Friendship and colleagues (2003) applied a retrospective quasi-experimental design, given that analysis was conducted after the Enhanced Thinking Skills (ETS) intervention took place, using administrative data. The program was implemented in a prison setting, and study participants were selected if they met the following criteria: adult males, who had been released for at least 2 years following a sentence of at least 2 years. Results from the similar Reasoning and Rehabilitation (R&R) program, which led to the development of ETS, were analyzed together. Treatment group participants must have voluntarily participated in either R&R or ETS between 1994 and 1996; however, program completion was not an eligibility requirement for inclusion in the study. A total of 667 made up the treatment group.
Control group participants met the same criteria, but had never participated in either program. A total of 1,801 made up the control group. Study participants were selected using the Inmate Information System (IIS), an internal database used by the prison service. A systematic matching method was applied using the following variables: current offense, sentence length, age at discharge, age at sentence, year of discharge, number of previous sentences, and probability of reconviction score measured using the Offender Group Reconviction Scale (OGRS).
There were significant differences between the treatment and control groups on all matching variables, which is a limitation of the non-randomized experimental design.
Logistic regression was used to determine if there was a relationship between the variables in the model and reconviction among the sample. A forward-selection stepwise method was applied.
Sadlier and colleagues (2010) used a retrospective quasi-experimental design to examine ETS. This evaluation differed from the previous 2003 study by Friendship and colleagues by including dynamic risk factor variables and other rich static risk factors into the model to improve matching between study groups. One-year reconviction outcomes for the treatment group were compared with those in a matched comparison group who did not participate in ETS. Pairs were matched across 42 variables using radius matching. There were no significant differences between treatment and control groups.
Data was collected from four administrative data sets. The first set included the Surveying Prisoner Crime Reduction (SPCR) prisoner survey sample, which measures dynamic risk factors such as attitudes about offending, drug use, and motivation. The second set came from the Offending Behavior Programme Interventions Database, which measured when the offender participated in ETS. The third set, from Offender Assessment System (OASys), provided data from a needs and static risk assessment that included information such as age, gender, marital status, and family criminal history. Lastly, the Police National Computer (PNC) measured criminal history and reconviction data. The final sample consisted of 2,771 individuals, 87 percent of which were men and 13 percent of which were women. Of this total sample, 257 prisoners were treated through ETS participation; this left 2,514 individuals from which to select matches via propensity score matching.
Travers and colleagues (2013) used an observational evaluation to compare reconviction outcomes of 17,047 male prisoners serving a minimum sentence of 1 year who had participated in ETS with those of a 19,792-participant control group who had not participated in ETS. The program sample was 79.2 percent white, 13.3 percent black, 5.1 percent Asian, and 0.4 percent other. The mean age at release for the program sample was 30 years. The control group was 80.9 percent white, 11.3 percent black, 5.6 percent Asian, and 0.6 percent other. The mean age at release for the control group was 29 years.
Measurements were taken 2 years following release. Reconviction data was collected via the Home Office Police National Computer (HOPNC). A risk of reconviction score, within 2 years of release, was calculated using the Offender Group Reconviction Scale (OGRS). Chi-square tests were used to determine differences in reconviction rates among different risk groups.