Best practices in reducing recidivism
Summarizing a systematic review of reviews on intervention effectiveness (R004-VB-2026)

Table of contents

Introduction

What is recidivism

Any new criminal offence by a justice-involved clients (JICs) after an initial criminal justice intervention.

Recidivism can be measured using a variety of indicators for recontact between JICs and the criminal justice system, including rearrest, reconviction, or reincarceration.

Why study best practices in reducing recidivism

It is estimated that 35-50% of provincial and federal JICs were reconvicted of any offence within 3 years.Endnote i

Recidivism accounts for a large proportion of crimeEndnote iiEndnote iii. A single chronic JIC can cost between $12 to $17 million Canadian taxpayers dollars over 15 yearsEndnote iv. This cost includes expenses related to the victim, police, courts, prosecution, legal aid, corrections, and estimates of undetected crimes and their costs.

Identifying effective interventions to reduce recidivism could yield significant economic and social benefits.

Selecting the appropriate correctional interventions can be difficult because there is a large number of evaluations on many intervention types. Each evaluation applies different methodologies and sometimes produces different findings that may conflict with earlier research.

The current study addresses this knowledge gap by systematically synthesizing available research on the effectiveness of correctional interventions. This research was sourced from multiple jurisdictions around the globe. It allows for a clearer understanding of the types of correctional interventions that are most promising in reducing recidivism.

Methodology

Overview

This review of reviews (RoR) summarizes the effects of interventions designed to reduce recidivism among adult JICs. It includes systematic reviews and meta-analyses.

What is a systematic review of reviews

A systematic RoR is a method of synthesizing findings from multiple systematic reviews or meta-analyses.

Search strategy and inclusion criteria

Several scholarly databases were searched using key terms. To be included in the RoR, the report must have consisted of:

Search sample

A total of 141 eligible systematic reviews from international jurisdictions were analyzed, from which 513 effects were extracted.

  1. An effect is defined as the difference in recidivism between a group exposed to an intervention and a group without exposure (i.e., control)
  2. In systematic reviews with a meta-analysis, the impact of an intervention is also quantified by an effect size, which is a numerical measure that describes the magnitude of a difference.
    • There are multiple types of effect sizes, each with specific thresholds to categorize an effect as small, medium, or largeEndnote v

Recidivism categories

In this study, recidivism outcomes were categorized into the following types (which are not mutually exclusive):

General recidivism

The commission of any new criminal offence following prior involvement with the justice system, irrespective of the nature or category of the offence. This category includes results explicitly classified under the general recidivism category, as well as those in which the type of recidivism was not specified. In instances where systematic review findings were specified to a particular offence subtype – such as sexual, violent, intra-partner/domestic violence, drug-related offences – those results were classified exclusively in the corresponding subtype categories belowEndnote vi.

Sexual recidivism

The commission of a new sexual offence following prior justice-involvement, which includes both contact offences (e.g., sexual assault) and non-contact offences (e.g., voyeurism).

Violent recidivism

The commission of a new violent offence following initial prior justice-involvement, generally involving the use or threat of force against another person (e.g. assault, robbery, homicide).

Intimate partner violence/domestic violence recidivism (IPV/DV)

The commission of a new violent or sexual offence within a romantic or sexual relationship (i.e., intimate partner violence) or against a family/household member (i.e., domestic violence).

Drug-related recidivism

The commission of a new offence involving possession or trafficking of drugs, impaired driving due to alcohol or cannabis, and violations of release conditions that are related to drug use.

Methodological limitations

It is important to note that this RoR could not reliably evaluate the quality of interventions – specifically, adherence to principles of effective correctional intervention/treatment (i.e., Risk-Need-Responsivity [RNR]Endnote viii) – or the fidelity with which an intervention was delivered. As a result, the findings reflect the overall effect of correctional interventions on recidivism, without consideration of intervention/treatment quality or the implications of individual differences.

Study results

Across all effects, 60% demonstrated that interventions reduced recidivism

25% to 48% of effects demonstrated at least a moderate reduction in recidivism when considering the percentage of effects from systematic reviews with a meta-analysisFootnote 1.

Figure 1: Percentage of effects from systematic reviews with and without a meta-analysis related to correctional interventions demonstrating a reduction in each type of recidivismFootnote 2
Figure 1: Image description

Bar chart showing percentages of systematic reviews demonstrating at least a small decrease in recidivism across different types: General (60.9%), Sexual (53.1%), Violent (56.8%), IPV/DV (64.0%), and Drug-related (61.3%). Bars are dark blue with labels on the x-axis for recidivism types and y-axis representing percentage values.

Figure 2: Percentage of effects from systematic reviews with a meta-analysis demonstrated a small and at least a moderate reduction in recidivism following correctional interventions
Figure 2: Image description

Horizontal bar chart showing the percentage of effects indicating a small decrease and moderate decrease, across five types of recidivism: General, Sexual, Violent, IPV/DV, and Drug-related. Dark blue bars represent small decreases and teal bars represent moderate decreases. Percentages are displayed at the end of each bar.

  • General recidivism: small decrease = 56.2% reduction; moderate decrease = 26.1% reduction
  • Sexual recidivism: small decrease = 56.3% reduction; moderate decrease = 31.3% reduction
  • Violent recidivism: small decrease = 64.0% reduction; moderate decrease = 42.3% reduction
  • IPV/DV (intimate partner violence/domestic violence) recidivism: small decrease = 50.0% reduction; moderate decrease = 48.0% reduction
  • Drug-related recidivism: small decrease = 56.3% reduction; moderate decrease = 25.0% reduction

Cognitive behavioural therapy (CBT) and related approaches like cognitive skills training and behavioural therapy, were associated with small to moderate reductions across all types of recidivism.

Some examples of effective interventions for reducing recidivism are:Footnote 3

Table 1: Was this correctional interventions associated with a reductions in general, sexual, violent, IPV/DV, and drug-related recidivism, according to systematic reviews with and without a meta-analysis?

How to read this table:

  • Yes indicates evidence of reduced recidivism
  • No indicates no reduction
  • n.a. indicates that there was no report on the intervention and recidivism.
  • Results marked with an asterisk (*) should be interpreted with caution due to a limited number of observed effects (fewer than 3).
Intervention General Sexual Violent IPV/DV Drug
Aftercare Yes n.a. n.a. n.a. n.a.
Anger management Yes n.a. Yes n.a. n.a.
Behavioural therapy Yes Yes Yes n.a. n.a.
Bootcamps No n.a. n.a. n.a. n.a.
Cognitive Behavioural Therapy Yes Yes Yes Yes Yes* (interpret with caution)
Cognitive skills training Yes Yes Yes Yes n.a.
Counselling Yes n.a. No Yes Yes
Diversion Yes n.a. Yes n.a. Yes
Duluth model No Yes Yes* (interpret with caution) Yes n.a.
Education/vocational training Yes n.a. n.a. n.a. Yes
Group therapy Yes No Yes Yes Yes* (interpret with caution)
Medical/surgical Yes n.a. Yes* (interpret with caution) n.a. n.a.
Multisystemic therapy Yes Yes No* (interpret with caution) n.a. n.a.
Offence-specific programming Yes n.a. n.a. No* (interpret with caution) Yes* (interpret with caution)
Other psychotherapy Yes Yes Yes Yes Yes
Pharmacology n.a. Yes* (interpret with caution) Yes* (interpret with caution) n.a. No
Psychoeducation n.a. Yes* (interpret with caution) Yes* (interpret with caution) Yes n.a.
Psychosocial skills training Yes n.a. n.a. Yes* (interpret with caution) No* (interpret with caution)
Restorative Justice No Yes* (interpret with caution) Yes n.a. n.a.
Social skills training Yes n.a. n.a. Yes n.a.
Supervision n.a. n.a. n.a. No n.a.
Other Yes Yes n.a. Yes* (interpret with caution) Yes
Unspecified Yes n.a. n.a. Yes n.a.
Note: The "Other" category included interventions mentioned by fewer than 10 eligible systematic reviews and did not fit into any other categories. These included schema therapy, risk management training, detoxification, and self-help. The "Unspecified" category included cases in which "intervention" used as an umbrella term, without specification of type.
Table 2: Was this correctional interventions associated with at least a moderate reduction in general, sexual, violent, IPV/DV, and drug-related recidivism, according to systematic reviews with a meta-analysis?

How to read this table:

  • Yes indicates evidence of at least a moderate reduction of recidivism
  • No indicates either no reduction, a small reduction, or no report on the relationship between the intervention and recidivism.
  • Results marked with an asterisk (*) should be interpreted with caution due to a limited number of observed effects (fewer than 3).
Intervention General Sexual Violent IPV/DV Drug
Aftercare No No No No No
Anger management Yes No Yes No No
Behavioural therapy No Yes Yes No No
Bootcamps No No No No No
Cognitive Behavioural Therapy No Yes Yes Yes No
Cognitive skills training No No Yes Yes No
Counselling No No No Yes No
Diversion No No No No No
Duluth model No Yes* (interpret with caution) Yes* (interpret with caution) No No
Education/vocational training No No No No No
Group therapy No No Yes Yes No
Medical/surgical Yes No Yes* (interpret with caution) No No
Multisystemic therapy No No No No No
Offence-specific programming Yes No No No No
Other psychotherapy No Yes Yes Yes No
Pharmacology Yes Yes* (interpret with caution) Yes* (interpret with caution) No No
Psychoeducation No Yes* (interpret with caution) Yes* (interpret with caution) Yes No
Psychosocial skills training No No No Yes* (interpret with caution) No
Restorative Justice No No No No No
Social skills training No No No Yes No
Supervision No No No No No
Other Yes No No Yes* (interpret with caution) Yes
Unspecified Yes No No Yes No
Note: The "Other" category included interventions mentioned by fewer than 10 eligible systematic reviews and did not fit into any other categories. These included schema therapy, risk management training, detoxification, and self-help. The "Unspecified" category included cases in which "intervention" used as an umbrella term, without specification of type.

All intervention settings were at least somewhat conducive to reducing general recidivism

Interventions delivered in almost all settings were associated with at least a moderate reduction in sexual, violent, and IPV/DV recidivism.

Drug-related recidivism, on the other hand, only evidenced recidivism reduction from interventions applied in court and community settings.

Table 3: Were interventions in this setting associated with a reduction in recidivism, based on systematic reviews with and without a meta-analysis?

How to read this table:

  • Yes indicates a reduction in recidivism.
  • No indicates no reduction in recidivism.
  • n.a. indicates that there is no report on the relationship between interventions in that setting and recidivism.
  • Results marked with an asterisk (*) should be interpreted with caution due to a limited number of observed effects (fewer than 3).
Setting General Sexual Violent IPV/DV Drug
Custody Yes Yes Yes Yes n.a.
Community Yes Yes Yes Yes Yes
Court Yes n.a. Yes n.a. Yes
Residential Yes Yes* (interpret with caution) Yes* (interpret with caution) Yes* (interpret with caution) n.a.
Hospital Yes n.a. Yes Yes* (interpret with caution) No* (interpret with caution)
Table 4: Were interventions in this setting associated with at least a moderate reduction in recidivism, based on systematic reviews with a meta-analysis?

How to read this table:

  • Yes indicates at least a moderate reduction in recidivism.
  • No indicates either no reduction in recidivism, a small reduction in recidivism, or no reports on the relationship between interventions in this setting and recidivism.
  • Results marked with an asterisk (*) should be interpreted with caution due to a limited number of observed effects (fewer than 3).
Setting General Sexual Violent IPV/DV Drug
Custody No Yes Yes Yes No
Community No Yes Yes Yes Yes
Court No No Yes No Yes
Residential No Yes* (interpret with caution) Yes* (interpret with caution) Yes* (interpret with caution) No
Hospital Yes No Yes Yes* (interpret with caution) No

The evidence suggests that correctional interventions have the potential to be effective across settings at reducing recidivism and should be offered beyond the supervision or incapacitation element of a sentence.

Conclusion

The goal of this study was to understand which types of correctional interventions most effectively and consistently reduce recidivism.

The study results demonstrate that a wide range of correctional interventions could potentially reduce recidivism more effectively than incarceration or community sentences with limited rehabilitative focus, such as bootcamps and intensive supervision. They suggest that "what works" to reduce recidivism is evidence- based therapeutic, rehabilitative, and educational approaches to corrections, grounded in cognitive behavioural techniques.

However, the practical significance of the correctional interventions remains uncertain. While a substantial portion of findings (~70%) from meta-analyses indicated that these interventions are effective, they typically result in only small reductions in recidivism. In addition, a small percentage of findings have reported that certain interventions may lead to increases in reoffending. These adverse outcomes may be associated with highly punitive approaches such as bootcamps and intensive supervision. Further research is needed to better understand the underlying factors contributing to the suboptimal effectiveness of correctional interventions.

CBT and related approaches may be particularly effective interventions for reducing recidivism because of their:

  1. Ability to directly target procriminal attitudes and ways of thinking (one of the domains most highly related to criminal conduct, per the need principle), and
  2. Direct correspondence to the responsivity principle – that is, the most effective correctional interventions adopt cognitive behavioural and social learning approaches
Date modified: