Reform Models & Programs
Where this research is used
This topic brings 91 research collections together into one overview. GPS puts it to work in three ways:
- Intelligence topics. GPS's investigative pages on Georgia's prisons build on this research. These draw on it most directly:
- GPS articles. We use this research when we write our own articles. 111 articles connected to this topic are listed below.
- Lighthouse. Lighthouse, GPS's public AI assistant, uses this research and our intelligence topics to answer people's questions.
Key Findings
Critical data points synthesized across multiple research collections. Tap a card for the full finding, GPS commentary on why it matters, and its source.
GDC's Inmate Services Impact Report (edition covering Fiscal Year 2021 releases) reports three-year felony reconviction rates of 26.06% for the general population, 20.98% for educational programming, 13.64% for vocational programming, 22.93% for cognitive programming, and 27.77% for residential substance abuse treatment.
The recidivism rate for vocational program completers in Georgia is 13.64%, roughly half the general rate of 26%.
Lower recidivism among people who finish vocational programs marks one of Georgia's own programs with a measurable result, while the state's direct spending on its programs amounts to a fraction of one percent of the prison budget.
Our reading of the finding above. It is not part of the source.- Research Compilation: Georgia Incarceration Trends & Population Data
GDC's FY2025 Annual Fiscal Report states that during FY25, the EBP model was expanded to McRae Women's Facility, bringing total offender enrollment in EBP to 2,917.
RAND's meta-analysis found that, based on the higher-quality research studies, inmates who participated in correctional education programs had a 43 percent lower odds of recidivating than inmates who did not, out of 267 primary empirical studies of which 58 met all three eligibility criteria.
GDC's press release states that the Lee Evidence Based Prison's Peer-to-Peer learning site saw a 62% reduction in disciplinary reports (DR) from December 2015 to February 2016.
The administration of Governor Gavin Newsom is spending approximately $239 million to remake San Quentin State Prison into a Scandinavian-style 'rehabilitation center.'
The Rehabilitation That Does Not Exist
Georgia's prison system spends well over a billion dollars a year and lists rehabilitation among its core functions. The most authoritative outside assessment of what that money buys arrived in 2024, when the U.S. Department of Justice concluded that educational and vocational programming in Georgia prisons had been 'slashed rather than expanded,' and that conditions in most facilities were so chaotic and violent that meaningful programming participation was effectively impossible (#5256, #2125). The same investigation found that understaffing left prisons without enough staff 'to prevent or even respond to the most blatant gang activities, let alone provide programs' (#5261).
The appropriations moved in the opposite direction from the programs. Georgia Department of Corrections State General Funds rose from $1.07 billion in FY2020 to $1.42 billion in FY2024, according to the Governor's Budget Report covering amended FY2026 and FY2027. Yet no dedicated line items for comprehensive reentry programming, transition planning, or post-release support services are visible in publicly available department budget documents (#2111). The agency is funded as a confinement operation and asked, in mission statements, to behave like something else.
Individual programs have also proven fragile. Georgia State University shut down its prison education program in March 2024 — a program that cost $180,000 a year and served 60 students (#5257). The institutional architecture of reform has weakened too: the Georgia Council on Criminal Justice Reform, created under Governor Nathan Deal in 2013 and once a national model for evidence-based policymaking, saw its influence wane after Deal left office in 2019 (#3891). The earlier Justice Reinvestment Initiative did produce measurable movement — a 6% reduction in the prison population (#2112) and a 10% decline in commitments to prison and probation for affected offenses, from 28,604 in 2012 to 25,784 in 2015 (#14330) — but that machinery is no longer driving programming dollars.
What the Evidence Actually Says Works
Outside Georgia's walls, the evidence base for correctional programming is unusually deep — and unusually specific about effect sizes. Meta-analyses consistently show cognitive-behavioral programs reduce recidivism by 20–30% relative to control conditions, with the largest effects for higher-risk participants and high-fidelity implementation (#4832, #13637). The Campbell Collaboration review by Lipsey, Landenberger and Wilson found average reductions of roughly 25–30% across well-implemented programs, and — a finding that matters for procurement — that brand-name curricula such as Reasoning and Rehabilitation, Moral Reconation Therapy and Thinking for a Change do not outperform generic cognitive-behavioral treatment (#7385).
The strongest single body of evidence concerns education. The RAND Corporation's 2013 meta-analysis found that people who participated in correctional education had 43% lower odds of recidivating than those who did not (#30698, #4845), translating into a 13 percentage-point reduction in the risk of reincarceration (#30700, #29049), 13% higher odds of post-release employment (#29046, #30924), and roughly $5 in avoided reincarceration cost for every $1 spent (#30700, #4847). Average program cost per participant was $1,400 to $1,744 a year (#29051). Washington State's Institute for Public Policy puts the benefit-cost ratio for postsecondary correctional education at $14.51 per dollar, vocational education at $4.18, and combined education at $8.04 (#29064, #29065, #29066). The National Institute of Justice's CrimeSolutions registry rates postsecondary correctional education and corrections-based vocational training both 'Promising' (#28981, #28978).
Later and more skeptical syntheses pull the numbers down. Sprick Schuster and Stickle's meta-analysis of 750 papers found that any form of educational participation reduces the likelihood of recidivism by 14.8% overall, but only 11.3% when restricted to the highest-quality studies, and about 7% for studies published since 2010 — with college education producing the largest effect at 27.7% and adult basic education the smallest at 6.3% (#34499, #34500, #34501, #34502, #34505). The reentry field is weaker still: a Department of Justice meta-analysis found reentry programs reduce recidivism by about 6% on average (#7710), an NIJ evaluation of seven Second Chance Act demonstration programs found no reduction at all in re-arrest, reconviction or reincarceration over 30 months (#28977), CrimeSolutions rates the Transitional Jobs Reentry Demonstration 'Ineffective' (#28979, #28980), and NIJ itself cautions that no reentry program yet meets the evidence-based standard through replicated randomized trials (#7712). The Risk-Need-Responsivity framework that organizes all of this holds that services should target higher-risk people, address criminogenic needs, and match intervention style to the individual — and that low-risk people can be harmed by intensive programming (#4854).
Georgia's Program Inventory and What Its Own Outcomes Show
Georgia's flagship vehicle is the Evidence Based Prison model: a two-year program, codified in a policy effective February 20, 2025, that requires participants to have two to five years remaining to their tentative parole month or maximum release date (#34312, #34308, #34313). In FY2025 the department reported expanding the model to McRae Women's Facility, bringing total enrollment to 2,917 people (#34260); Hancock State Prison alone accounts for 1,007 of them (#23729). Before that expansion, five facilities were designated and operating as evidence-based prisons — Lee, Hancock, Smith, Coastal and Autry — with Washington and McRae slated to join (#34244). The Lee program, which began in June 2015, is described by the department as a two-year, four-phase program that 'encourages offenders to change the way they think' (#34267).
Beneath that umbrella sit the cognitive-behavioral curricula — Moral Reconation Therapy and Thinking for a Change are the department's stated core programs at state prisons, private prisons and the Metro Reentry Facility (#13705) — plus 12 Residential Substance Abuse Treatment programs across prisons and probation treatment centers (#70), funded in part by $5.7 million in prison savings reinvested after the 2012 reform package (#30670). RSAT is defined as a nine-month modified therapeutic community combining cognitive-behavioral skill-building with pro-social and pro-sober recovery (#34349). The Faith and Character-Based Initiative, implemented in 2004, now operates at 18 sites, with a 12-month dormitory curriculum and a 24-month prison curriculum (#34111, #34124).
The department's own reconviction tables are the most useful outcome data it publishes, and they are more ambiguous than the program brochures. For the FY2021 release cohort, the three-year felony reconviction rate was 26.06% for the general population, 20.98% for educational programming completers, 22.93% for cognitive programming completers, 13.64% for vocational programming completers, and 27.77% for residential substance abuse treatment completers (#28961). The FY2017 cohort shows the same shape: 28.65% general, 24.93% cognitive, 22.61% educational, 20.17% vocational, and 29.02% for RSAT (#28944). Vocational completion is consistently associated with the largest apparent advantage — a 13.64% reconviction rate, roughly half the general rate (#3358, #2124) — and RSAT completion with the smallest, a result the 2024 Senate study committee attributed to the severity of untreated addiction among participants (#7560).
These are completer rates. They compare people who finished a program to everyone else, including people who never enrolled and people who dropped out, and they carry no control for self-selection; RAND's own meta-analysis identifies exactly this limitation as the central weakness of the correctional education literature (#30926). Georgia's earlier experiments show both the promise and the ceiling: the Reasoning and Rehabilitation curriculum produced statistically significant treatment effects of 17% on returns to prison and 19% on re-arrests among class completers in Phase I of the Georgia Cognitive Skills Experiment (#16342, #16343), but the overall 4% difference between experimental and control groups was not statistically significant (#16344), and the hypothesis that high-risk participants would benefit most was not supported (#16390).
Peer Mentoring: Georgia's Most Promising and Least Evaluated Experiment
Peer mentoring is where Georgia has done the most in-house invention. The R.I.S.E. — Reinforcing, Instructing, Supporting, and Encouraging — statewide mentoring program was established by standard operating procedure effective August 30, 2019, as a voluntary initiative to identify and train incarcerated mentors (#34245, #34251). Eligibility is stringent: a high school diploma or GED, enrollment in or completion of the Pathfinders Program, no high disciplinary report for at least three years, no moderate report for one year, no documented gang activity for five years, and active engagement in a case plan (#34246). Mentors are unpaid, participation is 'completely voluntary,' and it is expressly 'not a grieveable issue' (#34253). By FY2021, 182 people statewide had been identified as trained R.I.S.E. mentors (#34178), against a departmental target that 10% of the evidence-based prison population serve as mentors (#34309). Separately, 1,196 people were working as education, chaplaincy and other programming assistants and leaders in 2016 (#34345).
The Forensic Peer Mentor program, run by the Georgia Mental Health Consumer Network with the Department of Behavioral Health and Developmental Disabilities and the Department of Corrections since 2014, is the most carefully tracked variant (#34216, #34281). It requires a Certified Peer Specialist or CARES credential plus a 40-hour, five-day forensic training (#34220, #34385). As of April 2023, 14 training sessions had produced 240 new forensic peer mentors, each able to work with about 20 people at a time (#34386). In 2021 the program served 623 returning citizens with behavioral health needs across prisons, day reporting centers and mental health courts (#34367). A DBHDD annual report recorded 898 people served with two re-arrests and zero reconvictions, zero parole or probation revocations, and zero psychiatric readmissions (#34238) — striking numbers that the program's own advocates would concede are not a controlled comparison.
Georgia's program-level operational data are similarly encouraging and similarly limited. The department reported a 62% reduction in disciplinary reports at the Lee Evidence Based Prison's Peer-to-Peer learning site between December 2015 and February 2016 (#34172, #34388), and 2,485 completions in FY2024 of the TOPUCU orientation program delivered by peer mentors in intake dormitories (#34299). The caution comes from the systematic review literature. Bagnall and colleagues found most included studies were of poor methodological quality, found only weak evidence that peer mentoring improved health behaviors, treatment adherence, drug taking or re-offending (#34464), and found that the strongest and most consistent evidence concerns benefits to the peer worker rather than the recipient (#34463) — with negative health effects reported for prisoners trained as Listeners and befrienders because of the emotional burden of absorbing other people's problems (#34465). Britain's Justice Data Lab, analyzing 395 St Giles Trust peer advisors, found a reoffending rate 3.3 percentage points lower than comparable offenders and reoffences occurring 30 days later on average, but cautioned that for every 100 advisors the program may decrease proven reoffenders by up to 7 or increase them by up to 1 (#34475, #34477).
National Models: California, Scandinavia, and What Travels
The most-copied reform framework in American corrections is now the California Model, which the California Department of Corrections and Rehabilitation describes as built on four pillars: dynamic security, peer mentorship, normalization, and becoming a trauma-informed organization (#34540). Peer mentorship is defined as training incarcerated people 'to use their lived experiences to mentor and support their peers' (#34542); normalization aims 'to bring life in prison as close as possible to life outside of prison' (#34543). California has put real money behind the branding: approximately $239 million to remake San Quentin into a rehabilitation center (#6065), an institution designated the flagship of the model since its 2023 transformation (#34845).
The outcome evidence for the Scandinavian-adjacent approach is strongest where it has been tested experimentally. A three-year randomized controlled trial of the Vera Institute's Restoring Promise, run with the South Carolina Department of Corrections, found that participants were 73% less likely to be convicted of a violent infraction (#33236) and had 83% lower odds of a restrictive housing stay (#7757). Michigan's 'Little Scandinavia' unit, opened in May 2022, reported almost no violent episodes in 2024 even as violent incidents rose 22% in other Michigan prisons (#33268). Pennsylvania announced expansion of its Scandinavian-inspired approach to three new facilities after a randomized study at SCI Chester showed promising early results (#33359). Oregon's resource team model was associated with a 73.8% decline in mean quarterly use-of-force incidents in special housing units and a 23% drop in staff sick leave (#34949, #34948). A 2022 study of Italian prisons found that switching from closed-cell to open-cell regimes for a year reduced recidivism by about 6 percentage points, with the largest effects for people with low education (#18998).
What travels least well is the evaluation culture. California's Legislative Analyst's Office states that the department 'generally lacks robust evaluations of the actual cost-effectiveness of its parole rehabilitation programs' (#28991), and the Sacramento Bee reported that staff buy-in remains the 'biggest obstacle' to rolling out the California Model (#6079). The same caution applies to Georgia's ability to import any of it: a model that depends on officer role change, uncrowded housing units and sustained program funding is being proposed to a department that DOJ found could not staff basic programming (#5261).
Turning Programs Into Release: Earned Time and the Incentive Gap
A program that carries no consequence for release date is a class, not an incentive. Georgia does operate one such mechanism: the Performance Incentive Credit program, under which a person can accumulate up to twelve months of credit off their length of stay through educational, vocational and treatment programming, work details and good behavior (#34537). The credit is real but discretionary — the Parole Board determines which parole-eligible people who have earned credit may be released earlier on parole (#12140) — and the department frames it as preparing people for success while improving use of prison beds at a cost saving to taxpayers (#11672).
Other states have gone considerably further. Kentucky allows a credit of up to 90 days for each completed drug treatment, evidence-based or life skills program approved by the department (#30886). Illinois multiplies good time by 1.5 for successful completion of education, vocational, treatment and work programs (#31142). Louisiana's Certified Treatment and Rehabilitation Program allows up to 540 days of credit in total, including 180 days for workforce development work release (#30922); 5,185 people released in 2023 earned the credit, averaging 227.9 days each, and 45.8% of all good-time releases earned some (#30919, #30920, #30921). California's statute defines approved rehabilitation programming broadly — academic, vocational, vocational training, anger management, social life skills and substance abuse programs (#30826).
The release-side reforms that produce the largest population effects are not program credits at all. Mississippi's presumptive-parole statute requires release to supervision on the parole eligibility date without a board hearing if the person meets case plan requirements, no victim has requested a hearing, there is no serious violation report in six months, and a discharge plan is approved (#30062). California's Elderly Parole Program makes anyone 50 or older who has served 20 continuous years eligible for suitability review (#29898, #29903). Georgia has no equivalent presumptive or age-based mechanism in the record assembled here; its earned-time debate remains confined to program credits that the Board may or may not choose to honor.
What the Numbers Do Not Show
The largest gap in Georgia's rehabilitation record is not a missing program. It is the absence of any published evaluation capable of separating a program's effect from the characteristics of the people who complete it. Georgia publishes completer reconviction rates (#28961) but no matched comparison, no intention-to-treat analysis and no cost-effectiveness estimate. This is a national pattern rather than a Georgia peculiarity — California's Legislative Analyst's Office says the same of its parole programs (#28991) — but it is more consequential in a state that spends $1.42 billion in general funds a year on corrections.
The evaluation literature explains why the favorable numbers should be read cautiously. RAND's meta-analysis states that the dearth of experimental designs makes causal inference difficult and that people who elect to participate in education may differ in unmeasured ways from those who do not (#30926). The GRIP program's academic evaluation notes its sample was not statistically representative and was formed through self-selection that may have significantly influenced results, and recommends a future random-assignment study (#34972). The Prison University Project evaluation identifies the same problem — identifying an appropriate comparison group is the primary challenge when participation is voluntary (#34975). A commentary on the Campbell review of therapeutic communities observed that only two evaluations used random assignment, leaving the rest vulnerable to selection bias (#34485).
There are also gaps in the data Georgia simply does not publish. The department reports a 41% reduction in its extended restrictive housing population since January 2017 — 671 people — without ever stating the population before or after, a delta with no endpoints (#19298). Its seven-facility Tier I and II designation rests on fact sheets from 2018 and 2019 that may be stale (#19366), and its own procedures assert that offender records are 'confidential state secrets' under O.C.G.A. § 42-5-36(c), routing individual-record requests to legal services (#19296). The No Turning Back Recovery Program was reported in 2007 as still tracking participants to determine whether they completed parole successfully, meaning post-program outcome data were not then available (#13003). The record is not that Georgia's programs fail. It is that Georgia has arranged its data so that no one outside the agency can say whether they succeed.
Related Topics
Explore related areas of research.
Contributing Collections
Research collections that contribute data to this topic.
Sources
100 cited sources across all contributing collections.