Reentry & Rehabilitation
A Sense of Purpose as a Driver of Rehabilitation in Incarcerated People: An Evidence Brief
This evidence brief synthesizes theoretical, philosophical, and empirical literature on whether a sense of purpose drives rehabilitation among incarcerated people. It finds strong meta-analytic support for correctional education (43% lower recidivism odds, RAND) and CBT (25% recidivism reduction), moderate support from therapeutic communities and mentorship programs, but a critical gap in direct measurement linking purpose as a psychological construct to post-release outcomes. The Beaudry et al. (2021) Lancet Psychiatry RCT-only meta-analysis, which found effects largely disappear in larger trials, represents the most serious empirical challenge. For Georgia advocacy, the brief notes that the official GDC 3-year felony reconviction rate of ~25–31% and the BJS 9-year rearrest rate of 83% measure different things, and both are true.
Key Findings
The most impactful data from this research collection.
83%
83% Nine-Year Rearrest Rate for Released Prisoners
StatisticGeorgia Recidivism Flat Despite Doubled Spending
Finding$5.00
$1 Prison Education Saves $4–$5 Reincarceration
StatisticNo Study Links Purpose to Recidivism Outcomes
Data gapAll Data Points
58 verified data points extracted from primary sources.
BJS 5-year rearrest rate for 2005 state prisoner releases Statistic
Of 404,638 state prisoners released in 30 states in 2005, 67.8% were arrested within 3 years and 76.6% were arrested within 5 years; 55.1% had a reconviction within 5 years.
76.6% vs. percent rearrested within 3 years
BJS 9-year rearrest rate: 83% Statistic
The 9-year follow-up of the same 2005 release cohort (analytic sample 67,966 records, representing 401,288 releases) found 68% rearrested within 3 years, 79% within 6 years, and 83% within 9 years.
83% vs. percent rearrested within 3 years
BJS 9-year cohort accumulated ~2 million arrests Statistic
The 401,288 releasees in the BJS 9-year follow-up accumulated approximately 2.0 million arrests, averaging 5 arrests per person; 60% of these arrests occurred in years 4–9, meaning shorter follow-ups substantially undercount reoffending.
2,000,000 arrests accumulated vs. average arrests per person
GDC 3-year felony reconviction rates 2011-2022 Statistic
The Georgia Department of Corrections reports three-year felony reconviction rates for all inmate facilities of roughly 26–28% from 2011–2019, dipping to 23.9% (2018), 25.3% (2019), 29.9% (2021), and 31.1% for the 2022 release cohort.
31.1% vs. percent felony reconviction (2018 cohort, lowest)
GDC transition center recidivism lower, private prisons higher Statistic
Transition center releases have lower three-year felony reconviction rates (12–20%) while private prison releases are higher (~32%).
CSG Georgia 3-year felony reconviction rate: 27% Statistic
The Council of State Governments puts Georgia's three-year felony reconviction rate at exactly 27 percent, which is lower than the national average for that metric.
27% vs. lower than national average
GDC recidivism definition excludes technical violations and post-3-year reoffending Data gap
The official GDC reconviction rate does not take into account offenders who commit technical violations while on parole, nor offenders who recidivate after three years. National BJS data show that when technical parole/probation violations are inclu…
Georgia 30% recidivism rate unchanged for a decade despite doubled corrections spending Finding
The Georgia Council on Criminal Justice Reform acknowledged that Georgia's 30 percent recidivism rate had remained virtually unchanged for 10 years prior to the 2010s reforms despite a doubling of corrections spending.
RAND: 43% lower odds of recidivism for correctional education participants Statistic
Davis et al. (RAND Corporation, 2013) meta-analyzed 57 recidivism studies and found inmates who participated in correctional education programs had ~43% lower odds of recidivating than non-participants, translating to a 30% versus 43% three-year rec…
43% vs. percentage-point absolute reduction (30% vs 43%)
RAND: 13% higher odds of post-release employment for education participants Statistic
The RAND meta-analysis found odds of post-release employment were ~13% higher for education participants, though RAND notes only one study on this was rated higher quality, so this estimate is less robust.
13%
RAND: $1 in prison education saves $4-$5 in reincarceration costs Statistic
The RAND meta-analysis found that $1 invested in prison education saves $4–$5 in reincarceration costs over the first three years.
$5.00 vs. dollars saved per dollar invested (lower estimate)
RAND caveats: weak designs and selection effects in underlying studies Methodology note
RAND itself flags that many of the underlying studies in the correctional education meta-analysis have weak designs, and participants may differ from non-participants on unobserved motivation (selection effects partially controlled for in higher-qua…
RAND 43% figure is foundation for Second Chance Pell Grant restoration Policy
The RAND 43% lower odds of recidivism figure is the most-cited recidivism-reduction statistic in U.S. corrections policy and was the empirical foundation for restoration of Pell Grant access for incarcerated students (Second Chance Pell).
Wilson, Gallagher & MacKenzie: vocational/education participants ~1.5x more likely employed Statistic
Wilson, Gallagher & MacKenzie's earlier meta-analysis (2000) of corrections-based education, vocation and work programs found participants ~1.5 times more likely to be employed and modestly less likely to recidivate, though they cautioned that selec…
1.5x times more likely to be employed
Mitchell et al.: TC programs consistently show modest recidivism and drug use reductions Statistic
Mitchell, Wilson & MacKenzie (2007/2012) Campbell Systematic Reviews meta-analysis of 74 evaluations of incarceration-based drug treatment found TC programs consistently showed modest reductions in post-release recidivism and drug use; 30 of 35 TC e…
30 of 35 TC evaluations showing significant treatment effect
Mitchell et al.: average odds ratios for TC treatment effects Statistic
Average treatment-effect odds ratios across the 66 evaluations in the 2007 Mitchell, Wilson & MacKenzie review: ~1.37 for recidivism reduction and ~1.28 for drug-use reduction. TCs outperformed group counseling, boot camps and narcotic maintenance a…
1.4 odds ratio for recidivism reduction vs. odds ratio for drug-use reduction
Beaudry et al. 2021: RCT-only meta-analysis shows effects attenuate in larger trials Finding
Beaudry et al. (2021) Lancet Psychiatry systematic review and meta-analysis of 29 RCTs (9,443 participants) of psychological interventions in prison found OR = 0.72 (95% CI 0.56–0.92) across all RCTs, but when restricted to studies with >50 particip…
Beaudry et al.: TC-specific RCT evidence (2 RCTs only) Statistic
In the Beaudry et al. (2021) meta-analysis, therapeutic communities specifically (only 2 RCTs) showed OR 0.64 (0.46–0.91).
0.6 odds ratio for TC RCTs
Landenberger & Lipsey: CBT reduces recidivism by 25% Statistic
Landenberger & Lipsey (2005) meta-analysis of 58 experimental/quasi-experimental studies found mean recidivism was 25% lower in CBT treatment than control groups — a reduction from a mean recidivism rate of .40 in control groups to a mean rate of .3…
25% vs. from .40 control to .30 treatment mean recidivism rate
CBT effects largest for higher-risk offenders and high-quality implementation Finding
Independent moderators of larger CBT effects included treatment of higher-risk offenders, high-quality implementation, and inclusion of anger control and interpersonal problem-solving components.
Brand-name CBT programs do not outperform generic CBT Finding
Lipsey, Landenberger & Wilson (2007) Campbell review confirms CBT reduces recidivism by an average of ~25–30% across well-implemented programs; brand-name programs (e.g., Reasoning & Rehabilitation, Moral Reconation Therapy, Thinking for a Change) d…
InnerChange Texas: selection-effect critique (completers vs. controls) Case detail
Johnson & Larson (2003) evaluated the Texas InnerChange Freedom Initiative and found IFI graduates had lower rearrest and reincarceration rates. However, the original evaluation counted only program graduates (~75 of ~177 starters), and the comparis…
Duwe & King 2013: Minnesota IFI reduces rearrest and reconviction but not technical violations Finding
Duwe & King (2013) used Cox regression on 732 offenders released 2003–2009 in Minnesota and found IFI participation significantly reduced rearrest, reconviction, and new-offense reincarceration, though not technical-violation reincarceration. The au…
California Arts-in-Corrections: 88% favorable outcomes vs 72.5% for all parolees Statistic
The original 1987 California Department of Corrections study (n = 177) found that at six months post-parole, Arts-in-Corrections participants had 88% favorable outcomes vs. 72.5% for all parolees; the difference grew with time at risk. Americans for…
88% vs. percent favorable outcomes for all parolees
RTA program-reported <3% return-to-prison rate (with caveats) Statistic
Rehabilitation Through the Arts (RTA) at Sing Sing reports a
3%
Arts program evaluations are almost all observational and often small Methodology note
Almost all arts-program evaluations are observational, often small, and frequently program-reported. Van der Meulen & Omstead (2021, The Prison Journal) explicitly critique the rehabilitation-and-recidivism framing imposed on arts evaluations. The s…
Sells et al. 2020: peer-mentored reentry RCT shows lower recidivism Finding
Sells et al. (2020) conducted a small RCT (n = 55 men) of peer-mentored community reentry vs. standard services. The mentored group had significantly lower recidivism in logistic regression controlling for LSI-R risk and demographics. The mentorship…
Arches Transformative Mentoring: 69% lower felony reconviction at 12 months, 57% at 24 months Statistic
Arches Transformative Mentoring (NYC, ages 16–24) evaluation by Lynch et al. (Urban Institute, 2018) reports felony reconviction rates among Arches participants are 69 percent lower 12 months after beginning probation and 57 percent lower 24 months …
57% vs. percent lower felony reconviction at 12 months
Reker 1977: inmates score lower on Purpose-in-Life test than normative samples Statistic
Reker (1977) studied 48 male inmates at a Canadian federal penitentiary. Inmates scored significantly lower on PIL than non-incarcerated normative samples (inmate mean ~100; Crumbaugh's normal-sample mean 112.4 SD 14.1; clinical mean 92.6 SD 21.3). …
100 PIL mean score (inmates) vs. PIL mean score (normal sample)
PIL psychometric reliability in inmates Statistic
Reker (1977) found split-half reliability of .85 (corrected .92) and 12-week test-retest reliability of .68 for the Purpose-in-Life test in an inmate population, establishing it as psychometrically sound in this context.
0.9 corrected split-half reliability vs. 12-week test-retest reliability
Critical data gap: no prospective study links baseline PIL/MLQ in inmates to recidivism Data gap
A targeted search did not locate any large prospective study linking baseline Purpose-in-Life or Meaning-in-Life Questionnaire scores in incarcerated populations to subsequent recidivism. The inmate-purpose literature is dominated by cross-sectional…
Aliakbari Dehkordi et al. 2020: group logotherapy RCT increases hope in imprisoned women Finding
An RCT (n = 90) of group logotherapy vs. control among imprisoned women in Shiraz, Iran found logotherapy significantly increased hope (Miller Hope Scale, p = .001). The outcome measured was hope, not recidivism.
Van Dierendonck 2023: d = 0.44 effect size for interventions targeting Ryff well-being dimensions Statistic
Van Dierendonck's 2023 meta-analysis of interventions targeting Ryff psychological well-being dimensions reports an overall effect size of d = 0.44 in controlled trials. While not specific to prisoners, this establishes purpose as a measurable, modi…
0.4 Cohen's d effect size
Netto, Carter & Bonell 2014: zero RCT-level studies of GLM-based interventions Data gap
Netto, Carter & Bonell (2014) conducted the first systematic review of GLM-based interventions and found zero studies meeting inclusion criteria for RCT-level evidence.
Mallion et al. 2020: GLM interventions at least as effective as standard relapse prevention, with better engagement Finding
Mallion, Wood & Mallion (2020) systematic review of 17 GLM studies (only one RCT, all targeting men who had sexually offended) found GLM-consistent interventions were 'at least as effective as standard relapse prevention programs, whilst enhancing p…
Zeccola et al. 2021: not enough evidence to confirm GLM efficacy on recidivism Data gap
Zeccola, Kelty & Boer (2021) screened 1,791 articles and found only 6 met inclusion criteria for GLM recidivism evaluation — all observational designs, no RCTs. In half the studies GLM did not increase recidivism risk; in half there was some evidenc…
Andrews, Bonta & Wormith 2011: GLM adds nothing unique beyond RNR Quote
Andrews, Bonta & Wormith (2011) defended RNR and concluded: 'At the present time, there is nothing unique in GLM other than the encouragement of weak assessment approaches and the addition of confusion in service planning.' They argue enhancing pers…
Maruna: redemption scripts and generativity as central to desistance Finding
Maruna's Liverpool Desistance Study (fieldwork 1996–1998; n approximately 65) identified two prototypical self-narratives: a 'condemnation script' (persister's account of being doomed by uncontrollable forces) and a 'redemption script' in which the …
Sampson & Laub: marriage, employment, and military service as positive turning points Finding
Using reconstructed Glueck data on 500 delinquent men followed to age 70, Sampson and Laub developed the age-graded theory of informal social control, identifying marriage, stable employment, and military service as positive turning points that stre…
Giordano: four-stage cognitive transformation theory of desistance Finding
Giordano, Cernkovich & Rudolph (2002) argued that Sampson and Laub overweight external turning points and underweight agency. Their theory of cognitive transformation identifies four sequential shifts: (1) general openness to change, (2) exposure an…
Good Lives Model: 10-11 primary human goods Finding
Ward's Good Lives Model posits 10 primary human goods (later proposed as 11 by Purvis 2010): life (healthy living), knowledge, excellence in play, excellence in work, excellence in agency (autonomy/self-directedness), inner peace, relatedness, commu…
Ryff's six dimensions of psychological well-being include Purpose in Life Finding
Carol Ryff's six-dimension model of psychological well-being — Self-Acceptance, Positive Relations, Autonomy, Environmental Mastery, Purpose in Life, and Personal Growth — has been validated across hundreds of studies and explicitly includes purpose…
SDT: autonomously motivated treatment participants have better outcomes Finding
Treatment court studies (Morse, Cusack & Andiloro 2014; Wild, Yuan, Rush & Urbanoski 2016) found that autonomously motivated participants in drug treatment courts had better engagement and lower relapse than externally pressured participants. A Hong…
Van Damme et al. 2023: detained adolescents' treatment motivation aligns with GLM primary goods Finding
Van Damme et al. (2023) found in detained adolescent boys that factors which raised treatment motivation aligned with GLM primary goods ('excellence in work and play,' 'excellence in agency,' 'relatedness').
Interpretive note: felony reconviction vs. rearrest measure different things Methodology note
Felony reconviction within 3 years is not the same as rearrest within 9 years, which is not the same as any return to incarceration. The Georgia ~25–31% figure and the BJS 83% figure are measuring different things; both are true. For Georgia advocac…
Selection-effect lesson from InnerChange Texas Methodology note
The InnerChange Texas controversy is the textbook illustration of why 'completers vs. controls' comparisons systematically overstate program effects, since program completion is itself a marker of motivation, stability, and lower baseline risk. Most…
Reverse causation caveat in desistance research Methodology note
People who are already on a desistance trajectory may construct redemption scripts as a way of making sense of changes that have other causes (aging, fading peer effects, neurobiological maturation through the mid-20s, lifestyle changes). The Liverp…
Publication bias in faith-based and arts-in-prison literature Methodology note
Both faith-based and arts-in-prison literatures are dominated by program-affiliated researchers, and null findings appear to be underpublished.
GLM evidence base confined mostly to sex offender treatment Methodology note
Most rigorous GLM evaluation has been with sex offenders; most CBT evaluation has been with mixed adult populations; most desistance theory has been built from predominantly white and male samples. Application to women, racial minorities, and people…
Structural factors cannot be overcome by individual purpose alone Finding
McNeill, Baldry, and others have argued that an excessive focus on individual purpose and identity can responsibilize people for outcomes shaped by housing, employment, racial discrimination, and supervision conditions that no internal transformatio…
Net assessment: purpose thesis is strong on theory, moderate on proxies, weak on direct measurement Finding
The thesis that purpose is a key driver of rehabilitation is defensible but at a calibrated strength: strong on theory and qualitative mechanism, moderate when triangulating across proxy outcomes, weak on direct longitudinal measurement. No single s…
Proxy problem: hard outcome data measure program participation, not purpose directly Data gap
Almost all hard outcome data measure program participation, education, or employment, not purpose as a directly measured construct. When RAND reports a 43% reduction in recidivism odds for education participants, we do not know how much of that effe…
OJJDP Ohio youth mentoring: reductions not statistically significant Finding
Stout et al. (OJJDP-funded evaluation of six Ohio youth mentoring programs) found that while some reductions in recidivism were found, the differences were not statistically significant in the parole sample.
Frankl: meaning as primary motivational force Quote
Viktor Frankl argued that 'striving to find a meaning in one's life is the primary motivational force in man' — a drive distinct from Freud's pleasure principle and Adler's striving for power. His clinical method, logotherapy, treats existential vac…
Recommended Stage 1: education + CBT + peer-mentoring continuity Policy
For Georgia program design, Stage 1 (year 1) should anchor any new program in education + CBT + structured peer-mentoring continuity from inside to outside the prison — the three components carrying the heaviest meta-analytic evidence. Target high-r…
Recommended Stage 3: close the measurement gap with validated purpose scales Policy
Stage 3 (years 1–3): Administer validated meaning/purpose scales (Purpose-in-Life test; Meaning in Life Questionnaire; Ryff Psychological Well-Being scales) at baseline, mid-program, and release. Track for 3 and 5 years. This is the single research …
Benchmark: if 3-year reconviction exceeds GDC ~27% without adjustment, reanalysis needed Policy
If 3-year felony reconviction in program graduates exceeds the GDC ~27% baseline without selection-effect adjustment, intention-to-treat reanalysis is essential before scaling.
BJS 5-year reconviction rate: 55.1% Statistic
Of 404,638 state prisoners released in 30 states in 2005, 55.1% had a reconviction within 5 years.
55.1%
Sources
43 cited sources backing this research.
Primary
Official report
Primary
Academic
Secondary
Academic
Tertiary
Official report
Primary
Academic
Primary
Academic
Primary
Academic
Primary
Academic
Primary
Academic
Primary
Academic
Primary
Data portal
Primary
Data portal
Primary
Official report
Primary
Academic
Primary
Academic
Primary
Academic
Primary
Official report
Secondary
Journalism
Primary
Academic
Secondary
Journalism
Primary
Official report
Primary
Official report
Primary
Press release
Secondary
Official report
Primary
Academic
Primary
Official report
Secondary
Official report
Primary
Academic
Primary
Academic
Secondary
Official report
Primary
Academic
Primary
Academic
Primary
Academic
Primary
Academic
Primary
Academic
Primary
Academic
Key Entities
Organizations, people, facilities, and other named entities referenced in this research.
Arches Transformative Mentoring
[program]
Bureau of Justice Statistics
[organization]
California Arts-in-Corrections
[program]
Carol Ryff
[person]
Council of State Governments
[organization]
Fergus McNeill
[person]
Georgia Council on Criminal Justice Reform
[organization]
Georgia Department of Corrections
[organization]
Good Lives Model
[program]
InnerChange Freedom Initiative
[program]
John Laub
[person]
Liverpool Desistance Study
[program]
Peggy Giordano
[person]
RAND Corporation
[organization]
Rehabilitation Through the Arts
[program]
Risk-Need-Responsivity Model
[program]
Robert Sampson
[person]
Second Chance Pell
[program]
Shadd Maruna
[person]
Tony Ward
[person]
Viktor Frankl
[person]
Related Topics
Research topics that draw on data from this collection.
Parole & Sentencing
Georgia's parole and sentencing system is defined by extreme incarceration rates, a parole board that denies release to the vast majority of eligible prisoners, and a surging population of aging inmates whose continued confinement costs taxpayers hundreds of millions of dollars annually. With 528,000 residents under criminal justice supervision — the most of any state — and a felony probation population that is the largest in the nation, Georgia has constructed one of the most expansive carceral systems in the world. The data reveals a system that incarcerates at near-record rates, releases at declining rates, and invests almost nothing in the reentry infrastructure that would reduce the recidivism it claims to prevent.
1,832 data points
Policy & Advocacy
Georgia Prisoners' Speak documents a prison system whose policy architecture — from $0.54-per-meal food budgets to a 50% correctional officer vacancy rate — systematically produces violence, illness, and recidivism while shifting hundreds of billions of dollars in costs onto families and taxpayers. Reform advocacy must contend with a $1.8 billion annual corrections apparatus that prioritizes surveillance contracts and sentence length over rehabilitation, reentry, or basic constitutional standards of care. This page synthesizes the evidence base for legislative, budgetary, and structural reforms across nutrition, staffing, communications, solitary confinement, parole, post-conviction relief, and decarceration.
3,003 data points
Recidivism & Reentry
Georgia releases 14,000–16,000 people from its prisons each year into communities with minimal preparation, support, or resources — yet the state's official recidivism rate of 25–27% obscures a far grimmer reality: when technical violations, arrests, and extended measurement windows are factored in, the true return-to-incarceration rate approaches 50%. With 528,000 Georgia residents under criminal justice supervision and an incarceration rate of 881 per 100,000 — higher than any nation on earth except El Salvador — the state's failure to invest meaningfully in reentry is not merely a policy gap but a documented engine of mass incarceration costing taxpayers $1.8 billion annually.
1,405 data points
Reform Models & Programs
Georgia's prison system spends more than $1.8 billion annually while delivering rehabilitation outcomes that rank among the worst in the nation — a structural failure made visible by comparing GDC practices against evidence-based national models. From Scandinavian-inspired residential units to California's court-mandated programming overhaul, proven reform frameworks exist at scale; Georgia has largely refused to adopt them, even as its prisons recorded at least 100 homicides in 2024 and a recidivism rate that mirrors the national average of 76.6% rearrested within five years. This page synthesizes what works, what Georgia does instead, and the fiscal and human cost of that gap.
3,436 data points