LAW 676 | Spring 2026

Professor Tammi Walker, JD, PhD

2026-April-14

Tuesdays & Thursdays | 9:00–10:15 AM | Room 204

Agenda

  • Today: Desistance and Reentry
  • Thursday: Meet at 8:40 at the Juvenile Justice Center 2225 E. Ajo Way
  • Next week: Group Presentation — review the syllabus & the presentation rubric; you will also get the chance to rate yourself and your partner/group members on the preparation of this presentation

Learning Objectives

By the end of this class, you should be able to:

  1. Define desistance as both event and process, and connect it to the age-crime curve
  2. Explain how offending trajectories challenge the system's ability to predict persistence
  3. Evaluate what the first six months of reentry reveal about supervision, services, and gaps
  4. Assess whether current reentry programming delivers on the constitutional promise of rehabilitation

Roadmap

Last class: Sbeglia's data — 5.2% recidivism among 287 released juvenile lifers. The system predicted irredeemable; the data says otherwise.

Today: If 95% desist, what does the science tell us about how — and does the reentry infrastructure support it?

  1. The science of desistance — what it is and what drives it
  2. Offending trajectories — the prediction problem
  3. The first six months — what reentry actually looks like
  4. Programs and systems — what works, what fails, and the gap

Part I: The Science of Desistance

What Is Desistance?

  • Event: The cessation of criminal behavior — the last offense

  • Process: The maintenance of non-offending over time — identity transformation, maturation, sustained change

  • These are not competing definitions. The event is visible; the process explains it.

What the Process Looks Like

Desistance-as-process involves psychological transformation, not just behavioral compliance:

  • Identity shift: The youth stops seeing themselves as "someone who offends" and begins constructing a prosocial self-concept — one that is incompatible with criminal behavior

  • Narrative reorientation: Youth develop redemptive self-stories — "I need to be better," "I want to make my family proud" — that reframe past behavior as situational, not definitional

What the Process Looks Like (cont.)

  • Agency: Youth who desist emphasize their own role in making change happen — distinguished from passive compliance with court orders

  • Separation of past and future self: Creating cognitive distance from a criminal identity — "that was me then, not me now"

Desistance is not something that happens to youth. It is something youth do — through identity work the system rarely supports.

The Age-Crime Curve

  • Offending peaks in late adolescence (ages 17–19) and declines into the mid-twenties as psychosocial maturity increases

  • This is a normative developmental trajectory — the majority of adolescents who offend will stop without intervention

  • Driven by both internal factors (identity transformation, maturity, agency) and external factors (relationships, opportunities, community supports)

Desistance is the expected outcome. Persistence is the exception that requires explanation.

Is the Curve Universal? — Steffensmeier et al. (2025)

56 countries, 17th–21st century data

  • The adolescent-peaked curve reflects US and Western patterns — not a biological universal

  • Outside the West: postadolescent peaks (mid-twenties to thirties) and spread-out age distributions

  • Cross-national variation tracks social and cultural factors — collectivist societies, age-graded expectations, surveillance practices — not neurobiology alone

If the curve is socially shaped, the system isn't just responding to desistance — it is producing the conditions for it or against it.

Psychosocial Maturity — The Engine of Desistance

Three domains drive the developmental process:

  • Temperance: Self-control, impulse regulation, rational decision-making

  • Responsibility: Independence, self-reliance, identity formation

  • Perspective: Future orientation, consequence consideration, ability to hold multiple viewpoints

Key finding: Adolescents have adult-like cognitive capacity for deliberative reasoning by age 15–16 — but struggle with impulse control in emotionally arousing contexts

Part II: The Prediction Problem

Four Offending Trajectories — Cauffman et al. (2024)

Crossroads Study: N=1,216 boys, 9-year follow-up from first arrest

  • Stable Low (55%): Very low offending throughout the entire 9-year period — the majority

  • Escalating (23%): Low offending at baseline, gradually increasing over the study period

  • Short-Term Recidivist (15%): High offending in the first 2 years, declining to low levels by ~2.5 years

  • Persistently High (7%): Consistently high offending throughout the entire 9-year period

The 7% Problem

  • The persistent (7%) and short-term recidivist (15%) groups looked identical at baseline

  • No measured baseline factors could distinguish them at intake

  • The system cannot reliably identify who will persist and who will desist on their own

Miller assigned sentencing judges the task of distinguishing "transient immaturity" from "irreparable corruption." The data says the distinction cannot be made prospectively.

What Predicts Trajectory — and What Doesn't

Significant predictors:

  • Peer delinquency — strongest social predictor across all groups
  • Exposure to violence — elevated in all non-stable-low groups
  • Callous-unemotional traits — higher in short-term recidivist and persistent groups
  • Physical aggression, substance use, police legitimacy perceptions, age at arrest

What did NOT predict trajectory membership:

  • Race/ethnicity — despite massive racial disparities in the system

Decision Point

If the strongest predictor of trajectory is peer delinquency, and the system concentrates justice-involved youth together in facilities — is the system supporting desistance or undermining it?

The Snare Hypothesis

  • Youth processed informally after first arrest were significantly more likely to be in the stable low trajectory group

  • Youth processed formally were more likely to end up in escalating, persistent, and short-term recidivist groups

  • This held even after controlling for 21 other risk factors

Justice system involvement itself may trap youth in offending patterns they would otherwise outgrow.

Agentic Moves — What Youth Actually Do to Desist

Desistance isn't passive. Youth make intentional choices:

  • Peer navigation — fading away from antisocial peers
  • Supportive relationship cultivation — building prosocial connections
  • Time structuring — filling time with constructive activities
  • Goal pursuit — education, employment, concrete plans

Agentic Moves (cont.)

  • Creating sanctuaries — safe environments away from risk
  • Identity work — practicing prosocial self-conceptions
  • Environmental modification — changing physical and social surroundings

Which of these can a youth do from inside a locked facility? Which require community?

What Youth Need From Others

Agentic moves don't happen in a vacuum. The research identifies relational conditions that enable them:

  • Family: Supportive relationships providing emotional grounding and motivation for change; parental monitoring combined with warmth — not surveillance alone

  • Mentors: Trusted adults who model prosocial behavior and serve as "hooks for change" — catalysts for identity transformation

What Youth Need From Others (cont.)

  • School attachment: Bonding with teachers is more predictive of reentry success than academic performance alone

  • Employment: Functions as both practical support and a source of prosocial identity — "I am someone who works," not "I am someone who offends"

  • Community connections: Provide what Giordano et al. call "hooks for change" — prosocial opportunities that matter only if the youth is cognitively open to seizing them

Part III: The First Six Months

Who Comes Home

Demographics of youth returning from out-of-home placement:

  • Predominantly ages 16–17, male (85%), and racial/ethnic minorities (69%) — despite race not predicting offending trajectory

  • Most common offenses: crimes against persons (38%) and property crimes (22%)

  • Over 50% meet criteria for one or more psychiatric disorders; 33% have a learning disability

  • Duration of confinement varies widely — from under 30 days to over a year — across detention centers, group homes, residential treatment, and secure institutions

Community Reentry — Chung et al. (2007)

Pathways to Desistance: N=413 serious juvenile offenders, Maricopa County AZ + Philadelphia PA

Outcomes at 6 months:

  • 35% experienced formal system involvement (rearrest or readmission)
  • 27% self-reported antisocial activity
  • 80% achieved engagement in school or work
  • 71% of eligible youth attended school
  • 42% worked 21+ hours per week

Supervision vs. Services — The Gap

  • 85% received court supervision; only 35% received community-based services

  • When provided, service intensity averaged ~8 contacts/month — double the supervision rate

  • 15% received no supervision at all — often older youth who aged out

The system surveils far more than it supports. 85% are watched. 35% are helped.

Duration vs. Intensity — What Works

Duration reduced negative outcomes:

  • 44% reduction in odds of system involvement per additional month
  • 53% reduction in odds of antisocial activity per additional month

Intensity increased positive outcomes:

  • 85% increase in odds of engagement per additional contact/month
  • Peer deviance: 56% reduction in odds of engagement per unit increase
  • Caring adults: 60% increase in odds of engagement per additional adult

Arizona vs. Philadelphia — Two Models

Maricopa County Philadelphia
Model Corrections-control Court-control
System involvement 46% 31%
No supervision 28% 11%
Jurisdiction ends Age 18 Age 21

Arizona's earlier cutoff left 28% of youth with no supervision at all — nearly triple Philadelphia's rate

Reentry Challenges — Four Domains

  • Mental health & substance use: Over 50% of facility youth meet criteria for psychiatric disorders; treatment continuity disruptions during transition increase relapse and recidivism

  • Education: Slow record transfers, credit transfer problems, inflexible enrollment — less than half enrolled in school within 6 months

  • Housing: Significant homelessness rates; available housing concentrated in disadvantaged areas

  • Employment: Criminal records create hiring barriers; incarceration disrupts job readiness skills

Part IV: Programs and Systems

The Needs-to-Services Gap

Even when case management identifies needs, services don't follow:

  • Only 42% of youth with education/employment needs were matched with services

  • Only 21% of youth with substance use needs were matched with services

  • Only 33% of youth with family needs were matched with services

The system diagnoses what it cannot deliver.

Systems Barriers — O'Neill et al. (2017)

Intra-organizational (within agencies — found in 15/17 studies):

  • Slow record transfers, inflexible policies, large caseloads
  • Unclear release requirements, lack of youth involvement in transition planning

Inter-organizational (between agencies — found in 11/17 studies):

  • Poor communication between corrections, schools, mental health, and housing
  • Duplicated planning, unsafe housing placements, vertical service gaps

The Resilience Portfolio Model — Rock et al. (2025)

Three domains of resilience that support desistance:

  • Regulatory processes (38 studies): Coping strategies, internal motivation, readiness for change

  • Meaning-making strengths (39 studies): Sense of self, future orientation, cultural identity, narrative processes

  • Relationships and social ecology (68 studies): Family, peers, mentors, educational and vocational opportunities

The domain with the strongest association with positive outcomes — meaning-making — receives the least research attention and the least programmatic investment.

The OJJDP Reentry Model

Services begin before release, continue through transition, extend into community

Two distinguishing features:

  1. Youth receive services AND supervision — not one or the other
  2. Intensive during incarceration, transition, AND community phases

National baseline:

  • 55% rearrest within 1 year; 24% reincarceration
  • 48,000+ youth in residential facilities on any given day
  • 33% have a learning disability; 51% meet criteria for 1+ psychiatric disorder

What Works — Program Evidence

Meta-analytic findings:

  • Small but significant recidivism reduction for programs incorporating CBT or skills training
  • Programs targeting older youth (>16.5 years) showed stronger effects
  • Well-implemented programs outperform poorly implemented ones — fidelity matters
  • Programs more effective for youth who committed violent offenses (supporting RNR principle)

Caution: Overall reentry program effects on recidivism were not statistically significant

Named Programs — The Evidence

Program Key Finding
Project BUILD (Cook County) 33% recidivism vs. 57% control at 1 year
MST-FIT 30% lower felony recidivism hazard at 36 months
Operation New Hope 35% unsuccessful at parole vs. 53% control
Wayne County Second Chance 4.3% recidivism vs. 9.5% control at 2 years

All share common features: multi-phase (begin before release), multi-domain (address more than one need), intensive (high contact frequency)

Extended Juvenile Jurisdiction

  • Extended juvenile jurisdiction allows juvenile court supervision into early adulthood (often to age 21–25)

  • Blended sentencing combines juvenile and adult system consequences

  • The Chung data: Maricopa's cutoff at age 18 left 28% with no supervision — vs. Philadelphia's extension to age 21 (11% unsupervised)

The jurisdictional endpoint is not an abstraction — it is the point where the state abandons its obligation.

Decision Point

You're designing Arizona's juvenile reentry system from scratch. You have the Chung data, the trajectory research, and the program evidence. What are your three non-negotiable design principles — and which existing barrier is the hardest to overcome?

What Remains

  • Desistance is the norm. 55% never seriously reoffend after first arrest. The 7% who persist cannot be identified in advance.

  • The system surveils more than it supports. 85% supervised, 35% served. The needs-to-services gap is structural.

  • Duration prevents relapse. Intensity promotes engagement. Both require resources the system doesn't allocate.

  • The domain that matters most — meaning-making — gets the least investment.

The constitutional promise is rehabilitation and the possibility of change. The developmental science confirms that change is the default. The reentry infrastructure does not deliver on either.

A 1-page executive summary summarizing key claims, citing at least one primary legal authority and one secondary source ✔️ Clearly state their argument and support claims with sources ✔️ Use at least one legal case or statute and one secondary legal source ✔️ Anticipate counterarguments and prepare for audience Q&A ✔️ Use visual aids (charts, case comparisons, policy timelines) to clarify key points ✔️ Maintain clear organization and transitions between sections

Callback to WK12 DY2 closing: the trilogy lit the spark, the states carry the fire — but who's left in the dark? Today's answer: the kids coming home. Sbeglia proved the developmental science predictions hold. Today we ask: if the science is right, is the system built to deliver?

Desistance is not the same as "not getting caught." It requires both behavioral change and the internal processes that sustain it. The distinction matters for policy: if desistance is only an event, you just wait. If it's a process, you can support it — or obstruct it.

This is the theoretical core students need before the data. If desistance is just "stopping," the system's job is surveillance. If desistance is identity transformation, the system's job is creating conditions for that transformation.

Amemiya et al. (2017) found youth with "psychological reorientation" narratives reported more agentic moves and better outcomes. Agency vs. compliance is the key distinction: court-mandated behavior change doesn't stick without internal motivation. This pays off later when Rock et al. shows meaning-making is the strongest resilience domain but gets the least investment.

This reframes the entire system. The juvenile justice apparatus is built around predicting and preventing the exception — but the baseline is that most youth age out of offending. Connect to Roper's developmental science: the same immaturity that reduces culpability also predicts desistance.

This is not a rebuttal of the age-crime curve. It's a refinement. The curve holds in the US context your students will practice in — the Cauffman data confirms it. But Steffensmeier shows the shape is contingent, not inevitable. The implication strengthens the deck's argument: if desistance patterns are socially shaped, then system design choices (diversion vs. formal processing, services vs. surveillance, extended jurisdiction vs. cutoff at 18) aren't just responding to biology — they're shaping the trajectory. This also inoculates against the oversimplified claim that "brains mature, so everyone desists." The science is more nuanced than that.

This is the "context-dependent maturity" concept. Cold cognition (reasoning in calm settings) matures early. Hot cognition (resisting impulses under pressure) matures late — mid-twenties. This is why adolescent offending clusters in peer-driven, emotionally charged situations — not in planned, deliberative crimes. Connect to Roper's three hallmarks: immaturity, susceptibility to peer influence, transient personality. All three are explained by the hot/cold cognition gap.

The percentages are critical: 55% never really offend again after first arrest. Another 15% recidivate briefly and then stop. Only 7% persist. The system is processing 100% of these youth but can identify the 7% at a rate no better than chance — more on the next slide.

This is the empirical demolition of the sorting problem from last class. Miller asked judges to do something the best longitudinal data says is impossible — predict at the point of sentencing which youth will persist. Cold call: "If 15% and 7% look identical at baseline, what does that mean for risk assessment instruments?"

The race finding is crucial: individual-level risk does not explain system-level racial disparity. If race doesn't predict offending trajectory but does predict system involvement, the disparity is system-generated. Connect to WK12 DY2: 70% of JLWOP recipients are children of color. 76.3% of Sbeglia's PA sample was Black.

DECISION POINT. ~3 min. Target: "Congregate care creates the conditions the data identifies as the strongest risk factor. The system's default intervention may be criminogenic." This sets up the reentry data — the question is whether post-release services can overcome what incarceration may have reinforced.

"Snare" — the system catches youth and holds them in patterns of offending through labeling, peer exposure, disrupted prosocial development, and collateral consequences. This is not an argument that the system always harms. It is an argument that formal processing carries costs the system rarely accounts for — and that diversion should be the default for first-time offenders. Connect to the age-crime curve: if 55% are stable low regardless, formal processing is net-negative for the majority.

This list matters because it implies a service design. Effective reentry programming should support these specific moves — not generic "supervision."

Cold call the highlight-q question. Most require community. The implication: reentry services must begin before release. Identity work connects to the resilience portfolio model later — meaning-making is the strongest predictor but least invested in.

Each of these is a relational precondition for the agentic moves. Peer navigation requires alternative peers. Identity work requires adults who reflect a non-criminal identity back. "Hooks for change" is Giordano et al.'s (2002) framework: structural opportunities only matter if the youth is cognitively open to them. Relationships create that openness.

The Chung data in Part III will show that caring adults produce a 60% increase in engagement odds — this slide sets up why. Sanctuaries require safe spaces maintained by caring adults. Each relational domain maps to an agentic move: school attachment enables goal pursuit, employment enables time structuring, community connections enable environmental modification.

These numbers set the stage: the reentry population is overwhelmingly minority, overwhelmingly male, and overwhelmingly carrying unmet mental health and educational needs. The race disparity here is the system-level complement to the Cauffman finding: race doesn't predict trajectory, but it predicts who the system processes — and therefore who needs reentry services. The 50% psychiatric disorder rate and 33% learning disability rate mean reentry is not just a supervision problem. It is a service delivery problem.

Two stories in the same data: 35% system involvement looks like failure. 80% school/work engagement looks like success. Both are true at the same time for overlapping populations. This is Maricopa County — our jurisdiction. The Arizona-specific patterns matter.

This is the central finding for reentry policy. The system's default mode is surveillance, not support. Community-based services are the exception, not the rule. The 15% with no supervision: in Maricopa County, Arizona juvenile authority ended at 18 (vs. 21 in PA). Youth who aged out fell through entirely. CBS disparity: minority youth and those from disadvantaged neighborhoods were less likely to receive community-based services — even when controlling for offense severity.

Duration prevents relapse. Intensity promotes connection. These are different mechanisms serving different functions. The caring-adults finding connects directly to the agentic moves: "supportive relationship cultivation" requires adults to cultivate relationships with. The peer deviance finding reinforces the congregate care problem: facilities concentrate deviant peers; effective reentry requires navigating away from them.

The Maricopa data is directly relevant: this is where our students will practice. The jurisdictional age gap (18 vs. 21) produces a measurable difference in outcomes. Extended jurisdiction isn't just policy preference — it's empirically supported. Connect to Bassett: Arizona's structural gaps aren't limited to sentencing. They extend through reentry.

Each domain is a potential snare: the youth who can't enroll in school because records won't transfer is the youth who falls back on antisocial peer networks. The youth without stable housing is the youth who can't create a "sanctuary." The domains interact: mental health treatment requires stable housing. Employment requires education. Each gap cascades.

This is the implementation gap. Risk-needs-responsivity (RNR) frameworks require accurate assessment AND service delivery. Most systems do the first part reasonably well and fail at the second. The substance use gap (21%) is the most alarming given that 34% of facility youth meet criteria for substance use disorder.

"Vertical service gaps" — services available at the facility level don't exist at the community level, or vice versa. A youth receiving CBT in the facility may find no CBT provider in their neighborhood. The intra/inter distinction matters for reform: you can fix within-agency problems with better management. Between-agency problems require structural coordination that most jurisdictions lack.

91% of the sample studied relationships and social ecology. That's where the research goes. But Rock et al. found the strongest association with positive outcomes was in meaning-making — the least-studied domain.

DISCUSSION. ~2 min. The system invests in supervision (relationships/ecology) and CBT (regulatory). It largely ignores identity work, future orientation, and narrative processes — exactly what the agentic moves literature says youth do on their own when they successfully desist. "What would a meaning-making intervention look like for a 17-year-old coming home from a facility?"

The three-phase model (incarceration → transition → community) is the standard. Most programs only operate in one phase. The "services AND supervision" requirement directly responds to the Chung finding: 85% supervised, 35% served. Family visitation during incarceration produced rapid declines in depressive symptoms — one of the strongest single-intervention findings.

The "not statistically significant" finding is important context. Reentry programs as a category are not proven to reduce recidivism — but specific program types (CBT, well-implemented, targeting older/violent youth) do show significant effects. The RNR finding: programs matched to risk level and criminogenic needs outperform generic programming. One-size-fits-all fails.

Project BUILD is the strongest single finding: nearly halved recidivism at one year. Community-based, began pre-release, combined skills training with service linkage. MST-FIT (Multisystemic Therapy — Family Integrated Transitions): family-based, addresses the social ecology around the youth, not just the youth. Common thread: programs that look like the OJJDP model — three phases, services + supervision, intensive — outperform programs that only operate in one domain.

Extended jurisdiction directly addresses the age-crime curve: if offending peaks at 17–19 and desistance requires support through the mid-twenties, ending supervision at 18 abandons youth during the highest-risk period. Arizona context: A.R.S. § 8-202(H) caps juvenile court jurisdiction at age 19 (or 17.5 for certain DCS-involved youth). Legislative proposals to extend to 21 have stalled.

DECISION POINT. ~5 min. Target answers should reflect the data: (1) services must begin pre-release; (2) extended jurisdiction beyond 18; (3) meaning-making/identity work, not just supervision. The hardest barrier: inter-organizational coordination — no single agency owns reentry, and Arizona's fragmented system (corrections, courts, schools, mental health) has no coordination mandate. If participation is strong, let this run. If not, pivot to small groups.

This is the summary, not the close. The close is the next slide. Each bullet maps to a section of the class. Students should be able to connect each back to the underlying data.

Close on the gap. Do not resolve. The trilogy said youth are different. The science says youth desist. The system says "we know" — and then builds surveillance, not support. Looking ahead: collateral consequences — what happens to the youth who desist but carry the record forward.