Evaluation of Correctional Reintegration Programs

Anne Kelly
Commissioner

Date

Bruno Jean
Director General, Strategic Policy and Planning

Date

Authors:^{Footnote 1}

Prince Baffoe, Junior Evaluation Analyst

Nicholas Chadwick, A/Evaluation Officer

Rebecca Cherner, Evaluation Officer

Adam Crawford, A/Evaluation Officer

Sidikat Fashola, Evaluation Officer

Sylvie Gaudreault, A/Director of Evaluation

Aileen Harris, Evaluation Manager

Sara Johnson, A/Director of Evaluation

Patrick Savoie, Evaluation Analyst

Eliza von Baeyer, Senior Evaluator

Kayla Wanamaker, Evaluation Analyst

Evaluation Team Members:

Lauren Crossing, Junior Evaluation Analyst

Karen Koundakjian, Senior Evaluator

David Myers, Evaluation Analyst

Anthony Stock, Evaluation Officer

The evaluation team is grateful for the assistance provided by the Correctional Operations and Programs Sector and Women Offenders Sector. The support and continued collaboration of both the Reintegration Programs Division and Women Offenders Sector was instrumental in the completion of this evaluation. The evaluation team would like to thank all members of the Consultative Working Group, including representatives from CSC's Performance Measurement and Management Reports, Corporate Services, Indigenous Initiatives Directorate, and Research Branch for their contributions throughout the evaluation.

The evaluation team would like to express its appreciation to all those who participated in the collection of data and contributed valuable information, including Wardens, institutional staff members, and regional program managers. We would also like to extend our gratitude to all inmates for sharing their overall experience with correctional programs through their participation in the interviews.

We would also like to thank the several branches and sectors of CSC who helped provide data for this evaluation, including staff in the Research Branch (and in particular, Shanna Farrell-MacDonald for her work in data extraction), Financial Planning and Budgeting, Resource Management Branch, Learning and Development, and HR Corporate Reporting.

Finally, the evaluation team would like to thank everyone else who contributed to this evaluation, whose names do not appear here.

This evaluation addressed the correctional programs delivered by the Correctional Service of Canada (CSC). Specifically, the evaluation focused on the Integrated Correctional Program Model (ICPM) and the Women Offender Correctional Programs (WOCP).

In 2010, CSC began transitioning from delivering a traditional multi-program correctional program model to using an integrated multi-target or holistic program model for both men and women. In contrast to the traditional model, where individual programs focused on specific offence histories (e.g., substance abuse programs, violence prevention programs), the integrated model was designed to target multiple program need areas within the same program. This program also incorporates a harm reduction model to address substance use needs, whereby offenders are encouraged to select their substance use goals, focus on enhanced awareness through goal setting, and incorporate other supportive services and strategies. The main correctional program streams for men include mainstream programming without culture or sex offender-specific components, a stream that offers culture-specific programming designed for Indigenous offenders, a mainstream sex offender (SO) stream, and a SO stream that offers culturally-specific programming for Indigenous offenders. The main correctional program streams for women include a mainstream program, a culturally-specific program for Indigenous offenders, and a SO stream.

The evaluation covered the period from 2013-14 to October 2018, with variations across evaluation questions and data sources. The scope of the evaluation included: 1) the continued relevancy and need for correctional programs; 2) the effectiveness of correctional programs (i.e., program access and delivery as well as the extent to which programs are achieving their expected results); and 3) the efficiency of correctional programs.

Data were collected through literature review, extraction and analysis of administrative data, interviews with offenders, and surveys with staff. The key findings from each section are highlighted below.

There is a continued need for CSC to provide correctional programs to federal offenders. CSC's correctional programs are aligned with priorities, roles, and responsibilities of CSC and the federal government.

Timely Access. CSC does not have a definitive and standardized definition of timely access. Offenders are generally enrolled in a main correctional program before their full parole eligibility date (FPED) and about half are enrolled before their day parole eligibility date (DPED). Women had more timely access to correctional programs than men. Overall, there were no significant differences in enrollment and time to start correctional programs between Indigenous and non-Indigenous offenders.

Time to Complete Programs. Over half of offenders completed a main program by FPED, whereas a quarter completed a main program before DPED. Women completed their correctional programs more quickly than men did. There was no difference in the time to program completion for Indigenous and non-Indigenous offenders.

Engagement and Satisfaction. Many offenders described the main program as engaging. Most offenders were satisfied with the information provided in the programs, however, staff were less satisfied with the program content. Many offenders and half of the staff were satisfied with how the information was communicated.

Most participants of an Indigenous correctional program described the information provided in the program and the way it was communicated as culturally relevant and appropriate; in contrast, about a third of staff who delivered these correctional programs^{Footnote 2} agreed that the information and its communication were culturally relevant and appropriate. Staff suggested adapting the content to increase its relevance.

Program Completion. Most offenders had completed a primer or engagement program and a main program, with few non-completions. According to the data extracted from the Offender Management System (OMS), non-completions were primarily due to reasons unrelated to correctional program participation (including if an offender is deceased, if they cannot participate due to responsivity needs or for outside court or hospital).

Program Alignment with Risk Need Profiles of Offender Population. Men offenders’ risk and need profiles are being correctly identified, and they are generally being assigned to the proper correctional program intensity and stream.^{Footnote 3} When an override is granted, it is most commonly to override an offender to a higher intensity program. Offenders and staff generally agreed that the program addresses offenders’ risk factors.

Offender and Staff-Identified Barriers to Timely Program Completion. According to staff, some barriers to program access include a lack of resources, particularly human resources. In addition, staff and offenders reported that a lack of program availability, delayed program starts, and operational and population management constraints interfered with timely completions.

Institutional Outcomes. The program completers had similar results with respect to non-random urinalysis tests before and after program participation, and there was no clear pattern for random urinalysis test results. Overall, program completers formed the highest percentage of offenders who had no violent, drug, and other charges either before or after their program, in comparison with non-completers and eligible non-participants. Fewer completers had an increase or decrease in charges post programming, although completers generally had fewer charges both prior to and post program completion.

Discretionary Release. Compared to eligible non-participants, men offenders who completed a correctional program were granted discretionary release more often, even when accounting for risk relevant differences between the groups. The increased rates of discretionary release were observed for both Indigenous and non-Indigenous men. For women, when risk relevant differences were considered, program completers were granted a discretionary release more often than eligible non-participants, however, this result was not statistically significant. Although not statistically significant, results indicated that Indigenous women tended to receive discretionary release less often than non-Indigenous women.

Community Outcomes for Men. Community outcomes during the first release were examined as an index of program effectiveness. Regression analyses were conducted to account for the presence of risk relevant differences between program completers and eligible non-participants. Figure E.1 below summarizes the results for men for each program stream examined across the various community outcomes.^{Footnote 4} Findings generally indicated that program completers had lower rates of any revocation than eligible non-participants. Lower rates of any revocation were observed for both Indigenous and non-Indigenous men program completers. Given that any revocation represents the most common reason for discretionary release to be terminated, these findings are promising. For the remaining community outcomes, program effectiveness was mixed. Although not statistically significant, the results generally supported program effectiveness for revocation with offence and revocation with a violent offence. In contrast, results suggested that eligible non-participants had lower rates of the substance use outcome (i.e., either a positive urinalysis or a suspension due to a breach of a substance use related supervision condition) than program completers, although this finding was not statistically significant.

A chart depicting the likelihood of experiencing various community outcomes for men correctional program completers, relative to eligible men non-participants, across correctional program streams (all programs, ICPM-MT-moderate, ICPM-MT-high, AICPM-moderate, ICPM-SO-moderate).

All programs:

A chart depicting the likelihood of experiencing various community outcomes for women correctional program completers, relative to eligible women non-participants, for all programs, WOMIP, and AWOMIP.

All Programs:

The expenditures required to deliver correctional programming, as well as staffing and training resources, were examined to assess the efficiency of correctional programs.

Correctional Programing Expenditures. Table E.1 presents the direct costs of program delivery per participant and completer for ICPM (which includes AICPM) and WOCP (which includes AWOCP). The cost per participant for ICPM programs was approximately 14% higher than the cost per participant for WOCP. Additionally, WOCP had a higher rate of completion than ICPM, which led to a greater discrepancy in the costs per completer ($7,331 per ICPM completer vs. $5,576 per WOCP completer).

The lack of availability of precise financial data limited the cost-effectiveness analysis to the overall programming level (i.e., collapsed across all programming streams). Further, a no-cost comparison group of eligible non-participants was required for the cost-effectiveness analysis, which prevented analyses for women since all women are referred to engagement (readiness) sessions, and therefore have associated programming costs. Overall, ICPM was found to be cost-effective according to an examination of the direct costs associated with program delivery and first-release outcomes for program participants and eligible non-participants. For every offender who received programming, there was an approximate savings of $5,675 in avoided readmission costs, compared to eligible non-participants. The conclusion for the cost-effectiveness of correctional programming, however, was based on data obtained during the evaluation and may not generalize to all correctional program streams and intensity levels.

Staffing and Training Resources. A combination of quantitative and qualitative (i.e., interviews with staff) data generally indicated that there is a sufficient number of trained staff to deliver correctional programming. However, there is potentially an opportunity to increase the number of Aboriginal Correctional Program Officers (ACPOs), as evidenced by staff interviews and a higher vacancy rate among ACPO positions. Reliable quantitative data was not available for Elder involvement in the delivery of correctional programming, underscoring the need to enhance information gathering on this issue. Nearly all Correctional Program Officers (CPOs) and ACPOs were considered to have met their training requirements, and required training appeared to be offered in a timely manner. However, interviews with staff suggested that the content of the training could be enhanced to better equip CPOs/ACPOs with the knowledge required to deliver programs.

Based on these findings, recommendations and management action plans (MAPs) were put forward to address the key issues identified, such as adopting a standardized definition of timely access to programs, increasing the relevance of the program content and delivery for the Indigenous streams for men and women, conducting additional research to understand the effect of correctional programs on community outcomes, reviewing the impact of the newly implemented program referral criteria on the number of overrides for women, improving the availability of data related to correctional program expenditures, and reviewing the training protocol for CPOs.

There is a continued need for CSC to provide correctional programs to federal offenders.

CSC's correctional programs align with CSC's and the federal government's priorities, roles, and responsibilities. The delivery of effective correctional programs contributes to the overall priority of a just, peaceful, and safe society

CSC does not have a definitive and standardized definition of timely access. Staff provided varied definitions that defined timely access in relation to parole eligibility dates, in consideration of an offender's level of need and sentence length, or access to programs as early as possible in an offender's sentence.

Based on the indicators used in this evaluation, most offenders enrolled in a main program before their full parole eligibility date (FPED) and about half are enrolled before their day parole eligibility date (DPED). Women offenders had more timely access to programs than men offenders as they were enrolled in and began their programs more quickly. The hybrid programs were associated with quicker access. Overall, there were no significant differences in enrollment and time to start programs between Indigenous and non-Indigenous offenders.

According to staff, some barriers to program access included a lack of resources, particularly human resources, and insufficient program availability. Many of the 20 Indigenous offenders who were interviewed and had wanted to participate in an Indigenous program reported that they had not taken an AICPM or AWOCP as the programs were unavailable or not offered in a timely manner.

Over half of offenders completed a main program by FPED, whereas a quarter completed a main program before DPED. Women completed their programs more quickly than men, and hybrid programs were completed more rapidly than other men's moderate programs. There was no difference in the time to program completion for Indigenous and non-Indigenous offenders for men and women offenders combined.

Staff and offenders reported that a lack of program availability and delayed program starts interfered with timely completions of programs, as did operational and population management constraints. Staff also described offender-related factors and lack of resources as barriers to timely program completion.

Many offenders described the main program as engaging. Most offenders were satisfied with the information provided in the programs, however, staff were less satisfied with the program content. Many offenders and half of the staff were satisfied with how the information was communicated. Suggested improvements included changes to the content, such as a) increasing its relevance to offenders, and b) reducing repetition, simplifying the material, and reviewing it for errors.

Most AICPM and AWOCP participants described the information provided in the program and the way it was communicated as culturally relevant and appropriate. A third of staff who delivered these programs agreed that the information and its communication were culturally relevant and appropriate to a large/very large extent and around 40% agreed to a moderate extent. Staff suggested adapting the content to increase its relevance to the cultural background of the participants.

Most offenders had completed a primer or engagement program and a main program, with few non-completions. According to the data extracted from OMS, non-completions were primarily due to reasons unrelated to program participation (such as the offender is deceased, cannot participate due to responsivity needs, for outside court or hospital).

Men offenders’ risk and need profiles are being correctly identified, and they are generally being assigned to the proper program intensity and stream. When an override is granted, it is most commonly to override an offender to a higher intensity program. The concordance between program need and program assignment could not be assessed for women offenders, due to the recent implementation of the INCP screen.

Offenders and staff generally agreed that the program addresses offenders’ risk factors. In order to better address offenders’ risk factors, the most common suggestion was to adapt the program content. Overall, the frequency and length of the program were deemed appropriate by offenders, given their assessed level of risk.

The non-random urinalysis test results for the main program completers were generally similar in the 6 months prior to a main program and the 6 months following a main program. There was no clear pattern with the random urinalysis test results.

Most program completers did not have violent, drug, or other charges both before and after a main program. In comparison to the non-completers and the eligible non-participants, program completers did not increase or decrease in violent, drug, or other charges after the main program. Women program completers and women eligible non-participants had similar patterns of changes in violent and other charges, although a higher percentage of women in the eligible non-participant group had no drug charges.

Most of the staff reported that participation in correctional programming was related to decreased incidents in the institution, while about half of offenders thought that it had a positive impact on institutional security. Most offenders agreed that they had applied the skills learned in programs within the institution. Further, qualitative findings indicated that according to staff and offenders, additional skills could be taught to offenders to improve institutional security.

Across all program streams, men program completers were granted discretionary release more often than men eligible non-participants. Although not statistically significant, the results suggested that women program completers were granted discretionary release more often than women eligible non-participants.

Generally, staff and offenders perceived that participation in correctional programs had a positive impact on the ability of offenders to obtain discretionary release.

Overall, men completers, in particular those who participated in the multi-target moderate programs, were revoked for any reason less often than eligible men non-participants. The lower likelihood of any revocation was observed for both Indigenous and non-Indigenous men program completers. Although not statistically significant, results indicated that program completers overall tended to have a revocation with an offence less often than eligible non-participants. In contrast, program completers were revoked more often than men offenders with no-intent-to-treat (regardless of the type of revocation).

The findings related to the effect of program completion on substance use outcome were mixed. Results suggested that program completers more often had a substance use outcome, in comparison to eligible non-participants, although these findings were not statistically significant. Notably, men in the ICPM-MT high intensity program were significantly more likely to have a substance use outcome.

Although not statistically significant, results indicated that women completers of WOMIP and AWOMIP were revoked for any reason more often than eligible non-participants.

Separate models could not be conducted for Indigenous and non-Indigenous women due to sample size. However, the overall models that accounted for Indigenous ancestry indicated that Indigenous women tended to be revoked for any reason more often than non-Indigenous women, although this was not statistically significant.

While the findings suggest that program completers had a substance use outcome more often than eligible non-participants, the results were also not statistically significant.

More than half of the women who completed programming were overridden into the program as they did not initially meet program referral criteria. Override completers had lower rates of any revocation compared to women who initially met program referral criteria, but when risk relevant differences were controlled for, both groups experienced a comparable rate of revocations for any reason.

Offenders and staff generally perceived that correctional programs provided and effectively taught the correct tools and skills needed for reintegration. While most offenders indicated that they learned important skills necessary for reintegration, approximately half indicated that they anticipated challenges when applying these skills, with the most common concern referring to applying the skills in a different environment than which they learned (e.g., from the institution to the community). Nonetheless, most offenders and many staff agreed that programs will have a positive impact on an offender's reintegration.

Overall, for men who were identified as having a program need for general violence, program completers were revoked for any reason less often than eligible non-participants.

A similar finding was obtained for men offenders with a program need for substance use, whereby program completers were revoked for any reason less often than eligible non-participants. Conversely, eligible non-participants had a substance use outcome while in the community less often than program completers, although this finding was not statistically significant. This suggests that, among men with a program need for substance use, correctional programming appears to be effective at reducing revocations, but does not impact the likelihood of a substance use outcome in the same way.

Although not statistically significant, program completers with a program need related to family violence and program completers with a program need in sexual offending had a revocation for any reason less often than eligible non-participants with a program need related to family violence or sexual offending.

Overall, for women identified as having a program need for substance abuse, program completers and eligible non-participants had comparable rates of any revocation and a substance use outcome. The pattern of results remained consistent when comparing Indigenous women with non-Indigenous women.

Staff most commonly agreed that correctional programs sufficiently addressed specific offending behaviours related to general crime, general violence, and sexual offending. However, fewer than 60% agreed that substance use was sufficiently addressed and less than half reported the same regarding family violence.

Several offenders reported a responsivity need that interfered with their ability to participate in a correctional program. Although those with reading and writing barriers often had their needs addressed, fewer than half of offenders with mental health, intellectual or learning disability, anxiety/hesitance (for men only), or a brain injury agreed that they received accommodations, tools, or support to help them participate despite these needs. Staff also reported having access to limited tools to address offenders’ needs. Offenders and staff provided suggestions regarding possible accommodations.

Correctional programming for men (overall, across ICPM/AICPM) was found to be cost-effective according to an examination of the direct costs associated with program delivery and first-release outcomes for program participants and eligible non-participants. For every offender who received programming, there was an approximate savings of $5,675 in avoided readmission costs, compared to eligible non-participants.

Cost-effectiveness could not be examined for women's correctional programming since all women are referred to the engagement program and the current evaluation required a comparison group with no exposure to correctional programming (i.e., a no cost comparison group). However, it was found that the cost per participant for women's correctional programming was lower than the cost per participant for men's correctional programming.

While many staff who were interviewed indicated that there was a sufficient number of trained CPOs given the number of offenders requiring programs, only about a third of staff agreed that the number of ACPOs was sufficient. A comparison of the number of funded positions to active employees identified a vacancy rate for ACPO positions of 11%, suggesting there may be an opportunity to increase the workforce.

Reliable quantitative data was not available for Elder involvement in the delivery of correctional programming, underscoring the need to enhance information gathering on this issue.

Staff and offenders tended to agree that the number of program sessions that require an Elder was appropriate. However, staff did highlight challenges with Elder availability, resulting from a shortage of Elders or Elders having limited time for a given program, due to competing requests for involvement.

Many staff who received correctional program facilitator training agreed that it was provided in a timely manner. However, only about half of program managers and program facilitators strongly agreed or agreed that the content of the ICPM/WOCP training provided CPOs and ACPOs with the knowledge required to deliver programs.

Further, only a quarter of staff who had received training and delivered a program, or worked as a program manager, described the quality review process as effective/very effective in ensuring that programs are delivered appropriately.

It is recommended that efforts continue to be supported to enhance the INCP screen and amend policy to make the completion of the INCP screen mandatory.

It is recommended that clearly articulated guidelines for defining timely access to correctional programs with respect to program enrollment and completion dates be established and added to the Commissioner's Directives on correctional programs.

It is recommended that RPD:

• Identifies the best practices that allow for timely enrollment and completion of programs delivered by CSC and those offered in other jurisdictions.
• Considers how these can be applied to the men's programs with lengthier wait times and completion times.

It is recommended that ICPM and WOCP content be reviewed, and if required, its content should be simplified and streamlined.

It is recommended that CSC increases the relevance of the Indigenous correctional programming streams (AWOCP/AICPM) to Indigenous offenders through consultation with Indigenous Initiatives Directorate, as well as consideration of feedback from staff and offenders outlined in this evaluation.

It is recommended that CSC conducts research on the relationship between timely access to programs and post-release outcomes for both men and women to determine the optimal timing of program delivery throughout an offender's sentence.

It is recommended that research be conducted in the following areas:

• Conduct a study examining violent and sexual reoffending for men offenders wherein the sample size of offenders who have completed programs is increased and the follow-up period is extended. This study should occur in 5 years to ensure adequate follow-up.
• In the interim, if feasible, examine changes over time in pre and post-program measures related to violent and sexual offending for men offenders to determine if program participation is related to reductions in the likelihood of violent and sexual offending.
• Conduct a replication study of substance use outcomes separately for men and women offenders identified as having a substance use need. Consideration should be given to expanding the substance use outcome to account for changes in the severity of substance use over time, and whether returns to custody or new offences are directly related to substance use.

It is recommended that CSC examines the volume of overrides used to refer women offenders to correctional programs (both AWOCP and WOCP) and the justifications for the overrides. Further, CSC should examine the community outcomes for women offenders who received an override relative to women who initially met program referral criteria, and determine whether modifications to the program referral criteria are warranted.

It is recommended that CSC identifies how correctional program officers address the various responsivity needs of men and women offenders that may interfere with their ability to participate in programs.

It is recommended that RPD reviews the regional recording practices of financial resources associated with delivering correctional programs. The results of the review should inform new strategies, if required, to ensure accurate and consistent recording of resource allocations.

It is recommended that RPD examines the content and format of the training protocol to identify whether there are opportunities to enhance:

• The knowledge and skills of CPOs/ACPOs to assist in effectively delivering correctional programming, possibly through providing additional facilitation and practical training.
• The usefulness of the quality review process, possibly by increasing the timeliness of the review or adapting the method of assessment.

• Signatures
• Contributions
• Acknowledgments
• Executive Summary
• List Of Findings
• List Of Recommendations
• Table Of Contents
• List Of Tables
• List Of Figures
• List Of Acronyms
• 1.0 Introduction
  • 1.1 Overview Of Correctional Programs
    • 1.1.1 Correctional Program Model
    • 1.1.2 Program Referral
    • 1.1.3 Program Logic Model
  • 1.2 Scope Of The Evaluation
  • 1.3 Evaluation Questions
• 2.0 Methodology
  • 2.1 Data Sources
    • 2.1.1 Literature And Document Review
    • 2.1.2 Data Collection For Fifes 2, 3, And 4
    • 2.1.3 Data Collection For Fife 2
    • 2.1.4 Data Collection For Fife 3
    • 2.1.5 Data Collection For Fife 4
  • 2.2 Data Analysis
    • 2.2.1 Data Analysis Of Interview And Survey Data (Fifes 2, 3, And 4)
    • 2.2.2 Data Analysis Of Admission Cohort (Fife 2)
    • 2.2.3 Data Analysis Of Institutional Outcomes (Fife 3)
    • 2.2.4 Data Analysis Of Post-Release Outcomes (Fife 3)
    • 2.2.5 Data Analysis For Cost-Effectiveness (Fife 4)
  • 2.3 Limitations
• 3.0 Findings
  • 3.1 Fife # 1: Relevance Of Correctional Programs
    • 3.1.1 Recommendations: Relevancy
  • 3.2 Fife # 2: Effectiveness Of Correctional Programs: Program Access And Delivery
    • 3.2.1 Literature Review
    • 3.2.2 Timely Access
    • 3.2.3 Timely Completion Of Programs
    • 3.2.4 Engagement And Retention
    • 3.2.5 Program Alignment With Risk Need Profiles Of Offender Population
    • 3.2.6 Recommendations: Access And Delivery
  • 3.3 Fife # 3: Effectiveness Of Correctional Programs: Program Outcomes
    • 3.3.1 Literature Review
    • 3.3.2 Correctional Programs And Institutional Behaviour
    • 3.3.3 Correctional Program Completion And Discretionary Release
    • 3.3.4 Impact Of Programs On The Likelihood Of Revocation
    • 3.3.5 Integrated Model And Specific Offending Behaviours And Substance Abuse
    • 3.3.6 Responsiveness To The Special Needs Of Offending
    • 3.3.7 Summary
    • 3.3.8 Recommendations: Program Outcomes
  • 3.4 Fife # 4: Efficiency Of Correctional Programs
    • 3.4.1 Literature Review
    • 3.4.2 Are Csc's Correctional Reintegration Programs Delivered In Cost-Effective Manner?
    • 3.4.3 Given The Number Of Offenders, Are There Sufficient Staff Trained To Deliver Correctional Programming?
    • 3.4.4 Is There Sufficient, Efficient And Appropriate Training For Cpos/Acpos?
    • 3.4.5 Recommendations: Program Efficiency
• 4.0 Conclusion
• Appendix A: Correctional Program Referral Process
• Appendix B: Offender Population
• Appendix C: Correctional Programming Needs
• Appendix D: Program Categories
• Appendix E: Discretionary Release
• Appendix F: Community Outcome
• Appendix G: Program Need
• Appendix H: Case Study Results
• Appendix I: Cost Effectiveness
• References

• Table E.1. Cost of Correctional Programming for Men and Women: FY 2017/2018
• Table 1. Number of Sessions for Main Programs
• Table 2. Characteristics of Interviewed Offenders and Offenders in Custody in FY2017-2018
• Table 3. Region and Positions of Staff who were Sent Survey and those who Participated
• Table 4. Determination of Program Intensity
• Table 5. Characteristics of Institutional Outcomes Cohort
• Table 6. Characteristics of Men in the Post-Release Outcomes Cohort
• Table 7. Characteristics of Women in the Post-Release Outcomes Cohort: Updated Study Period
• Table 8. Characteristics of Post-Release Outcomes Sample for Cost-Analysis
• Table 9. Program Need Area Criteria
• Table 10. Number of Offenders with an Identified Need for Programming
• Table 11 Percentage of Offenders Who Enrolled In Their Main Program Prior to DPED, by Main Program Type
• Table 12. Offenders who Enrolled in Main Program, the Percentage Prior to FPED, by Main Program Type
• Table 13. Median Number of Days From Admission to Custody to Program Start Date by Program Type
• Table 14. Staff Perceptions of Reasons for Which Program Streams are Inaccessible to Indigenous and Non-Indigenous Offenders
• Table 15. Median Number of Days From Admission to Custody to First Main Program End Date by Program Type
• Table 16. Of Offenders who Completed a Main Program, the Percent who Completed Prior to DPED by Program Type
• Table 17. Of Offenders who Completed a Main Program, the Percent that Completed Prior to FPED by Program Type
• Table 18. Status of Primer and Engagement Program Assignments
• Table 19. Status of Main Program Assignments
• Table 20. Status of Institutional Maintenance Program Assignments
• Table 21. Status of Motivation Module Assignments
• Table 22. Hybrid Program Assignments
• Table 23. Reasons for Program Non-Completion
• Table 24. Reasons for Program Non-Completion Reported by Offenders and Staff who Selected Voluntary or Involuntary Withdrawal, or 'Other' Reasons for Non-Completion
• Table 25. Refused or Positive Non-Random Urinalysis Outcomes 6 Months Prior to and Following Main Program^a
• Table 26. Refused or Positive Random Urinalysis Outcomes 6 Months Prior to and Following Main Program^a
• Table 27. Violent Charge Outcomes 6 Months Prior to and Following Main Program^a
• Table 28. Drug Charge Outcomes 6 Months Prior to and Following Main Program^a
• Table 29. Other Charge Outcomes 6 Months Prior to and Following Main Program
• Table 30. Relationship Between Study Group and Discretionary Release for Men
• Table 31. Relationship Between Study Group and Discretionary Release for Non-Indigenous And Indigenous Men
• Table 32. Relationship Between Study Group and Discretionary Release for Women
• Table 33. Occurrence of Community Outcomes Among Men Eligible for Inclusion in Survival Analysis
• Table 34. Relationship Between Study Group and Any Revocation
• Table 35. Relationship Between Study Group and Any Revocation for Indigenous and Non-Indigenous Men
• Table 36. Relationship Between Study Group and Revocation with Offence
• Table 37. Relationship Between Study Group and Revocation with Offence for Indigenous and Non-Indigenous Men
• Table 38. Relationship Between Study Group and Revocation with Violent Offence
• Table 39. Relationship Between Study Group and Substance Use Outcome
• Table 40. Relationship Between Study Group and Substance Use Outcomes for Indigenous and Non-Indigenous Men
• Table 41. Occurrence of Community Outcomes Among Women Eligible for Inclusion in Survival Analysis
• Table 42. Relationship Between Study Group and Any Revocation: Women
• Table 43. Relationship Between Study Group and Revocation with a New Offence: Women
• Table 44. Relationship Between Study Group and Substance Use Outcome: Women
• Table 45. Relationship Between Treatment Group and Community Outcomes for Men with a Family Violence Program Need
• Table 46. Relationship Between Study Group and Community Outcomes for Men with a General Violence Program Need
• Table 47. Relationship Between Treatment Group and Community Outcomes for Men with a Sexual Offending Program Need
• Table 48. Relationship Between Treatment Group and Community Outcomes for Men with a Substance Abuse Program Need
• Table 49. Cost of Correctional Programming for Men and Women: Fy 2017/2018
• Table 50. Inputs for the Cost Analysis of Men's Correctional Programming
• Table 51. Count of Funded Cpos And ACPOS Positions by Region
• Table 52. Count of Active Institutional and Community Cpos and Acpos by Region
• Table 53. Count of Active Institutional Cpos and Acpos by Region
• Table 54. ICPM/AICPM And WOCP/AWOCP Training Programs Delivered to CPOS and ACPOS
• Table 55. Initial ICPM/AICPM And WOCP/AWOCP Training Programs Delivered to CPSO and ACPOS in 2017-2018 by Region

• Figure 1. Shifting from a Multi-Program Model to an Integrated Model (CSC, Reintegration Programs Division [RPD], 2016)
• Figure 2. Correctional Programs Logic Model
• Figure 3. Program Need Assessment Process
• Figure 4. The Number of Programming Needs at Admission for Men Offenders by Need Area
• Figure 5. Number of Men Offenders Identified for ICPM Participation at Admission by ICPM Stream
• Figure 6. The Percent of Offenders Enrolled in a Main Program Prior to DPED
• Figure 7. Percent of Offenders Enrolled Prior to FPED
• Figure 8. Mean Number of Days to Main Program Start Date
• Figure 9. Mean Days from Admission to Main Program End Date
• Figure 10. Percent of Offenders who Completed Main Program Before DPED
• Figure 11. Percent of Offenders who Completed a Main Program Prior to FPED
• Figure 12. Offenders’ Self-Reported Levels of Engagement
• Figure 13. Staff Perceptions of Offender Engagement
• Figure 14. Offender and Staff Level of Satisfaction with Information in Programs
• Figure 15. Offender and Staff Satisfaction with how Information is Communicated
• Figure 16. Staff and Offender-Reported Reasons for Program Non-Completion
• Figure 17. Program Need Assessment Process
• Figure 18. Staff Perceptions that Correctional Programs Address Specific Dynamic Risk Factors
• Figure 19. Staff Perceptions of Impact of Correctional Programs on Offenders Obtaining Discretionary Release
• Figure 20. Rate of Any Revocation within 1 Year of Release by Override Status and CRI Level for Men
• Figure 21. Rate of Any Revocation within 1 Year of Release by Override Status and CRI Level for Women
• Figure 22. Staff Perceptions of Extent to Which Correctional Programs Address Offending Behaviours/Needs
• Figure 23. Offenders' Self-Reported Responsivity Factors that Interfere with Program Participation and Learning
• Figure 24. Percentage of Offenders' Who Perceive that Accommodations Are Provided for Their Responsivity Needs
• Figure 25. Staff Perceptions of CPOs/ACPOs Having Tools to Accommodate Offender Needs
• Figure 26. Staff Perceptions of Challenges Related to CPO and ACPO Turnover

The Evaluation Division has undertaken an evaluation of Correctional Service of Canada's (CSC) correctional programs.^{Footnote 6} CSC defines a correctional program as "a structured intervention that targets empirically-validated factors directly linked to offenders’ criminal behaviour, in order to reduce reoffending" (CSC, 2018a). Effective correctional programming is an essential component in CSC's ability to fulfill its mission of contributing to public safety by assisting offenders in becoming law-abiding citizens.

The primary objectives of this evaluation are to review the relevancy and need for correctional programs, assess the effectiveness of correctional programs (with a focus on program delivery and program outcomes), and examine overall program efficiencies.

In 2010, CSC began transitioning from delivering a traditional multi-program correctional program model to using an integrated multi-target or holistic program model for both men and women.^{Footnote 7} This transition began with the implementation of Women Offender Correctional Programs (WOCP) and Aboriginal^{Footnote 8} Women Offender Correctional Programs (AWOCP) from 2010 to 2012, as well as the Integrated Correctional Program Model (ICPM) for men offenders from 2010 to 2017. In contrast to the traditional model, where individual programs focused on specific offence histories (e.g., substance abuse programs, violence prevention programs), the integrated model was designed to target multiple program need areas within the same program. To address substance use, ICPM integrates a harm reduction model whereby offenders are encouraged to select their substance use goals, focus on enhanced awareness through goal setting, and incorporate other supportive services and strategies (e.g., Opioid Agonist Treatment, needle exchange program). Through various harm reduction principles, the objective of the model is not necessarily to abstain from substance use, but to reduce the harmful use of substances. Individuals are asked to clearly explore the impact of any use upon their goals during the program sessions, the development and review of the crime process, and while identifying risk factors related to their crime and corresponding personal targets. This integrated model is illustrated in Figure 1 for the men's programming.

An image depicting the six programs that have now transitioned into one Integrated Model.

According to the National Correctional Program Referral Guidelines (CSC, 2018b), correctional program effectiveness requires matching the stream and intensity of the program to an offender's level of risk. During the offender intake assessment process, an offender's level of risk and need areas are determined through a combination of actuarial risk assessments, supplementary assessments as necessary (e.g., a sex offender-specific assessment), and a review of their criminal history. The 2015 version of the guidelines (CSC, 2015a) specified that program intensity was generally based on level of risk, as identified by the Revised Statistical Information on Recidivism Scale^{Footnote 11} (SIR-R1) for non-Indigenous men and the Custody Rating Scale^{Footnote 12} (CRS) for women and Indigenous men, the Dynamic Factor Identification and Analysis, Revised (DFIA-R)^{Footnote 13} for women, plus the Static-99R^{Footnote 14} for SOs. The Stable-2007^{Footnote 15} was added as another measure of risk for SOs in 2017 (CSC, 2017). Since January 2018, program intensity has generally been decided based on the results of the Criminal Risk Index^{Footnote 16} (CRI) for all offenders, in addition to the Static-99R and the Stable-2007 for male SOs (see Appendix A for additional information regarding the assessment tools and procedures for the program referral process).

Based on the results of the actuarial risk assessment tools, moderate risk offenders are referred to moderate intensity programming, and high risk offenders are referred to high intensity programming. Low risk offenders who meet the established override criteria may also be considered for participation in moderate intensity programming. Through participation in programs of appropriate intensity, offenders can address criminogenic needs, prepare for successful reintegration, and ultimately reduce their risk of reoffending (Andrews & Bonta, 2010a; CSC, 2009; Landenberger & Lipsey, 2005; Usher & Stewart, 2014).

A logic model for correctional programs is presented in Figure 2. The logic model is a visual representation of the inputs, key activities, outputs, as well as the immediate, intermediate, long-term, and ultimate outcomes related to the delivery of correctional programs.

The components of a logic model can be defined as follows (Treasury Board of Canada, 2016):

• Inputs: the resources (legislation, financial, human, policies, and protocols) used in a program that ensure the delivery of the intended results of a program;
• Key activities: actions associated with achieving program objectives (e.g., delivering awareness and motivation sessions, introducing concepts of crime process/offence path and risk factors/personal targets, introducing self-management tools and skills);
• Outputs: immediate results of implemented program activities (short-term achievements of the program, deliverables) (e.g., sessions are completed by offenders; detailed crime process is established and personal targets identified [men]; preliminary development of self-management/healing plan [women]);
• Outcome: achievements of program objectives/impact of a program's outputs (e.g., increased knowledge and skills to manage risk factors in the community and for successful reintegration into the community; decrease in institutional security incidents and charges; increase in conditional releases);
• Impact (ultimate outcome): long-term achievements of program objectives (e.g., reduced reoffending and safe and timely reintegration of offenders into the community). It should be noted that impacts at a population-level can seldom be attributed to a single program, however, a specific program may, together with another program, contribute to impacts on a population.

An infographic depicting the logic model of Correctional Programs, highlighting the inputs, key activities, outputs, and various outcomes.

Inputs: Legislation, financial resources, human resources, policies, and protocols.

Continuum of Care divided into three components: Readiness programs, Main programs, and Maintenance programs. Each component is comprised of unique key activities, outputs and short-term and long-term outcomes; these components combine to achieve the ultimate outcome of correctional programs, which is to contribute to reduced re-offending and facilitate the safe and timely reintegration of offenders into the community.

Key Activities

The scope of the evaluation was determined through consultation with key stakeholders. The stakeholders who were approached for consultation included the Correctional Operations and Program Sector, Women Offender Sector, Policy Sector, the Senior Deputy Commissioner's Office, and the Performance Measurement and Evaluation Committee (PMEC). It was determined that the evaluation would focus on 1) the continued relevancy and need for correctional programs, such as alignment with government priorities and consistency with federal roles and responsibilities; 2) the effectiveness of correctional programs (i.e., the extent to which programs are achieving their expected results); and 3) the efficiency^{Footnote 17} of correctional programs, including the ICPM for men offenders and the WOCP model for women. The evaluation focused on the delivery of correctional programs within institutional settings. The IICP was considered out of scope for the evaluation. However, IICP streams were included in some analyses in which they were aggregated with other programs. The WSOP was out of scope and not included in the evaluation, with the exception that five staff questionnaire respondents had delivered the WSOP.

The following questions are addressed in this evaluation:

• Do correctional programs continue to address a demonstrable need within federal corrections?
• How do correctional program objectives align with departmental priorities and federal government priorities?
• Does the delivery of correctional programs align with the roles and responsibilities of CSC and the federal government?

• Are offenders being granted timely access to programs (including Indigenous offenders being granted timely access to culturally-specific programs and programs overall)?
• Are correctional programs engaging and retaining offenders?
• Do programs offered align with the risk and need profiles of CSC's offender population?

• Does participation and/or completion of correctional programs impact institutional behaviour (e.g., institutional incidents)?
• Does participation and/or completion of correctional programs increase the likelihood of obtaining discretionary release?
• Does participation and/or completion of correctional programs impact the likelihood of returning to custody and the likelihood of reoffending?
• Does the integrated model address substance abuse and specific offending behaviours (e.g., family violence)?
• Are programs responsive to the special needs of offenders (e.g., those with mental health care needs, learning disabilities)?

• Are CSC's correctional programs delivered in a cost-effective manner (i.e., cost per offender, cost-benefit analysis)?
• Given the number of offenders, are there sufficient staff trained to deliver correctional programming?

An examination of the peer-reviewed literature and internal documents on correctional programming was conducted (e.g., CSC policies, legislation, evaluation reports, research reports, and other operational documents). These documents were consulted in order to assess current literature on timely access to correctional programs; engagement and retention in programs; alignment of programs with risks and needs of offenders; whether the integrated model addresses specific offending behaviours; meeting the responsivity needs of offenders; and the impact of correctional programs on institutional behaviour, discretionary release, and revocation of release.

Interviews with Offenders. Data from interviews with offenders were included to answer questions regarding program access and delivery (FIFE 2), program outcomes (FIFE 3), and efficiency (FIFE 4). Semi-structured interviews captured the perspectives of offenders regarding the delivery of correctional programs and the perceived impact of program participation. The questionnaire, developed by the evaluation team in collaboration with the Consultative Working Group, addressed evaluation questions and indicators identified in the Terms of Reference for the evaluation. The interview guide included both open and close-ended questions (e.g., 5-point Likert-type scales, categorical multiple choice questions). In accordance with the principles of Gender-Based Analysis Plus,^{Footnote 18} the guides were adapted to include questions that were specific and relevant to gender considerations and Indigenous offenders. Offenders were eligible to be interviewed if they had participated in any component of an ICPM or WOCP program.

In each region, interviews were conducted in one minimum, medium, and maximum men's institution and the regional women's institution, with the exception of Pacific region, where the visit to a men's maximum institution was not possible due to a lockdown.^{Footnote 19} Offenders at one men's healing lodge were interviewed. Within each institution, data extracted from the Offender Management System (OMS) was used to randomly select a list of offenders who had been enrolled in a correctional program, stratified by Indigenous ancestry. In order to maximize participation in the interview process, offenders were informed of the evaluation in advance by institutional staff and staff were provided posters to share with the inmate committee and to display within the institutions. While in the institutions, evaluation staff coordinated with institutional staff identified by the wardens and offenders were approached for interviews in the order their names appeared in the list. Additionally, offenders who were not on the list, but who expressed interest in participating, were interviewed. Interviews occurred in November and December 2017 and were conducted in French and English in Québec and Atlantic regions and in English in Ontario, Prairie, and Pacific regions.

A total of 209 offenders across all regions were interviewed in November and December 2017. Twenty-nine of the interviews (14%) were conducted in French and occurred in the Québec and Atlantic regions. Table 2 presents the characteristics of the interviewed offenders in comparison with those of offenders serving sentences of two years or more in custody across all CSC institutions during the 2017-2018 fiscal year (FY), the period in which data were collected for this evaluation.^{Footnote 20} The proportion of women and Indigenous offenders who were interviewed appeared greater than the proportion of these subpopulations in custody in 2017-18 as they were oversampled intentionally to ensure that there was sufficient representation of women and Indigenous participants. The proportion of offenders in the 31-40 age range was higher in the group of interviewed offenders. The regional representation of interviewed offenders also differed from the 2017-2018 in custody population as the interviews were divided relatively evenly across the five regions to ensure adequate response rates from each region. The representation across security level was also intentionally more evenly distributed among interviewed offenders.

All except three of the offenders (99%, n = 206 of 209) reported having participated in a primer/engagement program during their current sentence. Of those who started a primer/engagement program, all except one had completed it (99.5%, n = 204 of 206) and another offender was unsure. Almost all offenders (97%, n = 202 of 209) had begun participating in their main program during their current sentence. Most of the offenders had completed all of the modules of their main program (82%, n = 164 of 199).^{Footnote 21}

The ICPM program into which offenders were commonly placed was ICPM-MT-Moderate (28%, n = 58 of 209), followed by ICPM-MT-High (16%, n = 33), AICPM-Moderate (14%, n = 29), AICPM-High (8%, n = 17), ICPM-SO-Moderate (7%, n = 15), ICPM-SO-High (1%, n = 2). The WOCP programs into which offenders had been placed were WOMIP (14%, n = 29 of 209), AWOMIP (10%, n = 21), WOHIP (3%, n = 7), and AWOHIP (3%, n = 6). Note that offenders could report participation in multiple programs. Two participants reported ‘other’ programs, including one women who completed a Women's Engagement Program, but was not referred to a main program.

Questionnaires Completed by Staff. Data from staff questionnaires were included to answer questions regarding program access and delivery (FIFE 2), program outcomes (FIFE 3), and efficiency (FIFE 4). An electronic questionnaire was designed to collect the perspectives of institutional and regional staff who were familiar with correctional programs. The questions focused on the delivery of programs, training, and perceived program impacts. It was designed by the Evaluation team in collaboration with the Consultative Working Group. The questionnaire was developed using Snap Survey software and included open and close-ended questions. Respondents were routed to specific questions depending on their position and experience with delivering correctional programs since July 1st, 2017 (a date selected to ensure experience with delivering the current model of programs).

Data extracted from the Human Resources Management System (HRMS) were used to identify staff members with the following job positions working within an institutional setting: Parole Officers, CPOs, ACPOs, CPO Assessors, Program Managers, Assistant Wardens of Intervention, Elders, and Wardens. Additionally, Regional Program Managers, Regional Administrators of Assessment and Interventions, and Regional Administrators of Aboriginal Initiatives were identified. An invitation to complete the questionnaire was distributed by email on February 23rd, 2018 to potential staff participants in English and French. The questionnaire was available online until March 12th, 2018.

Invitations were originally sent to 1,392 employees,^{Footnote 22} although the email invitations to nine people could not be delivered and alternate contact information could not be located. A total of 1,388 staff, including Elders, were sent the invitation by email, and paper copies of the surveys were sent to four Elders who did not use email. The number of surveys sent to staff in each region and type of position are reported in Table 3.

A total of 341 individuals completed the staff questionnaire (of the 1,383 who are assumed to have received the survey), thus 25% of those who were sent information about the evaluation completed the survey. Data from 20 respondents were excluded as the respondents were not currently working within an institution or regional headquarters, therefore the responses for 321 participants were analyzed. Most staff completed the survey in English (82%, n = 264).

Of staff respondents who reported their gender, three-quarters were women (73%, n = 228), a quarter were men (26%, n = 81), and two (1%) endorsed the 'Other' category (e.g., transgender, agender, two-spirit). Of the staff who reported their age, the greatest proportion were 36 to 45 (39%, n = 120) or 46 to 55 years old (33%, n = 101). Others reported that they were 26 to 35 (17%, n = 51), 56 to 65 (11%, n = 34), and 66 years and older (1%, n = 3).

Many staff respondents were working in a men's institution (75%, n = 242). Staff also worked in a men's regional reception centre or intake unit (10%, n = 32), women's institution (9%, n = 28), regional treatment centre (5%, n = 17), regional headquarters (4%, n = 12), men's healing lodge (2%, n = 8), and women's healing lodge (1%, n = 2). Respondents worked primarily at medium security (39%, n = 126) or multi-level (38%, n = 123) institutions, followed by minimum (17%, n = 54) and maximum (12%, n = 40) security institutions (categories were not mutually exclusive). A breakdown of staff respondents by region and position is presented in Table 3. The regional distribution of the survey recipients and survey respondents was similar. However, it appeared that a greater proportion of CPOs and ACPOs completed the survey compared with the proportion that were sent the survey. In contrast, it appeared that a smaller proportion of Parole Officers and Aboriginal Liaison Officers completed the survey compared with the proportion that were sent the survey. The high levels of participation by CPOs and ACPOs and the lower levels of participation by Parole Officers and Aboriginal Liaison Officers might reflect the extent to which staff in these positions perceive correctional programs as relevant to their role.

Half of staff respondents had delivered an ICPM program or WOCP since July 1^st, 2017 (51%, n = 164). July 1^st, 2017 was selected as the cut-off date to identify staff with recent experience with ICPM/WOCP programs and, as the programs were implemented nationally by that date, to ensure that their exposure to correctional programs included the ICPM and WOCP versions. These staff members had a range of experience in delivering ICPM/WOCP programs, as some reported five or more years of experience (35%, n = 55 of 157), and others indicated that they had been involved for 6 months to 2 years (30%, n = 47), 2 to 5 years (24%, n = 38), and less than 6 months (11%, n = 17).

Of those who indicated which streams they had delivered, the most common were ICPM-MT-Moderate (30%, n = 48 of 161) and ICPM-MT-High (25%, n = 40). Other men's program streams that were delivered by staff included AICPM-Moderate (14%, n = 22); ICPM-SO-Moderate (12%, n = 19); AICPM-High (8%, n = 13); AICPM-SO-Moderate (5%, n = 8); ICPM-SO-High (4%, n = 7); and AICPM-SO-High (3%, n = 4). The women's program streams that had been delivered by respondents since July 1st, 2017 included WOMIP (10%, n = 16), WSOP (3%, n = 5), AWOMIP (3%, n = 5), WOHIP (2%, n = 3), and AWOHIP (1%, n = 2). Six (4%) staff respondents had taught the Adapted Program since July 1st, 2017. Of the staff who reported which component they had delivered since July 1st, 2017 (n = 157), the most common were main (82%, n = 129) and primer or engagement programs (69%, n = 109). Staff had also taught maintenance or self-management (40%, n = 62) and hybrid programs (9%, n = 14).

Creation of Admission Cohort. Information on men and women offenders used in the FIFE 2 analyses of timeliness of program participation, engagement, and retention (e.g., sentences and programs) was obtained from the OMS Data Warehouse, which is an electronic system containing offender file information. Data were extracted on August 17^th, 2018.

In order to analyze the timeliness of participation in correctional programs, as well as the engagement and retention of offenders in programs, two admissions cohorts of federal men and women offenders admitted to CSC custody on a Warrant of Committal (i.e., first term of their sentence) were created. Two separate admissions cohorts were used as ICPM was rolled out at different times across regions.^{Footnote 23} The two admission cohorts were created based on the following parameters:

1. The first cohort consisted of men and women offenders admitted to institutions in the Atlantic, Québec, Ontario, and Pacific regions between April 1^st, 2016 and March 31^st, 2018 (i.e., FY 2016-17 - 2017-18). If they were transferred to the Prairie region before ICPM implementation (i.e., July 1^st, 2017), they were excluded from the analysis.
2. The second cohort consisted of men and women offenders admitted to institutions in the Prairie region between July 1^st, 2017 and March 31^st, 2018.

The cohorts were merged. Offenders were excluded if their FPED fell after the data extraction date (August 17^th, 2018) to ensure that every offender in the admissions cohort had sufficient time to enroll in and complete their respective programs. Moreover, offenders who were released from CSC custody before their FPED for reasons outside of CSC's control (e.g., court order, bail, death) were excluded from the analysis. The IICP was considered out of scope for the evaluation. However, the IICP streams were included in some analyses in which they were aggregated with the other programs. Due to the low number of program participants, the WSOP was not included in the analyses.

Program assignment data were extracted from OMS for every offender in the cohort, where the program assignment date or program start date fell within an offender's incarceration period (i.e., between their admission date and first release date). If they had not been released by the time of data extraction, all of their program assignments and/or enrollments until the end of the study period (i.e., August 17^th, 2018) were captured. Only assignments and enrollments to institutional programs were included in the analysis. Correctional program assignment data for each unique offender in the cohort were summarized and manipulated to respond to the above-noted performance indicators.

Data Extracted from OMS Data Warehouse for Admission Cohort.

Data Definitions: The definitions of the terms related to program assignment and the data extracted from the OMS Data Warehouse are provided below.

• Assessed Program Need: An offender's assessed program need is based strictly on actuarial risk assessments conducted at intake (e.g., previously the SIR-R1, CRS, and the dynamic need level, and the CRI since January 2018).
• Actual Program Need: An offender's actual program need is based on actuarial risk assessments, but incorporates professional judgement. If a Parole Officer does not agree with an offender's assessed program need, they can request a program stream and/or intensity override. All assessed needs and overrides are reviewed by the Correctional Intervention Board, at which point an offender's actual program need is identified and recorded in OMS.
• Override: An override occurs when a staff member did not agree with the assessed needs, and an override was approved to another stream and/or lower or higher intensity level to better meet the offender's needs.
• Target Program: The target program is selected in OMS based on the offender's actual program need.
• Program Need: An offender is considered to have a program need if they have a target program other than No Program (e.g., MT Moderate Intensity) identified under program intensity target in the INCP screen in OMS, or if they were assigned to a program (had an accepted referral). This means the Correctional Intervention Board approved an actual program need, at which point a target program was identified. See Figure 3 below for visual depiction of the Program Need Assessment Process.
• Program Assignment: An offender is considered to be assigned to a program if they have a valid program start date or an assignment status date with a status of waitlisted or temporarily reassigned.
• Program Enrollment: An offender is considered to be enrolled if they had a valid program start date.
• Program Completion: A correctional program is considered to be complete when the program assignment has a valid start and end date and an assignment status of successful, or attended all sessions.
• Program Non-Completion: A correctional program is considered to be incomplete if an offender has an assignment status of:
  1. Offender Suspended: indicates the offender was assigned to a program and will not be reassigned without a new placement decision by the Correctional Intervention Board. This status is normally used to identify disciplinary concerns that have arisen during the program assignment or when the offender's quality of participation does not meet participation, behavioural, or other related program expectations, including poor attendance. The status Suspended is also used when an offender withdraws from an assignment specified in the correctional plan.
  2. Offender Transferred: indicates the offender was transferred outside the current facility but was assigned to, and participated in, a program immediately prior to transfer from the facility.
  3. Program Cancelled: indicates an offender was participating in the program (assigned or temporarily reassigned status) and the program was cancelled while the offender was participating in it. The cancellation of the program was due to circumstances beyond the control of the offender.
  4. Offender Released: indicates an offender was assigned to and participated in a program up to and immediately prior to the time of a scheduled release from incarceration on day parole, full parole, or statutory release.
  5. Program Assignment Transferred: the offender is assigned or temporarily reassigned to a program and permanently leaves the program to participate in a different assignment at the same site, or to participate in the same program at a different security level unit as a result of a movement within a clustered site.
  6. Program Incomplete: the offender was participating in a program but did not complete the assignment and/or was removed from the program for reasons unrelated to program participation. This status is typically used when an offender is deceased, is unable to meaningfully participate in the program due to responsivity needs, for outside court, or outside hospital. It is also used when an offender refuses to attend all sessions of the refuser/dropout stream of the motivational module.

Program Need Assessment Process. The process to identify an offender's correctional program need is illustrated in Figure 3 and described in the following text.

A flow chart depicting the Program Need Assessment Process.

In addition to the offender interviews and staff questionnaires, quantitative data from OMS were used to examine institutional and post-release outcomes in FIFE 3. The creation of the cohorts used for these analyses are described below.

Creation of Cohort for Institutional Outcomes. The data used to analyze institutional outcomes (e.g., minor and serious drug, violent and other charges, and urinalysis outcomes) were extracted from the OMS Warehouse on February 1^st, 2019.

The dataset included first-term offenders who started an ICPM/WOCP main correctional program during the following study periods:

1. between April 1^st, 2015 - December 31^st, 2017 in the Atlantic, Québec, and Pacific regions;
2. between April 1^st, 2016 - December 31^st, 2017 in the Ontario region; and,
3. between July 1^st, 2017 and December 31^st, 2017 in the Prairie region.

These dates were selected based on ICPM implementation dates, and were all prior to the introduction of a new program need assessment tool, the CRI, in January 2018.

The cohort included three groups: 1) program completers, 2) non-completers who enrolled in a program, but did not complete it for administrative/population management or offender-related reasons, and 3) offenders who were never assigned to an ICPM/WOCP main program, but met criteria for program participation (eligible non-participants). As much as feasible, the program referral criteria from the May 11^th, 2015 (CSC, 2015a) and January 23^rd, 2017 (CSC, 2017a) versions of the National Program Referral Guidelines (Guidelines 726-2) were used to identify the eligible non-participant groups. The Program Identification Tool is used by staff to determine the most appropriate program stream and intensity level for each offender. It is an automated report that compares the offender's results on actuarial tools and number of violent offences with the program selection criteria in order to determine the most appropriate program intensity and stream for the offender as per the information Table 4 below.

Program timelines had to be estimated for eligible non-participants to ensure that rates of institutional outcomes were compared during similar periods in the offender's sentence. An estimated program start date was calculated for each eligible non-participant in the following manner:

• The estimated program start date for offenders with an indeterminate sentence was calculated by: 1) determining the average point in a sentence in which program completers with an indeterminate sentence started a main program: (days from admission to program start date)/(days from admission to FPED); 2) using the average point in a sentence to begin programs, an estimated program start date was calculated for eligible non-participants.
• The estimated program start date for offenders with a determinate sentence was calculated by: 1) determining the average point in a sentence in which program completers started a main program: (days from admission to program start date)/(days from admission to warrant expiry date [WED]); 2) using the average point in a sentence to begin programs, an estimated program start date was calculated for eligible non-participants.

The average point in a sentence to begin programs was calculated separately for men and women. The average length of program (in days) was calculated separately for men and women and was added to the estimated start date to determine the estimated program end date for eligible non-participants. The program start date for non-completers was based on the date that they actually started their program. The estimated program end date for non-completers was calculated by adding the average length of program to the program start date. Eligible non-participants were included in the sample if their estimated start date occurred within the study period.

Only those offenders who were incarcerated during the 6 months before and 6 months after their program participation (program completers), actual start date and estimated completion date (non-completers), or estimated program dates (eligible non-participants) were included in the analyses of institutional outcomes. Program participants who completed two main programs within the study period were excluded. Offenders were excluded if they had participated in an ICPM program during the ICPM pilot.

Characteristics of the Institutional Outcomes Cohort. In total, 2,859 program completers, 441 non-completers, and 464 eligible non-participants were included in the analysis. Most of the offenders in the three groups were male. A third of the non-completers were Indigenous, in contrast with 3% of the eligible non-participants. Completers were most commonly from Ontario, and non-completers and non-participants from Québec. Participants were, on average, in their mid to late thirties (see Table 5).

Creation of Release Cohort for Post-Release Outcomes. The data used to analyze discretionary release, revocations, and the impact of correctional programs on specific offending behaviours, were extracted from the OMS Data Warehouse. The data were extracted on November 26^th, 2018, and the maximum follow-up date for outcomes was October 14^th, 2018. Since the implementation of WOCP occurred before ICPM, the timelines for inclusion in the dataset differed for male and female offenders.

The release cohort for men offenders consisted of those who had a first term release between April 1^st, 2015 and December 31^st, 2017 in Atlantic, Québec, and Pacific regions; April 1^st, 2016 to December 31^st, 2017 in Ontario; and June 1^st, 2017 to December 31^st, 2017 in the Prairie region. The timelines for inclusion differed across regions according to the dates of ICPM implementation.

Men program participants included the offenders who were assigned, enrolled, and completed their assigned ICPM main program. Offenders were identified as program participants if their ICPM main program completion and subsequent release occurred:

1. between April 1^st, 2015 and December 31^st, 2017 in the Atlantic, Québec, or Pacific region;
2. between April 1^st, 2016 and December 31^st, 2017 in the Ontario region; and
3. between June 1^st, 2017 and December 31^st, 2017 in the Prairie region.

Data for the offenders who completed the ICPM-MT-Moderate, ICPM-MT-High, ICPM-SO-Moderate, and AICPM-Moderate were included in the analyses. The outcomes of ICPM-SO-High, AICPM-SO-Moderate, AICPM-SO-High, AICPM-High, and IICP participants were excluded due to small sample sizes.

The release cohort for women offenders consisted of those who had a first term release between May 1^st, 2013 and December 31^st, 2017 across all regions. Women program participants included the offenders who were assigned, enrolled, and completed their assigned WOCP main during the study timeframe. Data for the offenders who completed WOMIP and AWOMIP were included in the analyses. The outcomes of WOHIP and AWOHIP, as well as WSOP participants were excluded due to small sample sizes.

Program participants were flagged as completers or non-completers. Analyses indicated that the non-completer groups (which included non-completions due to offender-related and administrative reasons) yielded insufficient sample sizes to be included in the analysis as comparison groups. For men and women, there were two comparison groups of offenders who did not participate in an ICPM/WOCP main program. One group of non-participants consists of those who met the program referral criteria but did not enroll in the main program (eligible non-participants) and the other included non-participants who did not meet the program referral criteria (no-intent-to-treat).^{Footnote 28} Similar to the institutional outcomes cohort, the program referral criteria from the May 11^th, 2015 (CSC, 2015a) and January 23^rd, 2017 (CSC, 2017a) versions of the National Program Referral Guidelines (Guidelines 726-2) were used to identify the eligible non-participant and no-intent-to-treat groups.

Offenders were excluded if they had: completed a correctional program under the old model or participated in the ICPM pilot; were admitted prior to June 30^th, 2009; were released prior to implementation of ICPM/WOCP within their region; or were released on a long-term supervision order.

Characteristics of Men in the Post-Release Outcomes Cohort. In total, 1,705 program completers, 800 eligible non-participants, and 2,393 no-intent-to-treat non-participants were included in the analysis. The characteristics of the offenders are presented separately in Table 6 for those who had a completed CRI and those who did not, as some analyses excluded offenders without a CRI.^{Footnote 29},^{Footnote 30},^{Footnote 31} Almost one fifth of the eligible non-participants were Indigenous, in contrast with 13% (n = 218) of the program completers and 8% (n = 189) of the no-intent-to-treat non-participants. Completers were most commonly from the Ontario region, whereas the eligible and no-intent-to treat non-participants were most commonly from the Québec region. The offenders in the three groups were, on average, in their late thirties to early forties. The length of time from admission to release, on average, was 560 days for program completers, 671 days for the eligible non-participants, and 531 days for the no-intent-to-treat non-participants. Almost 60% of program completers were rated as having a moderate level on the CRI at intake and a quarter had a high level, whereas about half of the eligible non-participants had a moderate level on the CRI and half had a high level, and almost two-thirds of the no-intent-to treat non-participants had a low level on the CRI.

Approximately two-thirds of eligible and no-intent-to-treat non-participants, and three-quarters of program completers, were rated as having medium motivation^{Footnote 32} at intake. Approximately two-thirds of program completers participated in an institutional or community-based maintenance program, in contrast with the eligible and no-intent-to-treat non-participants where few participated in a maintenance program. None of the program completers and few of the no-intent-to-treat non-participants completed an ICPM community program, in contrast with one-third of the eligible non-participants. Offenders who did not participate in ICPM programs in the institution, but who met the criteria for a program, are presented the opportunity to participate in the community program, which explains this latter result. Furthermore, the non-participation of program completers in a community program is to be expected as they completed a main program in the institution. Low participation among the offenders in the no-intent-to-treat group is not surprising either, as they would not meet referral criteria and are low risk. In addition, the result regarding maintenance program participants is not surprising given that only those who completed an ICPM main or community program are eligible to participate in maintenance programs.

With respect to differences between offenders who had a CRI and those who did not, a higher percent of Indigenous offenders had a completed CRI. Across the three groups, a higher percentage of offenders in the Pacific region had a completed CRI. The number of days from admission to release was higher for those with a CRI compared with offenders without a CRI. Additionally, more offenders without a CRI were rated as having high levels of motivation at intake compared with those with a CRI.

Characteristics of Women in the Post-Release Outcomes Cohort. In total, 723 program completers, 71 eligible non-participants, and 264 no-intent-to-treat non-participants who were women comprised the cohort available for analysis (N = 1,058). Table 7 presents the characteristics of the women. Approximately one-third (34%; n = 246) of program completers were Indigenous, in contrast with only 9% (n = 6) of the eligible non-participants and 18% (n = 47) of the no-intent-to-treat non-participants. Women serving their sentence in the Prairie region made up the largest portion of the sample (36%; n = 380 of 1,058), particularly for program completers (35%) and no-intent-to-treat non-participants (45%). Eligible non-participants were most commonly from the Ontario region. Age at release, on average for each group, ranged from early thirties (eligible non-participants M = 33) to mid-forties (no-intent-to-treat M = 45). Length of time from admission to release, on average, was 431 days for program completers, 668 days for the eligible non-participants, and 361 for the no-intent-to-treat non-participants. Notably, a large percentage of women did not have a CRI score, ranging from 33% of the program completers to 60% of the women in the no-intent-to-treat group.^{Footnote 33} Approximately one-third (35%) of the completers had a moderate level of risk at intake as identified by the CRI. Women in the eligible non-participant and no-intent-to-treat groups most commonly had a low level of risk on the CRI at intake. Program completers were split between medium (49%) and high (50%) motivation at intake, while many of the eligible non-participants (59%) and the no-intent-to-treat groups (62%) were rated as having high motivation at intake.

Nearly two-thirds of the no-intent-to-treat group (62%; n = 164) and 68% (n = 491) of program completers completed a self-management program, in contrast with 37% (n = 26) of the eligible non-participants. Although the self-management program is offered to women in the community, there is no program available that is equivalent to the men's community program.

In addition to the offender interviews and staff questionnaires, quantitative data from OMS and Financial Systems were used to examine the cost-effectiveness of correctional programs. Data from HRMS were used to examine the number of CPOs/ACPOs, as well as information on the number of CPOs/ACPOs who received training for the various correctional program streams.

Release Cohort for Post-Release Outcomes. The release cohort that was established for FIFE 3 served as the foundation for the cohort to be used to derive an estimate of the effectiveness of programming, which is needed in the overall cost-effectiveness calculation. Men participants (completers and non-completers) across all streams were considered in this cohort, as financial data (see below) did not allow the costs associated with each program stream to be separated. This resulted in the inclusion of ICPM-SO-High and AICPM-MT-High program participants to the original FIFE 3 cohort.^{Footnote 34} There were no AICPM-SO-Moderate, AICPM-SO-High, or IICP-Moderate participants eligible for inclusion in this cohort.

For the purposes of the cost analysis, it was primarily of interest to compare the outcomes of treatment participants to eligible non-participants. The composition of the eligible non-participant group differed from the sample used in FIFE 3, as it was desirable to establish a group of offenders where it was reasonable to assume zero costs associated with programming while in custody. As a result, only those eligible non-participants from FIFE 3 that had no exposure to the institutional primer program were included in the study group. Lastly, a fixed follow-up of 12 months was required to standardize the estimate of the effectiveness of programming between the two groups.

Financial Data. The financial expenditures directly related to institutional correctional program delivery in 2017-2018 were examined separately for men and women programs. Correctional program management related costs (e.g., salaries of regional administrators and program managers) were excluded from the cost of program delivery, as it is not possible to separate the costs allocated to men and women programs, and it is consistent with how the financial reports are populated in the Integrated Corporate Reporting Tool (ICRT).

The cost of maintaining an offender (COMO) was used to estimate the cost associated with a readmission for men only. Cost-effectiveness could not be examined for women's correctional programming since all women are referred to the engagement program and the current evaluation required a comparison group with no exposure to correctional programming (i.e., a no cost comparison group). The average institutional COMO was calculated by considering the costs associated with minimum, medium, and maximum security institutions for men during FY 2016-2017.^{Footnote 35}

Length of Readmission. Given that revocation for any reason on first release was examined as the index of program effectiveness, determining the average amount of time incarcerated following a revocation of a first release was necessary. The average length of readmission for all offenders who were initially released, were then readmitted, and then released on a 2^nd term during 2016-2017 and 2017-2018 was obtained from OMS.

Number of Program Participants. Data from CSC's ICRT were examined to determine the number of unique offenders who had either participated (i.e., enrolled but did not necessarily complete) or completed an ICPM readiness program, a main program, or an institutional maintenance program during FY 2017-2018. Global counts of enrolment across ICPM components were derived to remain consistent with the financial data available for the cost analysis. Offenders who participated in programs that were delivered in the community did not contribute to this overall count.

Human Resource Data. The number of CPOs and ACPOs at the end of FY 2017-2018 and FY 2018-2019 working within the institution or the community was extracted through CSC's HRMS. Information was available on the number of funded CPO and ACPO positions, the number of active substantive employees^{Footnote 36} and the number of employees acting in the CPO/ACPO position in each of the regions. Institutional level data and the number and type of program streams delivered by each CPO/ACPO was not available for analysis.

The number of CPOs and ACPOs who completed ICPM/WOCP initial training in FY 2017-2018 and FY 2018-2019 was also available in HRMS. It should be noted that the HR data does not include the specific program(s) that a given CPO/ACPO administers throughout the year. The number of training sessions entered into HRMS and delivered to CPOs and ACPOs from 2015-2016 to 2017-2018 was examined to assess the frequency and availability of training.^{Footnote 37}

Characteristics of Post-Release Outcomes Sample for Cost-Analysis. In total, 1,118 offenders participated in any ICPM program stream (n = 1,046 completers) and were eligible to be included in the 12-month follow-up analysis. There were 212 eligible non-participants who did not have exposure to the readiness program and who had a possible follow-up of at least 12 months. The characteristics of these subsamples are presented in Table 8. Given the minimal difference in sample size between the program participants (which included program completers and program non-completers) and solely program completers, the characteristics of program participants are reported here. A greater proportion of program participants were scored as having high motivation at intake (17%; n = 196) compared to eligible non-participants (4%; n = 9). Approximately 55% (n = 612) of program participants were rated moderate on the CRI and nearly a quarter (24%; n = 268) were rated high, whereas about half (49%; n = 103) of the eligible non-participants were rated as moderate and 37% (n = 78) were rated as high. Eligible non-participants were slightly more than a year older, on average, than program participants (39 vs. 38) and tended to be held in custody nearly 300 days longer, on average (863 days vs. 580 days).

The offender interview data were entered into Snap Survey software and exported into SPSS and Microsoft Excel. The Evaluation team analyzed the responses to the open-ended questions using an iterative and inductive process to identify relevant themes. Responses to the close-ended questions were analyzed using descriptive analysis techniques. Data from staff questionnaires were analyzed using the same process as used for the offender interview data.

Data extraction from the OMS Data Warehouse occurred through use of SAS. SAS was then utilized to manipulate and modify data as needed to answer the evaluation questions. OMS data were analyzed using SAS software or SPSS version 25. To analyze qualitative and quantitative data, the following qualifiers were used to indicate the weight of emerging themes and to meaningfully interpret evaluation results: a few/small number of refers to less than 25% of the sample, some refers to 25% to 45% of the sample, about half refers to 46% to 55% of the sample, many refers to 56% to 75% of the sample, most refers to over 75% of the sample, and almost all refers to 95% or more of the sample.

The admission cohort was primarily analyzed descriptively using frequencies, cross tabulations, percentages, medians, and means. Where appropriate, chi-square analyses were used to statistically compare frequency counts and percentages across groups and t-tests were used to compare means.

Institutional outcomes were explored by comparing outcomes on institutional indicators in the 6 months before and after main program participation (or before and after estimated program dates) of three groups: 1) main program completers, 2) program non-completers for administrative/population management and offender-related reasons, and 3) eligible non-participants.

Descriptive data (percentages and frequencies) were reported for refused or positive non-random urinalysis tests, as well as for refused or positive random urinalysis tests. Due to the distribution of the data, it was not possible to conduct statistical analyses.

Six-month pre and post-program participation outcomes were explored for:

• Number of minor and serious^{Footnote 38} violent charges for which the offender was found guilty;
• Number of minor and serious drug charges for which the offender was found guilty; and
• Number of minor and serious other charges for which the offender was found guilty.

The outcomes were examined to determine if the offenders had:

1. no change in charges from the 6 months before a main program to the 6 months following a main program (did not receive any charges before and after);
2. no change in charges from the 6 months before a main program to the 6 months following a main program (received one or more charges both before and after);
3. an increase in charges (no charges before a main and a minimum of one charge after the program); or
4. a decrease in charges (received one or more charges in the 6 months before a program and no charges after the program).

Outcomes of the participants of the following programs were analyzed using chi-square analyses in comparison with the non-completers and non-participants: ICPM-MT-Moderate, ICPM-MT-High, moderate SO programs (ICPM-SO-Moderate and AICPM-SO-Moderate), hybrid programs (hybrid ICPM-MT and hybrid AICPM), adapted programs (ICPM-MT-Moderate adapted and ICPM-SO-Moderate adapted), ICPM-SO-High, AICPM-MT-Moderate, AICPM-MT-High, WOMIP and AWOMIP. However, the violent and drug charge outcomes of WOMIP and AWOMIP, and the drug charges for the adapted programs, could not be examined using chi-square analyses since they did not meet the required statistical assumptions. In those cases, descriptive information was presented without statistical results.

The outcomes of three groups of offenders were compared: 1) program completers, 2) eligible non-participants, and 3) non-participants with no-intent-to-treat.^{Footnote 39} In an effort to isolate the relationship between program participation and post-release outcomes, analyses for both men and women controlled for the effects of the following covariates: CRI level at intake, age at release, number of days from admission to release, motivation level at intake, a flag for participation in a maintenance programs (held in the institution or in the community) and a flag for participation in the ICPM community program (men only). This enhanced the confidence that any observed relationship between the study group and outcome was truly due to the program, rather than the result of pre-existing differences on the covariates. In addition, in certain analyses, Indigenous ancestry was included as a covariate. For discretionary release outcomes, participation in the ICPM community program and participation in a maintenance program were not used as covariates, given that they would have occurred after the release decision.

The outcomes were explored for the following programs:

• ICPM-MT-Moderate (included hybrid ICPM-MT-Moderate);
• ICPM-MT-High;
• ICPM-SO-Moderate (AICPM-SO-Moderate and high intensity programs were excluded due to low number of participants);
• AICPM-Moderate (included hybrid AICPM-MT Moderate; high intensity excluded due to low number of participants);
• WOMIP (high intensity excluded due to low number of participants); and
• AWOMIP (high intensity excluded due to low number of participants).

In order to answer the question "Does participation and/or completion of correctional reintegration programs increase the likelihood of obtaining a discretionary release?", logistic regressions were conducted to determine the relationship between program participation and receiving discretionary (day parole or full parole) or statutory release. Logistic regression is the appropriate regression analysis to conduct when the dependent variable is dichotomous (e.g., yes or no). Logistic regression is used to explain the relationship between one dependent dichotomous variable and one or more independent variables. The key measure interpreted from a logistic regression is the odds ratio (OR). An OR measures the effect that independent variables have on an outcome in relative terms, which allows the comparison of the intervention group of a study relative to the comparison group. If the odds of the outcome is the same in both groups, the ratio will be 1, which implies there is no difference between them. However, if the OR is greater than 1, then the control group(s) (i.e. those who do not receive the intervention) have increased odds of the outcome relative to those with the intervention. If the OR is less than 1, then those with the intervention have increased odds of the outcome relative to the control group(s). In order to examine the impact of program completion relative to the two comparison groups, the program completer group was used as the reference group to which both comparison groups would be compared. For ease of interpretation, the inverse of the odds ratio (1/OR) was calculated, which reverses the direction of the effect, making it the effect of completing a program on the odds of obtaining a discretionary release relative to being in either comparison group.

In order to answer the question "Does participation and/or completion of correctional reintegration programs impact the likelihood of a revocation for any reason and/or revocation with an offence", Cox regression survival analysis predicting time from first release to offenders’ first outcomes following that release were conducted to identify the relationship between program participation and the following outcomes:

• Revocation for any reason (with or without offence);
• Revocation with offence;
• Revocation with a violent offence;
• Substance use related post-release outcome (includes suspensions due to a breach of a substance use related condition and/or positive urinalysis results in the community); and
• Revocation with a sexual offence (only examined for sexual offender program analysis).

Cox regression (or proportional hazards regression) is a method for investigating the effect of several independent variables (i.e., covariates) on the time to a specified event. The method assumes that the effects of the predictor variables upon survival are constant over time. The key statistic interpreted in the results is the hazard ratio (HR), which is a comparison of the probability of events in an intervention group to the probability of events in a comparison group. Similar to logistic regression, outcome variables must be dichotomous. This analysis is used to see if individuals receiving a main program experience a community outcome (e.g., any revocation) faster or slower than those not receiving a main program. In order to examine the impact of program completion, in comparison to the two comparison groups, the program completer group was used as the reference group, however for ease of interpretation, the inverse of the hazard ratio (IHR; 1/HR) was calculated. In addition to interpreting statistically significant effects, the direction of the effect (e.g., supporting treatment) for non-significant findings was reported when it appeared that the two groups meaningfully differed in the likelihood to experience the outcome. This threshold was set at an HR or IHR equal or less than 0.80, which represents a difference in the likelihood between treatment and comparison groups of at least 20%. If the estimated effect did not surpass this threshold, the two groups were considered to have a comparable likelihood of experiencing the outcome.

Outcomes were examined following first release only. Outcomes occurring following a subsequent release (on discretionary or statutory release or following the WED) were not examined due to the low occurrence of these events.

Due to small sample sizes (n = 9), the outcomes for those who completed the ICPM-Adapted program were unable to be analyzed separately.

With respect to the question "Does the integrated model address substance use and specific offending behaviours (e.g., family violence)?", Cox regression analyses were conducted for sub-groups of offenders who were identified as requiring programming to address the need areas listed in Table 9.

In addition, a separate set of analyses were conducted on offenders who were overridden. Offenders were identified as having received an override if they did not meet the initial program referral criteria, but had completed a moderate or high intensity correctional program. This method of identifying an override differs from that used for the admissions cohort analyses. Offenders who were overridden and completed a program were compared to offenders who completed a program but were not overridden (i.e., they met the program referral criteria). Note that those who participated in a program following an override were included in the main program completers group used in the analyses mentioned above. Due to small sample size, a descriptive analysis of outcomes was performed for men who received an override versus men who initially met program referral criteria. For women, there was an even distribution of program completers who were overridden versus those who initially met program referral criteria, which allowed for the use of survival analysis, controlling for relevant covariates, to complement the descriptive analysis.

The approach to assessing cost-effectiveness outlined below is based on that used for the CSC evaluation of education programs (Richer, McLean-McKay, Bradley, & Horne, 2013), which was originally developed by the RAND Corporation (Davis, Bozick, Steele, Saunders, & Miles, 2013). The analysis focuses on the direct costs of correctional programs and incarceration. The following elements were included in the calculation: index of program effectiveness, cost of readmission, and cost of programming per participant.

Under this model, programs are considered cost-effective when the combined cost of delivering programs and readmission for program participants is less than the cost of readmission for non-program participants. To demonstrate the potential cost savings associated with delivering programming to offenders, the costs associated with readmission for 100 eligible non-participants was compared to the costs associated with programming and readmission costs for 100 program participants,^{Footnote 40} after considering the rate of any revocation for each group.

Program effectiveness was assessed by comparing the likelihood of a revocation for any reason within 12 months of release for program participants (i.e., regardless of completion status) and eligible non-participants who did not have exposure to the institutional readiness program. Separate analyses were also performed restricting the programming group to only those who completed the main program that they enrolled in. Logistic regression was used to estimate the relationship between the study group (i.e., program participant vs. eligible non-participant) and the likelihood of a revocation for any reason, while accounting for the following risk-relevant covariates: CRI level at intake, Indigenous ancestry, motivation level at intake, age, and days between admission and release. Prior analyses presented in the previous chapter also controlled for the effects of participating in a maintenance program, but due to the reduced sample size among the eligible non-participants, and the limited opportunity to complete the program, examination of this covariate was not feasible. The likelihood of a revocation for any reason for each study group, while accounting for the differences on these covariates, was used as the metric of program effectiveness. Descriptive rates of any revocation for program participants or completers versus eligible non-participants were also examined, but more weight was given to the findings from the logistic regression due to the observed risk relevant differences between the groups.

In order to fulfill the broad scope of the evaluation, there were several methodological challenges and decisions that needed to be considered. The methodology allowed for a rigorous analysis of the evaluation questions that sought to reduce or mitigate as many limitations as possible. The following core limitations that relate to data collection for each of the chapters should be considered alongside the conclusions drawn from the evaluation.

The utilization of quantitative and qualitative data provided a comprehensive assessment of the evaluation questions. However, the qualitative interview data with offenders was limited due to the selection process and oversampling. The offenders who agreed to participate after being randomly selected, or those who approached interviewers to participate, could have differed from other offenders who did not agree to participate. Women and Indigenous offenders were oversampled in an attempt to ensure that there was a sufficient number of respondents to analyze the questions separately. Further, survey respondents were evenly distributed across the regions and security levels. This meant that the study sample was not representative of the in-custody population, potentially limiting the generalizability of the findings.

Lastly, although qualitative data provided access to rich information, and was often used to supplement quantitative analyses, some findings were based solely on qualitative data derived from small samples (e.g., identified barriers to timely access, barriers to timely program completion). Analysis of further data on these questions was outside the scope of the evaluation, but such analysis could help to further inform these areas.

The examination of the relationship between study group and institutional outcomes was limited by the requirement to have 6 months pre and post program participation (or similar timeframes for non program participants). This biased the sample to those with longer sentences, who began programming later in their sentence, and remained in custody after the completion of their programming. Although it was important to establish a consistent window of examination for all study groups, the results pertaining to the relationship between study group and institutional outcomes may not generalize to all program participants.

Due to sample sizes, the analysis of community outcomes could not include a matched sample of offenders who required the specific program stream. Although risk relevant differences between the groups were controlled for in the analysis, the ability to test the direct effect of each program stream was somewhat limited, and in some cases, not possible due to small sample sizes. Examinations of program effectiveness by Indigenous ancestry tended to be limited by small sample sizes, which affected the stability of the estimate of the relationship between study group and outcomes.

The recent rolling implementation of ICPM created challenges with maintaining regional representativeness among the data that contributed to the analyses. There was only 6 months of data available from the Prairie region, which resulted in an under-representation of Indigenous offenders in the evaluation sample. The recent implementation also limited the ability to examine longer-term community outcomes. This resulted in some community outcomes (e.g., revocation with violent offence) occurring infrequently, which limits the examination of program effectiveness. To provide a preview of the potential long-term effects of program participation, a case study of an earlier release cohort from the Pacific region, where ICPM was first implemented, was conducted to examine community outcomes beyond the first release.

Additionally, although the substance use outcome provided a useful analysis of substance use behaviour following release, the variable was limited in a number of aspects. First, the sample that was used for analyses involving the substance use outcome was not preselected based on whether they were required to submit urinalysis or whether they had a substance use related condition. That being said, analyses were conducted for the overall sample, and for a sample who were identified as having a treatment need for substance use, so it is likely that the offenders in these groups had relevant conditions to follow. Second, positive urinalysis results that contributed to the substance use outcome did not specify the type of substance use. Although this is a limitation, as use of some substances is not illegal, it is important to highlight that many substance use outcomes (approximately 70%) were comprised of suspensions due to a breach of a substance use related condition. It is also important to consider that ICPM integrates a harm reduction model that promotes a collaborative and goal-oriented approach to substance use. Harm reduction principles are used to address a number of substance use concerns, ranging from promoting abstinence to less harmful use. Given that information pertaining to substance use frequency and severity was unknown, one indicator of problematic behaviour that was used was a suspension due to a breach of a substance use related condition. While examining this outcome is warranted (recognizing that the results should be interpreted with caution), future research and evaluations should consider additional substance use outcomes that may more adequately reflect the harm reduction model supported by CSC. These outcomes could include assessing the severity of substance use over time and whether substance use was related to the current criminal behaviour or return to custody.

Lastly, another limitation corresponds to the fact that additional services (i.e., interventions other than correctional programming) were not examined in the current evaluation. Although research has demonstrated that effective correctional programming plays an integral role in offender rehabilitation (Landenberger & Lipsey, 2005; Lipsey, Chapman, & Landenberger, 2001; Usher & Stewart, 2014), there are a number of other services and interventions that are provided to federally sentenced men and women, both within the institutions and in the community, that aid in the preparation for successful release. Some of these additional services include employment and employability programs, educational programs, chaplaincy, family visits, mental health programs, counselling, and social programs. Notably, research has found that additional correctional services are associated with reduced revocations even after controlling for factors related to offending and participation in correctional programs for both men and women (Wilton, Nolan, Stewart, & Thompson, 2015; Wilton & Stewart, 2015). As such, future research and evaluations should consider the effects of additional services and supports that offenders receive and how these supports may further add to correctional programs in terms of successful outcomes in the community.

The examination of cost-effectiveness for delivering correctional programming was limited primarily by the available data. The coding of the financial data did not permit examination of each individual program stream. As a result, cost-effectiveness for men was estimated at the overall level, collapsing across program stream and intensity. Further, given that the evaluation methodology required a comparison group that incurred no costs related to correctional programming, an examination of cost-effectiveness could not be completed for women's correctional programming (i.e., all women are referred to the engagement program). The requirement for a zero cost comparison group also limited the representativeness of the sample for men offenders included in the cost analysis. Since most eligible non-participants participated in a primer program, it is apparent that the comparison group used in the cost analysis does not necessarily reflect a group of offenders with no exposure to correctional programming. Additionally, the model used to evaluate cost-effectiveness required estimates of several inputs (e.g., program effectiveness, cost of programming, cost of a revocation), each of which could be defined in numerous ways. In an attempt to ensure that the findings pertaining to cost-effectiveness were valid, several iterations of the cost-effectiveness model were performed. That being said, the conclusions for the cost-effectiveness of correctional programming are based on the data obtained during the evaluation and may not generalize to all correctional program streams and intensity levels. It is anticipated that when the coding of financial data associated with correctional programs improves, more precise estimates of cost-effectiveness will be achievable.

Another component of program efficiency is staff resourcing. Notably, the evaluation was limited in the ability to assess whether there was a sufficient number of staff available to deliver correctional programming. A count of active institutional CPOs/ACPOs was examined in relation to the total number of offenders in custody to approximate the ratio of correctional program staff to offenders. However, it was not possible to reliably examine the number of CPOs/ACPOs available to deliver a specific stream of correctional programming in a given institution due to the availability of recorded staffing information. Given the operational challenges associated with delivering correctional programs (i.e., changing demand for specific streams due to current offender population), the evaluation was unable to definitively determine whether there were an adequate number of CPOs/ACPOs for the number of offenders requiring correctional programming.

The evaluation found that correctional reintegration programs at CSC are relevant and respond to the needs of federal offenders. Positive impacts associated with completion of correctional programs were noted in a variety of areas. Findings suggested that program intensity and streams appropriately matched an offender's program need and that programming tended to be offered in a timely manner. For men offenders, completion of programs was associated with increased rates of discretionary release and decreased rates of any revocation. Findings tended to suggest that women program completers were granted discretionary release more often, but did not have lower rates of revocation, relative to eligible non-participants. Lastly, correctional programs were delivered in a cost-effective manner for men. The cost-effectiveness of women's programs was unable to be assessed. Several key areas were identified to improve the delivery and effectiveness of correctional programs, such as:

• Adopting a standardized definition of timely access to programs;
• Increasing the relevance of the content and delivery of the Indigenous programming streams;
• Conducting additional research to understand the effect of correctional programs on community outcomes;
• Reviewing the impact of the newly implemented program referral criteria on the number of overrides, particularly for women;
• Improving the availability and quality of data related to correctional programs, such as program expenditures; and,
• Reviewing the training protocol for CPOs.

This evaluation will assist CSC in enhancing the delivery and effectiveness of correctional programs to all offenders with a programming need. Moreover, the findings serve as a foundation for the evidence supporting the ICPM/WOCP models of programming, which will assist with continuing to monitor the results moving forward.

The following information is pulled directly from GL 726-2 and outlines the principles of the correctional program referral process.

1. Correctional program effectiveness requires matching the intensity of the correctional program(s) to an offender's level of risk.

2. Program intensity is generally determined by the results of the CRI.^{Footnote 114} For men's SO programs, the Static-99R and the Stable-2007, in combination with the CRI, generally serve as the determinants of program intensity for male SOs.

3. In the case of male offenders who meet the sexual offence criteria pursuant to CD 705-5 - Supplementary Assessments, a SO assessment will be conducted prior to an offender starting a main correctional program. This assessment will be conducted by the Correctional Program Officer/Assessor and will consist of administering the Static-99R and Stable-2007 pursuant to GL 726-3 - National Correctional Program Management Guidelines. In the case of women offenders who meet the sexual offence definition pursuant to CD 705-5 - Supplementary Assessments, a psychological risk assessment with the focus on sexual offending must be made available prior to the woman's SO program start date.

4. When determining an Aboriginal offender's correctional program needs, the offender's Aboriginal social history must be considered and documented in the decision-making process. For offenders who have expressed an interest in following a healing path, an Elder Review will be completed pursuant to CD 705-5 - Supplementary Assessments.

5. When determining appropriate referrals for Aboriginal offenders who wish to participate in Aboriginal correctional programming, referrals to national Aboriginal correctional programs should take precedence over referrals to the non-Aboriginal correctional program equivalent.

6. Referrals to national correctional programs should be the preferred intervention where appropriate and available. Referrals to non-standardized local or regional programs should only be used as a substitute for a national correctional program where no reasonable alternative exists. Offenders may only be assigned to one correctional program, including maintenance/self-management programs, at any given time.

7. Correctional planning requires that referrals to correctional programs:

1. Prepare offenders for timely and safe reintegration;
2. Ensure that the intensity and type of correctional program(s) selected is based on risk and needs;
3. Give priority to offenders serving sentences of four years or less for correctional programs during intake;
4. Include appropriate culturally-based correctional programs for Aboriginal offenders who wish to participate in Aboriginal correctional programs;
5. Include gender-informed correctional programs designed for women offenders;
6. Be achievable and available within the length of the sentence, and take parole eligibility dates into consideration;
7. Provide a justifiable rationale for any override to a correctional program, and document this rationale in the OMS;
8. Consider the offender's specific mental health care needs and/or physical disabilities.