Reform Models & Programs

What the Evidence Actually Shows: Rehabilitation That Reduces Recidivism

The national evidence base for correctional rehabilitation is extensive and consistent. Cognitive-behavioral programs remain the most rigorously evaluated intervention in correctional settings. The Thinking for a Change (T4C) curriculum, developed by the National Institute of Corrections, produced a 23% recidivism rate among participants compared to 36% in matched control groups during a six-month follow-up — a statistically significant reduction that has been replicated across multiple jurisdictions (Evidence-Based Rehabilitation Curricula). Trauma-informed care, mentorship pipelines, and structured cohort models show similarly durable results when implemented with fidelity and continuity.

While the long-term recidivism data from the Bureau of Justice Statistics makes the stakes undeniable — 76.6% of state prisoners released in 2005 were rearrested within five years and 83% within nine years, with 60% of all reoffending arrests occurring in years four through nine — more recent national figures show that roughly 62% of people released from prison are rearrested within three years, and 39% return to prison in that same period (Brennan Center, March 2026). With approximately 450,000 people released from prison each year, designing interventions that produce enduring change is an urgent priority. These numbers are not an argument against rehabilitation; they are an argument for sustained, evidence-based intervention that begins at intake and extends through reentry. The research consensus is that programs work best when they are structured around fixed cohorts, use peer mentorship pipelines, address cognitive distortions and trauma simultaneously, and are tied to measurable milestones rather than seat-time. A DOJ meta-analysis confirms that reentry programs reduce recidivism by about 6% on average, with larger effects when treatment begins in prison and continues into the community.

Program structure matters as much as curriculum content. Models like the Prison Entrepreneurship Program (PEP) in Texas reach more than 6,000 men across 80 TDCJ units annually through a fixed-cohort, tiered format that moves participants from screening through leadership and into post-release mentorship networks (Prison Program Structure Models). The Bard Prison Initiative enrolls 400 full-time students across seven New York prisons in degree-granting programs with near-zero recidivism among graduates. Michigan’s Vocational Village demonstrated that 2019 graduates had a recidivism rate 6.5 percentage points lower than the state’s overall rate that year. The common thread across high-performing programs is institutional commitment: dedicated space, protected program time, staff trained in facilitation rather than custody, and outcome tracking that feeds back into program design. Norway’s approach illustrates what is possible when these principles are fully funded: the country’s reconviction rate has fallen to 18% within two years of release and 25% after five years — down from a pre-reform 60–70%. Bastøy Prison, a low-security island facility holding about 115 men with roughly 69–72 staff members (only 3–5 remaining overnight), achieves a recidivism rate of just 16%, compared to Norway’s national average of 20% (Comparative Solutions Evidence Base).

The Case for Compassionate Release of Aging Prisoners

The evidence on aging prisoners strongly supports expanded use of compassionate and geriatric release. The U.S. Sentencing Commission found that the recidivism rate of older federal offenders (21.3%) was less than half that of offenders under age 50 (53.4%) over an eight-year follow-up. Vera Institute data show that arrest rates drop to just over 2% for ages 50–65 and near zero above age 65. A Massachusetts DOC study of 2019 releases found a three-year recidivism rate of just 10% for women aged 55 and older and 12% for men in the same age group. Despite 45 states and the federal government having compassionate or geriatric release laws on the books, the mechanisms are rarely used because of political and procedural barriers. Meanwhile, the cost differential is stark: the ACLU estimated that releasing an aging prisoner saves states on average $66,294 per year, with a minimum of at least $28,362, and the DOJ found that a federal medical-center prisoner cost $57,962 versus a $28,893 average in 2013. Freeing low-risk elderly individuals through robust release mechanisms thus not only cuts correctional spending but poses minimal public safety risk.

Staffing as the Foundation for Rehabilitation

Adequate staffing is a critical enabler of any rehabilitative effort. Pennsylvania demonstrated that a targeted recruitment strategy can quickly reduce dangerous vacancy rates: its correctional-officer vacancy rate dropped from 10.5% to 4.8% in two years, supported by a dedicated Recruitment and Retention Division that held more than 750 job fairs and events in 2024 alone. Yet filling posts is only the first step; Pennsylvania still saw a roughly 21.6% increase in violent incidents in 2024, underscoring that numbers alone do not change culture. Alabama’s experience illustrates the risk of hiring pipelines that fail: after March 2023 salary increases (trainee starting salary near $57,000, up about $20,000) made officers 28% less likely to resign and saved an estimated $7.9–$10 million in voluntary-turnover costs, average annual hires collapsed by 50%, and correctional-officer staff still declined 55% over nine years. North Carolina continues to report an average officer vacancy rate around 30%, with some facilities near 60%.

Internationally, Norway’s prison system is designed for a 1:1.1 staff-to-inmate ratio and spends $127,671 per year per inmate, compared to roughly $25,000 in the United States. A smaller-scale domestic experiment reinforces the same theme: Pennsylvania’s “Little Scandinavia” unit at SCI Chester operates with a 1:8 officer-to-resident ratio, versus 1:128 in the rest of the facility. The renovation cost about $300,000–$310,000, and per-inmate daily costs are roughly 1.5 times those of double-celling — a relatively modest investment for a model that prioritizes normalized human interaction and programming time, two elements consistently linked to better outcomes.

Decarceration and Public Safety

The most robust finding in the entire evidence base is that decarceration can proceed without increasing crime. Between 1972 and 2009 the U.S. prison population grew nearly 700%, but from 2009 to 2021 it declined 25% even as violent crime reported to police fell to half its 1990s level by year-end 2024. New York more than halved its prison population from 1999 to 2023 while its violent crime rate fell 34%, outpacing the national 28% decline. New York City saw an even sharper pattern: between 1996 and 2014 the serious-crime rate fell 58% while the combined jail and prison incarceration rate fell 55%, with the number of residents behind bars declining by 31,120. The Netherlands reduced its prison rate by approximately 46% from 2005 to 2016, closing roughly half its prisons as crime fell. These examples — alongside New Jersey and California — show that population reductions driven by policy and rehabilitation do not compromise public safety when implemented thoughtfully.

Reducing Solitary Confinement While Improving Safety

The data on solitary confinement reinforces the same theme: reducing isolation does not cause violence explosions, and in many cases improves safety. North Dakota achieved a 74.28% reduction in the use of solitary confinement between 2016 and 2020, including a 99% drop in monthly solitary sanctions at one facility (JRCC) and a 59.1% drop at the state penitentiary. Oregon’s Resource Team approach, which engages people with extensive solitary histories (average 9.7 prior admissions), cut disciplinary infractions by 55.7% and assaults by 73.9% among those with at least three interactions. Staff use-of-force in Oregon’s Behavioral Health Unit fell nearly 86% from 2016 to 2021. California’s Ashker settlement moved more than 1,512 people out of solitary and cut Pelican Bay’s long-term isolation population from 513 to just 2 — a 99.6% reduction — with no reported violence surge. Statewide, California’s total SHU population fell 65% between December 2012 and August 2016 (from 9,870 to 3,471) and continued dropping to 594 by June 2018. These outcomes align with the fiscal reality: solitary confinement requires significantly more staff resources per incarcerated person, and avoiding the litigation costs that flow from isolation-driven harm — like the $4.55 million in plaintiffs’ attorney fees from the Ashker case — is a concrete financial benefit.

Independent Oversight and Accountability

Rehabilitation-focused reform also requires institutional accountability to ensure that programs operate in safe, transparent conditions. Independent corrections oversight bodies, such as New Jersey’s Office of the Corrections Ombudsperson, provide a cost-effective model. For about $2.8 million per year and 26 staff, New Jersey’s office oversees 9 state prisons and roughly 13,000 incarcerated people, with subpoena power, unannounced inspection authority, and confidential, privileged communications, all advised by a citizens’ board. Similarly, Washington’s Office of the Corrections Ombuds oversees 11 state prison facilities incarcerating about 13,075 people. These modest investments — less than the cost of a single conditions settlement — can catch systemic failures before they escalate. By contrast, HM Inspectorate of Prisons in England and Wales published 64 reports in 2024–25 and found that 30 of 32 closed prisons were rated poor or insufficiently good for purposeful activity; positive random drug-test rates frequently topped 30%, and violence rose 55% at Lowdham Grange, illustrating what happens when monitoring is absent or ignored. In the U.S., the false-positive rate of colorimetric field drug tests — as high as 33% in the Colorado DOC and up to 38% in some contexts — coupled with the fact that 89% of surveyed prosecutors accept guilty pleas without confirmatory lab testing, means that roughly 30,000 people are falsely implicated annually. These errors can funnel individuals into punitive settings that disrupt program access and erode trust in the system. Independent oversight, fair disciplinary procedures, and accurate evidence are therefore not separate from rehabilitation — they are foundational to maintaining an environment in which rehabilitative programming can take root and succeed.