Absence of conflict of interest.
Willison, J. B., Roman, C. G., Wolff, A., Correa, V., & Knight, C. R. (2010). Evaluation of the Ridge House residential program: Final report. Washington, DC: Urban Institute
- The study’s objective was to examine the impact of Ridge House, a faith-based halfway house program, on recidivism outcomes.
- The authors used a nonexperimental design (propensity-score matching) to create a comparison group of nonparticipants who were similar to program participants. The authors estimated the program’s effects by comparing these groups’ post-release recidivism outcomes using administrative data from the National Crime Information Center at the Federal Bureau of Investigation.
- The authors found that the odds of rearrest or reconviction were 75 percent higher for those who participated in the Ridge House program than those in the comparison group.
- The quality of causal evidence presented in this report is moderate because it was based on a well-implemented nonexperimental design. This means we are somewhat confident that the estimated effects are attributable to the Ridge House program, but other factors might also have contributed.
Ridge House program
Features of the Intervention
The Ridge House program is located in Reno, Nevada. It is a 90-day faith-based halfway house program offering substance abuse treatment and employability training to Reno parolees. During the study period, the program operated six facilities with a combined occupancy of 38 beds. The program accepts most formerly incarcerated people except high-level sex offenders or serious violent offenders, but this does not necessarily mean it excludes murderers. Parolees must apply to Ridge House before their scheduled release date. Prison-based caseworkers provide inmates with Ridge House program applications, assist them with the application process, and coordinate with program staff. Ridge House accepts applicants on a first-come, first-served basis. Applicants transferred to Ridge House are assigned schedules, which require them to be out of bed by 5 a.m., depending on the location and distance of their employer. They are scheduled to work from 8 a.m. to 4 p.m. and are to report back to their respective houses by 5 p.m. Residents must take turns preparing dinner for the entire house and performing clean-up duties. All residents are required to return to their home by 6 p.m., when all housemates share dinner together. In addition to the employment, clients must participate in groups and meetings that are held between 6:30 and 8:00 p.m., with weekly house meetings held on Tuesdays. Available classes include parenting, computer literacy, money management, and others.
Features of the Study
The authors used a statistical approach called propensity-score matching to compare post-release recidivism outcomes between the treatment and comparison groups. This resulted in 156 people in the treatment group, however, and just 86 people in the comparison group, which was too small to generate enough power to test for statistically significant differences in recidivism. Therefore, the authors included in the analysis an expanded group of people initially identified as eligible for the research but who were not approached to participate. The expanded comparison group consisted of 375 people who could not initially be reached for study participation or were paroled outside or Reno, which raised the total to 461 comparison group members. Outcomes were observed for about 18 months following the end of recruitment, allowing for a follow-up period of at least 12 months for all people in the sample.
- The odds of rearrest in any given month were 75 percent higher for those who participated in the Ridge House program than those in the comparison group, indicating that the time until rearrest was shorter for Ridge House participants. This finding is statistically significant.
- There were no statistically significant associations between participation in the Ridge House program and the prevalence or number of re-arrests.
Considerations for Interpreting the Findings
Although the people in the original comparison group were paroled in the Reno area as the people in the treatment group were, people in the expanded comparison group could have been paroled outside this geographic area, including to Las Vegas, Nevada, or out of state to California. The expanded comparison group could have differed from the treatment group in other ways as well. The authors addressed these potential differences in the impact analysis by accounting for the demographic characteristics and prior arrests of all sample people as well as for being a member of the expanded comparison group. This well-implemented design leads to a moderate evidence rating.
The authors imputed the date of release from prison for people in the expanded comparison group because this information was not available. This imputation is important to the construction of outcomes measures because it defines the period over which outcomes are observed.
Because the authors did not have outcomes data on conviction or custody, they could not determine whether people were “on the street” or incarcerated at a given point. Thus, if a rearrest was not observed for an individual during a given period, the reason might be that the individual was incarcerated and did not have an opportunity to be rearrested or the individual was in the community and was not rearrested. Conviction data would have helped provide a stronger indicator of recidivism if they had been available.
Causal Evidence Rating
The quality of causal evidence presented in this report is moderate because it was based on a well-implemented nonexperimental design. This means we are somewhat confident that the estimated effects are attributable to the Ridge House program, but other factors might also have contributed.