Evidence-based practices in corrections are supervision and treatment methods whose effectiveness at reducing reoffending has been confirmed through controlled research rather than tradition or intuition. The field is anchored by a framework called Risk-Need-Responsivity, delivered mainly through cognitive-behavioral programs, guided by validated risk assessments, and reinforced by structured incentives. When agencies follow the research, recidivism drops. When they ignore it, well-meaning programs can make outcomes worse.
What “Evidence-Based” Means Here
In 1974, sociologist Robert Martinson reviewed 231 correctional treatment studies and concluded that “with few and isolated exceptions, the rehabilitative efforts that have been reported so far have had no appreciable effect on recidivism.” That finding, reduced to the catchphrase “nothing works,” reshaped policy for more than a decade and accelerated a shift toward purely punitive sentencing.
The conclusion was narrower than the slogan. Martinson found that most programs as they were run at the time didn’t work, not that rehabilitation was impossible. In the 1980s and 1990s, psychologists Don Andrews and James Bonta began asking a sharper question: which programs work, for whom, and why? The patterns they identified became the Risk-Need-Responsivity model, and decades of meta-analyses built on it. That research base is what “evidence-based” refers to in modern corrections.
The Risk-Need-Responsivity Framework
Risk-Need-Responsivity structures programming around three principles. Agencies that follow all three see recidivism reductions. Agencies that follow one or two often don’t.
Risk: Match Intensity to Likelihood of Reoffending
The intensity of supervision and treatment should match the person’s risk level. High-risk individuals receive the most intensive services. Research on treatment dosage suggests moderate-risk individuals may need around 100 hours of programming to see results, while high-risk individuals with multiple needs often require 200 hours or more.1Civic Research Institute. Exploring the Risk-Dosage Relationship in Offenders Classified as Neurotic
Getting this backwards is worse than doing nothing. Placing low-risk individuals in high-intensity programs pulls them away from stable jobs, families, and prosocial peers and puts them alongside higher-risk participants. One study found a 17 percent increase in recidivism for lower-risk individuals subjected to intensive rehabilitation supervision, while the same program produced a 20 percent reduction for higher-risk participants.2United States Courts. Adhering to the Risk and Need Principles The instinct to “do something” for everyone often causes real harm.
Need: Target the Right Problems
Not every problem a person has drives criminal behavior. The need principle separates criminogenic needs (factors directly linked to offending) from non-criminogenic needs (real problems that don’t drive criminal conduct). Programs aimed at low self-esteem or generalized anxiety consistently fail to reduce reoffending. Effective programs focus on the specific thought patterns, associations, and habits that make new offenses more likely.
Responsivity: Deliver It in a Way People Can Absorb
Well-targeted programming still fails if it’s delivered poorly. General responsivity means using cognitive-behavioral and social learning approaches, which work for the broadest range of people. Specific responsivity means adapting delivery to the individual: simplified materials for someone with a learning disability, trauma-informed methods for abuse survivors, culturally relevant content for particular populations.
The Central Eight Criminogenic Needs
Andrews and Bonta identified eight factors most strongly linked to criminal behavior. Programs that target these produce the largest recidivism reductions; case plans that ignore them are essentially guessing.
- Criminal history. The single strongest predictor. It can’t be changed, but it determines how intensive services need to be.
- Antisocial personality pattern, including impulsivity, aggression, and seeking immediate gratification regardless of consequences.
- Procriminal attitudes, such as beliefs that rationalize illegal behavior (“everyone steals,” “the system is rigged so the rules don’t apply to me”).
- Procriminal associates: close relationships with people who support or participate in criminal activity, combined with few prosocial connections.
- Substance abuse, particularly where it intersects with impulsivity or criminal peer networks.
- Family and relationship problems, including conflict, instability, and lack of positive support.
- Lack of prosocial recreation and heavy engagement in high-risk settings.
- Employment and education deficits: unstable work history, low satisfaction with legitimate work, and insufficient education or vocational skills.
The first four, often called the “Big Four,” are the strongest predictors. Effective programs prioritize criminal thinking, antisocial associations, and personality characteristics over factors that feel important but contribute less to reoffending risk.
Risk Assessment Tools and Their Limits
Standardized instruments translate the framework into scores that guide decisions. The Level of Service Inventory-Revised (LSI-R) measures ten domains, including criminal history, education, employment, family circumstances, and attitudes, through 54 scored items. The Correctional Offender Management Profiling for Alternative Sanctions (COMPAS) produces both a risk score and a criminogenic needs profile.3SAGE Journals. The LSI-R and the COMPAS – Validation Data on Two Risk-Needs Tools In federal prisons, the Bureau of Prisons uses the Prisoner Assessment Tool Targeting Estimated Risk and Needs (PATTERN), developed under the First Step Act, which classifies each person as minimum, low, medium, or high risk and is reassessed periodically.4Federal Bureau of Prisons. PATTERN Risk Assessment
These tools improve on unstructured professional judgment, but they aren’t neutral. A widely cited 2016 analysis of COMPAS found that Black defendants were nearly twice as likely as white defendants to be incorrectly classified as high risk when they did not go on to reoffend, with a false positive rate of 44.9 percent compared to 23.5 percent. White defendants were more likely to be labeled low risk and then commit new offenses. The tool’s overall accuracy was about 61 percent, roughly the same as untrained volunteers making predictions from basic biographical information.
If an algorithm drives supervision intensity, programming assignments, or release eligibility, biased outputs can entrench the inequities the system claims to address. Some jurisdictions now require transparency disclosures and independent audits of automated decision tools, and there is growing pressure to validate instruments across racial and demographic groups before deployment. A risk score is one input in a structured decision, not a verdict.
Cognitive Behavioral Programs That Work
Cognitive behavioral therapy (CBT) programs are the workhorses of evidence-based corrections. They rest on a simple premise: criminal behavior is driven in part by distorted thinking, and those patterns can be identified, challenged, and replaced through structured practice. A meta-analysis published through the Office of Justice Programs found that CBT-based treatment reduced general recidivism risk by 23 percent and violent recidivism by 28 percent. Among people who completed the full program, recidivism dropped by 42 percent overall and 56 percent for violent offenses.5Office of Justice Programs. A Systematic Review and Meta-Analysis on the Effectiveness of CBT-Informed Anger Management The gap between enrollment and completion is one of the clearest findings in the field. Starting a program helps. Finishing it matters far more.
Thinking for a Change
Thinking for a Change (T4C), developed by the National Institute of Corrections, is one of the most widely adopted CBT curricula in American corrections. The current version (T4C 5.0) includes 25 core lessons that blend cognitive restructuring, social skills training, and problem-solving exercises. Sessions run about two and a half hours each, delivered two to three times per week. Because every group member must practice new skills through role-play before the group moves on, a single lesson often takes two sessions. Facilitators are instructed never to stack lessons back-to-back; participants need time to apply what they’ve learned before taking on another skill.6National Institute of Corrections. T4C 5.0 Introductory Information The program is used in prisons, jails, and community supervision settings across the country.7National Institute of Corrections. Thinking for a Change
Moral Reconation Therapy
Moral Reconation Therapy (MRT) walks participants through 12 steps of moral development using a structured workbook. Exercises focus on confronting distorted beliefs, repairing damaged relationships, and building the capacity to delay gratification. Participants work through steps at their own pace, completing homework between meetings and presenting it for peer feedback. Completing all 12 steps typically requires about 24 group sessions. A meta-analysis of 33 studies found that MRT participants reoffended at a rate roughly one-third lower than comparison groups over a three-year follow-up, with the strongest effects in incarcerated settings.8National Library of Medicine. A Randomized Controlled Trial of Moral Reconation Therapy
Both curricula rely on repeated practice. Participants role-play real conflicts, rehearse alternative responses, and get feedback from facilitators and peers. The structured format matters because it forces facilitators to cover the same ground in the same order rather than improvising.
How Agencies Deliver It Well
The National Institute of Corrections developed eight principles for effective correctional interventions, giving agencies a checklist for translating research into operations.9National Institute of Corrections. Evidence-Based Practices in Corrections Some mirror the RNR model. Others address how programs are actually run day to day.
One of the most counterintuitive principles involves the ratio of positive reinforcement to negative sanctions. Long-term behavior change is most likely when positive responses outnumber punitive ones by at least four to one. Correctional systems have historically operated the opposite way, defaulting to sanctions for noncompliance while taking compliance for granted. Flipping the ratio doesn’t mean ignoring violations. It means that for every consequence imposed, the system recognizes at least four instances of the person doing the right thing. Verbal praise, reduced reporting requirements, extended privileges, and small tangible rewards all reinforce that compliance has value.
Motivational interviewing is another core principle. People ordered into treatment by a court rarely arrive eager to change. Staff trained to resolve ambivalence rather than fight it produce measurably better results. Research on justice-involved populations found programs using motivational interviewing produced a statistically significant reduction in recidivism and higher session attendance than programs without it.10National Library of Medicine. Effectiveness of Motivational Interviewing with Justice-Involved People The technique relies on reflective listening, recognizing a person’s own statements in favor of change, and avoiding confrontation. Certification programs typically require around 38 hours of instruction and demonstration of competency through coded practice sessions.
Graduated Sanctions and Incentives
Evidence-based supervision replaces one-size-fits-all responses with structured decision tools. A graduated sanctions matrix plots the severity of a violation on one axis and the person’s risk level on the other, producing a recommended response ranging from a verbal warning or increased reporting up to short-term detention or revocation. Matrices can include override provisions for stabilizing factors like steady employment or destabilizing ones like an escalating pattern of violations.
Incentive systems work on multiple levels. Low-magnitude rewards include verbal praise for routine compliance, small tangible items like bus passes or phone cards, and recognition during supervision check-ins. Moderate rewards include reduced reporting frequency, extended curfews, modest gift cards, and supervised social events. High-magnitude rewards include lifted travel restrictions, peer mentor roles, help with fines or educational costs, and formal graduation ceremonies. Point systems and randomized prize drawings add unpredictability that behavioral research shows increases engagement. When the four-to-one reinforcement ratio becomes embedded in a structured system rather than left to individual officer judgment, application is more consistent and outcomes are better.
The First Step Act and Earned Time Credits
The First Step Act of 2018 wrote evidence-based practices into federal law for the first time. It required the Attorney General to develop and publicly release a risk and needs assessment system that classifies each federal prisoner by recidivism risk, identifies criminogenic needs, and assigns appropriate programming.11Office of the Law Revision Counsel. 18 USC 3632 – Risk and Needs Assessment System The resulting tool, PATTERN, must be reassessed periodically using dynamic indicators of progress and regression, so a person’s classification can improve over time rather than being locked at intake.4Federal Bureau of Prisons. PATTERN Risk Assessment
The law created a financial incentive for participation. Every eligible person who successfully participates in evidence-based programming earns 10 days of time credits for every 30 days of participation. Those classified as minimum or low risk who maintain that classification over two consecutive assessments earn an additional 5 days, for a total of 15 days per 30 days of programming.12U.S. Sentencing Commission. First Step Act Earned Time Credits The credits apply toward early transfer into prerelease custody or supervised release.11Office of the Law Revision Counsel. 18 USC 3632 – Risk and Needs Assessment System People serving sentences for certain serious offenses are ineligible, and anyone subject to a final order of deportation cannot apply credits toward early release.
Return on Investment
Community supervision costs roughly one-tenth of what incarceration does. In fiscal year 2024, detaining a person before trial and incarcerating them post-conviction was about ten times more expensive than supervising them in the community, accounting for housing, monitoring, treatment, and administrative costs.13United States Courts. The Public Costs of Supervision Versus Detention Even residential reentry centers ran about nine times the cost of community supervision.
The Council of Economic Advisers calculated returns for specific program types. Substance abuse treatment returned an estimated $5.27 for every taxpayer dollar spent, driven by $3.31 in reduced crime costs and $1.96 in lower long-term incarceration expenses. Mental health programs returned $1.47 per dollar, with a 21 percent median reduction in recidivism. Educational programming showed inconsistent results, but the break-even threshold was low: these programs only needed to reduce recidivism by about 2 percent to pay for themselves.14Trump White House Archives. Returns on Investments in Recidivism-Reducing Programs These figures count direct system costs and crime reduction only. They don’t capture the economic value of formerly incarcerated people reentering the workforce, paying taxes, and supporting families.
Fidelity: Why Two Copies of the Same Program Get Different Results
A well-designed program delivered poorly produces poor results. Fidelity, the degree to which a program is actually delivered the way it was designed and tested, is the biggest implementation challenge in the field. When facilitators skip lessons, shorten sessions, or improvise on content, the program being delivered is no longer the program the research validated. Researchers call this “program drift,” and it happens gradually enough that agencies often don’t notice until outcomes deteriorate.
The Correctional Program Checklist, developed by the University of Cincinnati, is the most widely used fidelity tool. It evaluates programs on capacity (leadership, staff qualifications, quality assurance) and content (whether assessment and treatment delivery match evidence-based standards), producing an overall rating from high to low adherence. Programs receiving low ratings get specific recommendations for improvement.
Ongoing data collection keeps fidelity from being a one-time checkbox. Administrators tracking session attendance, completion rates, assessment score changes, and post-program recidivism can spot problems early. Some agencies review recorded sessions to give facilitators direct feedback. The difference between a program that works on paper and one that works in practice almost always comes down to whether anyone is watching closely enough to catch the drift before it consumes the results.