the 51 independent variables considered were statistically significant predictors of reincarceration: failure to complete the sex offender treatment program, a history of parole violations, and young age at first arrest. Model 4 identified factors that distinguished those reincarcerated due to a technical violation from those reincarcerated due to a subsequent non-technical conviction. Three of the 51 independent variables were significant predictors of parole violation: failure to complete a sex offender treatment program, the use of digital penetration at the time of the sexual offense, and previous parole violations. The prediction was somewhat better for paroled inmates who returned for technical violations (74.8 percent) than for non-technical returnees (63.6 percent). 8 tables, 2 figures, and 49 references