No information on the intervention is provided in the study, other than that it is a work release program in Florida.

Basic

The study population consisted of males released from Florida State prisons. Limited demographic and criminal history characteristics were reported.

This study utilized a quasi-experimental design. The sample (N=1,210) was drawn from a population of prisoners released in Florida between March 1 and May 31, 1978. The treatment group consisted of 690 individuals who participated in a work release program, 458 of whom successfully completed the program and were released from community centers. Although not specified by the researcher, it appears that the comparison group included 520 observations.

No analysis of pre-existing differences between the treatment and comparison groups was presented. Two reentry outcomes, recidivism and employment, were measured. Recidivism was defined as prevalence of re-arrest and was tracked for 24 months after release, while employment was operationalized as successfully obtaining employment (either part-time or full-time) and was measured over the course of 5 quarters. Statistical analyses for both outcomes involved the use of two-step, multiple regression models, which accounted for a number of control variables, including race, age, marital status, and previous arrests.

Self-selection bias is a potential limitation to this study. Because participation in work release is voluntary, the treatment group may have had inherent and unmeasured characteristics that differed from those of the comparison group. It also appears that members of both groups may have participated in other reentry programs, such as vocational training or prison industries, which might affect post-release outcomes.

The quality of intervention implementation was not discussed in this study.

• A chi-square test and multivariate regression model found no evidence that work release significantly impacted the chances for rearrest over the 2-year follow-up period.
• A multivariate regression model found no significant differences in employment success between the treatment and comparison groups over the 5-quarter follow-up period. This model controlled for members of the treatment group who had a job at the time of release because of the intervention. A significant effect was found when members of the treatment group who did not successfully complete work release were excluded from the analysis; however, this analysis is problematic as it may introduce selection bias (given that those who completed the program may differ from those who did not complete the program).