The Influence of Correctional-Based Programs on Mental Health Outcomes and Recidivism for Women Inmates: A Meta-Analysis

Background: The percentage of women prisoners increased from four to nine percent since 1988. Women are incarcerated at a rate 1.5 times that of men. Women incarcerated for drug offenses increased more than 5,000 percent since 1983. Despite these increases, and research indicating that women's needs differ from men's, little is known about the effectiveness of correctional-based interventions for women inmates. This study uses meta-analytic techniques to examine correctional-based treatments and their associations with mental health and recidivism outcomes in an effort to suggest best practices and appropriate policies.

Methods: Two researchers searched databases for studies that tested any intervention for female inmates. We included studies with experimental, quasi-experimental, and one-group pretest-posttest research designs with sufficient information to calculate effect sizes. If information was missing, the researchers attempted to contact the author(s) before excluding the study. The researchers developed coding sheets and two independent coders had inter-rater reliability of 93 percent. Differences were resolved through discussion and agreement.

Hedges g effect sizes were calculated for mental health outcomes and odds ratio effect sizes for recidivism outcomes. Overall effect size estimates were calculated for outcomes grouped by type (mental health, recidivism). Effect sizes for one-group pretest-posttest studies were pooled separately.

Results: Twenty-six articles were collected and twelve were included. Effect sizes for the independent groups design studies ranged from 0 to 1.13 for mental health outcomes, and odds ratio effect sizes from .34 to 1.29 for recidivism outcomes. There were three studies containing one-group pretest-posttest designs, all with mental health outcomes, with effect sizes ranging from .27 to .57.

The mental health random effects model for independent groups design studies shows a weighted mean effect size of .46 (p<.05; CI .09 to .84). Between-study variance is significant (Q=21.63, p<.05) indicating heterogeneity among the studies' effect sizes.

For one-group pretest-posttest studies, the mental health random effects model shows a weighted mean effect size of .30 (p<.05; CI .19 to.42) There was heterogeneity among the studies' effect sizes (Q=1.55; p=.46).

Random effects model for recidivism shows a weighted mean odds ratio effect size of .66 (p=.28; CI .31 to 1.41). However, offenders in treatment groups were 34 percent less likely to recidivate.

Implications: Findings suggest that correctional-based treatment is associated with reduction in mental health problems for women prisoners. Results were not as encouraging for programs aimed at lowering recidivism, but recidivism rates for women in the treatment groups were lower than women in comparison groups. Future research should examine individual and contextual characteristics, timing, and type of intervention.

Most states are cutting correctional programs due to economic hardships. However, these programs appear to be helpful with mental health problems associated with re-incarceration. Re-incarceration is significantly more expensive than the costs of providing correctional-based mental health treatment.

Correctional-based treatment does not seem sufficient in decreasing recidivism for women included in the studies. Perhaps social workers working with released women should provide and advocate for mental health services, which coupled with correctional-based treatment, may reduce recidivism rates.