Study 1
To determine the impact of the new protocols on parolee recidivism, Meredith and Prevost (2009) used a stratified sampling design, whereby parole field offices were selected for participation. New parolees who entered supervision were assigned to either the experimental or comparison group. The treatment group was supervised using the new protocols, while the comparison group was supervised using the old protocols, respectively. Field offices were selected based on three criteria: 1) the ability to supply sufficient cases for evaluation, 2) the ability to provide both required treatment programming options to parolees assigned to treatment, and 3) their geographic locations. Twelve offices were selected as treatment offices, which represented urban, rural, and suburban settings in the given geographic regions of the state. All parolees in the 12 treatment offices who were placed on parole beginning March 1, 2007, and who also met the new protocol criteria, were supervised under the new practice protocols (described in the Program Description section). The comparison group included all parolees beginning supervision during the same time period in the remaining eligible offices who also met the new protocol criteria. Parolees who transferred in or out of the offices were eliminated from the study.
A total of 629 parolees were eligible for inclusion in the treatment group, and a total of 1,335 parolees were eligible for inclusion in the comparison group. Parolees in both groups were followed for up to 12 months, or until their parole supervision terminated. All interactions in the Parole Field Log of Interaction Data were monitored daily for identification of patterns that were included in the new supervision protocol. A monitoring system tracked email transmissions, receipt, and acknowledgement to ensure the protocol system was being implemented as intended.
The treatment and comparison groups were similar in terms of demographics, offense type, and prior criminal history. The treatment group was 89 percent male and 11 percent female; the comparison group was 91 percent male and 9 percent female. Both the treatment and comparison groups were 30 percent white and 70 percent nonwhite. The offense type for both groups was similar in terms of violent personal offenses, nonviolent personal offenses, property offenses, drug offenses, and other offenses. Additionally, the percentage of arrests prior to parole supervision and prior prison incarcerations was the same for both the treatment and comparison groups. However, it is important to note that the two groups differed on parole status; the treatment group was more likely to have their parole revoked than the comparison group.
Data was analyzed at 6- and 12-month time points to determine if there were any significant differences between the treatment and comparison groups that could be due to the protocol changes. Statistical significance was assessed with t tests and the Pearson Chi-square test for bivariate analyses, and through linear and nonlinear regression-based techniques for multivariate analyses. The outcomes of interest included arrest for new misdemeanor offense, arrest for a new felony offense, revocation of parole, unemployment throughout parole, a positive drug test, and arrest for a technical violation.