Study 1
Bonta and colleagues (2011) conducted a randomized controlled trial (RCT) using probation officers (POs) from three Canadian provinces (British Columbia, Prince Edward Island, and Saskatchewan). POs were actively supervising adults on probation (18 years and older) and were randomly assigned to either the Strategic Training Initiative in Community Supervision (STICS) training (the treatment group) or to the no-training condition (the control group). Of the original 80 officers, this study included 52 officers (treatment group n = 33; control group n = 19).
Of the 52 participating POs, 67.5 percent were women, 77.4 percent were white, 8.1 were Aboriginal, and 14.5 percent were from a variety of ethnic backgrounds. The average age of officers was 38.9 years. There was no statistically significant difference between treatment and control groups in baseline demographic characteristics. Twenty-eight officers (18 from the treatment group; 10 from the control group) did not submit any posttraining data (i.e., failed to submit at least one audiotaped session with the accompanying research documents on them assigned to their caseload after group assignment) and were excluded from the analysis. Attrition rates between the treatment (35.3 percent) and control groups (34.3 percent) were not statistically significant.
Following training, all 52 POs recruited two medium- and four high-risk clients as new cases. These individuals were assigned to community supervision (i.e., prospective clients). Prospective clients (n = 143) agreed to have their community supervision sessions audiotaped at three points [at the beginning of supervision after the risk-need assessment was completed (Tape 1), at 3 months (Tape 2), and 6 months into supervision (Tape 3)]. There were 295 posttraining audiotapes, comprising 220 treatment sessions (98 in Tape 1; 71 in Tape 2; 51 in Tape 3) and 75 control sessions (42 in Tape 1; 22 in Tape 2; 11 in Tape 3).
Of the 143 prospective clients, 100 clients were supervised by STICS treatment officers, and 43 were supervised by control officers. Treatment clients were 83.0 percent male, 71.0 percent white 28.0 percent Aboriginal, and 1.0 percent were from a minority ethnic background. The average age of treatment clients was 35.3 years. Control clients were 93.0 percent male, 67.4 percent white, 23.3 percent Aboriginal, and 9.3 percent were from a minority ethnic background. The average age of control clients was 32.6 years.
Of the recruited and submitted individuals, the researchers took a random sample of four medium- or high-risk persons, under the supervision of the probation officer, 1 year prior to entering the project and assigned them as retrospective clients (n = 185). There were 120 retrospective clients supervised by 29 of the 33 treatment officers and 65 retrospective clients by 17 of the 19 control officers. Some officers were not able to submit clients, as they were either not on staff or in their present position 1 year prior to entering the project. No adjustments were made.
Outcomes of interest for this CrimeSolutions review focused on PO behavior that included discussion of any and all criminogenic needs and effective correctional skills at posttraining, and recidivism rates at the fixed 2-year follow up. Effective officer skills and discussion of criminogenic and non-criminogenic needs were measured through skill-items with an intraclass correlation coefficient (ICC) of 0.60 or greater. Skills were defined using a 24-item assessment, rated on a 7-point scale ranging from 1 (very poor) to 7 (very good) on four constructs (structuring skills, 8 items; relationship skills, 5 items; behavioral techniques, 7 items; and cognitive techniques, 4 items) of retrospective clients. A score of 0 was given if the skill did not take place at all over 5-minute intervals. Recidivism was measured through reconviction data collected from the Royal Canadian Mounted Police’s Criminal History Records.
Chi-square analysis was used to evaluate differences in officer behavior from pre- to posttraining (6-month time point) of treatment POs, compared with control POs, and for between-group comparisons of retrospective and prospective clients for recidivism rates. Cox regression analysis was used to address potential selection bias and to control for varying follow-up times and two key client risk characteristics for recidivism: 1) scores on the Criminal History subcomponent of the Level of Service Inventory–Revised; and 2) age. The revised Criminal History component (10 items) was used to create uniformity across the three provinces, as each province had a different risk-need assessment instrument; however, the classification instruments all measured the seven criminogenic needs. The study authors did not conduct subgroup analyses.
Study 2
Bonta and colleagues (2019) conducted an RCT using the same protocol and training manuals (as closely as possible) as those in the earlier evaluation (Study 1) by Bonta and colleagues (2011). There were, however, two important differences that this study addressed: 1) POs were randomly assigned to the experimental or control conditions, but their clients were not; and 2) the coding of POs intervention skills was set to a higher standard by only including items that met a minimum interrater reliability. This study also introduced procedures to better ensure the random assignment of people on probation.
In this study, POs volunteered from four offices in the Canadian province of Alberta and were asked if they would volunteer (n = 60). Of the 60 POs, 36 volunteered for the study. Prior to group assignment, 30 POs submitted a total of four pretraining audio recordings (one recording on each of the four different clients who were assessed as medium or high risk). Following pretraining data submission, POs were randomly assigned to either the STICS training (treatment) or control (no training with probation as usual) condition. Of the 36 officers, 27 provided audiotapes and were included in the study analysis. There were 15 officers in the STICS (treatment) group and 12 in the control group.
Treatment POs received 3.5-days of STICS training and clinical support that included monthly meetings, participation in individual clinical feedback from submitted audiotapes, and a 2-day refresher course. The control POs had 1-hr monthly meetings to discuss their progress in recruiting clients and complete audio recordings and other research documents; however, supervision of people on probation was as usual. Audiotapes were conducted at three points over 6 months (at start of supervision, at 3 months, and at 6 months). Treatment POs submitted 160 recordings on 81 clients and control POs submitted 123 recordings on 42 clients. There were no statistically significant differences between the treatment and control groups in the number of recordings submitted or the average number of recordings per PO.
Clients (i.e., people on probation) were assigned to POs by the office manager/delegate who referred a new admission to the first PO on the roster that was able to take another client based on their caseload. The intake of clients was exclusively dependent upon probation dispositions from the courts (which provided a better representation of the probation population for the jurisdiction). Over the course of 18 months of data collection, this procedure was repeated for each new client until each officer included in the study had two medium- and four high-risk clients with a minimum 12-month probation order (this was a goal set by the study).
Of the 81 STICS treatment people on probation, 87.3 percent were male, 51.3 percent were white, 41.0 percent were Aboriginal, and 7.7 percent were other. The average age of treatment clients was 34.4 years. Of the 42 control people on probation, 82.9 percent were male, 61.0 percent were white, 29.3 percent were Aboriginal, and 9.8 percent were other. The average age of control persons was 32.6 percent. Approximately 95 percent of the people on probation included in the study were categorized as medium or high risk by the Service Planning Instrument (SPIn), There were no statistically significant differences between the STICS people on probation and the control group.
Outcomes of interest included 2-year recidivism rates for medium- and high-risk persons, and POs’ effective correctional skills and discussion of any and all criminogenic needs at the posttraining session (sessions taped at the 6-month time point). The Service Planning Instrument (SPIn), measured clients’ risk, needs, and protective factors. This was a full version, 90-item risk/needs assessment instrument used by the province that included 11 content areas (e.g., mental health, attitudes, substance abuse, aggression/violence). Officers who conducted the assessments were provided with information on the risk level of the client (low, medium, high), areas of concern, and case planning targets. Recidivism was defined as a new conviction within 2 years of the client’s date of first recording. Recidivism data were collected from the criminal records obtained by the Royal Canadian Mounted Police. The interclass correlation coefficient (ICC) at 0.60 (the minimum level) was used at the item level rather than the skill area level to code (the presence or absence) of discussion of criminogenic and non-criminogenic needs and effective correctional skills. This study used chi-square and cox regression analyses to compare the difference between the trained POs and control POs. The study authors did not conduct subgroup analyses.
