The researchers utilized a quasi-experimental design with a matched comparison group to evaluate the impact of the ComALERT program, a program for parolees in Brooklyn, New York, that provides substance abuse treatment, transitional employment, transitional housing, and referrals for other services as needed. Both the treatment and comparison groups consisted of parolees released between October 1, 2004, and October 1, 2006…. (more)

The researchers utilized a quasi-experimental design with a matched comparison group to evaluate the impact of the ComALERT program, a program for parolees in Brooklyn, New York, that provides substance abuse treatment, transitional employment, transitional housing, and referrals for other services as needed. Both the treatment and comparison groups consisted of parolees released between October 1, 2004, and October 1, 2006. The comparison group was developed using propensity score methods to match individuals to treatment group participants on several characteristics, including criminal history, race, age, and gender.

Recidivism was measured as re-arrest, re-conviction, and re-incarceration (defined as any return to state prison), each of which was tracked at 6, 12, 18, and 24 months after release. The researchers created a proportional hazards Weibull model incorporating data from each of these four time periods to assess differences between groups in the likelihood of recidivism over the full 2-year follow-up period. The model controlled for race, gender, age, and criminal history. In addition to recidivism outcomes, the authors conducted a survey of study participants enrolled between 2004 and 2007 that provided information on employment. The survey had a low response rate, with 128 out of 448 responding in the treatment group and 113 out of 448 responding in the comparison group (or slightly more than 25% in each group).
(less)

Despite the use of a comparison group that was matched to the treatment group on a number of variables, the groups may differ on other, unobserved characteristics that may contribute to differences in recidivism rates. In particular, an unspecified number of parolees who were mandated to participate in the ComALERT program did not attend, and these individuals were excluded from the study. To be included in the treatment group, parolees … (more)

Despite the use of a comparison group that was matched to the treatment group on a number of variables, the groups may differ on other, unobserved characteristics that may contribute to differences in recidivism rates. In particular, an unspecified number of parolees who were mandated to participate in the ComALERT program did not attend, and these individuals were excluded from the study. To be included in the treatment group, parolees must not only have been assigned to ComALERT but also must have attended an orientation and completed at least an initial psychosocial assessment. Thus, the treatment group may have consisted of a higher proportion of individuals who were motivated to participate in programming than the comparison group because of the removal of these potentially less-motivated individuals from the treatment group. However, all of those who completed at least the orientation and psychosocial assessment were included in the treatment group for analysis purposes, regardless of whether they later dropped out of the program.

In terms of employment outcomes, the low response rates to the follow-up survey limit the generalizability of the findings, given that those who responded to the survey may have had employment and earnings outcomes that were different from those who did not respond. However, because the response rate was similar for both the treatment and comparison groups, the findings from the survey still likely reflect an actual difference between the groups, at least for those individuals who were able to be contacted and who chose to respond to the survey.
(less)

Based on a Weibull proportional hazards model incorporating data collected at 6, 12, 18, and 24 months after release, ComALERT participants exhibited significantly lower rates of re-arrest and re-conviction than the comparison group over the 2-year follow-up period, but no significant differences were found with respect to re-incarceration rates. Additionally, the researchers found that participation in ComALERT was associated with increased … (more)

Based on a Weibull proportional hazards model incorporating data collected at 6, 12, 18, and 24 months after release, ComALERT participants exhibited significantly lower rates of re-arrest and re-conviction than the comparison group over the 2-year follow-up period, but no significant differences were found with respect to re-incarceration rates. Additionally, the researchers found that participation in ComALERT was associated with increased levels of employment and earnings, though these findings are likely related to the fact that transitional work was provided as a part of the program, so these outcomes may speak more to the implementation of the program than to its effectiveness independent of the services it provides.

• The treatment group was 26% less likely to be re-arrested than the comparison group, a significant difference (p<.01).
• The treatment group was 25.8% less likely to be re-convicted than the comparison group, also a significant difference (p<.05).
• No significant difference was found between groups with respect to re-incarceration rates (ComALERT participants’ likelihood of re-incarceration over the follow-up period was 2.2% lower than that of the comparison group, a nonsignificant difference).
• Significantly more ComALERT participants reported being employed over the study period than comparison group members—74.6% compared to 13.6% (p<.05). These results include both transitional jobs that were provided as part of the ComALERT program and independently obtained employment.
• Employed ComALERT participants had significantly higher average weekly earnings than employed comparison group members—$390.31 compared to $266.14, a difference of $76.75 (p<.05). These results include both earnings from transitional jobs that were provided as part of the ComALERT program and earnings from independently obtained employment.
  (less)