Reform Models & Programs

The Gap Between the Model and the Institution

On paper, the Georgia Department of Corrections runs a rehabilitation apparatus. Its Reentry and Cognitive Programming unit lists Moral Reconation Therapy and Thinking for a Change as core cognitive-behavioral programs at state prisons, private prisons, and the Metro Reentry Facility (#13643, #13705). It operates 12 Reentry and Cognitive Programming centers with 2,344 transitional beds, 346 of them designated for women, providing roughly six months of work release before parole (#13683, #4957, #4975). Its "Evidence Based Prisons" model enrolled 1,007 people at Hancock State Prison alone in FY2025, part of 2,917 systemwide (#23729). Its own written policy, SOP 107.14, requires quality assurance evaluations at least every two years confirming programs are "delivered in the manner in which it was designed," plus a Georgia Program Assessment Inventory scoring compliance with evidence-based principles and corrective action plans for critical findings within 30 days (#13707).

Then there is the money. Georgia's state general fund appropriation for the Department of Corrections in FY2025 was $1.42 billion (GPS budget dataset, Governor's Budget Report, Amended FY2026 and FY2027, p. 152). The statewide vocational education budget for FY2025 was $172,000 — $3.44 per incarcerated person (#5252). Prison education as a whole represents 0.11% of what one GPS collection describes as the $1.8 billion corrections budget, or about $40 a year per person, $0.11 a day (#5266). Publicly available budget documents carry no dedicated line items for comprehensive reentry programming, transition planning, or post-release support (#2111), though the FY2026 amended and FY2027 budgets do contain small facility-level programming adds: $150,000 for a peer-led pilot at Autry State Prison, $93,179 (FY2026) and $39,786 (FY2027) for additional programming at Metro Reentry Facility (GPS budget dataset). Set against the $50,067,051 the state appropriated for Transition Centers in FY2027 (#5573), the largest of those is well under half of one percent.

The federal government's 2024 findings describe what that combination produces. The Department of Justice investigation found educational and vocational programming had been "slashed rather than expanded," and that conditions in most facilities were so chaotic and violent that meaningful participation in programming was effectively impossible (#2125, #5256). Investigators put the cause partly in staffing: "prisons do not have enough staff to prevent or even respond to the most blatant gang activities, let alone provide programs" (#5261). Marquis Reshawn Jefferson died May 11, 2022, at Washington State Prison; GPS's registry lists the cause as homicide and his age as 26. Reporting on his death found the prison was so understaffed that no one was watching the dorm when he was attacked — other incarcerated people carried his body to the door before staff noticed (#1308). A program model cannot operate in a building where that is possible.

There is at least one documented counter-example. The same Department of Justice findings identified Walker State Prison as the exception among 17 facilities visited: more security staff positions filled, more rehabilitative programming, less violence and contraband, and no reported homicides in several years (#13717, #6658). Walker's program list includes Motivation for Change, Matrix, Gateway, Residential Substance Abuse Treatment, and Prison Fellowship Mentoring (#13706). The contrast matters because it locates the failure in delivery conditions rather than in the model itself. So does institutional fragility: Georgia State University shut down its prison education program in March 2024 — a program that cost $180,000 a year and served 60 students — in the same period that federal Pell Grants for incarcerated students were restored nationally on July 1, 2023, with programs expanding in 44 states (#5257, #5267).

What the Evidence Supports — and Where It Stops

Cognitive-behavioral programs are the most replicated intervention in corrections. Meta-analyses consistently find 20–30% reductions in recidivism against control conditions, with the largest effects for people at higher risk and where implementation fidelity is high (#4832, #7384). A 58-study meta-analysis by Landenberger and Lipsey found mean recidivism 25% lower in treatment groups — .40 in controls against .30 in treatment (#7383). The Lipsey, Landenberger and Wilson Campbell review put the average at roughly 25–30% and found that brand-name curricula — Reasoning and Rehabilitation, Moral Reconation Therapy, Thinking for a Change — do not outperform generic cognitive-behavioral treatment (#7385). The Risk-Need-Responsivity framework that organizes the field holds that services should target higher-risk people, address criminogenic needs, and match intervention style to the individual, and warns that intensive programming can harm low-risk people (#4854). Best practice gives high-risk people multimodal, longer-term, intensive services and low-risk people minimal exposure (#4969).

Education carries the strongest benefit-cost case. RAND's 2013 meta-analysis found correctional education participants had 43% lower odds of recidivating — a 30% versus 43% three-year rate, a 13-percentage-point absolute difference — and 13% higher odds of post-release employment (#4845, #7373, #4846). Every $1 spent on correctional education returns $4–$5 in avoided reincarceration (#4847); direct program costs run $1,400–$1,744 per person and savings run $8,700–$9,700 per participant over three years (#1218, #1219). Even partial participation reduces recidivism compared with none (#1251). Vocational education is associated with a 15.6% decrease in recidivism odds and 28% higher employment odds (#4850, #4840). Postsecondary education has the largest per-person benefit of any program — roughly $16,908 per student in a 2023 American Journal of Criminal Justice meta-analysis, against $3.05 returned per dollar for vocational education (#13635, #13634) — and prison-earned bachelor's degrees produced 42% higher callback odds than GEDs in a 2024 audit study (#4848). The Washington State Institute for Public Policy ranks correctional education, especially postsecondary, as the highest-return adult corrections program (#13638). The Vera Institute of Justice estimated that expanding postsecondary education in prisons would save states a combined $365.8 million a year (#13650).

The specific curricula in Georgia's inventory sit inside this literature. Moral Reconation Therapy moves participants through 12 progressive steps in an open-ended group format, and its evidence base runs to more than 200 published outcome studies, with effects reported for periods as long as 20 years (#4829, #4827, #4828). Thinking for a Change is 25 scripted lessons delivered over 12–30 weeks; a 2009 evaluation found 23% recidivism among participants against 36% in a control group at six months (#4826, #4825). Reasoning and Rehabilitation is 35 sessions across nine modules, with a meta-analytic reoffending reduction of about 14% and a shortened 16-session R&R2 version (#4859, #4831, #4860). Trauma-specific programming is thinner but real: Seeking Safety, a present-focused treatment for co-occurring trauma and substance use, has more than 40 published outcome studies and can be delivered by paraprofessionals (#4839); SAMHSA's six principles of trauma-informed care are the common framework (#4858), and trauma-informed care is universal across the successful California programs examined in GPS's California collection (#4924). Restorative justice reduces recidivism with stronger effects for violent offenses, and victim satisfaction runs 80–90% (#4837, #4836).

The honest reading of the evidence has to include its edges. The National Institute of Justice cautions that no reentry program yet meets the "evidence-based" standard through replicated randomized trials, and that results are mixed (#7712); a Department of Justice meta-analysis put the average reentry effect at about 6%, larger where treatment begins in prison and continues into the community (#7710). RAND itself flags weak designs and selection effects in the studies beneath its 43% figure (#7376). The Texas InnerChange Freedom Initiative shows the trap: graduates had lower rearrest and reincarceration rates, but the original evaluation counted only graduates — about 75 of 177 starters — and once dropouts were included the program group did no better than the comparison group, and on some measures worse (#7386). Mentoring carries its own warning: MENTOR's Elements of Effective Practice, 16 elements covering design, screening, training and monitoring, states plainly that poorly implemented mentoring can be harmful (#4863, #4855), and the credible-messenger evidence, while promising, is largely correlational (#7716). Arts programs reduce in-prison infractions; recidivism claims for them rest on weaker designs (#7390). The Good Lives Model, which underlies much desistance-oriented programming, still lacks peer-reviewed evidence sufficient to confirm its effect on recidivism (#7400, #7413).

Georgia's Reform Archive, 1972–1987

Georgia has run this experiment before, more than once, and the evaluations survive. The Youthful Offender Act of 1972 created an indeterminate sentence of zero to six years whose termination depended on the person's assessed treatment needs (#13814). It was operationalized through a written contract signed by the incarcerated person, an institutional representative, the Youthful Offender Division, and the Parole Board (#13866), supported by a Performance Plan and a Performance Activity Description specifying what had to be completed to earn release (#15096, #14474). The theory of change was explicit: make release contingent on completing a specific set of programs and services (#15410). A companion demonstration, the Motivation in Offender Rehabilitating Environments Project, opened August 1, 1972, at the Georgia Training and Development Center in Buford using a point-based token economy in which passing a test at a given level earned points negotiable for privileges (#13815, #13816) — the ancestor of the modern, differently named M.O.R.E. program.

The results were uneven and then routinized. Earned Time participants sustained academic program completion rates of 21% to 27% across four quarters, with a first-year cohort climbing from 17% to 39% (#14951). The evaluation also recorded structural problems: time off the end of a sentence may be too far away to function as a reinforcer (#13822); removing good time in an automatic-award system becomes an aversive control procedure with little regard for its effect on rehabilitative programs (#13860); and justice-model critics argued for determinate sentencing, an end to coerced treatment, and voluntary program participation (#13820, #13861). By 1978 evaluators concluded the Youthful Offender Program had become "routinized in a traditional prison mold": contracting was restricted to programs the prisons already ran (#15415), vocational counseling was oriented toward satisfying contract stipulations rather than long-term career goals (#15534), and the program "appears to lack the envisioned reforms in correctional practice" (#15519, #15402). GII Vocational Rehabilitation clients returned to prison at 12.5% within one year, 22.1% within two, and 27.3% within three (#15470, #15474, #15477); one-year returns rose from 10% to 13% to 20% across the FY1974–1976 exit cohorts, a change the report called "perhaps the most alarming" (#15480), and the specialized programs were described as "losing ground" against the rest of the prison population (#15483). The 1978 evaluators recommended implementing the Act rather than repealing it (#15523); the Act was eventually discontinued (#18303).

Intensive Probation Supervision, launched in 1982, is the archive's clearest success. It began as a pilot in 13 of Georgia's 45 judicial circuits and had expanded to 33 circuits by the end of 1985 (#13971, #13972, #13973). Of 2,322 people admitted between 1982 and 1985, 1,952 were successfully diverted from prison and 16% were terminated and returned to prison for technical violations or new crimes (#13974, #13979, #13990). Sixty-eight percent were still on probation under IPS or regular caseloads and 15% had completed their sentences (#13987, #13988). People convicted of drug offenses did better under IPS than under regular probation, with a 90% success rate in the 18-month follow-up (#14043), even though the program initially considered discouraging their admission (#14086). Risk score was the strongest predictor of failure (#14046). The evaluation's most durable finding is a design constraint rather than an outcome: "the near impossibility of separating treatment from enforcement" (#14062). Judges became the program's strongest supporters because it increased local control and because an officer could be reached who had recent, direct contact with the person supervised (#14074, #14089).

Shock probation and split sentencing round out the period. Georgia operated shock probation programs in three correctional institutions, most often by taking a probationer on the verge of revocation through a mock entry into the institution so the person could see what imprisonment actually looks like (#14419, #19178). A 1978 report catalogued the five functions Ohio's parole authority attributed to the device — impressing seriousness without a long sentence, allowing release of people the institution finds more amenable to community treatment than the court expected, compromising between punishment and leniency, protecting against socialization into the prison's harder culture, and permitting community treatment while still imposing a deterrent sentence (#14673) — and recommended expanding Georgia's programs into additional institutions (#14433). It also recommended that evaluative mechanisms be built into every split-sentence and shock-probation program created or modified, so effectiveness could be measured (#14435). The record reviewed here contains no subsequent evaluation of Georgia's shock probation programs.

The Cognitive Skills Experiment, 1997–2001

Georgia's best-documented rehabilitation program is also its most instructive. In 1997 the Georgia Board of Pardons and Paroles began training parole officers as cognitive coaches and implemented Ross and Fabiano's Reasoning and Rehabilitation curriculum — the first known statewide effort to use cognitive skills training with people on parole (#19224, #16295, #16294). At its peak the agency reported 1,050 participants in cognitive skills programs in FY2005, with 135 active coaches who were parole officers or chief parole officers (#10702); 30 officers were trained to teach Reasoning and Rehabilitation in FY2002 alone, alongside 22 coaches trained for Thinking for a Change (#13495). The department also delivered a 60-hour intensive version described as building skills in anger management, decision-making, and communication (#13493), and reported an 18% recidivism reduction for class graduates against control groups (#11128, #11109).

The controlled evaluation is the part worth reading closely. Phase I used discrete-time survival analysis controlling for background characteristics and found statistically significant treatment effects of 17% on returns to prison and 19% on re-arrests for class completers compared with a control group (#16342, #16343). But the same Phase I analysis found that the 4% lower return-to-prison and re-arrest rates for the experimental group as a whole were not statistically significant (#16344). That gap between "completers" and "everyone assigned" is the same methodological fault line that sank the Texas InnerChange evaluation (#7386), and it is the single most important number on this page to read correctly: the 17% figure describes people who finished the course, not people who were sent to it.

Implementation data explain part of the distance. Of the Phase I experimental group, 59.6% completed the course; in Phase II, 62.7% did — 382 people (#16281, #16279, #16280). Among those who actually began, completion ranged from 33% to 100% across program groups, with a mean of 72.0% and a median of 73.2% (#16282, #16283, #16284). The curriculum called for classes of four to six members (#16384, #16326); actual groups formed by random assignment ran from six to 28, with a typical class of 11 (#16553). Observers rated class structure less than satisfactory in 41.3% of program groups at session 17 (#16286), though coaches met prosocial interpersonal expectations in 78.3% of groups at that point and 91.1% by session 30 (#16288, #16289). Screening was supposed to exclude people with IQ scores below 80, histories of sexual offending, and debilitating substance abuse (#16325, #19230); despite that, 10.2% of the experimental group were inappropriate for the program, and if the 299 low-risk participants are counted, over half the group — 54% — could be considered inappropriate (#16547, #16548).

What the experiment actually demonstrated is more interesting than a headline recidivism figure. On the one hand, fidelity was unusually high: the evaluators concluded the program was "exceptional in correctional programming history" because it rated high on adherence to modeling, reinforcement, practice, and interpersonal sensitivity, strongly suggesting Reasoning and Rehabilitation was administered according to its underlying theoretical design — something most correctional programs fail (#16561). Reported use of taught skills rose from 71.0% of participants at time 1 to 94.3% at time 5, and the increase was not an artifact of attrition (#16512). On the other hand, two core predictions of the Risk-Need-Responsivity model failed. Analysis did not support the hypothesis that high-risk participants would benefit more than low-risk participants (#16390), and the Phase I outcome study found that low-risk people on parole also benefited from the program (#16391) — the opposite of the standard warning that intensive programming can harm low-risk people (#4854). The evaluators' own recommendations were practical: adopt attendance and completion requirements, since completion is a key component of program success and Phase II completion was only marginally better than Phase I (#16320); reduce group size to the recommended maximum of six (#16319); and improve how coaches structure sessions (#16321).

The Reentry Machinery: Housing, Treatment, and Transition

Georgia's reentry programs are real, numerous, and small relative to the population they serve. The Max-Out Reentry program, run through Department of Corrections transitional centers with specialized parole officers assigned to implement it, is open to people within 12 to 18 months of their maximum release date (#9681, #9860). In FY2014, 494 people were placed, 326 were active as of June 30, 161 of them referred by the Parole Board, and 42 people previously denied parole were granted parole after referral; 241 completed the program and were released at the end of their court-ordered term (#9593, #9594, #9596, #9598, #9600). Participants must work, with a portion of earnings covering program costs, and non-compliant participants can be returned to a secure prison (#9759, #9760). In FY2013, the program's first seven months of operation, 367 people passed through it; 71 failed and were returned to a secure facility (#12472, #12471).

The state's umbrella reentry framework, the Georgia Prisoner Reentry Initiative, launched in November 2013 and was formally created in 2014, organized around three phases — Getting Ready, Going Home, and Staying Home — with a mission of a seamless plan of services from prison entry through aftercare (#9707, #14366, #9861, #9708). Its stated goal was a double-digit reduction in recidivism, to be achieved by examining assets and barriers in risk and need assessment, programming, housing, and employment (#9709). The record reviewed here contains no outcome evaluation of the initiative.

Residential addiction treatment shows the clearest throughput trend, and it is downward. No Turning Back, which accepted its first participants in February 2005 at the Whitworth Parole Center, is a five-month, 60-slot program for people on parole assessed as both high-risk and high-need, delivered under contract with Spectrum Health Systems using a research-based curriculum (#10912, #10913, #12665, #10124, #12966). Reported admissions and graduations include 146 admissions and 131 graduates in FY2010 (#10186, #10187), 160 and 147 in FY2011 (#9965, #9966), 156 admissions with 159 graduates and 19 unsuccessful discharges in FY2012 (#10350, #10351, #10352), 115 admissions, 74 graduates and 23 discharges in FY2013 (#12501, #12502, #12503), and 195 admissions, 154 graduates and 44 discharges in FY2014 (#9630, #9631, #9632). Its successor at Treutlen Probation Detention Center, the New Direction Recovery Program, is used by the Parole Board as an alternative to revocation for people who failed community-based treatment and supervision (#11552). It reported 115 referrals and 116 graduates in FY2019, its first full year (#12306), 127 referrals and 120 graduates in FY2020 (#12081, #12082), then 36 referrals in FY2022 (#11485), 55 referrals and 50 graduates in FY2023 (#11275), and 36 referrals with 14 graduates in FY2025 (#8592).

Community-based treatment is the larger channel and operates at a different scale. In FY2014, 7,740 people on parole were referred to drug treatment programs, 7,905 were enrolled, and 4,139 completed successfully — a 60% completion rate — delivered by 56 part-time contracted substance abuse counselors (#9622, #9623, #9624, #9625, #9627). Completion rates were 67% in FY2011 and 66% in FY2013 (#9957, #12496). A successful discharge requires attendance at a minimum of six consecutive sessions and continued negative drug tests (#9777). The Treatment Aftercare for Probationers and Parolees program supplies a specially assigned mental health caseworker who coordinates medical and social service appointments and helps the parole officer track behavior during the transition (#11125, #13499); in FY2014 officers supervised a monthly average of 556 people with mental health disorders enrolled in treatment upon release (#9618). Housing is the other binding constraint: the Reentry Partnership Housing program houses people who remain in prison after the Parole Board has authorized release solely because they have nowhere to live, paying providers $600 a month for up to three months, capped at $1,800 per person (#9768, #9615). In FY2014 there were 28 providers and 488 placements, and 86% of placements since FY2011 were either on active parole or successfully discharged (#9610, #9611, #9614); the program generated $1,966,316 in cost avoidance in FY2013 (#12484).

Two program families sit outside the treatment pipeline. The first is incentives: Georgia's Performance Incentive Credit program lets people earn earlier parole release by completing rehabilitative programming while incarcerated, with each point equal to one month that a tentative parole month can be moved earlier (#11548). Awards declined from 37,012 in FY2019 to 26,855 in FY2023 (#11333). The second is peer support. 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 2017, requires Certified Peer Specialist or CARES credentials plus a 40-hour forensic training and works with people within 18 months of release; it operates in five prisons, five day reporting centers, three mental health treatment courts, and two state hospital forensic units (#13718). The state reports more than 550 people served in FY2017–2018 with two re-arrests, three psychiatric readmissions, and one revocation since inception — self-reported figures (#13719). A newer peer-led model, FORGE at Dooly State Prison, launched in 2025 as a 36-week, three-phase program with gate requirements including 85% attendance and scored co-facilitation, drawing its curriculum from Thinking for a Change, Moral Reconation Therapy, Seeking Safety, the SAMHSA trauma-informed framework, and MENTOR's Elements of Effective Practice (#13720). Alongside these, the Parole Board reports partnerships with more than 70 faith-based housing providers and more than 200 non-housing faith-based service providers (#9616), and its community mentor program supplies volunteers who provide what the agency describes as a continuing apprenticeship in life skills (#11168, #11230).

Other States, Other Results

The strongest current model for system-wide change is Maine, which reorganized its entire correctional system starting in 2022 around normalization, humanization, and destigmatization, in a state that incarcerates fewer than 2,000 people (#4717). Maine's three-year recidivism rate fell from 30.5% in 2017 to 21.4% in 2022 — nearly a one-third reduction (#4695). Resident-on-resident assaults fell 40%, assaults on staff fell 36%, staff use-of-force incidents fell 69%, and self-inflicted injuries at Maine State Prison fell 84% (#4696, #4697, #4698, #4699). Nearly 50% of Maine's incarcerated population receives medication-assisted treatment for opioid use disorder, against less than 1% of the federal prison population in 2021 (#4702), and the state offers education from GED through master's degrees through the University of Maine (#4706). The Brennan Center separately documented that Maine's expansion of education, job training, and mental health support contributed to a 40% decrease in prison violence (#5024).

The best-evidenced single intervention is Vera's Restoring Promise, which pairs adults 25 and older as mentors with young adults 18 to 25 in dedicated housing units; it has established 9 units in 6 states since 2016 (#4959). A randomized trial at South Carolina sites found a 73% reduction in the odds of violent incidents and an 83% reduction in the odds of restrictive housing stays (#7756, #7757), results that held after controlling for custody level, education level, pre-treatment outcomes, length of time in study, race, and age — and with no significant difference between applicants who did not get spots and non-applicants, ruling out self-selection (#4677). South Carolina's programming reforms more broadly produced a 73% reduction in violence write-ups and an 83% reduction in restrictive housing stays (#5025, #5026). Michigan's Vocational Villages, operating in three facilities with capacity for roughly 600 students across 13 trade programs for people within 12 to 24 months of release, saw 2019 graduates recidivate 6.5 percentage points below the state rate (#4722, #7755). California issued 299 Innovative Programming Grants after Brown v. Plata (#5245), and its flagship programs report outcomes far below the state's 41.9% general recidivism rate: GRIP 421 graduates released since 2012 with 2 returns, and Project Rebound a 0% recidivism rate among its students (#4878, #4888). The distinction between these and Georgia's posture is one of portfolio breadth and continuity rather than of a single superior curriculum.

Texas supplies both the entrepreneurship model and the caution. The Prison Entrepreneurship Program invites more than 6,000 men across 80 prison units annually into a fixed-cohort program with a Baylor University-certified mini-MBA (#4939, #4940, #4981), has graduated more than 3,500 participants since its founding in 2004 (#4945), and reports roughly 7% of graduates returning to prison within three years against 23% of Texas male releasees — from a pool of about 2,000 applications and about 500 acceptances a year (#13645). That selectivity is the finding as much as the outcome. At system scale, Texas invested $241 million or more in evidence-based programs and graduated sanctions and avoided $3 billion or more in projected prison construction and operating costs, while its parole revocation rate fell 46% (#1221, #1224, #1222). Louisiana passed 10 reform laws in 2017, reduced its prison population 9% and its probation and parole population 12%, and shifted $30 million into community-based services (#1225, #1226, #1227, #1228). The Pew-supported Justice Reinvestment Initiative worked with more than 30 states (#1230).

Georgia's own justice reinvestment record is real but stalled. H.B. 265 in April 2011 formed the Special Council on Criminal Justice Reform for Georgians; H.B. 1176 in 2012 passed unanimously (#14364, #8345). Commitments to prison and probation for affected offenses fell from 28,604 in 2012 to 25,784 in 2015, a 10% decline (#14330); post-reform admission cohorts served less time before release at one, two, and three years than the 2012 cohort (#14375); and probation terms ended earlier (#14376). The state reinvested $57 million of savings into accountability courts, substance abuse treatment, and community supervision improvements, and achieved a 6% reduction in prison population without increasing crime (#2114, #2112, #2115). Senate Bill 174 in 2017 added behavioral incentive dates that automatically end some probation terms after three years (#14369). But the Council on Criminal Justice Reform's influence waned after Governor Deal left office in 2019 (#3891), and the reforms never touched the program-funding line items that determine what happens inside the buildings. The Scandinavian-inspired experiments now expanding in Pennsylvania and California are instructive for the same reason: Little Scandinavia opened in March 2020 with six men serving life sentences as mentors (#6180) and uses a contact officer model adapted from Norway that required revising the state's anti-fraternization policy (#4716, #6190); Pennsylvania announced expansion to three additional facilities, including a maximum-security site and a women's prison, in March 2025 (#6152, #6183). The researchers running it stress that results so far are early and mostly non-causal (#7775), and the broader Nordic comparison is a poor transplant argument — those outcomes reflect much shorter sentences, far smaller populations, and stronger social-welfare baselines (#7678). Staff buy-in, not curriculum, is reported as the biggest obstacle to the California rollout (#6079).

Step-Down and Gang Exit: The Programming Georgia Doesn't Have

Georgia's Tier Segregation System was implemented in August 2013 as a managed pathway from administrative segregation to lower security levels (#20803), and the Segregation Transition Education Program is its step-down mechanism. The written policies are more generous than the reputation of the units. Tier II STEP guarantees a minimum of four hours out of cell per day, seven days a week, plus at least one hour of exercise daily; cells may be single or double occupancy; telephone calls are limited to 15 minutes; visits are non-contact but last at least two hours; and privileges are the same as general population (#19283). Tier III STEP, written the same day, is textually near-identical with one substantive difference: all cells are single occupancy (#19284). The four hours are not four hours of association or programming — the Tier II policy counts shower and feeding time toward the total (#19283). The Tier III Program itself is a minimum 13-month, 390-day sequence requiring at least 60 days in E Wing, 60 in F Wing, and 90 days each in D, C, and B Wings before a person is considered for Tier III STEP with close security status (#20784, #20787).

Georgia reports that its extended restrictive housing population fell 41% — 671 people — between January 2017 and the end of FY2018, achieved partly through STEP, specialized protective custody housing, specialized mental health treatment units, administrative reviews, and discipline changes (#19298, #20780). The disclosure is shaped as a delta with no endpoints: the department reports how many people left and what percentage that was, and never states the population before or after (#19298). Its own table lists 119 STEP beds without splitting them between Tier II and Tier III, so the tier composition of step-down capacity is unknown (#20854). Other states publish more. In Oregon's randomized study of the Step Up Program against the Intensive Management Unit, 10.0% of the Step Up group maintained a mental health classification against 53.9% of the Intensive Management Unit group (#20825). Virginia's Restrictive Housing Reduction Step-Down Program, developed in 2011 at Red Onion and Wallens Ridge, saw the share of participants with a mental health diagnosis drop from 47% in February 2016 to 32% in July 2017 (#20827, #20829). North Carolina's rehabilitative diversion unit operates with approximately 500 participants at Marion Correctional Facility (#20824). The National Institute of Justice concluded that introducing step-down and gang-exit programs is "a positive move toward jointly reducing the influence of gangs and overuse of restrictive housing" (#3698) — while also noting that no program to date has been rigorously shown to permanently remove people from prison gangs (#3699).

The gang-exit half of that pair is where Georgia is most exposed. Texas offers GRAD, a structured nine-month, three-phase exit pathway combining substance abuse and chaplaincy work, cognitive intervention and anger management, and half-day work with unit programs (#3687), plus RP-GRAD for returning gang-validated people who would otherwise be auto-placed in restrictive housing (#3719). Arizona's Security Threat Group program produced declines of over 50% in assault, drug violations, threats, fighting, and rioting among segregated gang members, a 30% system-wide reduction in total rule violations, and an estimated 22,000 prevented violations (#3682, #3683, #3684) — but among people who formally renounced membership, disciplinary reductions were not statistically significant, suggesting renunciation must be paired with robust post-exit support and monitoring (#3714). An Ohio study found that gang affiliates leaving restrictive housing fared worse afterward, with higher misconduct, indicating that separation without programming and structured reintegration may delay rather than prevent violence (#3713). Georgia has no structured gang renouncement or exit program at all (#3715); its state-level gang strategy is prosecution rather than correctional management, with the department's internal role limited to intelligence gathering and reactive lockdowns (#3709). And the leading academic study of Texas GRAD explicitly does not answer whether the renunciation process works — no rigorous recidivism or before-and-after violence evaluation exists (#7668). The reform literature here is a map of what has been tried, not a record of what has been proven.