The Transition from Prison to Community

Shifts in the Rehabilitative Project

US prisons were historically intended for two distinct purposes: to punish and to rehabilitate. At their origin in the nineteenth century, penitentiaries in the Northeast were designed for moral improvement, offering a humane alternative to the stocks, flogging, and the wide use of capital punishment (Rothman 2002b; Rotman 1998). By the early twentieth century, rehabilitation through work and education—components of what David Garland (2001) called “penal welfarism”—had become essential to good correctional practice (Rothman 2002a).

Support for rehabilitation was often unsteady within penal institutions and collapsed in the period of the prison boom. The abandonment of rehabilitation is often dated from the publication of Robert Martinson’s review of correctional programming in the 1970s (Lipton et al. 1975) and his subsequent appearance on 60 Minutes (Cullen 2013). During the episode, Martinson remarked that the rehabilitation programs “simply have no fundamental effect on the recidivism rate of people who go through … the system” (quoted in Cullen 2013, 327). Doubts about the effectiveness of rehabilitation were echoed a few years later in a report by the National Research Council (Martin et al. 1981).

The rejection of correctional programming in the 1970s was followed by another shift in research consensus in favor of rehabilitation two decades later. A modern theory of rehabilitation emerged that emphasized changing opportunities, behavior, and peer groups. Education and work programs—the mainstays of prison programming for a century—gained theoretical support from life course criminology that found that stable jobs provided structure and routine in daily life and diverted workers from peers who themselves may be involved in crime (Laub and Sampson 2003; Sampson and Laub 1993; Uggen 2000). Parallel to academic currents, correctional researchers and practitioners developed a theory of rehabilitation that emphasized principles of risks, needs, and responsivity (MacKenzie 2006, chap. 4). In this framework, treatment was to be focused on those at highest risk of recidivism, tailored to “criminogenic need” addressing the traits most closely connected to crime, and programs should aim to modify characteristics that are responsive to change. Antisocial attitudes and behaviors and criminally involved peers are predictive of crime, and these have emerged as targets for case management and behavioral health programs.

Do modern rehabilitation programs reduce recidivism and improve social integration? Gerald Gaes and his colleagues (2000, 361) summarized the new consensus after Martinson: “Most correctional treatments for adult prisoners probably have modest positive effects.” Rehabilitation programs, they say, have larger effects for adolescents rather than adults, and community programs have larger effects than programs in prison. Work and education programs were often found to be associated with reduced recidivism (Bozick et al. 2018; Gaes et al. 2000). For example, a review of correctional program evaluations by the Washington State Institute for Public Policy shows that basic education, GED programs, post-secondary education, and correctional industries all pass a cost-benefit test, and the effects are especially large for post-secondary education (Bitney et al. 2017; on work and vocational programs, see Nur and Nguyen 2023; on prison GED programs, see Tyler and Kling 2007). These results are vividly illustrated in qualitative accounts of model programs, such as the Bard Prison Initiative, that provides post-secondary education in prison (Lagemann 2011).

Although the new consensus rallied behind rehabilitation, the field offers few well-powered randomized trials. In one of the rare randomized experiments that includes educational programming in prison (along with cognitive behavioral therapy, drug and alcohol treatment, case management, and post-incarceration support), one-year rearrest rates were lower (63 versus 72 percent), but employment outcomes were similar for treatment and control subjects (Cook et al. 2015). Behavioral interventions, such as cognitive behavioral therapy or motivational interviewing, are also associated with reduced recidivism. The US Department of Justice’s CrimeSolutions website, for example, classifies both kinds of interventions as “promising,” meaning they are supported by “moderate quality evidence with statistically significant average effect sizes” (National Institute of Justice 2023).

In an area of evaluation that often struggles with program fidelity and research design, the case for rehabilitation has been buttressed by meta-analysis that pools together large numbers of (often imperfectly designed) studies. Whatever the value of meta-analysis, it does consistently find that deterrence is less effective than rehabilitation at reducing recidivism. “Interventions that are punitive—that emphasize deterrence, discipline, or surveillance—have weak … effects on recidivism,” writes Francis Cullen (2017, 248).

Despite evidence of the positive effects of programs, policymakers of the 1980s and 1990s largely rejected rehabilitation and adopted incapacitation and deterrence as the main goals of penal policy. The rejection of rehabilitation is captured in figures on prison program participation from the Survey of Prison Inmates (formerly called the Survey of Inmates of State Correctional Facilities). Survey data from 1991 indicate that only a minority of people in state prison participated in drug, education, or vocational programs (table 2). By 2016, participation rates in all these areas had fallen by 20 to 35 percent. Prison jobs are more widely available than rehabilitative programs and two-thirds of survey respondents were involved in work assignments or employment programs in 1991. Participation in work programs by 2016 had fallen by about 25 percent. Declines in program participation were reported across all racial groups. Similar trends in program participation are reported by Phelps (2011) and Reich (2024).

The decline in prison programming had clear implications for reentry. As Faye Taxman and her colleagues (2014, 51) observe: “It is necessary for correctional programming to be spread throughout the correctional system. … A routine regime of treatment and programming is more likely to produce positive outcomes than programming that is rare or offered to few individuals within a prison or correctional setting. Essentially, what happens inside prison will affect what happens in the community; the result being that mass incarceration will have a long-term impact on offenders, their families, and communities.” Declines in program participation and the turn from rehabilitation reflected a deeper shift in the philosophy of correctional policy. Rehabilitation has a symbolic component, signaling society’s commitment to compassion and an individual’s capacity for change. Rehabilitation announces that we are “a civilized nation. … We are capable of turning our collective cheek in hopes of effecting redemption” (Cullen 2013, 308). Hopes for redemption dimmed during the prison boom, and many rehabilitative programs were shuttered.

In sum, support for the rehabilitative mission of the prison has fluctuated with penal policy. Research evidence suggests that prison programs can have modest effects on reducing rearrest and re-incarceration in the years after prison release. Still, support for rehabilitation declined with the punitive turn in criminal justice policy. Survey data indicate that only a minority of incarcerated people are enrolled in the main categories of rehabilitative programs, and prison program participation declined from 1991 to 2016.

The Warehouse Model and Prison Violence

While US prisons offer little support for rehabilitation, there is evidence that incarceration can be actively harmful. When prison populations were growing rapidly and overcrowding became persistent in many states, researchers described prisons as warehouses, operating chiefly as storage units for prime-age men from poor communities (Lynch 2009; Phelps 2011 discusses commentary on the warehouse prison). The loss of programs and ensuing inactivity, overcrowding, and the influx of new and younger prisoners have all been associated with violence in prison.

Violence, fear of violence, and victimization have been major concerns of prison researchers. Predatory violence where the strong exploit the weak, gang rivalries, management failures and disorder, and the self-regulation of trade in drugs and other contraband have all featured in discussions of prison violence. The salience of violence in prison life is reflected in the high levels of vigilance among incarcerated people reported in field studies. Gresham Sykes (1958, 78), in his classic study of New Jersey State Prison from the 1950s, describes the loss of security as one of the “pains of imprisonment” that “arouses acute anxiety” in a “tense and fearful existence.” Analyzing interviews with nine hundred respondents incarcerated in New York’s maximum-security prisons, Hans Toch (1977) finds the safety of the prison environment to be a dominant concern. Prison environments that feel dangerous and unpredictable create an “explosive” climate of “tension” (Toch 1977, 38–50). Craig Haney (2006, 172–73) summarizes the behavioral impact in which fear can lead to social withdrawal, hypervigilance, a tough exterior, and flat affect as incarcerated people try to avoid conflict.

Despite the well-documented salience of violence in prison life, accurate measurement is extraordinarily challenging. Estimates of the levels of violence in prison have relied on official misconduct data, self-reported victimization in survey studies, and reports of injury and death compiled by health officials. Bias in measurement, particularly the underestimation of violence, is deeply rooted in the power relations of incarceration. Although authority is pervasive in prison, custody staff are collectively distrusted, and violence is actively hidden and underreported. Official misconduct data are widely regarded to grossly underestimate true levels of prison violence (Bottoms 1999, 222; Cooley 1993), and where comparison is possible, self-reported victimization can exceed rates of official misconduct by a factor of ten or more (for example, Bottoms 1999). The power structure of the prison also distorts self-reports in surveys. Underreporting is likely where victimization elicits shame or embarrassment, as it may in the congressionally mandated surveys of sexual violence under the Prison Rape Elimination Act (PREA). When the Department of Justice adopted national standards for reporting and investigating sexual violence in prison, self-reported sexual victimization in PREA surveys increased threefold (Rantala 2018). In his fieldwork, Bruce Western (2018) finds that formerly incarcerated survey respondents speak more readily about violence in prison after they are released than while incarcerated. Injury surveillance systems and prison grievances offer alternative data collection strategies that, like self-reports, indicate higher levels of violence than indicated in official misconduct records (Butler and Kariminia 2006; Calavita and Jenness 2015). Butler and colleagues (2024) synthesize data collection methods in their account of a multi-strategy approach that combines information on institutional infractions, incident reports, and interviews with staff and incarcerated people.

Despite obstacles to measurement, and the pressures for downward bias, published rates of prison violence routinely exceed estimates of community violence. Levels of sexual violence self-reported in prison surveys are ten to one hundred times greater than the rates from community surveys (Western 2021, 107). Prison misconduct rates (themselves a significant underestimate) have been found to exceed community-based self-reports of victimization by a factor of five to ten, and non-firearm homicide victimization rates are twice as high in prison as in free society (Western 2021). In short, to the limited extent that rates of prison violence can be compared to community violence, evidence indicates substantially higher rates of violence in prison.

Does violent victimization affect reentry? Victimization studies find that depression, anxiety, post-traumatic stress, and suicidality are all more common among people with histories of child and adult victimization. Direct injury and disability, gastrointestinal, and cardiopulmonary problems are common physical health problems that follow violent victimization in the general population (Priester et al. 2016). In studies examining prison violence specifically, victimization is associated with post-release drug use, emotional distress, depression, and criminal offending (Hochstetler et al. 2004; Wooldredge 1999; Zweig et al. 2015).

In an institution organized around surveillance and control, whose staff are charged with the enforcement of order, violent victimization and protracted fears of violence are a common and harmful consequence of incarceration. Anxiety about violence is widespread in prison, safety is elusive, and victimization is common. There are relatively few studies on the impact of victimization after prison release, but the larger research literature on exposure to violence points to enduring physical and mental health effects.

Health and Health Care

Health in prison can also be negatively affected by exposure to, and the management of, disease. People at high risk of incarceration are, on average, in poor health. Housing a high-needs population, prisons are responsible for health care on a vast scale and deliver care with “often-dramatic variation” in quality and access to health services (Wildeman and Wang 2017). The impact of incarceration on health can be seen in studies of mortality and infectious disease. Mortality studies find clear evidence of race differences where, unexpectedly perhaps, Black men are at lower risk of death in prison than in the community. On the other hand, prison conditions facilitate the spread of infectious disease, notably including HIV and tuberculosis (Rich et al. 2016), hepatitis, and most recently COVID-19 (Wang et al. 2020).

The health needs of the prison population are reflected in the high cost of correctional health care. As prisons grappled with escalating populations, the costs of health care in prison were also climbing. Data for the period from 2001 to 2015 show a broad increase in health care costs. California, the leader on correctional health expenditures, spent an average of $21,848 on medical expenses per person incarcerated in state prison in 2015, compared to $7,807 in 2001 (2021 dollars; Huh et al. 2017; Schiff et al. 2014). Growth in health care costs is related to a complex combination of factors including growth in the prison population, aging of the prison population, and rising health care costs in the wider community.

Even without population aging, incarcerated people are generally in poor health and have high needs for health care. Epidemiological data indicate three areas in which incarcerated people are in worse health than the general population: chronic conditions, infectious disease, and mental illness. Poor physical and mental health is often traced to persistent poverty over the life course and risky health behaviors such as needle use, heavy alcohol consumption, and smoking (Fazel and Baillargeon 2011). Researchers observe chronic conditions like hypertension, asthma, and arthritis at rates around 50 percent higher in prison than in the community (Binswanger et al. 2009; Fazel and Baillargeon 2011). Tabulations by Raphael and Stoll (2013) also show that the prevalence of serious mental illness such as bipolar disorder and psychotic conditions like schizophrenia are around five times higher in prison than in the general population. Estimates of substance use disorders in prison populations vary between around 10 and 50 percent (Fazel et al. 2017), greatly exceeding the availability of treatment (Belenko and Peugh 2005).

Despite the relatively high burden of disease, mortality rates in prison are not uniformly high. Standardized mortality ratios for White incarcerated men have been estimated at around 1.2, indicating an age-standardized mortality risk in prison about 20 percent higher than in the general population (Patterson 2010; Rosen et al. 2011). But standardized mortality ratios for Black incarcerated men have been estimated at around 0.5, indicating a death rate for Black men in prison that is about half the death rate for those in the general population (see also Wildeman et al. 2016). Evelyn Patterson (2010) examines the contribution of violence to prison and community mortality rates and finds that the low prison homicide rate of Black men cannot explain the mortality gap between prison and community. Analysis of cause-specific mortality data for men incarcerated in North Carolina finds that the excess risk is associated with cardiovascular disease, cancer, and infectious disease (Rosen et al. 2011). Researchers speculate that access to health services during incarceration improves the everyday treatment of chronic conditions as compared to the quality of health care in free society (Patterson 2010; Rosen et al. 2011). Other characteristics of prison life such as regular meals and consistent housing may also help the management of chronic conditions.

Against evidence of reduced mortality, research shows that prisons have clearly harmed health through the transmission of infectious disease. High rates of HIV and hepatitis B and C have been widely documented in US prisons. Recent estimates indicate that HIV prevalence in prison exceeds community rates by a factor of three to five, and hepatitis B and C prevalence exceeds community rates by five to ten times (Bick 2007; Gough et al. 2010). Screening at prison intake suggests around 80 to 90 percent of cases were present before incarceration, with the remainder transmitted in prison, mostly through sexual activity and needle use. A related line of research examines outbreaks of infectious disease, focusing on the spread of tuberculosis, influenza, and varicella (Beaudry et al. 2020). Each of these infections is airborne and spread through aerosol transmission (droplets) and contact with surfaces. The congregate living areas, dining halls, and recreation areas that make up the physical plant of prisons facilitate the spread of airborne pathogens, particularly in overcrowded conditions. Population turnover raises the risk of both bringing infections in from surrounding communities and transmitting disease back to the communities from which the incarcerated population is drawn.

The significance of correctional facilities for the transmission of infectious disease was strikingly illustrated by the novel coronavirus pandemic. One of the best case studies of COVID-19 dynamics estimates the reproduction number, called in epidemiological models, in an unnamed county jail (Puglisi et al. 2021). The reproduction number quantifies the new infections associated with a single infected case. An indicates an outbreak where the prevalence of infection increases at an increasing rate. At baseline before an outbreak, the reproduction number is written, and describes the speed at which an infection spreads. Testing for the novel coronavirus in the study jail continued for eighty-three days in 2020, providing daily measures of new infections. At the onset of the outbreak in the jail, was estimated at 8.23, meaning that a single person with the novel coronavirus infected eight others. At this reproduction rate, the spread of infection in the jail population was explosive (Puglisi et al. 2021).

Explosive COVID-19 outbreaks assailed prisons and jails throughout 2020 (Wang et al. 2020). According to the COVID Prison Project, one in five prisoners contracted COVID-19 during the first year of the pandemic (COVID Prison Project 2020; Schwartzapfel et al. 2020). Facilities such as Rikers Island jail, in New York City; Cook County Jail, in Chicago; and Marion Correctional Institution, in Ohio, suffered ferocious outbreaks that resulted in dozens of fatalities among staff and incarcerated people. The outbreaks also contributed to wider community spread. For example, one study finds that one in six of all COVID-19 cases in Chicago were linked to people who were detained and released from Cook County Jail (Reinhart and Chen 2020).

Like exposure to violence, the transmission of disease underscores the influence of an institutional environment that creates significant obstacles to social integration. We will see later how poor health, including substance use disorders, is associated with extreme health risks immediately after incarceration. While the prison delivers health care at great scale and cost—with evidence of success for at least fractions of the population—it contends with an unhealthy ecology that readily facilitates the spread of disease.

Solitary Confinement

Research on outcomes like prison violence and health suggests how prison conditions influence well-being. Prison conditions themselves have been a focus of study, exemplified by research on solitary confinement that provides the limiting case of penal severity and isolation (C. Haney 2018; National Institute of Justice 2016; Smith 2006). Solitary confinement is “a mainstay of prison management and control” (Shames et al. 2015, 28) and is widely used in its official purposes to punish misconduct, to control conflicts such as gang rivalries, and to protect those who are unsafe in the general prison population because of, say, youth or gender identity (Kapoor and Trestman 2016, 200). Solitary confinement typically involves incarceration in a cell for twenty-two or twenty-three hours each day with a brief time out for recreation or showers. Usually, people incarcerated in solitary confinements are restricted from having visits, making and receiving phone calls, or participating in programs. David Cloud and his coauthors (2015, 19) describe the physical space of a solitary confinement unit: “The typical cell is 60 to 80 square feet, with a cot, a toilet, a sink, a narrow slit for a window, and sometimes a small, molded desk bolted to the wall. In many facilities, cells have a steel door with a small slot for delivering meals. … Some solitary confinement units are nearly silent except for sudden outbursts; others subject prisoners to incessant cacophony of clanking metal doors, jingling keys, booted footsteps, and distressed voices reverberating off thick walls.”

The census of prison facilities, conducted by the Bureau of Justice Statistics, shows that the population incarcerated in solitary confinement increased from 3.0 percent of the total prison population in 1979 to 5.7 percent in 2005. In the 1980s and 1990s, the use of solitary confinement expanded with the proliferation of super-maximum security prisons that hold their entire populations under twenty-three-hour lockdown (Reiter 2016). A 2011–2012 survey of a national sample of the US prison population showed that 18 percent of respondents had been held in solitary confinement in the past year, and of those, about half were in isolation for thirty days or longer (Beck 2015). In 2015, the United Nations adopted minimum standards for the treatment of prisoners that prohibited “prolonged solitary confinement,” defined as a period greater than fifteen consecutive days. Solitary confinement in the United States regularly fails the international standard. In Pennsylvania, for example, the median period of solitary confinement is about twenty-eight days (Western 2021, 114). The use of solitary confinement is marked by disparities of race, age, and the recency of prison admission (Beck 2015; Simes et al. 2022). Racial disparity in solitary confinement and racial disparity in imprisonment combine to create high levels of penal isolation among Black men in the general population. Period prevalence estimates using state prison data from 2007 to 2018 show that around one in ten of all Black men in Pennsylvania have been held in solitary confinement by age thirty-two (Pullen-Blasnik et al. 2021).

Mental illness and solitary confinement are closely connected with evidence of causality flowing in both directions. Prisons hold people—many with significant physical, mental, and behavioral health problems—in a physical space that is often overcrowded, rule-bound, and managed by staff with broad discretion. People with mental illness may fail to respond to orders or have difficulty following prison rules. In some cases, they may act violently toward other incarcerated people or staff. Prison officials then respond to prisoners with mental illness “as they do to other prisoners who break the rules. When lesser sanctions do not curb the behavior, they isolate the prisoners in the segregation units, despite the likely mental health impact” (Metzner and Fellner 2013, 317). Several leading Eighth Amendment cases thus describe how solitary confinement becomes used for incarcerated people with high needs for mental health care.³

While mental illness is a risk factor for solitary confinement, psychologists also find evidence of psychological damage caused by solitary confinement (Arrigo and Bullock 2008; Kapoor and Trestman 2016). Clinical assessments of people in solitary confinement report prisoners’ being in a mental fog and having obsessive thoughts, perceptual distortions, hallucinations, and other forms of distress (Grassian 2006; C. Haney 2018). Evidence for the negative effects of solitary confinement on mental health is especially strong for those who are in strict isolation for long periods and for those with prior mental illness (Arrigo and Bullock 2008).

A few observational studies, focused on recidivism, have investigated the relationship between solitary confinement and postprison outcomes. David Lovell and his colleagues (2007), analyzing data from Washington state, find significantly higher rates of subsequent felony conviction for those released directly from supermax confinement compared to state prisoners matched on demographic characteristics and criminal history. Daniel Mears and William Bales (2009) find similar results for supermax confinement in Florida. These findings are consistent with studies exploiting natural experiments that find that assignment to higher levels of prison security is associated with high rates of recidivism (Chen and Shapiro 2007; Gaes and Camp 2009). Few studies analyze broader measures of social integration, although mortality rates have been found to be significantly higher for former state prisoners who have been held in solitary confinement (Brinkley-Rubinstein et al. 2019).

In sum, research on rehabilitation programs, prison violence, health, and solitary confinement points to the harms of incarceration that create obstacles to integration after release. Correctional programs can be modestly effective, but relatively few incarcerated people are involved in programs, and program participation has been declining at least since the early 1990s. The weak rehabilitative environment accompanies a high risk of violent victimization, and trauma during incarceration may be a source of ongoing post-traumatic stress. Prisons also expose incarcerated people to high risks of infectious disease that create greater needs for health care after release. Finally, solitary confinement and other harsh prison conditions are common causes of psychological distress, especially for people with preexisting mental health problems.

Collateral Consequences

Community return after prison is shaped by a legal environment that imposes restrictions on people with criminal records. These restrictions, known as “collateral consequences,” are typically established by state statutes and govern rights in a wide range of areas including employment, housing, health care access, social services, and voting. Collateral consequences are defined by the American Bar Association (ABA) as “a penalty, disability, or disadvantage, however denominated, that is imposed by law as a result of an individual’s conviction for a felony, misdemeanor, or other offense, but not as part of the judgment of the court” (American Bar Association 2018 citing the Court Security Improvement Act of 2007). Collateral consequences are sometimes applied to specific offense categories, such as convictions for violence or drug crimes, but they are typically attached to conviction status without reference to the facts of any specific case and are undecided by any judicial process. Instead, collateral consequences effectively institutionalize criminal stigma by designating those with convictions as undeserving of legal rights. Although collateral consequences are linked to a criminal record and not imprisonment, all formerly incarcerated people carry the burden of collateral consequences into the reentry process.

A large legal literature documents the variety and harms of collateral consequences that accompany release from incarceration (Chin 2011; O’Hear 2019; Olivares et al. 1996; Pinard 2010). Research on collateral consequences has focused on legal barriers that impair social integration by, for example, reducing employment opportunities, access to health care, or welfare benefits. We document collateral consequences and related statutory provisions that influence the reentry environment in seven main areas: an overall count of collateral consequences, employment restrictions including Ban-the-Box laws, health insurance coverage, eligibility for welfare benefits, identification documents at prison release, driver’s license suspension rules, and voting rights. Beyond the scope of this review, there are also other kinds of legal entanglement—for example, court-ordered child support or outstanding court fees—that create additional obstacles to social integration (L. Haney 2018; Harris et al. 2010).

State and federal collateral consequences have been documented in the National Inventory of Collateral Consequences. The inventory records roughly 44,000 collateral consequences. Most of these are in the areas of employment (19,334 collateral consequences) and occupational licensing (13,791). About half of all collateral consequences are mandatory—for example, by prohibiting business licenses for people with criminal records. Most of the remainder are discretionary or require criminal background checks. About three-quarters of collateral consequences are imposed permanently (US Commission on Civil Rights 2019). Most of these consequences are triggered by a felony conviction and nearly a quarter apply to a felony conviction of any kind (Umez and Gaines 2021, 4). The number of laws and regulations requiring criminal background checks increased through the 1990s and 2000s. The adoption of background check laws was unrelated to violent crime or incarceration rates but was regionally patterned: the laws were adopted at relatively high rates in the South and the West (McElhattan 2022).

Collateral consequences regulate employment by preventing employers from hiring or retaining workers with criminal convictions, by barring access to occupational licenses, and by restricting self-employment by prohibiting business licenses. Most restrictions apply to work in health care, the public sector, and education. Most employment-related collateral consequences apply broadly to “any felony” or “any crime.” In thirty-three states, occupational licenses can be denied for an arrest, even if it did not lead to a criminal conviction (Sibilla 2020). Five states —Alabama, Alaska, Nevada, South Dakota, and Vermont—have wide bans on licenses that apply to all arrest records, without time limits or mitigation through programs or employment history, and without opportunities for appeal (Sibilla 2020, 1).

Since 2015, thirty-three states have either removed or limited licensing barriers for people with conviction histories, and fifteen states have blocked their licensing boards from denying licenses solely on the basis of moral turpitude provisions (Sibilla 2020). (Most states still use vague character criteria to deny licensing to the formerly incarcerated.) Even when explicit bans are dropped, formerly incarcerated people often struggle to pay fees for licenses or preparatory coursework (Umez and Pirius 2018). Touted as a model for other states, Indiana has implemented an overarching ban on denying a license based solely or partially on a criminal record. Indiana also bans consideration of arrest records and expunged records, prohibits discretionary character standards, and implements a five-year time limit on the consideration of any criminal history except for violent or sexual crime convictions (Sibilla 2020, 40). Applicants in Indiana can also appeal a license denial, placing the burden of proof on the state to provide clear and convincing evidence supporting its decision to deny a license.

In addition to statutory limitations on employment, employers have been found to widely use criminal background checks in hiring (Holzer et al. 2006). In response to the use of background checks, a civil rights organization called All of Us or None, led by formerly incarcerated advocates, launched a Ban-the-Box (BTB) campaign in Oakland, California, in March 2003. The campaign gave rise to a nationwide movement that successfully lobbied thirty-seven states and over 1,250 cities and counties to “provide applicants a chance at employment by removing conviction and arrest history questions from job applications and delaying background checks until later in the hiring process” (Craigie 2020; Avery and Lu 2021, 2). By 2023, sixteen states had adopted statewide bans on criminal background checks at the initial job application, and many other jurisdictions had passed county or municipal bans or bans for public sector hiring. In 2016, United States Office of Personnel Management adopted Ban-the-Box rules for the federal government. The rules went into effect in March 2017 and delayed the point at which federal agencies could ask job applicants to provide criminal record information (Office of Personnel Management 2016). This rule was superseded in December 2021, when the Fair Chance to Compete for Jobs Act went into effect. The Fair Chance Act prohibits (with some exceptions) federal agencies and contractors from requesting the disclosure of criminal history record information before a conditional offer of employment.⁴ Like the federal legislation, state and municipal regulations do not completely prohibit criminal background checks in hiring but just delay the use of criminal record information beyond the first stages of the hiring process.

Several studies have assessed the effects of Ban-the-Box legislation but the empirical results are mixed (see Raphael 2021 for a review). Terry-Ann Craigie (2020), analyzing data from the 1997 cohort of the National Longitudinal Survey of Youth, reports difference-in-difference estimates showing increased public sector employment among workers with criminal convictions in counties with Ban-the-Box regulations. Showing similar results, Daniel Shoag and Stan Veuger (2021) report difference-in-difference estimates of the impact of Ban-the-Box laws in high-crime census tracts. Analyzing nationwide data, they find a 3 to 4 percent increase in employment and wages following the adoption of Ban-the-Box laws. Employment gains are particularly large in the public sector. Several studies report spillover effects where employers respond to restrictions on background checks by discriminating against minority job seekers in general (Agan and Starr 2018; Doleac and Hansen 2020), although several articles find no evidence of spillovers (Burton and Wasser 2024; Flake 2018; Rose 2021a).

Like the legislative activity around criminal background checks in employment, health care insurance coverage through Medicaid after incarceration has also been a leading area of policy reform. Under the “inmate exclusion policy” of the Social Security Act, federal Medicaid funds cannot be used for health services for adults and children in “public institutions,” including prisons.⁵ During incarceration, states can choose to either suspend or terminate Medicaid payments. From the point of view of reentry, Medicaid enrollment can be reactivated more quickly under suspension rather than termination, although even suspension can delay coverage for several months after prison release (Wang et al. 2020, 66–69). Policy consensus federally and across the states appears to be moving toward Medicaid suspension and facilitating Medicaid enrollment just prior to or at the point of prison release. In 2020, the National Academy of Sciences reported that just nine states terminated Medicaid during incarceration (Wang et al. 2020, 68). The 2018 SUPPORT for Patients and Communities Act prohibited Medicaid termination for young adults under age twenty-one and required eligibility redeterminations prior to release. California has received a waiver for the Medicaid coverage of incarcerated youth, and by 2022 fourteen other states had applied. In that same year, the Medicaid Reentry Act was introduced in the US Congress. The bill would extend Medicaid coverage to all incarcerated adults within ninety days from release. By 2022, twelve states had applied for demonstration waivers to drop the inmate exclusion policy and extend Medicaid coverage prior to prison release (Roach and Baseman 2022).

Eligibility for safety net transfer programs has also expanded for people with criminal records. The 1996 Personal Responsibility and Work Opportunity Reconciliation Act (PRWORA) imposed a lifetime ban on Supplemental Nutrition Assistance Program (SNAP) and Temporary Assistance for Needy Families (TANF) participation for anyone with a felony drug conviction. States could opt out of the ban, and most states have modified or removed the prohibitions. By 2023, twenty-five states and the District of Columbia had opted out of bans on SNAP and TANF eligibility, and another four states had opted out of the SNAP ban and modified the conditions for TANF receipt. Another twenty states have modified the bans for both federal programs or just modified the ban for SNAP. Only South Carolina continues its full SNAP and TANF drug felony conviction ban (Love and Sibilla 2023).

The Government Accountability Office (2005) assessed the impact of the welfare bans finding that among those with felony drug convictions released from prison in 2001, about a quarter would otherwise have been eligible for SNAP and about 15 percent would have been eligible for TANF. Gustafson (2008, 672) reports that about 92,000 recipients were removed from the welfare rolls for drug convictions from 1997 to 2002.

To apply for jobs and housing and enroll in government programs, applicants often need to provide government identification. Most states are supposed to collect incarcerated people’s identification records (or at least ascertain where these records are held), such as Social Security cards, birth certificates, and drivers’ licenses. However, many people enter prison without access to these records or with expired documentation. Release procedures in some states provide identification documents before prison release. These state governments recognize that “access to official documentation, such as Social Security Cards, birth certificates and photo identification is critical to successful reentry” (New York Department of Corrections and Community Supervision 2022). Even with this acknowledgment, identification services during reentry vary widely across states, with only twenty states helping people obtain identification either at release or immediately after. Only four states—Mississippi, North Carolina, Nevada, and Wisconsin—specify the provision of drivers’ licenses, allowing people to immediately drive.

Drivers’ licenses are doubly important for the reentry process, by providing state identification and conferring a legal right to drive. Formerly incarcerated people are at high risk of drivers’ license suspension or revocation because of unpaid fines and court fees or prior arrests and convictions. In 2017, forty-three states and the District of Columbia suspended drivers’ licenses because of unpaid court debt. In nineteen jurisdictions, suspensions were mandatory. Since then, twenty-four states and the District of Columbia have passed reforms to curb drivers’ license suspension in response to advocacy efforts and investigative reporting into the effects of suspension practices (Hirsch and Jones 2021). Table A.6 describes these trends in more detail. Finally, legal restrictions on the right to vote represent a domain of collateral consequences that is closely connected to civic engagement after incarceration and has been an area of intense advocacy and contest in the political system (Manza and Uggen 2006). In most states, imprisoned US citizens have no right to vote, and a patchwork of felon disenfranchisement laws across the states variably restricts voting rights for those under community supervision or with prior convictions. A 2022 analysis estimates that 4.6 million Americans, or 2 percent of the voting age population, are ineligible to vote because of a felony conviction (Uggen et al. 2022). Laws restricting voting rights for criminal convictions originated in colonial times, although many contemporary voting bans date from the passage of the Fifteenth Amendment, ratified in 1870, that extended voting rights to Black citizens. Felon disenfranchisement laws passed after Reconstruction were specifically intended to suppress the voting rights of Black Southerners. The goal of racial exclusion was often touted in political debates about disenfranchisement in the post-Reconstruction period (Behrens et al. 2003). In 1901, for example, John B. Knox, president of the Alabama Constitutional Convention asked: “What is it that we want to do?… Why it is within the limits imposed by the Federal Constitution to establish white supremacy in this State. … The justification for whatever manipulation of the ballot that has occurred in this State has been the menace of negro domination” (quoted in Behrens et al. 2003, 571). In the modern era, perceptions of racial threat continue to drive support for felon disenfranchisement. Supporting the theory of racial threat, analysis of the period 1950 to 2002 shows that the adoption of felon disenfranchisement law is closely related to the size of the non-White prison population.

Advocacy campaigns and ballot initiatives have generally expanded voting rights for people with criminal records over the last two decades. In 2023, New Mexico and Minnesota restored voting rights to citizens on parole while Virginia instead reverted to requiring people to apply for rights restoration on an individual basis (National Conference of State Legislatures 2023). In Florida, voters passed a 2018 ballot measure that was intended to restore the voting rights of roughly 1.4 million formerly incarcerated citizens in the state, an estimated one of every four Black men in Florida (data tabulation from The Sentencing Project 2022). However, the legislature severely restricted re-enfranchisement just months later by requiring full payment of court debts before rights restoration (Sanchez 2022). Florida Governor Ron DeSantis also established an election police unit, specifically targeting formerly incarcerated people for allegedly committing fraud by trying to vote (Rozsa 2023).

Voting rights today remain widely withheld from incarcerated and formerly incarcerated citizens. In twenty-one states, voting rights are revoked upon incarceration and at release citizens can reregister to vote. In another fourteen states, disenfranchisement extends at least throughout the period of probation or parole. States that disenfranchise during community supervision often require the payment of outstanding fines, fees, or restitution before voting rights are restored. Another eleven states impose an indefinite period of disenfranchisement or require some other additional action before someone with a felony conviction history can once again exercise their right to vote. In only three jurisdictions—Maine, Vermont, and the District of Columbia—is a right to vote sustained during incarceration. In the District of Columbia, this right was restored in 2020. Because sentencing in the district results in Federal Bureau of Prisons incarceration, voter registration information and ballots must be disseminated and collected in facilities across the country for incarcerated DC residents.

The landscape of collateral consequences across fifty states shows strong regional variation (table 3; see tables A.1–A.7 for more detail). Ban-the-Box measures are concentrated in the Northeast. Western states widely allow motorists to keep drivers’ licenses despite nonpayment of court fees. Northeastern and Western states commonly provide state identification at discharge from prison. Northeastern states are most likely to reinstate Medicaid and retain SNAP eligibility after prison release. Southern states are more likely to deny voting rights to people with criminal convictions, as they have historically. Table 3 does not reflect a gradual movement to the liberalization of collateral consequences, particularly beginning in the 2000s. Medicaid and safety net programs have become more widely available, penalties for outstanding court financial obligations have been relaxed, and efforts to extend voting rights have been at least partially successful.

Probation and Parole

Supervision, control, and an elevated risk of re-incarceration continue after prison for those on probation or parole. The scale of community supervision has increased alongside the scale of incarceration, creating a secondary, community-based, system of penal control (Phelps 2020). Around 70 to 80 percent of all those leaving prison are released to the oversight of a probation or parole officer and remain at risk of returning to prison, not only for new crimes but also for violations of the conditions of release.

Probation and parole began as Progressive Era reforms of the late nineteenth century that aimed to provide community-based support for people with criminal convictions (Rothman 2002a). Probation is a judicial function, assigned by a court at sentencing, while parole is an executive function determined by prison authorities during incarceration. In practice, probation and parole operate in similar ways, involving a period of community supervision managed by a probation or parole officer. Historically, community supervision was conceived as a kind of individualized case management where probation and parole staff would help with employment, housing, and other supports needed for reintegration into society. Coinciding with the punitive turn in sentencing policy, probation and parole too became more punitive starting in the 1970s. Joan Petersilia (2003) describes a transformation in parole from social work to law enforcement in which the job of community supervision shifted from supporting reintegration to monitoring the compliance with conditions. Vincent Schiraldi (2023, 40) describes a similar “death of rehabilitation,” unfolding from the 1970s through the 1990s, in which “the rehabilitative ethic that probation and parole embodied ran headlong into this punitive shift, ushering in an era of rising caseloads and diminishing resources. Supervision agencies became either overtly punitive or, to a large extent, actuarial risk managers.”

The legal vulnerability of people on probation and parole is rooted in the legal conditions of release. Periodic surveys of standard parole conditions show a steady increase in new conditions from the early 1980s to the 2000s (Travis and Stacey 2010; Wiggins et al. 2022). Among the most common conditions are compliance with the law, reporting residence changes, restrictions on firearms, maintaining employment or programming, and regular reporting. Supervision has become more intrusive as drug testing, home visits and searches, and fee payment have become common since the 1980s (Travis and Stacey 2010; Wiggins et al. 2022).

The constraining force of conditions of release is amplified by case law that reduces the scope of Fourth Amendment protections against intrusions by police. Privacy protections conferred by the Fourth Amendment against searches and property seizures are denied to parolees, just as they are to incarcerated people.⁶ The Supreme Court has held that people on probation and parole are not extended privacy protections as other citizens would be, because “probation, like incarceration is a form of criminal sanction” and “parolees have [even] fewer expectations of privacy than probationers” (Congressional Research Service 2012 citing 483 U.S. 868 [1987]; Samson v. California, 547 U.S. 843, 850 [2006]). In these decisions, the Court treated probation and parole as more akin to incarceration than freedom, rendering Fourth Amendment protections largely inapplicable to people on community supervision.

While there is little systematic evidence of the crime-reducing effect of probation or parole, several studies show how community supervision operates as a kind of recidivism trap, returning people to incarceration not for new offenses but for “technical violations,” noncriminal violations of the conditions of release. In 2022, 469,000 individuals were admitted to state or federal prison and about a quarter of those admissions, 112,000, were for violations of conditional release. A number of studies document the relatively high risk of re-incarceration among parolees, where analysis aims to adjust for criminal history and involvement in crime (Harding et al. 2017; Rose 2021b).

The discretion involved in parole revocation raises the possibility of additional racial disparity and bias that exacerbate the large racial disparity in incarceration. Black probationers have been found to have relatively high rates of revocation but were less likely to be discharged early or to successfully complete the probation sentence, net of demographic characteristics, a risk score, and offense type (see also Olson and Lurigio 2000; Steinmetz and Henderson 2016; Vito et al. 2012). A detailed analysis of revocation in North Carolina by Evan Rose (2021b), found relatively high revocation rates among Black parolees who were unlikely to commit new crimes but were unable to pay supervision fees. An analysis of racial differences in probation revocation in Texas, Iowa, Oregon, and New York City from 2007 to 2010 found Black probationers experienced relatively high revocation rates in all jurisdictions, being 18 to 36 percent more likely to be placed in custody compared to White probationers (Jannetta et al. 2014). Between 20 and 49 percent of the observed racial disparity remained unexplained after researchers controlled for demographics, offense type, criminal history and other covariates. Similar results were obtained in the analysis of probation revocation from a large Southwestern state.

Conceived as a rehabilitative measure to provide community-based support for people with criminal convictions, community supervision became focused on monitoring compliance with a growing list of release conditions. The punitive character of parole is most clearly indicated by the increased risk of incarceration that also contributes to racial disparity in prison admissions.

The Reentry Policy Movement

In reviewing research on collateral consequences and community supervision, we find significant obstacles to social integration created by the process of criminalization and punishment, but we also see a liberalization of the reentry environment in the 2000s. These more recent policy shifts were part of a larger movement for reform that focused on the transition from prison to community to reduce the harms and perhaps the scale of incarceration.

By the early 2000s, support for rehabilitative programming had regained some momentum among policymakers and correctional practitioners. Programs for transition from prison to community became a focus of policy. Policymakers and scholars described the emergence of a “reentry movement” supported by government agencies, professional associations, and advocacy groups (Jonson and Cullen 2015; Travis 2007). Political opportunity for the proliferation of reentry programs at the state and local levels opened with a variety of federal initiatives that spanned Democratic and Republican presidential administrations. Under Bill Clinton, Attorney General Janet Reno urged the National Institute of Justice (NIJ) director, Jeremy Travis, to develop research and policy initiatives for people coming home from prison. NIJ subsequently sponsored the development of eight “reentry courts” and another five “reentry partnerships” between correctional agencies and community organizations.

Federal support for reentry policy went further under George W Bush. In 2003, the federal government allocated $110 million, across eighty-nine separate programs, in the Serious and Violent Offender Reentry Initiative (SVORI). SVORI programs were intended to reduce recidivism and improve employment, health, and housing. In his 2004 State of the Union address Bush spoke to the plight of people leaving prison and proposed a new federal program:

In the past, we’ve worked together to bring mentors to the children of prisoners and provide treatment for the addicted and help for the homeless. Tonight, I ask you to consider another group of Americans in need of help. This year, some 600,000 inmates will be released from prison back into society. We know from long experience that if they can’t find work or a home or help, they are much more likely to commit crime and return to prison. So tonight, I propose a four-year, $300 million Prisoner Re-Entry Initiative to expand job training and placement services, to provide transitional housing and to help newly released prisoners get mentoring, including from faith-based groups. America is the land of second chances, and when the gates of the prison open, the path ahead should lead to a better life. (Quoted in Jonson and Cullen 2015, 533)

In 2007, Congress passed the Second Chance Act to provide federal funding for reentry programs. Since 2010, annual Second Chance funding has varied between $60 million and $100 million, supplementing state and local support for social services for formerly incarcerated people.

The breadth of community-based reentry programs was cataloged by the encyclopedic Report of the Re-Entry Policy Council, which listed thirty-five separate policy areas along with recommendations for best practices and a long list of exemplary programs around the country (Re-entry Policy Council 2005). Policy recommendations and programming spanned the areas of health care, housing, continuity of care, employment opportunities, identification and program enrollment, and dozens of other topics. By 2011, the Obama administration had convened a Federal Interagency Reentry Council that was charged with coordinating the work of twenty federal agencies in support of the federal reentry effort and funding at the state and local levels. Bipartisan support for reentry policy was matched by public opinion data showing large majorities in support of social services for people coming home from prison (Jonson and Cullen 2015, 535–36). In the words of Joan Petersilia (2009, 255), reentry policy has had “staying power,” becoming institutionalized in correctional practice.

The reentry policy movement shifted the policy conversation around incarceration and provided new funding for reentry programs. Although public spending on incarceration still vastly exceeded the funding of reentry programs, support for reentry policy among elected officials softened the harsh rhetoric of the earlier tough-on-crime era. New programs and policies dulled the sharp edge of punitive excess in which incarceration was often intended to be severe and humiliating.

How much did the reentry policy movement change conditions on the ground for people who were leaving prison? Was recidivism reduced and social integration increased? It is difficult to answer this question in the aggregate. By the peak of mass incarceration in 2008, there was no national reentry policy but a vast piecemeal collection of public and private programs, some long-standing, some springing up with the influx of government funding. Some programs were mandated by courts or parole boards and imposed fees that were often paid to private vendors. Other programs were voluntary, offering support to those who asked. One window into the effects of large-scale reentry efforts is provided by three high-profile federally funded initiatives: SVORI, the Justice Reinvestment Initiative (JRI) supported by the Department of Justice’s Bureau of Justice Assistance (BJA), and the Project HOPE initiative that replicated a promising probation reform program from Hawaii.

Although fielded under the banner of a single policy initiative, SVORI consisted of eighty-nine separate reentry programs for people under age thirty-five who were convicted of serious violent crimes or who had long histories of repeated incarceration. Local programs received between $500,000 and $2 million in support of community-based social services for formerly incarcerated people who met the eligibility criteria. In many ways, SVORI represented the state of reentry efforts nationwide: highly decentralized and minimally funded. The aggregate effects of SVORI were assessed through an evaluation of thirteen sites. Program participants were contrasted with a randomized comparison group in two sites and a matched group of program-eligible respondents in the remaining eleven. SVORI participants received relatively high levels of social services support in the areas of employment, education, skills training, transitional services, and service coordination. Evaluation results indicated small program effects, and in general no substantial differences were found between treatments and controls in the areas of housing, drug use, self-reported recidivism, or re-incarceration (Lattimore and Steffey 2009).

The idea of justice reinvestment was originally motivated by research on the spatial concentration of incarceration and the identification of “million-dollar blocks.” Million-dollar blocks were small localities with high incarceration rates that accounted for millions of dollars in state prison budgets. Policy advocates Susan Tucker and Eric Cadora (2003) argued that crime and incarceration could be reduced by redirecting spending from incarceration to community investment in education, jobs, housing, and health care. The idea of redirecting savings from incarceration to communities was adopted by the JRI. Although not a traditional reentry program, JRI was part of a wave of reform that aimed to reduce the social costs of incarceration and perhaps reduce incarceration by increasing public spending on social programs. From 2007 to 2017, BJA in partnership with Pew Charitable Trusts and the Council of State Governments disbursed $160 million and provided technical assistance in thirty-five states to support the justice reinvestment model. The federal implementation of JRI departed significantly from its conceptual origins as a strategy for community investment. William Sabol and Miranda Baumann (2020) write that the evolution of JRI “represented a marked shift in the emphasis of reinvestment away from place- and community-based offender management to a state-level strategy that emphasized improving corrections departments’ practices through expanded use of evidence-based programming.” Their review of the impact of JRI found no evidence for reduced incarceration. Indeed, non-JRI comparison states experienced slightly greater declines in incarceration than JRI states. Sabol and Baumann (2020) also report that there were no systematic evaluations of the impact of JRI on crime or recidivism.

Project HOPE, or Hawaii’s Opportunity Probation with Enforcement, was a court-based anti-recidivism program that emphasized deterrence for people on felony probation. First developed in Honolulu in 2004, HOPE focused on “swift and certain sanctions.” HOPE probationers received random drug tests, and any violations resulted in immediate, but relatively short periods, of incarceration. HOPE was evaluated in a randomized trial, and probationers subject to swift and certain sanctions were found to have significantly fewer violations of their community supervision, fewer new arrests, and fewer days sentenced to incarceration (Hawken and Kleiman 2009).

Swift and certain principles caught the imagination of analysts and policymakers and have been influential for the correctional role in reentry policy. By 2014, these principles had been adopted in at least forty jurisdictions, and the conservative American Legislative Exchange Council developed model legislation for a “Swift and Certain Sanctions Act” (Duriez et al. 2014). One seasoned policy analyst wrote, “There aren’t any magic bullets that can end America’s continuing battle with crime and addiction. But HOPE comes closer than anything we’ve seen in a long time” (Adam Gelb quoted in Duriez et al. 2014, 57). In 2014, NIJ funded a replication study of HOPE in four other sites, in Arkansas, Massachusetts, Oregon, and Texas. Evaluation of the replication pooled samples across the sites and found that HOPE probationers were more likely to have violations and twice as likely to have drug violations despite a fivefold increase in drug testing. On recidivism outcomes, results were similar between treatments and controls. The evaluation team concluded “that there is little to support a conclusion that HOPE or HOPE-like programs will produce substantial improvements over PAU [probation as usual]” (Lattimore et al. 2018).

Although SVORI, JRI, and HOPE represent a small fraction of reentry policy efforts, their evaluations suggest the challenges faced by criminal justice policy in improving social integration after incarceration. SVORI, JRI, and HOPE all struggled to fundamentally reorient the work of prison, parole, and related agencies or to develop programs that consistently reduced new arrests or incarcerations.

Family

Family members have been found to provide the most immediate and substantial social support after incarceration (Berg and Huebner 2011; Harding et al. 2019; McKay et al. 2019; Western 2018). Families help their relatives who are coming home after incarceration by providing emotional support, material security, and network connections to employment and other opportunities. Family support is often foundational, promoting other kinds of social integration including employment (Sullivan 1989), stable housing (Sirois 2019), and desistance from crime (Sampson and Laub 1993). Although research focuses on the positive role of families in the reentry process, reuniting with family can nevertheless be marked by stress and conflict.

Many studies emphasize the temporal quality of reentry in which the first days and weeks after prison release are accompanied by feelings of anxiety, stress, and isolation among people leaving prison (Irwin 1970; Western et al. 2015). Called “the stress of transition” (Travis 2005) this period has been found to be associated with poor social integration months after release. Emotional support from families counteracts the stress of transition. Emotional support describes the warm and affectionate interactions that can provide a sense of connectedness, affirmation, and safety that make a person feel valued and loved (Sugie and Augustine 2020; Cobb 1976). Research on stress and coping argues that emotional support contributes to esteem, well-being, and overall health (Cobb 1976; House et al. 1988). Positive emotional connection is often expressed in intimate household relationships for relatives who return to a family home after incarceration. Shared meals, everyday conversation, and ritual events, such as welcome parties, can commemorate the transition from prison to community as a return “home,” a place where one is valued and belongs (Western 2018).

Families also make important contributions to material well-being by providing housing and financial assistance after incarceration. For example, in a major study of reentry among Michigan parolees, conducted by David Harding and colleagues (2014), about 60 percent of the sample first resided with family after leaving prison. Most of the family members who provided housing were a generation older than their formerly incarcerated relatives. Typically, formerly incarcerated men and women (with a median age around thirty-four) lived with their parents and grandparents straight after prison. Another reentry study, the Boston Reentry Study, interviewed men and women in the year after prison release in Massachusetts and found that study respondents were similarly supported, mostly by mothers, older sisters, and grandmothers. Living in a private household with older family members conferred not just shelter but also meals, clothing, sociability, and safety that was generally unmatched by the institutional alternatives provided by shelters and transitional housing programs (Sirois 2019; Western et al. 2015). The intergenerational and gendered character of family support after incarceration suggests how the costs of reentry are unevenly distributed through family networks.

Although older family members are a staple of emotional and material support, research on family relationships after incarceration often focuses on children and romantic partners (Braman 2004; Comfort 2008; Nurse 2002). This research has examined not the support received by formerly incarcerated parents but the levels of contact and support they provide to their co-parents and children (Kane et al. 2015; Western and Smith 2018). Relationships between formerly incarcerated parents, children, and partners can be fraught or contentious (McKay et al. 2019). In the case of formerly incarcerated fathers, mothers can act as gatekeepers determining if and when fathers can have contact with their children (Edin and Nelson 2013). In the Boston Reentry Study, two-thirds of respondents reported having biological children, and about 40 percent reported being social parents, feeling like a mother or father figure to other children. Two main factors in the Boston study were closely correlated with contact between children and formerly incarcerated parents: multiple-partner fertility and stable housing. Where parents had children with several different partners, they likely had contact with children of just one partner. Where formerly incarcerated parents lived in shelters or rooming houses, they were also in less regular contact with their children. Parents who lived in private households—typically with mothers, aunts, grandmothers, and older sisters—were more likely to receive visits from their children, and contact with children was generally more consistent (Western and Smith 2018).

Relationships with children and the obstacles to reunification have been a significant theme in research on women’s reentry (Richie 2001). Family and intimate relationships have often been sources of serious violence in the lives of incarcerated and formerly incarcerated women, and life histories of victimization are often related to drug use and involvement in drug distribution (Chesney-Lind 2002; Haney 2010; Sered and Norton-Hawk 2014). The close relationships between domestic violence and criminal justice involvement have been observed particularly for Black women (Gross 2015; Jones 2004; Richie 2012). Researchers also describe the tenuous connections of incarcerated mothers to their children. In many cases, mothers had lost custody of their children prior to incarceration, and reestablishing relationships with children and a positive sense of oneself as a parent were a significant focus of reentry (Leverentz 2014). The limited availability of stable housing, treatment services, and the scrutiny of child protective services, however, all present obstacles to safely taking custody of children. As Beth Richie (2001, 379) writes in her study of reentry among mothers: “Most women note that their children’s suffering continues after their mother is released, and the issues related to custody of children, repairing relationships, parenting, and family reunification and stabilization are urgent and stressful. … Often the legal, practical, and emotional challenges are so overwhelming, that women and their children never resolve the damage caused by the ruptured relationships.”

Family support after incarceration varies with age. Reentry studies find that family support is more common for younger people in their twenties and early thirties. Isolation, time alone, and friendlessness is more common for those who are older, in their forties and fifties (Crawley and Sparks 2006; Sirois 2019; Western et al. 2015; Wyse 2018). For older men and women, qualitative research suggests that long histories of incarceration, substance use, and mental illness fray family ties and contribute to social isolation.

Findings of the supportive role of older relatives and uneven involvement with children underline the complex character of family connection after incarceration. Family relationships are at once a resource and a brittle social tie that is easily broken and difficult to restore. The complex structure of kin relations, housing insecurity, the vulnerabilities of poor health or addiction, and recidivism all present obstacles to nourishing the family relationships that indicate social integration. Despite these challenges, intergenerational family support remains a key source of material security and emotional reassurance immediately after incarceration.

Housing

Housing insecurity and criminal justice involvement are intimately connected. People without housing often conduct their private lives in public space and become subject to police scrutiny, risking arrest. Housing insecurity and homelessness are common after incarceration. Stephen Metraux and Dennis Culhane (2004) followed a large cohort leaving prison in New York state for New York City from 1995 to 1998. Over a two-year follow-up period, 11 percent of the sample of 48,000 prison releasees stayed in the New York City shelter system. Consistent with the temporal character of reentry and the initial period of vulnerability, the likelihood of shelter use was highest in the first month after prison release. Housing insecurity, like family support, was age-graded, with men over age forty having the highest likelihood of shelter use. The rate of shelter use was also about 20 percent higher for Black men and women, compared to Whites and Latinos. Other studies also find a close relationship between shelter use, mental illness, and substance use disorders (Geller and Curtis 2011; Remster 2019). In this context, the shelter is not just an indicator of housing insecurity but also a marker of isolation for older men whose family relationships have often eroded owing to long struggles with addiction and illness.

Consistent with research on shelter use, both the Michigan and Boston reentry studies reported extreme levels of housing insecurity and residential mobility. In Michigan, parolees changed addresses 2.6 times a year. In Boston, only a fifth of the sample remained in the same residence through the first year after prison release, and half the sample moved three times or more (Sirois 2019). Amanda Geller and Mariah Curtis (2011), analyzing data on a disproportionately low-income sample of poor fathers with young children in the Fragile Families survey, report that the formerly incarcerated experience about twice as much housing instability as those who have never been incarcerated, controlling for age, race, and measures of human capital.

Rates of shelter use among formerly incarcerated people are much higher than in the general population, but only a relatively small fraction of people released from prison are in shelters on any given day. Around 70 to 75 percent of Michigan parolees were living in private households in the first six months after prison (Harding et al. 2013). Similarly, the Boston Reentry Study respondents were mostly living in private households with family and friends in the first year after prison release (Western et al. 2015).

Although formerly incarcerated men and women often live in private households, few live in their own independent households. In the Boston Reentry Study, in a sample of 122 men and women leaving state prison, none lived in their own independent housing in the year after prison. Besides the economic resources needed for independent living, landlords are reluctant to rent to tenants with criminal records (Leasure and Martin 2017). Discrimination based on a criminal record is not prohibited by the Fair Housing Act (FHA), but the Department of Housing and Urban Development (HUD) has recognized the adverse effects of this discrimination for people of color, which then does violate the FHA in cases of racially disparate impact (Department of Housing and Urban Development 2016).

Employment, Income, and Economic Well-Being

The first months after prison release are characterized by extreme economic hardship. Formerly incarcerated men and women experience high rates of unemployment and poverty and rely heavily on family support and safety net programs. Economic hardship after prison is closely correlated with a variety of other socioeconomic disadvantages. We have seen that incarceration is concentrated in highly disadvantaged communities, educational attainment in the prison population is very low, chronic health problems are common, incarcerated people often have little work experience, and many have histories of institutionalization prior to prison. Extreme socioeconomic disadvantage prior to incarceration helps explain extreme poverty and economic insecurity found by reentry studies in the first year or so after prison release. In considering economic well-being after prison, employment is often a major focus of research, but in social contexts of deep correlated disadvantage, employment is just one specific survival strategy among many.

The Michigan study of parolees offers a detailed picture of the patchwork of financial support and poverty-level incomes pieced together from families, social programs, and employment (Harding et al. 2014). Qualitative interviews with twenty-two respondents over a three-year period from prison release showed that “over one-quarter experienced a trajectory of persistent desperation and struggle, experiencing frequent periods of homelessness and housing instability, relying on short-term measures such as social support and social services to meet their most basic needs, and never attaining the stability of resources needed to make ends meet on a day-to-day basis. About a third attained some degree of stability but intermittently experienced periods of desperation and struggle for survival” (Harding et al. 2014, 461).

Similar results indicating the importance of welfare payments and family support were found in the Boston Reentry Study, where employment earnings accounted for less than half of all incomes earned after prison. In Boston, the “analysis shows that most prisoners leave incarceration for poverty in which housing is often insecure and incomes are supplemented by government programs. In the struggle to obtain community membership, former prisoners rely on mothers, grandmothers, and sisters to play a caring role to feed, provide for, and house their kin, now in their twenties and thirties. For those in their forties and older, family ties have deteriorated, and the state becomes the supporter of last resort” (Western et al. 2015, 1540–41).

The Boston and Michigan reentry studies indicate the importance of public assistance for economic security, but few studies systematically examine social program participation at a national level. One analysis, by Brielle Bryan (2023), analyzes welfare benefit receipt among formerly incarcerated respondents in the 1979 cohort of the National Longitudinal Survey of Youth. Safety net programs can be divided into contributory social insurance programs, like Social Security Disability Insurance (SSDI) and unemployment insurance (UI), and means-tested benefits, like Supplemental Security Income (SSI) or the Supplemental Nutrition Assistance Program (SNAP), for recipients living at or near the poverty line. Benefits from social insurance programs require a history of earnings and program contributions. Because many formerly incarcerated respondents in the NLSY had little employment history, they received relatively higher total payments from the anti-poverty programs, SSI and SNAP, but received less from social insurance programs like SSDI and UI whose payments are more generous.

While safety net programs provide a rudimentary level of economic stability, rising out of poverty after prison often depends on employment (Western 2018). Employment and its unequal distribution across the reentry population has been a major topic for research. Here researchers have focused on racial inequality and the impact of discrimination and social support (Lyons and Pettit 2011; Pager 2003; Western and Sirois 2019; Yang 2017). Much of the research in this area focuses on young men, whose labor market participation rates are highest upon reentry.

Employment has been a research focus, not only for its economic benefits, but for its wider social effects. Improved material well-being that comes with employment also offers pride and self-sufficiency (Leverentz 2014, 155–56; Western 2018, 83–90). The regular routines of a steady job and supervision in the workplace are sources of informal social control that reduce opportunities for crime. Thus, employment is also associated with reduced recidivism. In part because a stable job is associated with desistance from crime (Sampson and Laub 1993), steady employment has been viewed as “a critical dimension of successful prisoner reentry” (Baer et al. 2004).

Formerly incarcerated job seekers face at least three challenges to finding work (Western et al. 2001). First, there are significant deficits of human capital, health, and work experience that place formerly incarcerated job seekers at a disadvantage to those who have never been to prison. Second, incarceration can erode social network connections to job opportunities that are often important for access to skilled or stable employment. Third, the stigma of a criminal record, often discovered through background checks, can deter employers from hiring. A criminal record can signal dangerousness, dishonesty, and unreliability in the workplace and may expose prospective employers to liability for negligent hiring. These mechanisms help explain high levels of unemployment after incarceration and the low level of earnings when job seeking is successful (Apel and Sweeten 2010; Grogger 1995; Kling 2006; Lyons and Pettit 2011).

Much of the research on earnings and employment after prison aims to estimate the effect of incarceration on the labor market. Reentry research aims to explain variability in labor market experiences among those who have been incarcerated. Here, a variety of studies find that the mechanisms for poor labor market outcomes can work differently across racial groups. The labor market disadvantage of formerly incarcerated people of color has been traced to intensified criminal stigma and weak network connections to employment. Devah Pager’s (2007a; Pager, Western, and Bonikowski 2009) audit studies in Milwaukee and New York City find that callback rates for White job seekers with criminal records are two to three times higher than for Black job seekers who present equivalent résumés. Pager (2007, 115) argues that a criminal record reinforces racial stereotypes and Black job seekers with criminal records face “an intensification of stigma.” The audit studies shows that personal interaction with employers in a job interview can moderate the negative effects of a criminal record, but White applicants are given greater opportunities for personal interaction than Black (Pager, Western, and Sugie 2009, 201–2). For the audit research, a criminal record compounds the disadvantages of race in the labor market.

While the audit method examines an anonymized labor market in which employer and job seeker are unknown to each other, qualitative field studies emphasize personal connections. Mercer Sullivan’s (1989) field research with criminally involved White, Black, and Latino youth in New York City finds steadier employment and higher pay among Whites by the time they reach their early twenties. Although Sullivan’s three study groups all had similar levels of education, and all had trouble at school, the White subsample came from families and neighborhoods that were more affluent and in which employment was abundant. Family connections to jobs were decisive for the young White men. Young men in the Black subsample had few social ties to employment. Latinos found work more easily than Black youth but were largely confined to low-wage manual jobs (see also Black 2010). Sullivan (1989, 103) concludes that “personal networks, not human capital in the form of either education or work experience, accounted for most of the disparities between the neighborhood groups.” Sandra Smith (2007) similarly finds that network ties also shape employment opportunities after incarceration. In her analysis, Black men and women with prison records are more likely to seek work by themselves, and their family and friends are reluctant to recommend them. In the low-wage labor market where employers often rely on referrals, low-income Black job seekers with criminal records are disadvantaged by social networks that are uncooperative and distrusting.

Racial inequalities in employment for formerly incarcerated workers emerge in contexts in which discrimination and network effects have been observed more broadly in urban low-wage labor markets. Thus, William Julius Wilson (1996) describes the reluctance of employers to hire young Black men because of concerns about their reliability and honesty. Deirdre Royster (2003) in her study of young Black and White job seekers finds that Whites are more likely to be connected by teachers and older White men to steady, blue-collar jobs.

Economic life after incarceration is embedded in this larger racially differentiated structure of opportunities that confronts low-skill workers. The stigma and social network accounts of the economic disadvantage of Black and Latino workers after incarceration suggest a theory of compounded disadvantage in which racial inequalities in incarceration are amplified in the labor market after release (Lyons and Pettit 2011).

Evidence of this compounded disadvantage, or “racialized reentry,” was found in the Boston Reentry Study (Western and Sirois 2019). In the Boston study, half the sample were jobless in any given month and average earnings are well below the poverty level. Jointly modeling employment and earnings, Black and Latino respondents were found to have lower total earnings than Whites even after accounting for health, human capital, and criminal involvement. Indeed, White respondents had worse health and higher rates of drug addiction, so accounting for covariates increased the regression-adjusted race gap in earnings. In a striking illustration of the effects of social support, nearly all the non-poverty employment was concentrated among a small number of White men who were employed in union jobs in the construction industry. In these cases, the men had union cards before incarceration and in many cases family members paid union dues while their relatives were incarcerated.

In sum, acute economic hardship typically follows incarceration, and researchers report high levels of poverty and joblessness in the years after prison release. Much of this appears related to disadvantages of schooling, health, criminal involvement, and institutionalization. Those leaving prison are thus reliant on family support and government programs. Employment also offers economic support, and here, race differences in social connections and criminal stigma tend to result in weaker employment outcomes for formerly incarcerated Black and Latino workers.

Health

Research on health status in prison showed a population burdened by high rates of chronic health conditions, exposure to infectious disease, and high levels of substance use disorders and serious mental illnesses. Health problems at the point of prison release create significant obstacles to social integration. Chronic pain and disability may prevent employment and contribute to family conflict, and serious mental illness may be associated with homelessness. Drug and alcohol use and long histories of addiction have been prominent obstacles to life stability in qualitative studies (Black 2010; Harding et al. 2014; Kirk 2020; Miller 2021; Richie 2001; Sampson and Laub 1993; Western 2018). Addiction is associated with chronic conditions (such as liver disease and arthritis), unemployment and homelessness, social isolation, and new criminal justice system contact.

Evidence of physical health problems and failures of health care after prison release are provided by studies showing high rates of hospitalization and mortality in reentry populations. Emily Wang and her colleagues (2013) analyzed a large sample of Medicaid administrative records including a subsample who were released from correctional facilities. Compared to all Medicaid recipients, those released from prison were younger, more likely to be Black, and more likely to be eligible for Medicare because of physical disability. Formerly incarcerated Medicaid recipients were compared to unincarcerated recipients matched on age, other demographics, zip code, and disability status. The key outcome was hospitalization for an “ambulatory-care sensitive condition.” Such conditions tend to be chronic, preventable with primary care, and include, for example, diabetes, hypertension, asthma, bacterial pneumonia, and angina. In the formerly incarcerated sample, 1.4 percent were hospitalized in the first week after prison release compared to 0.6 percent in the comparison group. After ninety days, 8.3 percent of the incarcerated group had been hospitalized compared to 4.8 percent in the comparison group.

Acute health risks immediately after release from incarceration are also indicated in mortality studies. Ingrid Binswanger and colleagues (2009) analyzed data on a cohort released from prison in Washington state between 1999 and 2009. Linking prison data to death records from the National Death Index, 3.2 percent of the formerly incarcerated sample died at some point within an average 4.4-year follow-up period. Death rates were strikingly high in the first weeks after prison release. In the first week after release, the overall annual death rate was about 2,500 per 100,000, with about two-thirds of the fatalities due to drug overdose related to the use of pharmaceutical opioids. By the third week after prison release the overdose death rate had dropped to about 250 per 100,000. Very high rates of overdose mortality after prison release were also found in North Carolina data. Those data also showed relatively high mortality rates in the first year after prison release among those in solitary confinement (Brinkley-Rubinstein et al. 2019).

Physical and mental health problems are often analyzed separately, but studies of prison and reentry populations find that chronic conditions, chronic pain, and mental illness, including substance use disorders, often cluster together (Jahn et al. 2022; Link et al. 2019; Western and Simes 2019). Untreated comorbidities have been found to be associated with recidivism (Fahmy and Mitchell 2022; Link et al. 2019), suggesting the importance of primary care and mental health services particularly in the period of acute risk immediately upon community return. Acute health risks immediately upon community return and clustering of health problems have led medical researchers to describe reentry as a significant health event that can have serious effects on the well-being of people leaving prison (Wang et al. 2013, 2017). The timing of severe health problems so soon after prison release also underlines the utility of Medicaid reforms that would provide health care coverage prior to release from incarceration.

Recidivism and Desistance

Renewed contact with the criminal justice system, measured by arrest or re-incarceration, has been a key focus of reentry research and policy. While repeat offending and criminal involvement over the life course have been long-standing interests of criminology, the emergence of reentry as a field of study raises the question of whether the circumstances of community return may themselves contribute to renewed involvement in crime. From this perspective, desistance from crime is just one of several interrelated indicators of social integration after incarceration (Rosenfeld and Grigg 2022). In this section, we briefly introduce statistics on rates of arrest and incarceration following release from prison. We then examine evidence linking the social context of reentry to recidivism.

National statistics on recidivism after release from state prison have been compiled in a series of studies conducted by the Bureau of Justice Statistics. The most recent study provided a ten-year follow-up analysis of a sample of state prisoners released in 2008 (Antenangeli and Durose 2021). Ten years after release from prison, 82 percent of the sample had been rearrested at least once. The highest rates of rearrest were recorded by younger people (89.5 percent for those under twenty-four compared to 74.5 percent for those aged forty and older). The data also indicate a racial disparity where Black and Native prison releasees were rearrested at somewhat higher rates (85.6 and 84.7 percent) than Whites and Latinos (80.2 and 78.9 percent). Patterns of re-incarceration are similar, and about 61 percent of the those in the recidivism study sample returned to prison within ten years. Notably, statistics on re-incarceration include those who returned to prison on probation and parole violations, and not just those who were convicted for new offenses. Re-incarceration for violations underscores a basic limitation of research on recidivism: while arrest and incarceration statistics are interpreted to indicate new criminal activity, they also reflect the work of surveilling and sanctioning by police and parole officers.

A major challenge for research on recidivism involves distinguishing crime from criminalization. Although researchers often treat arrest and incarceration as measures of criminal behavior, criminal activity is rarely directly observed, and few studies examine the relationship between crime and post-incarceration police contact. The distinction between crime and criminalization raises methodological challenges for the measurement of crime that have been examined extensively in criminology (Huizinga and Elliott 1986; Junger-Tas and Marshall 1999). Official records as a source of information on offending face several limitations in the study of reentry. Formerly incarcerated people often return to disadvantaged communities where policing tactics can be intensive, creating frequent contact with local residents (Stuart 2016; Travis et al. 2012). A few articles on “system avoidance” show that formerly incarcerated people are strongly concerned by the possibility of police contact; this concern leads to avoiding institutional settings like schools and hospitals where police may be encountered (Brayne 2014; Haskins and Jacobsen 2017; for a discussion of policing and reentry see Travis et al. 2012). Some direct evidence on crimeless criminalization is provided by parole research on technical violations that find that around a third of revocations are for rule violations without a related arrest (Rosenfeld and Grigg 2022, 21). Police stops, searches, and parole revocations are also high-discretion decisions that may be affected by racial bias or statistical discrimination (Western et al. 2023, 66–75, 182–84). Criminalization, as distinct from crime, may be a relatively larger component of measured recidivism statistics for Black and Latino men and women, although there is little research directly on this question. Even if arrest and incarceration are, at best, noisy measures of new criminal activity, they have been found to be systematically related to several key conditions of reentry. Stable employment and family relationships have been associated with low rates of recidivism. Steady jobs and stable marriages are also seen as major steps along the life course in theories of criminal desistance.

The links between employment and reduced recidivism have been observed with a wide variety of research designs. Perhaps the leading study that integrates a theory of criminal desistance with empirical evidence on employment is provided by Robert Sampson and John Laub (Laub and Sampson 2003; Sampson and Laub 1993). Sampson and Laub reexamined case records on juvenile delinquency originally collected by Sheldon Glueck and Eleanor Glueck from a sample of White working-class boys living in the Boston area in the 1930s. The Gluecks’ data consisted of two independently drawn samples, one from a population of “persistent delinquents,” aged ten to seventeen incarcerated in the Massachusetts juvenile correctional system. The second sample of nondelinquents, also White males aged ten to seventeen, were drawn from Boston public schools where the respondents were matched by age, race, neighborhood, and measured intelligence. Interview and administrative record data was collected up to age thirty-two, and additional follow-up efforts culminated in interviews with surviving respondents at age seventy. With this long follow-up period, criminal desistance could be studied over the entire life course in a sample with substantial involvement in the adult penal institutions. Stable employment was argued to provide informal social control that monitored conduct, routinized social life, and invested community roles and values in men with histories of institutionalization (Sampson and Laub 1993, 140–43). Incarceration entered the analysis as a sample restriction, limiting analysis only on reform school respondents, and as a control variable, indicating adult incarceration. Regressions at ages seventeen to twenty-five, and twenty-five to thirty-two showed that respondents with stable jobs were less likely to be arrested.

Later observational studies have replicated the finding in formerly incarcerated samples of the association between stable employment and relatively low rates of rearrest and re-incarceration (Berg and Huebner 2011; Lockwood et al. 2012; e.g., Tripodi et al. 2010). Researchers have also studied the impact of local labor market conditions on recidivism, finding that employment and wage growth are associated with reductions in recidivism (Agan and Makowsky 2023; Schnepel 2020; Yang 2017). A limitation of the observational research, however, is that criminal involvement may undermine a steady attachment to the labor market. More generally there may be other factors, such as untreated addiction or propensity to antisocial behavior, that cause both unemployment and arrest. Employment and recidivism may be correlated, but the relationship may be spurious, resulting from the influence of unobserved behaviors.

The relationship between employment and recidivism has also been studied through the evaluation of reentry employment programs. Where people coming out of prison are randomly assigned to jobs or employment services, we can assess the causal effect of employment on recidivism, controlling for the influence of confounding variables (for a review of program evaluations, see Western 2008a). Rigorous randomized evaluations offer some support for the effect of transitional jobs on reducing recidivism, but results are often mixed or apply only to experimental subgroups (Bloom 2010; Visher et al. 2006). From 1975 to 1979, the National Supported Work Demonstration provided low-wage manual jobs in construction, light manufacturing, and public works for twelve months, and participants in the experiment were followed for three years. Among parolees in the sample, rearrest was significantly reduced by transitional employment, but this effect was found only among parolees over the age of twenty-six (Uggen 2000). In another randomized trial, the Center for Employment Opportunities (CEO) provided low-wage transitional jobs and job readiness training to men and women released from prison and living in New York City. Participation in CEO was associated with a reduction in arrest rates over a three-year follow-up period. These effects were largest for those who entered the program within three months of prison release (Redcross et al. 2012). A common result running through all evaluations of employment-based programs is the generally weak employment effects and the very low wages of formerly incarcerated men and women, regardless of whether they were employed in subsidized jobs (Cook et al. 2015). Jennifer Doleac’s (2023) comprehensive literature review concludes that mixed results for employment and related income-support programs may reflect the low economic value of program wages and the possibly negative effects of peer influence.

Observational studies find a stronger relationship between employment and desistance than the experimental program evaluations, but employment in observational studies is often quite different from employment in transitional jobs programs. Sampson and Laub (1993) emphasize that desistance is not the result of employment by itself but the informal social control of job stability that often accompanies long-term employment in skilled work. Qualitative studies have also emphasized the importance of access to skilled jobs obtained through referrals from family members (Black 2010; Sullivan 1989; Western 2018). These kinds of jobs, and the social control they provide, is very different from short periods of minimum-wage unskilled work that is typical of transitional jobs programs. In short, it is not employment by itself but the kind of employment and the larger social context in which it is embedded that may be most important for desistance. From this perspective, reentry research finds that employment is common after prison release, but it is more common in low-wage, secondary labor market jobs (Harding et al. 2019, 178–88) that provide little informal social control and do relatively little to provide and affirm a full and positive social role in one’s community.

The second major life course milestone associated with criminal desistance is a stable marriage. Like research on employment, evidence for the recidivism-reducing effect of marriage is provided by a large body of observational research. Laub and Sampson’s (2003) interviews with formerly incarcerated elderly men point to marriage as a turning point in the life course that promotes criminal desistance. Prospective studies also report a strong association between marriage and desistance (Kendler et al. 2017; King et al. 2007; Sampson et al. 2006; Warr 1998). The relationship between marriage and desistance is stronger for formerly incarcerated men than women. Deeper explorations of the association suggest it is not marriage by itself but the quality of social bonds that draw people out of criminally involved networks that are important. The challenges of causal inference are also substantial in observational studies of marriage, and researchers have examined within-person changes and selection on observables through matching designs (Sampson et al. 2006).

Marriage has been a focal point for research on desistance, yet, as we saw in the discussion of family support, reentry research indicates that social support and the strongest social bonds after incarceration are provided not by romantic partners but by older female relatives. A few studies have examined how broader family networks are associated with recidivism. The Urban Institute’s Returning Home study reported a positive relationship between family support and reduced re in the Chicago site, but this was not replicated in Baltimore (La Vigne et al. 2004; Visher et al. 2004). Mark Berg and Beth Heubner (2011), analyzing state parole data, find lower rates of rearrest among those with family support. The relationship is largely explained by higher rates of employment in the group with supportive families. A few studies suggest that family support, and emotional support in particular, works in gendered ways with the largest recidivism-reducing effects for women (Taylor 2015).

Like employment and family ties, neighborhood environments are closely connected to the transition from prison to community and are important sources of informal social control that can contribute to criminal desistance. Prison admissions and releases are highly spatially concentrated in a small number of neighborhoods characterized by high rates of poverty, crime, and racial segregation (Sampson 2012; Simes 2021). Spatially concentrated poverty, unemployment, and racial segregation, along with other markers of disadvantage, have been found to weaken informal monitoring of public spaces, undermine daily routines, and erode helping behaviors for people in trouble. These neighborhood environments also provide a criminogenic context to which people return from prison (Morenoff and Harding 2014).

Several studies have tried to assess the impact of neighborhood environment on reentry and recidivism (see the review by Morenoff and Harding 2014). Analyses at the census tract and county level have found higher rates of rearrest, parole violation, and re-incarceration in localities with higher levels of poverty, unemployment, and other measures of disadvantage (Hipp et al. 2010; Kubrin and Stewart 2006; Mears et al. 2008; Yang 2017). In one of the few studies to exploit quasi-experimental variation in neighborhood conditions, David Kirk (2020) followed parolees from New Orleans whose reentry was relocated because of the destruction of housing by Hurricane Katrina. Those who were relocated were significantly less likely to return to prison after eight years than those who remained in their old New Orleans neighborhood. Kirk (2020, 71) attributes part of the difference in outcomes to improvements in local economic conditions. The remainder he attributes to a break with a “criminogenic past” that includes social networks and opportunities for crime in the old neighborhood.

Recent research on safety net policies also suggests that the social policy environment likely influences recidivism. We saw that following the 1996 welfare reform, states varied significantly in extending SNAP and TANF eligibility to applicants with drug convictions. Medicaid coverage also varies across states for people released from prison. A burgeoning body of research shows that safety net programs reduce crime and arrest rates. For example, Jessica Simes and Jaquelyn Jahn (2022) exploit state variation in Medicaid coverage in their study of arrest trends in expansion and non-expansion counties. They find that Medicaid expansion through the Affordable Care Act is associated with a 20 to 30 percent reduction in the violent arrest rate, and a 25 to 40 percent lower rate of drug arrests. Similar findings are reported in other studies focused on state and longitudinal variation, and additional evidence suggests that expanded treatment for substance use disorders is one channel connecting health care coverage to reductions in crime and arrests (He and Barkowski 2020; Vogler 2020; Wen et al. 2017). Manasi Deshpande and Michael Mueller-Smith (2022) studied the criminal involvement of young people who lost eligibility for Supplemental Security Income (SSI) at age eighteen following the 1996 welfare reform. They found that continuing SSI payments reduced the number of criminal charges by 20 percent over the following two decades. Similar results have been reported for SNAP, where recidivism increased following the loss of eligibility as a result of the 1996 welfare reform (Jácome 2022; Tuttle 2019). Although these studies do not specifically examine the effects of social programs on reentry, they are suggestive of how alleviating poverty and poor health can prevent crime and arrest.

In sum, the reentry process—involving unemployment and economic insecurity, contentious relations with former partners, and returns to old neighborhoods—carries many of the risks of new criminal involvement, rearrest, and re-incarceration. Conversely, stable employment, family relationships, and supportive neighborhood and policy environments are associated with reduced rates of arrest and criminal involvement.