The Residue of Imprisonment: Prisoner Reentry and Carceral Gang Spillover

Abstract What happens to the gang ties of people when they leave prison and return to the community? There is much speculation but little empirical research concerning carceral gang spillover, which refers to the reproduction of prison gang associations, identities, politics, and structures in communities. This study examined continuity and change in gang embeddedness in a representative sample of 802 men in Texas interviewed in prison and reinterviewed twice upon release. Based on a series of multilevel models this study arrived at four main conclusions. First, prisons are a vector of gang activity. Embeddedness in gangs dropped markedly after release from prison and continued to recede with temporal distance from prison. Second, the residue of imprisonment was moderated by pre-release gang status. Negative linear trends were stronger for active gang members and weaker for former and never gang members. Third, a temporary surge in gang identification occurred for active and former gang members despite continued declines in gang associations, reflecting a behavioral-cognitive decoupling. Finally, the types of groups with which prisoners affiliated contoured trajectories of gang embeddedness. The declining significance of gangs with prisoner reentry opens productive lines of inquiry on carceral gang spillover and offers policy and practice guidance concerning reentering populations.

Nearly everyone who enters prison in the United States eventually leaves. 1 Each year about 600,000 people are released from federal and state prisons (Carson, 2021b).This transition is not a clean break, however.Former prisoners bring to the community the experiences, identities, norms, and relationships accrued while incarcerated, an imprisonment residual.The emergence of mass incarceration, or the quadrupling of the rate of people held in prisons over the last five decades, has meant that its collateral consequences have been imprinted on a much wider swath of people, families, and communities (Enns et al., 2019; National Research Council,   1 The Sentencing Project reported that in 2020 about 56,000 people were serving life without parole and 148,000 people were serving sentences of life with parole or virtual life (50+ years) (Nellis, 2021).From 2011 to 2018 between 3600 and 4500 prisoners died while incarcerated in federal or state prisons (Carson, 2021a).2014).Gangs rose to prominence in prisons throughout the country with mass incarceration, occupying a central place in the social order of prisons (Skarbek, 2014).Around 15% of the 1.5 million prisoners in the United States are affiliated with gangs (Pyrooz & Mitchell, 2020), though a much larger proportion are associated by virtue of hometown geography, housing assignment, or race/ethnicity (Bloch & Olivares-Pelayo, 2021;Bolden, 2020;lopez-Aguado, 2018).What happens to the gang ties of incarcerated populations when they leave prison and return to the community?
Carceral gang spillover refers to the reproduction of prison gang associations, identities, politics, and structures in communities.Jacobs (1974) and Moore (1978) first observed its occurrence in Chicago and los Angeles, respectively, through the migration of people via admissions and releases and/or migration of information via phone calls, letters, and visits.More recently, scholars have identified how prison gangs consolidate and project power over the streets (e.g., lessing & Willis, 2019;Skarbek, 2014) and how carceral policies and practices institutionalize the social order in communities (e.g., lopez-Aguado, 2018).The specter of cascading gang influences is further reinforced by qualitative retrospective accounts of rockier reentry experiences (e.g., lopez-Aguado, 2018; Scott, 2004) and quantitative prospective accounts of higher recidivism among former prisoners involved in gangs (e.g., Huebner et al., 2007;Pyrooz et al., 2021).i focus on one dimension of carceral gang spillover: the extent to which gang ties rise or fall with the migration of people from the prison to the street.
There is much speculation but little empirical research on continuity and change in gang involvement with prisoner reentry.leading panel studies of prisoner reentry have documented post-release trends in employment, family, housing, recidivism, and social supports (Harding et al., 2019;Mowen & Boman, 2018;Visher et al., 2011;Western, 2018), but their survey instruments lack measures of gang involvement.The prospective and retrospective studies that have concentrated on gangs have documented continued gang ties (lopez-Aguado, 2018;Moore, 1991;Scott, 2004;Trammell, 2012;Venkatesh & levitt, 2000), but lack repeated measures over the course of reentry to assess continuity and change.it remains an open question if gang ties outlast the transition from prison or if they are mere remnants of an institutionalized self.
in this study, i examine continuity and change in gang involvement with prison reentry using three waves of data from the loneStar Project, or the Texas Study of Trajectories, Associations, and Reentry, which consisted of a pre-release and two post-release interviews with a representative sample of 802 men exiting prison in 2016.i focus on continuity and change in gang embeddedness rather than membership since it better reflects the nuances of involvement in gangs in prison and on the street.Do levels of embeddedness in gangs change when people leave prison?i develop and test hypotheses by drawing on classic and contemporary theory and research on prison social organization and determine if continuity and change in gang embeddedness are moderated by pre-release gang status (i.e., active, former, and never in a gang) and the organizational structure of gangs with which prisoners affiliate.Together, these findings will aid in replacing speculation about carceral gang spillover with empirical research.

Prison, Gangs, and Reentry in the Era of Mass Incarceration
Understanding the realities of prisoner reentry and reintegration has become a major focus of social scientists over the last two decades (Jonson & Cullen, 2015;leverentz et al., 2020;Mears & Cochran, 2015;Middlemass & Smiley, 2019).Visher and Travis (2003, pp. 90-91) defined reentry as the "process of leaving prison and returning to free society" and reintegration as the "reconnection with the institutions of society."Recidivism receives outsized attention in the reentry process, and for good reason (Alper et al., 2018), but social scientists are increasingly concerned with the broader aims of reintegration within institutions, such as community, family, and work.Reintegration is best understood, as Visher and Travis argued, using a dynamic framework containing multiple stages and dimensions of influence.i focus on the relevance of one core stage of reentry, in-prison experiences, and one core dimension of influence, peer networks, both of which have been supported by recent theoretical and empirical advances (Hummer & Ahlin, 2018;Mowen & Boman, 2018;Taxman, 2017;Walters, 2016).Even as incarceration scholars are (re)discovering the significance of gangs in prison (e.g., Kreager & Kruttschnitt, 2018, p. 268;Wooldredge, 2020, pp. 170-71), little is known about the extent to which carceral gang spillover occurs.
Carceral gang spillover refers to the reproduction of prison gang associations, identities, politics, and structures in communities. 2it may occur through two primary modes of transmission.The first entails the communication of gang directives, norms, and rules originating inside of prison and disseminated outside of prison via mediated channels, such as phone calls, text messages, letters, and non-residents (e.g., attorneys, staff, and visitors), akin to airborne transmission.The second involves the dissemination of such information via the migration of people who transition from the prison to the street, akin to bloodborne transmission.Both modes are relevant to carceral gang spillover, as they are for criminology and public health generally (Papachristos, 2011).The focus of this study is on carceral gang spillover that may occur with release from prison, which takes on added significance as the cascading or receding of gang involvement could contour the process of prisoner reentry and reintegration. in the following sections, i develop hypotheses with respect to gang embeddedness, which refers to dynamic individual immersion within enduring deviant groups/ networks.Derived from the application of network theory to criminology generally and gangs specifically (Hagan, 1993;McCarthy & Hagan, 1995;Pyrooz et al., 2013), i focus on gang embeddedness because it offers four advantages to understand prisoner reentry and carceral gang spillover.First, by tapping associations, behaviors, influences, and positioning, it avoids making a crude in or out conceptualization and operationalization of gang membership (lopez-Aguado, 2018, p. 131;Rios, 2017) while broadening the scope of gang ties to include physical and social components that transcend prison.Second, it recognizes that there are nuances to gang involvement, 2 While carceral gang spillover specifies a directional relationship, prison to community, it is important to recognize that the relationship between gangs on the street and in prison is mutually reinforcing, as Jacobs (1974) and Moore (1978) observed in the earliest research on the topic.Still, concentrated power in prison permits outsized influence over the streets and new entrants to prison, which is supported by the bulk of the nascent evidence (Bloch & Olivares-Pelayo, 2021;Cruz & Rosen, 2022;lopez-Aguado, 2018;Pyrooz & Decker, 2019;Skarbek, 2014;Weide, 2015).Owing to study objectives, i focus on carceral gang spillover that occurs with prisoner reentry.akin to Thornberry et al. (2003, p. 6) declaration that "not all gang members are created equal."Third, embeddedness in gangs is not limited to bona fide gang members, as both former and even non-gang members can maintain non-zero levels of embeddedness by virtue of associations, behaviors, influences, and positioning (Decker et al., 2014;Papachristos et al., 2015).Finally, gang embeddedness appreciates multiplexity in that individuals are differentially integrated into overlapping networks of family, religious, work, and other types of relationships that bring meaning to lives (Hureau & Braga, 2018;Sierra-Arevalo and Papachristos 2015;Thornberry et al., 2003, p. 166).To the extent that carceral gang spillover occurs via the migration of people, it should manifest as higher levels of gang embeddedness among people transitioning from prison to the community.

Imprisonment, Reentry, and (in)Variances in Gang Embeddedness
People who affiliate with gangs are disproportionately incarcerated.Official records from correctional and police agencies in the United States indicate that gang affiliates are overrepresented in federal and state prisons relative to non-institutional populations by around a factor of 50 (Egley et al., 2014;Pyrooz & Mitchell, 2020).Two classic theories, both of which are rooted in the sociology of prisons pre-dating mass incarceration, offer much different explanations for gang activity in prisons, which in turn offer guidance to consider prisoner reentry and carceral gang spillover.
in deprivation perspective, gangs are a functional adaptation to the conditions of confinement.Classic works on prisoner society contended that the convict code and social organization peculiar to prisons emerged in response to the pains of imprisonment, or the loss of autonomy, identity and status, heterosexual relationships, material goods, and safety (Clemmer, 1940;Goffman, 1961;Sykes, 1958).Gangs proliferated in prison systems as informal controls predicated on reputational norms did not scale with the rise of mass incarceration (irwin, 1980;Skarbek, 2014).it is not as if the deprivations of the total institution subsided with mass incarceration-in fact, they likely worsened-thus ensuring a continued demand for outlets to blunt the indignities of imprisonment.Gangs present such a solution, as prisoners can band together under a collective banner to compensate for deprivations by offering selective incentives, such as companionship, goods, protection, and status (Bolden, 2020;lopez-Aguado, 2018;Pyrooz & Decker, 2019).As Jacobs (1974, p. 401) first put it, "the gang allows you to feel like a man; it is a family with which you can identify."The reprieve offered by the gang should be temporary, confined to the contours of imprisonment when the weight of deprivations is felt. in other words, when people leave prison, the pressure toward involvement subsides.There is some evidence in support of this (Pyrooz et al., 2017).young people from Philadelphia and Phoenix were more likely to self-identify as a gang member when they were incarcerated in jails and prisons than when they were in the community.
Embeddedness in gangs is not exclusive to people who identify as gang members.While many may "go it alone" behind bars, lopez-Aguado (2018) highlighted the significance of "cars" in institutional spaces, which refer to loose collectives of incarcerated people united by common features and interests to navigate the deprivations of imprisonment.in California, for example, cars reflect the broader carceral order organized around gangs, geography/hometown, and, especially, race/ethnicity-Surenos (i.e., Southern latino), Nortenos (i.e., Northern latino), Bulldogs (i.e., latinos from Fresno), Woods (i.e., White), or Blacks (i.e., Blood or Crip sets) (Bloch & Olivares-Pelayo, 2021;Goodman, 2008;Walker, 2016).While gang affiliation is inevitably linked to cars, not everyone who "runs with" a car identifies as a gang affiliate.Still, prisoners are expected to follow the norms and rules of the carceral order around gang politics, subjecting them to the wide-ranging influences, from eating to seating, riots to reprisals (Bolden, 2020;Maitra, 2023;Walker, 2016;Weide, 2022).institutional sorting practices for housing and work assignments further ensure that prisoners, regardless of whether they are active, former, or non-gang members, are physically and socially adjacent to dominant gangs (i.e., embedded).And since the structuring of the total institution diverges from that of the community (Maxson, 2012), the formal policies and informal politics that dictate encounters and interactions should recede with reentry.The fact that just about all prisoners could maintain some level of involvement in gangs-irrespective of gang status-motivates the first hypothesis: H1: Embeddedness in gangs declines with prisoner reentry.
The null hypothesis-embeddedness in gangs does not change with prisoner reentry-is equally relevant theoretically given its link to an alternative explanation for the social organization of prisons: importation.Whereas the deprivation perspective emphasizes context, the importation perspective emphasizes composition.in this view, gangs in prison are a product of gangs and criminalization on the street.irwin and Cressey (1962) countered the functionalist argument by contending that the culture and organization in prison are "not peculiar to the prison at all" (142), but instead reflect the biographies, culture, and relationships of people who are arrested, convicted, and, ultimately, imprisoned.Jacobs (1974) and Moore (1978) first observed continuities between street gangs in Chicago and los Angeles and prison gangs in illinois and California, respectively.yet as street gangs proliferated across the United States in the 1980s and 1990s, and young men were subject to mass criminalization, a street-to-prison gang pipeline emerged.The mass incarceration of gang populations was the product of criminal justice practice and policy, such as police crackdowns, task forces, gang units, sentencing enhancements, and criminal enterprise conspiracy statutes, combined with elevated violence in urban areas (Aspholm, 2020;Decker et al., 2022;Klein, 1995;Rios, 2011).
The exportation hypothesis is the reversal of importation-an imprisonment residual of associations, experiences, identity, and knowledge that migrates back to the community with carceral gang spillover through prisoner reentry.irwin and Cressey (1962, pp. 153-55) acknowledged this possibility, though Visher and Travis (2003) offered the framework to chart the contours of exportation.More recently, Hummer and Ahlin (2018) contended that exposure and adherence to the culture of prison violence is baggage carried by people long after returning to the community.Researchers have established how a variety of prison experiences, such as solitary confinement and visitation (e.g., Mitchell et al., 2016;Wildeman & Andersen, 2020), and prison programs, such as cognitive behavioral therapy and therapeutic communities (Jonson & Cullen, 2015), are consequential for reentry outcomes, including mortality and recidivism.
The exportation hypothesis anticipates that gangs in prison are also important.The emergence of gangs with corporatized structures and constitutions declaring lifelong affiliation coincided with the formerly incarcerated leading "street programs" or co-opting street gangs to do the bidding of the gangs in prison (Buentello et al., 1991;Gundur 2022;Skarbek, 2014;Venkatesh & levitt, 2000).And researchers have documented how former prisoners (re)connect with neighborhood street gangs, hanging out in parks and on stoops, recounting the "glory days" and "tales of prison life" (Moore, 1991, pp. 45-46;Venkatesh & levitt, 2000, p. 437). in criminal capital perspective (McCarthy & Hagan, 1995;Nguyen, 2020), if prisons are "higher education" for gangs, reentry is graduation, which is when material and status dividends of incarceration are realized thus inducing continuity, or equality, in gang embeddedness.

Gang Status, Group Organization, and Continuity and Change in Gang Embeddedness
Carceral gang spillover that occurs with release from prison could be moderated by the gang status of returning populations and the organizational structure of the groups with which they affiliate.There is considerable evidence in gang research that attitudes, behaviors, and experiences differ between and within people across active, former, and non-gang status (Clark et al., 2023;Melde & Esbensen, 2014;Pyrooz et al., 2021;Thornberry et al., 2003).Deprivation and exportation perspectives also anticipate that gang status could moderate trends in gang embeddedness but for different reasons.Both perspectives expect a gradient of embeddedness by gang status, particularly while people are incarcerated.That is, there should be rank-order differences in gang embeddedness, such that active gang members score the highest and never gang members the lowest, while former gang members fall in the middle, reflecting normative orientations, routine activities, preferences, and structural influences that shape social networks.Decker et al. (2014) observed this in a multi-site study.
yet the deprivation and exportation hypotheses offer different expectations of trajectories in gang embeddedness upon leaving prison.The deprivation hypothesis is that gang embeddedness falls with reentry regardless of gang status.With the subsiding of institutional pressures described above, returning populations are less likely to be induced to associate with gangs or follow their proscriptives, which should be experienced differently by gang status in part because baseline levels of gang embeddedness differ across active, former, and never gang members.Accordingly, the next hypothesis in this study is: H2: Negative gang embeddedness trajectories should be strongest for active and weakest for never gang members.
The exportation hypothesis, in contrast, anticipates that trends in gang embeddedness will be stable or invariant by gang status. in other words, the rank-order differences observed in prison will carryover after release, such that linear trends in gang embeddedness by gang status will be indistinguishable from zero since the context of the prison environment is not what is driving gang embeddedness.
The organizational structure of gangs also factors into trajectories of gang embeddedness.The classic view of prison gangs portrays their forms and functions as conspiratorial, hierarchical, monolithic, predatory, and rule bound (Camp & Camp, 1985;Crouch & Marquart, 1989;Diiulio, 1987;Jacobs, 1974). in fact, Venkatesh and levitt (2000, p. 435) contended that prison "played a critical role in the infrastructural development of Chicago's largest gang federations" (see also Aspholm, 2020).Even leading definitions of prison gangs include corporatized features as defining properties (e.g., lyman, 1989;Skarbek, 2014).Such features might apply to a special type of prison gang, but it stands in contrast to the wider literature on street gangs that reveals far greater variation in forms and functions (Pyrooz, 2022).Owing to questions surrounding the extent to which street gangs operated as organized drug traffickers or freelance dealers, researchers developed an organizational structure continuum using the ideal-types of instrumental-rational and informal-diffuse as the extremes of the distribution (Decker et al., 1998;Decker & Van Winkle, 1994;Hagedorn, 1994).Emerging qualitative and quantitative evidence reveals that such heterogeneity does indeed exist among gangs in prison (Gundur 2022;lopez-Aguado, 2018;Ortiz, 2018;Pyrooz & Decker, 2019;Tapia, 2014), and it should matter for continuity and change in gang embeddedness after release from prison.
All gangs have an interest in minimizing adverse selection and turnover (Densley, 2012), but there are major differences in entry and exit processes across types of gangs.instrumental-rational gangs, the archetypal prison gang (Buentello et al., 1991), typically recruit their members from large pools of cliques, screen for quality, and subject prospective members to lengthy periods of observation before they are "made" (Pyrooz & Decker, 2019;Skarbek, 2014).They also aim to limit disengagement because it is costly to replace members-constant recruitment is time-consuming and losing members damages reputations and makes gangs vulnerable to attacks.The first rule in the Mexikanemi's constitution is "blood in, blood out" (Fong, 1990, p. 40) and la Nuestra Familia explicitly states that "a Familiano will not be released from his obligations toward the organization because he is released from prison" (Skarbek, 2014, p. 113).Such proscriptives could delay disengagement from gangs long after leaving prison.
informal-diffuse gangs lack such features.They are structured horizontally rather than vertically, decision-making is decentralized, lack formal constitutions, and maintain more symbolic than entrepreneurial functions.The largest Texas prison gang, the Tangos, is linked to hometown and region. 3Membership is temporary, primarily as defense against the predatory prison gangs (Gundur 2022;Pyrooz & Decker, 2019;Tapia 2014).And although gangs on the street and in prison are not one in the same, there is evidence indicating the organizational structure of gangs is associated with prolonged membership (Cruz & Rosen, 2020;leverso & Matsueda, 2019;Melde et al., 2012). in sum, this leads to a third hypothesis: H3: People who affiliate with instrumental-rational gangs will maintain greater continuity in gang embeddedness after release from prison than those who affiliate with informal-diffuse gangs.

The Current Study
There is good reason to believe that embeddedness in gangs can both wax and wane with release from prison, but little is known about carceral gang spillover despite its potential significance for undermining successful reintegration and as a mechanism linking prisons and communities.This is due, in part, to reentry scholars having "almost completely ignored the participation of ex-prisoners in mid-range informal community organizations, such as street gangs" (Scott, 2004: 110).i use three waves of interview-based survey data in Texas to assess and explain continuity and change in gang embeddedness.These data were collected for the purpose of understanding prisons, gangs, and reentry near peak mass incarceration in the United States, offering an opportunity to uncover an imprisonment residual in an area that has eluded the expansive body of scholarship on prisoner reentry.
Texas is an ideal context to assess these hypotheses.it is the largest state correctional system in the United States with a large amount of annual turnover in the custodial population (Carson, 2021b).Prison gangs rose to prominence nearly three decades ago and maintain an established presence in correctional facilities (Bolden, 2020).A sizeable minority of prisoners are officially classified as active or former gang affiliates, which maps closely to self-reports (Pyrooz & Decker, 2019).There is heterogeneity in the organizational structures of gangs, ranging from cliques to security threat groups (i.e., STGs) (Buentello et al., 1991).The Texas prison system, and the attendant gangs, are racially and ethnically diverse, split about evenly between Black, latino, and White individuals (Trulson & Marquart, 2010).

Data
Three waves of interview-based survey data are used to examine gang embeddedness among 802 males who participated in the loneStar Project. 4 The sampling frame consisted of state prisoners scheduled for release from the largest release unit in Texas.A weekly list was provided by the executive services office from which disproportionate stratified random sampling was used to invite voluntary participation in the study.Consistent with the purpose of the loneStar Project, people who were classified by correctional authorities as maintaining non-zero levels of gang affiliation (confirmed, suspected, and ex-gang affiliates) were oversampled by a factor of five. 5The rate of response among those selected for participation was 61%, while the rate of cooperation among those contacted and invited to participate was 94%, both of which are consistent with comparable studies (e.g., Schubert et al., 2004).The correctional policy prohibited the provision of incentives at the pre-release interview, but compensation in the form of Walmart gift cards was offered at the post-release interviews.
The pre-release interview, Wave 1, occurred between April and December 2016 in two prison units. 6The first was a release unit where study participants were transferred before leaving custody, where 95% of interviews were conducted.The second was an administrative segregation unit, where the remaining 5% of interviews were conducted owing to correctional policy concerning gangs and high-custody level prisoners; all of these individuals were released from the first unit.Trained interviewers affiliated with universities associated with the loneStar Project used computer-assisted personal interviewing in both settings.
Post-release interviews occurred about two and 10 months after release, constituting Waves 2 and 3, beginning in May 2016 and concluding in February 2018.Geographic distribution of former prisoners across Texas required computer-assisted telephone interviewing, although face-to-face interviews were conducted if study participants were reincarcerated at the target date.The retention rates for Waves 2 and 3 were 66.3% (n = 532) and 64.1% (n = 514), and 78% completed at least one post-release interview.These retention rates are on par with prior large, longitudinal prisoner reentry studies in Texas and elsewhere (lattimore & Visher, 2020;Visher & la Vigne, 2020).Additional information about pre-and post-release interview operational procedures can be found in Fahmy et al. (2022) and Mitchell et al. (2022).
interviewing former prisoners is critical to understanding reentry, but it is well-known that this is a hard-to-reach population (Western et al., 2016).The potential for nonignorable bias in longitudinal research is high.The concern is that estimates of gang embeddedness and its correlates are biased in the observed data to the extent that high-risk individuals are excluded, such as active and former gang members.Prior research using the loneStar Project data found no statistical differences in retention rates between gang and non-gang members (Clark et al., 2020).Further, there were no differences in pre-release levels of gang embeddedness between the retained and attrited subsamples after adjusting for sociodemographics. 7Of the 622 retained study participants, 68% were interviewed at both post-release waves, 15% at Wave 2 only, and 17% at Wave 3 only.The analytic sample includes 1,632 person-wave interviews. 8

Gang Embeddedness
The outcome of interest concerns dynamic immersion within enduring deviant networks, drawing on the broader concept of criminal embeddedness (Hagan, 1993;Pyrooz et al., 2013).Embeddedness involves several time-varying dimensions, including gang associations, behaviors, influences, and positioning, which cut across prison and community life.i construct a scale of gang embeddedness using a mixed graded response model, derived from item response theory (iRT), based on seven items repeated at each of the three waves of data collection.Consistent with the conceptualization of gang embeddedness, these items were administered to all respondents irrespective of gang status.
The model was estimated across pooled waves of retained study participants and scores were generated using an empirical Bayes estimator and standardized to a mean of 0 and standard deviation of 1.9 Descriptive statistics for these items can be found in Table 1.

Imprisonment and Reentry
To test hypotheses derived from deprivation and exportation perspectives i compare levels of gang embeddedness at pre-and post-release interviews.First, a linear trend variable is created.Wave 1, or the pre-release interview, is coded 0, and Waves 2 and 3 are coded 1 and 2, respectively, to capture the direction, sign, and significance of any change in gang embeddedness and fix the model constant to the pre-release interview.Second, since the functional form of the reentry/embeddedness may not be linear, two dummy variables are used to distinguish between the post-release periods, proximal reentry (Wave 2 = 1; Waves 1 and 3 = 0) and distal reentry (Wave 3 = 1; Waves 1 and 2 = 0); Wave 1, or the pre-release interview, is used as the reference category.Finally, supplemental analyses pooled the post-release interviews together, testing whether gang embeddedness is higher in Waves 2 and 3 (=1) than in Wave 1 (=0).

Gang Status
Continuity and change in gang embeddedness with prisoner reentry are expected to be moderated by gang status.While gang status is dynamic in the life course, for the purposes of this study, i focus on gang status during the pre-release interview.Self-nomination was used to measure gang status, a method commonly used in gang research and found to be valid in community and institutional settings (Decker et al., 2014;Esbensen et al., 2001;Maxson et al., 2012;Pyrooz et al., 2020).All respondents were asked whether they had ever been in a gang and whether they were in a gang while incarcerated.Those who responded "yes" were later queried about their current gang status.Respondents who reported they had never been in a gang were recorded as non-gang members and respondents who reported that they were in a gang while incarcerated but have since left were recorded as former gang members.Respondents who reported that they were in a gang while incarcerated and have not left were recorded as active gang members.These measures are coded dichotomously (focal category = 1, reference category = 0) and mutually exclusive, with non-gang members serving as the reference category.

Gang Organizational Structure
The organizational structure of the groups with which people affiliate is also expected to moderate the relationship between prisoner reentry and gang embeddedness.A four-part typology of gang organizational structure was determined based on a combination of official and survey data to distinguish instrumental-rational and informal-diffuse gangs among respondents who reported having been affiliated with a gang while incarcerated at the baseline interview.
The Security Threat Group Management Office in the Texas Department of Criminal Justice is responsible on an ongoing basis for identifying and classifying groups of prisoners in state custody that pose a threat to the prison system.They differentiate between STGs and cliques.At the time of baseline data collection 12 gangs were designated as STGs.Correctional personnel indicated that this was based on a group's history and propensity for violence, centralized organizational structure, and threat posed to prison operations.An additional distinction is made among the 12 STGs.Seven of the gangs are believed to pose an elevated threat level due to their enhanced organizational structure, warranting placement of confirmed affiliates in administrative segregation, while the other five do not.10These distinctions lead to two parts of the gang typology based on self-reporting an affiliation with one of these 12 groups: STG: high organizational structure and STG: low organizational structure. 11he next part of the typology pertains to cliques.At the time of baseline data collection, there were 67 gangs designated as cliques that, while disruptive and require monitoring, did not meet all three criteria of STGs.Survey measures inquiring about the locus of gang influence-street only, prison only, or both-were presented to respondents who self-identified as having a history of gang affiliation.While nearly everyone (98%) with an STG affiliation (of both types) claimed their gang was active in both settings, there was considerable variation among those with a clique affiliation.Fifty-eight percent of affiliates of prison-oriented cliques indicated that their gang was only active in prison, while 41% indicated it was active on the street and in prison.None of the affiliates of street-oriented cliques indicated that their group was active only in prison; 66% indicated street-only while 35% indicated both contexts.
These distinctions lead to the final two parts of the typology based on self-reporting an affiliation with one of these cliques: Clique: prison-oriented and Clique: street-oriented.
Of the four gang classifications, the survey data showed that 15% of the gang subsample self-reported affiliating with high organizational structure STGs, 17% with low organizational structure STGs, 15% with prison-oriented cliques, and 7% with street-oriented cliques.Empirical validation of this typology was presented by Pyrooz and Decker (2019), who outlined the 38 gangs represented in the data that fell within each classification.Based on multi-level modeling nesting people within gangs, they found substantial heterogeneity in the communicative, expressive, instrumental, and profit-generation features across the gangs, but that the typological "scheme we use [d] to classify gangs accounts for nearly all of the variation between gangs" and "avoid[s] the need to rely on multilevel modeling" of group clustering (120).Given the object of the current study, the four-part typology is optimal to test the organizational structure hypothesis.

Control Variables
in-prison and post-release factors are included as control variables.Baseline self-reported sociodemographic factors include age in years at the time of release, racially Black, ethnically Hispanic, educational attainment in years, and father to any living children.An indicator of street-only membership in a gang was used to partition individuals who never claimed gang membership while incarcerated (i.e., typically adolescence-only gang involvement) from those who did and those who have never been in a gang.Criminal justice factors include the count of prior arrests, whether the conviction resulting in incarceration was violent, number of years in time served for the current sentence, the number of prison spells in Texas, and season of release (spring [reference], summer, or fall) to adjust possible for period effects.Together, sociodemographic and criminal justice controls variables address time-or rank-order stable heterogeneity that could alter the conclusions of the study, such as differential experiences in incarceration.Time-varying factors are also included, such as whether someone was reincarcerated at the post-release interview and the demeaned days elapsed between interviews.The specter of unobserved heterogeneity also exists, though tempered to some degree by rather short intervals between observations.Descriptive statistics can be found in Table 1.

Analytic Strategy
Three waves of panel data are used to examine continuity and change in gang embeddedness during the transition from prison to the community.Mixed effects modeling is used to correct for non-independence since people are nested within waves.Random intercepts give each respondent his own constant, or gang embeddedness score, at the Wave 1 pre-release interview.Gang embeddedness is regressed on linear and dummy variables of interview waves, controlling for time-stable and varying factors.These regressions are repeated for (1) the items in the embeddedness scale, (2) a moderator of baseline imprisonment gang status, and (3) a moderator of gang organizational structure.The moderation analyses are estimated in the full sample using multiplicative interactions and separately to condition estimates on the subsamples (reported in the Supplementary Appendix).Combined, these findings provide evidence for Hypotheses 1, 2, and 3. Maximum likelihood estimation is used in all models and robust standard errors are reported.The outcome is standardized, as are all non-binary independent variables, allowing for the interpretation of coefficients as standard deviation unit differences in gang embeddedness.All models were reestimated using generalized structural equation modeling to confirm the reliability of statistical significance.12All analyses were conducted in Stata 17.0.

Continuity and Change in Gang Embeddedness
Figure 1 contains the predicted values of gang embeddedness at each study wave derived from a series of mixed effects models that adjust for various sociodemographic, criminal justice, and survey-related factors (see Tables S1 and S2 in the Supplementary Appendix for the coefficients and standard errors).in the upper left panel are the findings for the gang embeddedness scale (standardized scaling), while the seven remaining graphs contain the findings for the items comprising the scale (original scaling).
Across the three waves, the linear trend in gang embeddedness was −0.239 (p < .001).levels of gang embeddedness were over one-quarter of a standard deviation lower in the post-release periods than while in prison (d = .288,p < .001).Temporal proximity to imprisonment mattered.Around two months after release, former prisoners' levels of embeddedness were .142standard deviations lower (p = .001);by 10 months post-release, they were a half-standard deviation lower (p < .001).The difference between Waves 2 and 3, over one-third of a standard deviation, was also statistically significant (p < .001).Combined, these findings offer support for Hypothesis 1, consistent with the deprivation perspective, which anticipates gang embeddedness reduces after release from prison.That these reductions in embeddedness are substantively large and grow with time stands in contrast to the exportation perspective, indicating the residue of imprisonment fades rather abruptly with reentry.
The remaining graphs in Figure 1 offer evidence of continuity and change in the individual items composing the gang embeddedness scale.The linear trends for four of the items are statistically significant and negative, consistent with the overall trend in gang embeddedness: attacks/threats (b = −0.023,p < .001),contact (b = −0.130,p < .001),friends (b = −0.226,p < .001),and influence (b = −0.095,p < .001).The linear trends for symbols and importance were indistinguishable from zero while positioning in gangs was positive and statistically significant (b = .073,p < .001).Both positioning in the gang and the importance of the gang are notable in part because the trends were non-linear, both of which reflected a proximal surge.For positioning, respondents scored .319standard deviations (p < .001)greater in the proximal reentry period than while in prison, then the effect reduced in the distal reentry period to only .098standard deviations (p=.022).For importance, respondents scored .116standard deviations (p < .001)greater in the proximal reentry period than while in prison, then the effect reduced to non-significant in the distal reentry period (b = −0.030,p = .383).Temporal proximity to imprisonment mattered for all of the remaining items (except attacks/threats), as they were statistically lower in Wave 3 relative to Wave 2. Overall, examining the individual items adds further support for the deprivation hypothesis, but also indicates that there are critical nuances to the relationship between reentry and embeddedness, particularly around associations, behavior, and identity.

Moderation in Trends of Embeddedness by Gang Status
Figure 2 differentiates patterns in gang embeddedness by self-reported pre-release gang status and Table 2 quantifies these patterns with multiplicative interactions (see Table S2 and Figure S1 in the Supplementary Appendix for gang status-partitioned model results).As expected, across all three periods of observation active gang members were the most embedded in gangs and never gang members the least, while former gang members scored between the two.For example, at the pre-release interview active gang members scored 1.885 greater standard deviations greater than never gang members and 1.522 greater than former gang members.
After release from prison, all three groups underwent a decline in gang embeddedness, as evidenced visually in Figure 2 and quantitatively in Table 2. Model 1 shows that the linear trend for each group was negative and statistically significant.Gang embeddedness fell by −0.182 units per wave for never gang members, −0.631 units for active gang members (−0.182 + −0.449), and −0.136 units for former gang members (−0.182 + .046).While the linear trend for former gang members differed from zero (p = .019)and the effect size was smaller than the linear trend for active -0.17 gang members (p < .001), it was indistinguishable from never gang members (p = .397),running counter to Hypothesis 2. This was primarily due to the proximal reentry period (Wave 2).Whereas active and never gang members reduced their gang embeddedness, it remained unchanged for former gang members owing to a short-term surge in positioning (b = .670,p < .001),symbols (b = .126,p = .009),and importance (b = .199,p = .007)(see Supplementary Appendix for complete results).

Moderation in Trends of Embeddedness by Gang Organizational Structure
Table 3 and Figure 3 display the findings for the mixed-effects models that differentiate trends in gang embeddedness by groups with various organizational structures.Never gang members constitute the reference group in these models and multiplicative interactions are reported (see Table S3 in the Supplementary Appendix).Three of the gang types scored higher in gang embeddedness at the baseline interview than never gang members; people who affiliated with high organization STGs were the exception, likely a reflection of policy resulting in their placement in administrative segregation.instrumental-rational gangs are expected to maintain sturdier trajectories of gang embeddedness than informal-diffuse gangs.The results were mixed.
On the one hand, the patterns observed among prison-oriented cliques and high-organization STGs were consistent with expectations of continuity and change.Prison-oriented cliques experienced a sharp downward and linear fall in gang embeddedness; whereas never gang members reduced their levels of embeddedness at −0.216 units per release wave, the people affiliated with prison-oriented cliques reduced at −0.518 units (χ 2 = 49.83,p < .001).This reduction was not only statistically distinguishable from never gang members, but it was also sharper than what affiliates of STG high organization (p < .001),STG low organization (p = .003),and street-oriented cliques (p = .001)experienced.in addition, people affiliated with high-organization STGs maintained stable post-release trajectories in gang embeddedness, falling at a rate of only −0.050 units per release wave, which was indistinguishable from zero (χ 2 = 0.56, p = .453). it further bears mention that while the trend in embeddedness for prison-oriented cliques was linear, it was not for high-organization STGs (or for that matter, low-organization STGs or street-oriented cliques).The initial surge of embeddedness in the proximal reentry period was followed by a larger retreat in the distal reentry period.
On the other hand, the patterns observed among low-organization STGs and street-oriented cliques were inconsistent with expectations.Neither group maintained a linear trend in embeddedness that differed statistically from never gang members.Respondents affiliated with low-organization STGs reported reductions in embeddedness of −0.214 units per wave, which was distinguishable from zero (χ 2 = 8.81, p = .003),running counter to expectations.in contrast, the affiliates of street-oriented cliques maintained continuity in gang embeddedness, falling at a rate of −0.122 units, which was indistinguishable from zero (χ 2 = 1.81, p = .179).Similar to high-organization STGs, the affiliates of both of these groups experienced a proximal surge in embeddedness after release from prison followed by a distal retreat. 13

Discussion
Gangs occupy a central place in the social order of prisons because they provide extra-legal governance, structure social relations, are disproportionately involved in violence, elicit administrative responses, and their influence and power can extend from the prison to the street.What remained unexplored, however, was if and how gang associations, behaviors, influences, and statuses also pervaded prisoner reentry, a core 13 When conditioning the models on the characteristics of the subsamples, the results were substantively similar despite differences in the point estimates (please consult Table S3 in the Supplementary Appendix).mode of transmission in carceral gang spillover.if gangs are composed of ties that bind individuals, do such ties diminish or persevere with release from prison?Despite the tremendous growth and expansion of research on prisoner reentry, prior studies have been unable to document and explain continuity and change in gang embeddedness, instead offering a snapshot circumscribed by prospective or retrospective accounts or ignoring these issues altogether.Owing to three waves of interview-based survey data collected among a large and representative sample of men before and after release from prison in Texas, it was possible to put anecdote and speculation to the test informed by classic and contemporary theory and research on gangs and prison social organization.Four main findings merit further consideration.
First, the empirical patterns observed in this study lead to the conclusion that prisons are a vector of gang activity.levels of gang embeddedness were highest when men were incarcerated and fell rather dramatically after they left.And the residue of imprisonment faded, even more, the further away temporally people got from prison, as evidenced by greater reductions in gang embeddedness when comparing the distal to the proximal reentry interviews.Moreover, embeddedness in gangs reduced with prisoner reentry for active, former, and even never gang members, which offers greater confidence in this conclusion.The two classic theories of prison social organization leveraged in this study emphasized contextual (deprivation) and compositional (importation/exportation) effects.The evidence comports with the arguments derived from deprivation theory.it is the context rather than the people that drives embeddedness in gangs.While there is little doubt that the indignities of imprisonment do not fully cease upon release from prison (Miller, 2021;Western, 2018), the pressures to associate and affiliate with gangs appear to attenuate as former prisoners return to the community and encounter the freedoms and challenges of reentry.A productive next step would entail systematic comparisons of the deprivations-autonomy, interpersonal relationships, material goods, safety, and status-experienced in institutionalized and non-institutionalized settings, including identifying which dimensions best account for reductions in gang embeddedness.Moreover, a complete test of these models must be based on data that measures embeddedness in gangs before imprisonment, thus warranting caution in the conclusions reached, though it bears mention that, to the best of my knowledge, such data do not exist.Still, these findings aid in understanding the residual gang ties of people who return from prison, which has largely escaped the literature on prisoner reentry as it has grown in scope and volume (National Academies of Sciences, Engineering, & Medicine, 2022;Fleisher & Decker, 2001).
Second, people experienced the imprisonment residual differently based on their gang status.On the one hand, the rank-order differences in gang embeddedness were stable across the three interviews: people who were active gang members scored the highest, never gang members the lowest, and former gang members fell in between.The gang was simply harder to shake for active and, to a lesser extent, former gang members, consistent with Moore's (1996) observations.lopez-Aguado (2018) offered a persuasive case for how the "Home Team" helped to overcome the reentry challenges experienced by former prisoners in Fresno.Others have either theorized or offered anecdotes of the gang helping to cope with the vicissitudes of reentry and leverage the status influx that corresponds with release from prison (e.g., Huebner et al., 2007;Moore, 1991;Venkatesh & levitt, 2000).
On the other hand, the stability of rank-order differences should not overshadow the fact that all three gang statuses maintained negative trends in their gang embeddedness trajectories.Moreover, the reductions were especially strong for active gang members, consistent with the research hypothesis.Although speculative, one possible explanation is offered by Scott (2004), who found that former prisoners in Chicago became keen to the seductions of the gang with age and reentry, recognizing that after release from prison, the benefits of material gain, sensualities of street life, and sought-after status were outweighed by exploitation and spoiled social bonds.That former gang members did not experience proportionally greater reductions in gang embeddedness than never gang members ran counter to the research hypothesis, likely a reflection of their familiarity with the gang and the perceived value and utility it brought to their lives, which must be considered in relation to the next point.
Third, the temporary post-release surge in gang identification occurred concomitant to a reduction in gang associations for active and former gang members, which could be considered behavioral-cognitive decoupling.About two months after release from prison, active and former gang members reported reductions in contact with gangs and friends in gangs, which continued to fall in the next interview period.yet positioning in the gang and the importance of the gang became more salient, rallying at the proximal interview before tailing off 10 months later.This is consistent with qualitative research on prison gangs and reentry (e.g., Jacobs, 1974;lopez-Aguado, 2018;Moore, 1991;Scott, 2004), which alludes to the comfort afforded by the gang in the presence of reentry challenges and the heightened street status of former prisoners.People may cling to the identity and meaning of the gang as they traverse the initial stage of reentry even as it misaligns with their actions.To be sure, disengaging from gangs is a process that involves the unwinding of embeddedness, and identity may not perfectly change in tandem with behavior (Densley & Pyrooz, 2019).These findings demonstrate the value of focusing on gang embeddedness, which recognizes multiplexity in gang involvement, while also revealing nuances in associations, behaviors, identity and status, and power in ways that bounded approaches cannot (e.g., Pyrooz et al., 2017, p. 297).What brings about these empirical observations is worthy of continued study.Since parole and reincarceration are ruled out as explanations of behavioral-cognitive decoupling, there is a need to search for factors outside of prison as sources of variation in trends in post-release gang embeddedness.lopez-Aguado (2018) and others point to the manifold burdens encountered in the reentry process, which could lead former gang members to leverage the gang as part of the reintegration strategy.
The final major point concerns the significance of gang organizational structure to contour trajectories of gang embeddedness, although the findings did not comport perfectly with the research hypothesis.The high-organization STGs and prison-oriented cliques performed as expected: continuity in gang embeddedness for the former and change for the latter.instrumental-rational gangs, such as the Aryan Brotherhood of Texas and Texas Syndicate, aim to protect their interests and impose rules on their members, with tentacles that extend from the prison to the street.informal-diffuse gangs, such as the various Tango sets and Peckerwoods, lack these features, instead viewing themselves more as affinity and defense groups (Bolden, 2020;Gundur 2022;Pyrooz & Decker, 2019).The gang embeddedness trajectories for low-organization STGs and street-oriented cliques behaved in unexpected ways: a negative linear trend for the former and a null linear trend for the latter.low-organization STGs, which were overwhelmingly a smattering of Blood and Crip gang sets, may lack the interest or power to shape the reentry process for their members, thus promoting change.The "street" component to street-oriented cliques, alternatively, could reflect the importance of community influences, as most former prisoners return to their communities, again pointing to the relevance of factors outside of prison (Kubrin & Stewart, 2006;leverentz, 2022;Wallace, 2015).The prison-street connection between gangs in prison and gangs on the street warrants continued inquiry.And the typological strategy used to measure gang organizational is also deserving of further scrutiny, as alternative approaches, such as social network analysis or qualitative case studies, could yield different results.Still, at this point, it appears that the organizational structure of groups matters in the reentry process.
The policy and practice implications of these findings are not subtle.The conclusion that prisons are a vector of gang activity puts the onus on correctional agencies to devise solutions to counter the grip of gangs on institutional life.Even though disengagement from gangs with prisoner reentry was the norm, active gang members, and, to a lesser extent, former gang members are more likely to recidivate post-release and encounter rockier reentry experiences than non-gang members (Pyrooz et al., 2021).Concentrating on the contours of joining and leaving gangs with prevention and intervention programming could assist in blocking on-ramps and closing off-ramps, though the evidence is far too sparse to identify "what works" (McGrath, 2020;Pyrooz, 2022).it appears that the realities of reentry overwhelm or displace embeddedness in gangs, though less so for some groups than others.The top priority should be to respond to the gang ties of former prisoners with service provisions and supervisory conditions that do not backfire to promote iatrogenic reengagement with gangs (Rubenson et al., 2021).it appears that special consideration should be aimed at the proximal reentry period, a likely window of opportunity to support former prisoners in conventional endeavors.
Recommendations for policy and practice are predicated on sound evidence, and given that this work is based on a representative sample in a single state, Texas, replications and extensions are warranted.Other features of carceral gang spillover require empirical inquiry, including how the migration of people from the prison to the street shapes gang behaviors, politics, and structures.And it is equally necessary to enumerate the various channels of airborne transmission of carceral gang spillover.Since gangs on the street and in prison are co-constitutive, it also means that it is also necessary to understand bidirectionality (Moore, 1991;Skarbek, 2014).lastly, it is important to understand the impact of behavior-cognitive decoupling on recidivism, victimization, and other outcomes of concern for criminologists.

Conclusion
A large proportion of the reentering population had ties to gangs while they were imprisoned.This study showed that embeddedness in gangs recedes rather than persists upon reentry to the community, suggesting while prisons are a vector for gang activity, former prisoners are remnants of their institutionalized selves.Disengagement from gangs occurred with prisoner reentry regardless of one's prison gang status or embeddedness in prison, and the residue of imprisonment further declined as people remained in the community.These findings should aid in tempering some concerns surrounding the specter of prison influence on community gang dynamics, though more research is warranted on carceral gang spillover.

Figure 1 .
Figure 1.Predicted gang embeddedness values at pre-and post-release interviews.Note: Wave 2 and 3 interviews occurred around 2 and 10 months post-release.Values for embeddedness by wave are derived from the linear predictions of two-level models (n = 622, n*w = 1632) with waves nested within persons fixing pre-release control variables (age, racially black, ethnically latino, educational attainment, father, street-only membership, prior arrests, violent conviction, years imprisoned, number of prison spells, season of release, reincarceration, and time elapsed between waves) at mean values.see table s1 for complete model results.

Figure 2 .
Figure 2. Predicted gang embeddedness values by pre-release gang status.Note: Wave 2 and 3 interviews occurred around 2 and 10 months post-release.Predicted values for embeddedness by wave are derived from Model 2 of table 2.

Figure 3 .
Figure 3. Predicted gang embeddedness values by gang organizational structure.Note: Wave 2 and 3 interviews occurred around 2 and 10 months post-release.Predicted values for embeddedness by wave are derived from Model 2 of table 3.

Table 1 .
Measurement timing and descriptive statistics for the study variables.

Table 2 .
Mixed effects maximum likelihood estimates of gang embeddedness moderated by pre-release gang status.

Table 3 .
Mixed effects maximum likelihood estimates of gang embeddedness moderated by gang organizational structure.