Minnesota DOC Releases Study on Impact of Prison-Based Sex Offender Treatment
In March 2010 the Minnesota Department of Corrections (DOC) released a report on the impact of in-prison sex offender treatment programs on recidivism rates. The results of the study “suggest that prison-based treatment in Minnesota produces a significant, albeit modest, reduction in sex offender recidivism.”
The report opens by noting that previous studies, which tended to show either no positive effect or a very small reduction in recidivism when comparing sex offenders who participated in in-prison treatment programs with those who did not, suffered from a lack of methodological rigor. The research weaknesses in those studies included a lack of random assignment or matching techniques so that nonequivalent comparison groups were used (in 84% of previous studies), and, most importantly, insufficient sample size (in 87% of earlier studies).
The March 2010 report examines recidivism rates through December 31, 2006 of sex offenders released from Minnesota state prisons between 1993 and 2003. After removing releasees who refused treatment and a matching group from the study, 1,020 releasees who had not been offered treatment were matched with a corresponding group of releasees who received in-prison treatment, for a sample size of 2,040.
To match individual treated vs. untreated releasees, the study took into account variables based upon the releasee’s race, age, criminal history, length of imprisonment, prison disciplinary record, type of release, year of release, community notification requirements and relationship with the victim, as well as the victim’s age and gender. Untreated releasees were paired with the closest possible match in the treated releasee group. Only matched releasees were used for the study.
Dividing the releasees into four groups – treatment completers, dropouts, participants (completed or not) and non-participants (never offered treatment) – the study reported a 7.1% rate of rearrest for a sex offense within three years for completers, compared with 11.6% for non-participants, 8.1% for participants and 10.6% for dropouts.
For the entire 14-year length of the study, sex offense rearrest rates were 13.4% for completers, 19.5% for non-participants, 14.2% for participants and 16.2% for dropouts.
“Controlling for other factors, prison-based treatment significantly reduced the hazard ratio [risk of recidivism] for a new sex offense rearrest, decreasing it by 27 percent,” the study found.
The research showed a 13.4% rate of rearrest for a violent offense within three years for completers, compared with 19.3% for non-participants, 14.4% for participants and 16.9% for dropouts. For the entire study period, violent rearrest rates were 29% for completers, 34.1% for non-participants, 30.8% for participants and 35.1% for dropouts. “The hazard ratio for a violent rearrest was 18 percent lower for treated sex offenders in the risk score model and 19 percent lower in the individual predictor model.”
The study also reported a 29.1% general rearrest rate within three years for completers compared with 38.5% for non-participants, 30.3% for participants and 33.1% for dropouts. Over the entire study period, general rearrest rates were 55.4% for completers, 58.1% for non-participants, 56.6% for participants and 59.3% for dropouts. “Participating in treatment had a statistically significant effect on general recidivism, reducing the hazard ratio for rearrest for any offense by 12 percent,” the report concluded.
The study admitted to a number of weaknesses. It did not use a randomized experimental design. Lack of data prevented controlling for the impact of post-release treatment. It could not remove from the untreated study group those sex offenders who would have refused treatment had it been offered to them. Further, an uncontrolled, unaccounted-for bias with a gamma value of as little as 1.02 would be sufficient to undermine the reported differences in recidivism rates.
That is the equivalent of the length-of-incarceration variable gamma value, the smallest gamma value among the controlled variables. Thus, an uncontrolled variable such as education level, employment history, economic class, ability to maintain a normal sexual relationship such as marriage, or community-based treatment could possibly invalidate the study’s findings.
Another factor not taken into account in the study was the changing nature of Minnesota’s prison-based sex offender treatment program. What had been four separate programs has, since 2000, been integrated into a single program that includes a substance abuse component. The different releasees went through different treatment programs depending on when and where they received in-prison treatment. The study also did not appear to account for the civil commitment of sex offenders who have completed their terms of imprisonment and who are then confined in a purported “mental health facility” which is similar to a prison. The study also did not note whether the refusal of sex offenders to participate in prison treatment programs is correlated to the fact that to participate they must admit guilt, and that any statements made in treatment programs can later be used to civilly commit them for the rest of their lives.
Despite its flaws, the study had one very significant finding: that the prediction of recidivism being used by the DOC to determine which sex offenders should be offered treatment was ineffective. The study also found that “[g]iven the state of research and practice in the field of sex offender treatment, the Sex Offender Treatment Program continues to be a work in progress.”
Source: “The Impact of Prison-Based Treatment on Sex Offender Recidivism: Evidence From Minnesota,” Minnesota DOC (March 2010), available at www.doc.state.mn.us.