An Equine-Facilitated Prison Based Program: Effects On Recidivism
Methods: We used administrative data collected over three years to compare recidivism rates of 206 adult men inmates who participated in an equine-facilitated vocational program with a comparison group of 216 adult men inmates who participated in other vocational programs. Recidivism was defined as an inmate’s return to custody of the State’s Department of Corrections. We hypothesized that after completion of the program participants will have lower recidivism rates than the comparison group. We used discrete event history analysis to assess recidivism risk, as well as propensity score matching to control for possible selection effects. The analysis included variables of recidivism, group membership (program/ comparison), race, repeated recidivism, duration of observation per event, age at release, and sentence length (years)..
Results: The racial distribution among sample participants was: 76.5% White; 23% Black; .4% Hispanic; and .2% other. The median time from release that recidivism occurred was 5.2 months (SD= 16.1), the mean age at release was 35 years (SD= 10.29), and the median sentence length was 10.5 years (SD= 12.06). The recidivism analysis consists of 524 person-time points, with no missing values. The program group consisted 294 (56%) person-time points, and the comparison group consisted 230 (44%) person-time points. 284 (54%) were events of primary recidivism, and 240 (46%) were events of repeated recidivism. 362 (69 %) were of no recidivism among: comparison (96, 18%), during program and pre-program events (266, 51%); 162 (31%) were of recidivism: post-program (28, 5%) and comparison (134, 26%). Propensity score matching results indicated that the model correctly classified 76% of the cases. Wald statistics indicated that group membership significantly predicts recidivism, with the odds ratio (the effect size) of about .07(p<.001), holding all other predictors constant. In conclusion, controlling for all independent variables, program group participants, after completion of the program, have a statistically significant decreased risk to recidivate.
Implications: Based on desistance theory and previous research, these finding support the possibility that reduced recidivism risk is related to role transformation in prisoners that resulted from becoming caretakers to the horses (e.g. from ‘victim to survivor to helper’). Future research should explore such underlying emotional processes of change in participant’s state. Findings help to refine and justify continuation of current programs. Furthermore, they can contribute to broader criminal justice issues of retributive versus rehabilitative policies and prison management practices. This study provides both a corrective to existing knowledge and a basis for developing policies and solid interventions for rehabilitating prisoners. Limitations concern the narrow definition of recidivism.