Skip to main content
PMC home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2020 May 1.
Published in final edited form as: Crime Delinq. 2018 Oct 13;65(5):681–704. doi: 10.1177/0011128718800286

Do We Have It All Wrong? The Protective Roles of Peers and Criminogenic Risks From Family During Prison Reentry

Thomas J Mowen 1, John H Boman IV 1
PMCID: PMC7034934  NIHMSID: NIHMS1037213  PMID: 32089563

Abstract

Prior work on the process of reentry from prison has highlighted the pivotal role that family and peers play during reintegration. Families are traditionally understood as important protective mechanisms against recidivism whereas peers are typically viewed as primarily criminogenic. Yet, drawing from differential coercion and social support theory, family and peer relationships can both be supportive (and protect against recidivism) and coercive (and contribute to recidivism). Using four waves of data from the Serious and Violent Offender Reentry Initiative, results of mixed-effects models demonstrate that family, but not peer, coercion relates to increased odds of reincarceration. Peer, but not family, social support relates to decreased odds of reincarceration. Findings suggest families are primarily criminogenic, whereas peers are protective during reentry.

Keywords: reentry, differential coercion and social support, family, peers, reincarceration

Introduction

Due to largely high rates of incarceration (Carson, 2018), the United States releases more people from prison on an annual basis than any other country in the world. Approximately 10,000 individuals are released from prison each week, and around 650,000 are released each year (Department of Justice, 2018). The process of returning home from prison—referred to as “reentry” (see Travis, 2005)—is a time that presents a unique set of challenges to returning persons. Nearly everyone who is incarcerated is eventually released (Hughes & Wilson, 2002). Consequently, developing a more comprehensive understanding concerning the respective roles of families and peers during reentry is a pressing matter for researchers, policy makers, and practitioners.

Existing studies clearly demonstrate the importance of family (Breese, Ra’el, & Grant, 2000; Comfort, 2008; Martinez & Christian, 2009; Naser & La Vigne, 2006; Petersilia, 2003; Shapiro & Schwartz, 2001; Visher & Courtney, 2007; Western, Braga, Davis, & Sirois, 2015) and peers (e.g., Binswanger et al., 2012; Boman & Mowen, 2017; Martinez & Abrams, 2013) during the reentry process. Families are almost always highlighted as important helping mechanisms. In addition to providing emotional support, families also play a key role by providing instrumental support—transportation, housing, and financial assistance—to those who are reentering society (Berg & Huebner, 2011; Visher & Courtney, 2007; Western et al., 2015). Although families are traditionally viewed as being beneficial to returning persons, peers are typically seen as factors that increase the likelihood of recidivism (e.g., Boman & Mowen, 2017). In addition to contributing to the person’s initial arrest, peer influence is often cited as a key risk factor for reentry failure (Andrews & Bonta, 2006; Andrews, Bonta, & Hoge, 1990). Prior work has shown that contact with criminal peers is a robust correlate of recidivism and rearrest (Cobbina, Huebner, & Berg, 2012; Davis, Bahr, & Ward, 2012) as well as substance use (Binswanger et al., 2012; Mowen & Boman, 2018b) following release from prison.

Although the common perspective on the roles of families and peers during reentry is grounded in a sizable amount of research (see Bahr, Harris, Fisher, & Armstrong, 2010; Boman & Mowen, 2017; Breese et al., 2000; Comfort, 2008; Martinez & Christian, 2009; Naser & La Vigne, 2006; Petersilia, 2003; Shapiro & Schwartz, 2001; Visher & Courtney, 2007; Western et al., 2015), the extent to which people “fail” the reentry process by being reincarcerated in the United States remains alarmingly high. Nearly half of all released individuals will be rearrested within the first year, and about two thirds of all individuals will be reincarcerated within the first 3 years after release (Department of Justice, 2018). This fact serves as a striking realization that researchers must continue to push the envelope by continuing to evolve the understanding of how social relationships affect recidivism during reentry.

One particular theoretical perspective that remains underutilized in the context of reentry research is Colvin’s (2000) and Colvin, Cullen, and Vander Ven’s (2002) theory of differential coercion and social support (DCSS). As an integrated theory of crime, DCSS allows for a bifurcated view into crime causation by emphasizing that supportive, protective factors constantly compete against “coercive factors,” or factors which increase (or “coerce”) people into committing crime. DCSS is an extremely valuable perspective to criminologists who study reentry because it allows for the same social relationship to simultaneously exert beneficial and detrimental impacts on criminal behavior. In other words, DCSS allows for families to not only be viewed as protective factors, but also risk factors. Similarly, DCSS provides the impetus to view peers as being more than risk factors by highlighting the possibility that they may be agents of support who decrease recidivism. This raises attention to the goals of the current study.

Using data from the Serious and Violent Offender Reentry Initiative (SVORI), the current study draws upon DCSS theory to investigate the extent to which families and peers may simultaneously serve as both supportive and coercive entities during reentry. Adhering to the main tenets of DCSS theory, we view family support and peer support as being factors that should decrease the likelihood of reincarceration. To capture coercion, we examine the extent to which family conflict and peer criminality impact the likelihood of reincarceration. Prior to discussing the specifics about our hypotheses and methods, we offer an overview of DCSS theory and discuss how extant findings on families and peers during the reentry process fit rather neatly into the purview of the theory.

DCSS

DCSS theory (Colvin, 2000; Colvin et al., 2002) is rooted in the notion that coercion is the major source of criminal behavior whereas social support is the major protective factor against offending. Thus, the theory is focally concerned with how the interplay between coercion and support affects crime. As an integrated theory, DCSS draws heavily from Cullen’s (1994) notion of social support and Agnew’s (1992) general strain theory. The social support element of the theory emphasizes that support can be derived from both micro-and macrosources. Regardless of its source, social support restrains individuals from offending by providing mechanisms that allow an individual to “cope with adversity through noncriminal means” (Colvin et al., 2002, p. 24). In addition to being a central component of general strain theory, the “coping” concept is key to how coercion is formulated in DCSS theory. Coercion refers to negative or criminogenic experiences that compel an individual to engage in deviance. Although coercion can also stem from both micro-and macro-level sources, Colvin et al. (2002) stressed that “coercive interpersonal relations constitute the most aversive and negative forces individuals encounter” (p. 22).

DCSS is an extraordinarily versatile theory as it allows for the same social factor to provide coercive and supportive influences. Although social support can be derived from a variety of sources, Colvin et al. (2002) identified families as the major source of support. To this end, they state “social support exist[s]… in the immediate interactions within families” (Colvin et al., 2002, p. 20). Thus, families play a key role in reducing crime through social support. At the same time, families can simultaneously contribute to crime by providing coercion through “aversive family interchanges” (Unnever, Colvin, & Cullen, 2004, p. 246). As we highlight in a moment, this comprehensive theoretical portrait is supported by research finding that family support can reduce crime during reentry (e.g., Visher & Courtney, 2007) but family conflict can cause it (Mowen & Visher, 2015).

DCSS’s theoretical comprehensiveness is also echoed when thinking about the roles of peers on behavior. To DCSS, peers who provide social support are theorized to reduce crime whereas peers who incite coercion cause crime. In the original conceptualization of coercion, (Colvin, 2000: see pp. 5, 36, 59–68, 72–81) deviant peer groups are coercive. In fact, Colvin draws directly from Sutherland’s (1947) differential association when he says that the term “differential” references the “degree” (p. 5) to which people are or are not exposed to a coercive deviant peer group (see supporting research by Costello & Zozula, 2018). Emphasizing the intertwined nature of the theory, Colvin’s line of thinking meshes with the revised version of general strain created by Agnew (2006), who noted that deviant peers are a source of strain (see also Agnew, 1991). As such, DCSS allows for peer groups to be both protective via support and harmful via coercion, thus making the theory extraordinarily diverse and closely tied to the meaning of the theory’s first word—”differential.” As we outline momentarily, this orientation is supported by research demonstrating that peers contribute to recidivism through coercion while also providing protection via social support (e.g., Visher & Travis, 2003).

Overall, the versatility of DCSS allows for the same social group to be viewed as factors that protect against crime via mechanisms of support or encourage offending through coercion. Furthermore, research on DCSS has demonstrated that it not only applies to life inside prison (Colvin, 2007) but also to individuals undergoing the reentry process as they can— and frequently do—experience support and coercion through a variety of social relationships (Day, Brauer, & Butler, 2015; Listwan, Colvin, Hanley, & Flannery, 2010). We now turn to the extant research on family and peer relationships during reentry and then place these findings within the framework of DCSS.

Family and Peers During Reentry

With few exceptions, families are traditionally viewed as important mechanisms of social support during reentry. Family support has been tied to a host of positive reentry outcomes including lower substance use and offending, improved mental health outcomes, and decreased reincarceration rates (Bahr et al., 2010; Mallik-Kane & Visher, 2008; Wallace et al., 2016; Western et al., 2015). Moreover, individuals returning from prison tend to highlight that family support was the most important factor in keeping them from going back to prison (Courtney & Visher, 2007). Studies have shown that families provide support to returning individuals by providing housing, transportation, and financial assistance (Western et al., 2015), connect individuals to employers (Berg & Huebner, 2011), and are far more likely to overlook an individuals’ criminal record than any other social group (Ekland-Olson, Supancic, Campbell, & Lenihan, 1983). In sum, the research clearly conceptualizes families as central mechanisms of social support.

Although families are typically viewed as important sources of social support, peers are typically regarded as primarily criminogenic. Examining a large sample of returning adults, Boman and Mowen (2017) found that individuals who associated with criminally active peers were likely to offend themselves. Similar results were found among returning youth— association with delinquent peers was associated with increased offending and substance use (Mowen & Boman, 2018a). The notion that peers are criminogenic has been highlighted in other work as well (e.g., Abrams, 2007; Breese et al., 2000; Visher & Travis, 2003). For example, in examining factors that promoted successful reentry among individuals undergoing reintegration in Illinois, Yahner and Visher (2008) found that returning individuals who had more antisocial peers were significantly more likely to be reincarcerated than those who reported having fewer, or no, antisocial peers. Having criminal peer associations is also regarded as one of the central factors that places an individual at an increased risk of recidivating (e.g., Andrews, Bonta, & Wormith, 2011).

Although the above reviewed studies provide strong evidence that families are supportive and peers are coercive in the reentry process, emerging research suggests that the opposite may also be true. Using data from the Returning Home Study, Mowen and Visher (2015) demonstrated that family conflict was significantly associated with increased odds of recidivism among men undergoing reentry and that family support played no significant role in this process. In another study using data from the SVORI, Wallace et al. (2016) found that family conflict was significantly associated with worse mental health outcomes among returning men and women. Other research has also demonstrated that family relationships can become strained during the reentry process (e.g., Western et al., 2015) and can create tension and stress among family members (Bahr et al., 2010; Grieb et al., 2014), likely resulting in strain and coercion.

Just as families may be coercive, emerging research has begun to highlight the ability of peers to offer social support during reentry. For example, Bahr et al. (2010) found that individuals were more likely to succeed at parole when they spent time engaging in enjoyable activities with friends. Like prior work, Bahr and colleagues hypothesized that peers would have a criminogenic effect among returning individuals. Despite having an empirically and theoretically informed hypothesis, their examination led Bahr and colleagues to conclude that those who successfully complete parole were likely to list their friends as a significant prosocial resource. Similarly, in interviewing returning individuals in Washington D.C., Solomon, Gouvis, and Waul (2001) found that previously incarcerated individuals highlighted the importance of having positive peer networks and supportive friends during reentry. Other research has also found that peers can provide anticriminogenic social support to returning persons (Hlavka, Wheelock & Jones, 2015; Martinez & Abrams, 2013; Mowen & Boman, 2018b). As such, peers may be more than just “bad” for those who are returning home.

When placed within the framework of DCSS, the conglomerate of findings reveals a natural tension that exists between the protective mechanisms of support and criminogenic influences of coercion that exist for both family and peers. It is clear that family support can restrain offending, but it is equally clear that family conflict may incite criminal behavior. Likewise, while criminal peer associations can coerce an individual into offending, support provided by peers may also protect against offending. Finally, it is entirely possible that the coercive and supportive mechanisms offered by both family and peers may interact in their effects on recidivism during reentry. Overall, this discussion raises attention to the goals of the current study.

Current Study

Drawing on the preceding discussion, the broad objective of this study is to investigate how supportive and coercive mechanisms from families and peers “compete” against each other during the process of reentry. Drawing on DCSS theory (Colvin et al., 2002) and using data from the SVORI project, we have three specific research questions.

Research Question 1: Do mechanisms of family and peer coercion relate to reincarceration?

We expect that higher levels of family conflict and peer offending will relate to increased odds of incarceration (Hypothesis 1).

Research Question 2: How do the supportive aspects of family and peer relationships relate to reincarceration?

Following the tenants of DCSS, we expect that peer and family support will both relate to decreased odds of incarceration (Hypothesis 2).

Our third research question relies heavily on the recognition that supportive and coercive mechanisms coexist. As a result, the supportive and coercive roles of families and peers should compete with each other in a meaningful way in the context of reincarceration. Following this, the third research question

Research Question 3: To what extent do supportive and coercive family and peer mechanisms jointly affect reincarceration?

Through the use of interaction terms, we expect that family and peer coercion will significantly reduce the effect of the prosocial aspects of peer and family support (Hypothesis 3).

Method

Data

Data for this project come from the SVORI. A multisite evaluation, SVORI was designed to examine the impact of enhanced reentry programming on outcomes pertinent to criminal justice, employment, education, health, and housing (National Institute of Justice, 2018). The enhanced reentry programs contained a variety of extra resources including substance abuse and mental health treatment, reentry planning, educational training, and programs to improve self-efficacy and decrease criminal attitudes. About half of the sample were designated SVORI participants whereas the other half did not receive SVORI designation (see Lattimore & Steffey, 2009 for an overview of SVORI methodology and data collection). Overall, data were collected between 2004 and 2007 from a total of 1,697 adult males across 12 different sites. Although sample selection varied across sites, to participate in SVORI, potential respondents needed to be serious of violent offenders.

One key advantage to this data set compared with other reentry data sets is that SVORI contains four waves of panel data, thus providing researchers the ability to examine longitudinal outcomes. At each wave, respondents completed face-to-face interviews, computer-assisted interviews, and oral swab drug tests with trained SVORI researchers. Administrative data were also collected from the National Crime Information Center. Wave 1 data were collected while the individual was still incarcerated (about 30 days prior to the scheduled date of release). Wave 2 data were collected 3 months postrelease, Wave 3 data were collected about 9 months following release, and Wave four data were collected 15 months following release from prison. During each wave, respondents were asked about a variety of experiences and characteristics including family and peer relationships, substance use, offending, marital and family factors, and treatment and programming. In the current study, we draw data from all four waves of the SVORI project.

Dependent Measure

The dependent measure in this study is reincarceration during Waves 2, 3, and 4. This binary measure has a mean of .234 and a standard deviation of .423. Importantly, respondents could report being incarcerated at one point in time but not another, making this dependent measure time variant. The within-individual standard deviation, which captures the time variant nature of reincarceration, is .246 (between individual SD = .374). Descriptive statistics for all measures used in this analysis are shown in Table 1.

Table 1.

Descriptive Statistics for the SVORI Sample.

M SD Minimum Maximum Within SD Between SD n
Dependent measure
 Reincarceration 0.234 0.423 0 1 0.246 0.374 1,346
Family and peer coercion
 Family conflict 6.367 1.922 3 12 0.982 1.738 1,282
 Peer crime 6.335 2.457 3 12 1.178 2.249 1,265
Family and peer support
 Family support 16.157 2.976 5 20 1.490 2.699 1,279
 Peer support 14.584 3.743 5 20 2.009 3.362 1,273
Time variant measures
 Employment 0.657 0.474 0 1 0.280 0.409 1,298
 Married/stable relationship 0.109 0.312 0 1 0.141 0.283 1,346
 Criminal offending 0.434 1.012 0 8 0.877 0.587 1,341
 Substance use 1.726 2.308 0 12 0.167 1.916 1,341
Time invariant measures
 White 0.340 0.473 0 1 — — 1,697
 Non-White 0.660 0.499 0 1 — — 1,697
 Age 26.590 7.460 15 68 — — 1,697
 Children 0.622 0.485 0 1 — — 1,697
 Length of incarceration (days) 918.30 932.35 44 9,486 — — 1,697
 Prior convictions 5.967 8.268 1 90 1,658
Primary conviction — — 1,697
 Violent 0.157 0.362 0 1 — — 1,697
 Drug 0.224 0.417 0 1 — — 1,697
 Property 0.119 0.323 0 1 — — 1,697
 Sexual offense 0.048 0.212 0 1 — — 1,697
 Other 0.450 0.498 0 1 — — 1,697
Programming and treatment
 Criminal attitude 0.438 0.496 0 1 — — 1,697
 Substance abuse treatment 0.429 0.495 0 1 — — 1,696
 Education 0.478 0.500 0 1 — — 1,697
 Case manager 0.535 0.499 0 1 — — 1,694
 Religious support 0.368 0.482 0 1 — — 1,697
SVORI participant 0.509 0.500 0 1 — — 1,697
Open in a new tab

Note. SVORI = Serious and Violent Offender Reentry Initiative, n = valid cases.

Differential Coercion Measures

Two of our key predictor variables capture family and peer coercion. First, family coercion is comprised of three items asking the respondent how much they agreed with the following statements: I tend to fight a lot with family members, I feel like I disappoint my family, and I am criticized a lot by my family. Possible responses ranged along a 4-point Likert-type scale (1 = strongly disagree, 2 = disagree, 3 = agree, 4 = strongly agree) and items were summed to create the family conflict scale. This average reliability across all waves is .745, indicating interitem consistency (Cronbach, 1951). The resulting measure of family conflict has a mean of 6.367, a standard deviation of 1.922, and ranges from 3 (very low conflict) to 12 (very high conflict). This measure is time variant as respondents could—and frequently did—report changes in family conflict across time.

Peer coercion is comprised of three items asking the respondent how many of their close friends are incarcerated, have assaulted someone, and sell drugs. Possible responses ranged along a 4-point Likert-type scale (1 = none, 2 = some, 3 = most, or 4 = all). The averaged alpha across all waves is .794, indicating acceptable reliability. The items were summed to create a scale of peer crime. This measure has a mean of 6.335, a standard deviation of 2.457, and ranges from 3 (no criminal peers) to 12 (all criminal peers). This measure is time variant and respondents reported significant changes in criminal peers across waves.

Social Support Measures

To create a measure of family social support, we draw data from five items asking the respondent if they had a family member who could provide help finding a place to live, help finding a job, substance abuse support, transportation, and financial support.1 Responses were captured using a 4-point Likert-type scale (1 = strongly disagree, 2 = disagree, 3 = agree, 4 = strongly agree). The averaged alpha across all waves is .881, indicating a strong level of reliability across items. Each item was summed to create a scale of family support. This measure has a mean of 16.157, a standard deviation of 2.976, and ranges from 5 (very low support) to 20 (very high support). This measure is time variant as respondents reported changes in family support across each postrelease wave.

To create the measure of peer social support, we use a similar set of items asked about friends. Specifically, respondents were asked five items about whether they had a friend who could provide: help finding a place to live, help finding a job, substance abuse support, transportation, and financial support. Responses fell along the same 4-point Likert-type scale as family support and the averaged alpha across each wave (.842) indicated strong consistency across the items. The mean for peer support is 14.584 with a standard deviation of 3.742 and a range from 5 to 20. Higher scores capture higher peer support, and this measure is time variant as respondents reported changes in peer support across time.

Time Variant Controls

To account for the influence of employment, we include a time variant measure indicating whether the respondent was legally employed at each wave (1 = yes). Overall, 65.7% of respondents reported being employed during the study time frame, though there was considerable variation in this mean (overall SD = .474). We also include measures to capture partnership status by including a time variant binary variable indicating if the respondent was married or in a stable relationship (1 = yes). Overall, 10.9% of respondents reported being married/in a stable relationship during the study time frame. We allow this measure to change across time as respondents could report being in a partnership in one wave and not another.

In addition to employment and partnership status, we include measures of criminal offending and substance use as they should be significantly predictive of reincarceration. To account for criminal offending, we draw from eight questions asking the respondent (1 = yes, 0 = no) if they had committed a violent crime, a crime against a person, carried a weapon, sold drugs, committed another drug crime (not including possession), driven under the influence of alcohol, committed a property offense, or some other lesser crime. We summed each item to create a variety index of criminal offending. The overall mean of this index is .434 with a standard deviation of 1.012 and a range from 0 (no offending) to 8 (all types of offending committed). This measure is time variant (within person SD = .877) as respondents could report committing offenses in some waves and not others.

To control for substance use, we draw data from 12 questions asking the respondents if they had used (1 = yes, 0 = no) tranquilizers, stimulants, steroids, marijuana, hallucinogens, cocaine, heroin, methamphetamine, inhalants, sedatives, pain relievers, or methadone since the last interview. As some of these substances can be prescribed, respondents were asked about illicit use only. Each item was summed to create a variety index of substance use. This measure has a mean of 1.726, a standard deviation of 2.308, and ranges from 0 (no substance used) to 12 (all substances used). Like criminal offending, this measure is time variant.

Time Invariant Controls

We also include controls for a wide range of time invariant measures. First, we include a measure indicating that the respondent was non-White (66.0% of the sample) in contrast to White (34.0% of the sample). We also include the age of the respondent at the time of their current term of incarceration. The average age of 26.59 years has a standard deviation of 7.46 and ranges from 15 to 68 years. Although the median (25 years) is close to the mean, this measure is skewed due to significantly higher numbers of respondents in their mid-20s relative to older respondents. As a result, we use the natural logarithm. We also include a measure capturing whether the respondent reported having any children at Wave 1. Overall, 62.2% of the sample reported having at least one child. We considered including this as a time variant measure, but very few respondents reported changes in parenthood or the number of children they had during the study time frame.

We also include a control for the total length of incarceration in days. This measure has a mean of 918 days, a standard deviation of 932 days, and ranges from 44 to 9,486 days. To account for the skew in the data (median = 656 days), we use the natural logarithm in the analysis. To account for criminal history, we also include the total number of prior convictions the individual reported receiving. This measure has a mean of 5.967, a standard deviation of 8.268, and ranges from 1 to 90. We examined the influence of outliers in this measure and note that the skewness in this measure did not significantly alter results with or without the outliers included. We also control for the primary conviction for the respondent’s current term of incarceration by including controls for those convicted of a violent (15.7%), drug (22.4%), property (11.9%), or sex crime (4.8%) in contrast to any other offense.

Prior work on the SVORI evaluation has demonstrated that some forms of programming and treatment have modest effects on recidivism (see Visher, Lattimore, Barrick, & Tuller, 2017). To control for this, we include binary measures indicating that the respondent reported participating in courses to change their criminal attitude (43.8% reported participating in courses), substance abuse treatment (42.9% reported receiving treatment), education classes (47.8% reported receiving education classes), having a case manager (53.5% reported having a case manager), or religious support (36.8% received religious support). Finally, we also include a measure indicating that the respondent was identified as a SVORI program participant to capture any differences between SVORI and non-SVORI participants.

Analytic Plan and Missing Data

Panel data such as the SVORI project require a statistical approach that accounts for a lack of independence across time. That is, individuals tend to be like themselves, causing panel data to violate the assumption of independence made by ordinary least squares (OLS) regression. A mixed-effects model—specifically, a two-level hierarchical model—introduces a random intercept to account for this similarity by nesting time within the individual. As the dependent measure is binary (1 = reincarcerated, 0 = not reincarcerated), we use a generalized linear mixed-effects model (Rabe-Hesketh & Skrondal, 2012). This results in the individual occurring at Level 2 (a between-person effect) with time occurring at Level 1 (a within-person effect).

One of the assumptions of the mixed-effects model is that the between-individual estimate exerts similar effects on the outcome measure as the within-individual estimate (Rabe-Hesketh & Skrondal, 2012). This assumption—called the assumption of equality—can be easily violated, thus introducing bias in the estimates. To examine this within the context of our study, we estimated each forthcoming model and performed a Hausman test of endogeneity. This test compares the fixed effects (within person) estimates to the random-effects estimates to examine whether there are significant differences in the estimates. Results yielded a nonsignificant test statistic, thus indicating this assumption was met and validating the mixed models.

As missing data have been well documented in the SVORI data (see Lattimore & Steffey, 2009), we also examined the prevalence and influence of missing data in our analysis. Prior research has demonstrated that missing data in the SVORI data set occurs at random (see Mowen & Boman, 2018b; Wallace et al., 2016). Of the 1,697 respondents interviewed at Wave 1, we use data from 1,156 respondents. Their patterns of missingness were tested to determine whether any of the measures used in our analysis were significantly related to, or predictive of, attrition. Results of this attrition analysis revealed nonsignificant t tests across all measures for individuals retained in the sample relative to those removed from the analysis. This indicates that missing data were not biasing results. The lack of a relationship between attrition and missingness supports much of the prior work using the SVORI data (e.g., Lattimore & Visher, 2009; Visher et al., 2017; Wallace et al., 2016).

To examine our research questions and hypotheses, we first present a model with family and peer differential coercion in Model 1, along with all control measures. In Model 2, we remove the coercion measures and replace them with variables capturing family and peer social support, along with all control measures. Next, in Model 3, we present results containing both coercive and supportive elements of family and peer relationships, along with all control measures. Finally, we present a matrix of the interaction terms (estimated from unreported models) to investigate the conditional relationship across each measure of coercion and support. To create the interaction terms, the measures were group-mean centered (Paccagnella, 2006) and included along with the original measures in the mixed-effects models.

Results

Prior to presenting the results of the mixed-effect regression models, we first explored whether there was multicollinearity present across the variables used in the analysis. Using Allison’s (1991) criteria of desirable variance inflation factors (VIFs) below 2.5 and tolerances above 0.40, the collinearity diagnostics appeared favorable. No VIF exceeded 1.4 and all tolerances were well above the 0.40 threshold. As such, multicollinearity does not appear to be significantly affecting the results of the forthcoming models (Ender, 2018).

Results of the mixed-effects regression models examining family and peer coercive factors, along with all control measures, are shown in Table 2. Overall, the model fit statistic indicates the model fits the data (χ2 = 109.99, p < .001). The intraclass correlation (ICC) demonstrates that about 42.4% of the variation in the odds of reincarceration occurs between individuals (at Level 2) with the remainder of the variation occurring across time within individuals (at Level 1). To allow for more direct comparability between the key theoretical predictors, we present standardized effects (via z scores) for the coercion and support measures.

Table 2.

Generalized Mixed-Effects Models Assessing Reincarceration (n = 1,156).

Model 1 Model 2 Model 3
95% CI
 95% CI
 95% CI
B Lower Upper B Lower Upper b Lower Upper
Family and peer differential coerciona
 Family conflict 0.699*** 0.500 0.896 — — — 0.628*** 0.423 0.832
 Peer crime 0.157 −0.019 0.332 — — — 0.155 −0.023 0.332
Family and peer social supporta
 Family support — — — −0.315*** −0.490 −0.140 −0.133 −0.315 0.050
 Peer support — — — −0.288*** −0.464 −0.112 −0.257*** −0.435 −0.075
Time variant controls
 Employment 0.429* 0.069 0.790 0.399* 0.038 0.759 0.483* 0.113 0.853
 Married/stable relationship −0.431 0.138 −1.001 −0.419 −0.983 0.146 −0.456 −1.038 0.125
 Criminal offending 0.497*** 0.324 0.669 0.592*** 0.417 0.768 0.509*** 0.333 0.686
 Substance use 0.345*** 0.198 0.492 0.389*** 0.242 0.537 0.335*** 0.186 0.485
Time invariant controls
 Non-White 0.527* 0.129 0.925 0.537* 0.140 0.934 0.506* 0.099 0.913
 Age (logged years) −0.443 −1.323 0.437 −0.655 −1.526 0.216 −0.515 −1.418 0.387
 Children 0.145 −0.241 0.530 0.160 −0.225 0.544 0.148 −0.247 0.543
 Length of incarceration (logged days) 0.073 −0.164 0.310 0.021 −0.215 0.257 0.053 −0.191 0.296
 Prior convictions −0.007 −0.029 0.015 −0.002 −0.023 0.020 −0.008 −0.030 0.014
Primary conviction
 Violent 0.068 0.270 0.789 0.113 −0.385 0.610 0.051 −0.461 0.562
 Drug −0.507* −1.001 −0.014 −0.358 −0.846 0.129 −0.470 −0.974 0.034
 Property 0.211 −0.341 0.764 0.341 −0.210 0.892 0.231 −0.335 0.797
 Sex crime −0.033 −0.909 0.843 0.073 −0.794 0.941 −0.011 −0.908 0.886
Programming and treatment
 Criminal attitude 0.321 −0.090 0.733 0.388 −0.023 0.798 0.369 −0.054 0.792
 Substance abuse treatment 0.145 −0.259 0.549 0.173 −0.229 0.575 0.164 −0.249 0.578
 Education −0.141 −0.533 0.251 −0.181 −0.571 0.209 −0.153 −0.554 0.247
 Case manager −0.083 −0.474 0.308 −0.017 −0.408 0.374 −0.043 −0.443 0.358
 Religious support 0.148 −0.219 0.515 0.168 −0.198 0.535 0.177 −0.199 0.554
SVORI participant −0.007 −0.379 0.365 −0.060 −0.431 0.310 −0.009 −0.390 0.371
Random intercept 0.886 0.306 1.467 0.908 0.236 1.490 0.961 0.380 1.543
ICC 0.424 0.430 0.443
Model fit χ2 109.99*** 102.47*** 108.02***
Open in a new tab

Note. b = regression coefficient; CI = confidence interval; ICC = intraclass correlation.

a

Standardized coefficients presented.

*

p ≤ .05.

**

p ≤ .01.

***

p ≤ .001.

In Model 1, results demonstrate that family conflict—as a coercive force—relates to significantly higher odds of reincarceration. That is, individuals with higher levels of family conflict report significantly greater odds of reincarceration. Converting the logged odds to odds ratios through exponentiation can help in understanding the magnitude of effect (see UCLA Statistical Consulting Group, 2018). Applying this to the results, findings reveal that a one standard deviation increase in family conflict relates to a 101% increase in the logged odds of reincarceration. After accounting for the coercive family effect, peer crime does not significantly relate to reincarceration. Results of the control measures largely support prior literature as employment is associated with a reduction in the odds of reincarceration, whereas criminal offending and substance use are robustly related to increased odds of reincarceration. Non-White individuals, compared with White individuals, report higher odds of reincarceration.

Model 2 of Table 2 examines the effect of family and peer social support on reincarceration. Overall, about 43% of the variation occurs between individuals. The remaining 57% of variance occurs within individuals over time. Like the prior model, the model statistic indicates strong fit to the data (χ2 = 102.47, p < .001). Results of this mixed-effects model demonstrate that both family and peer support relate to a reduction in the odds of reincarceration. Specifically, a one standard deviation increase in family instrumental support is associated with a 27.1% decrease in the logged odds of arrest while a one standard deviation increase in peer instrumental support is associated with a 25.0% decrease in the logged odds of arrest. Results of the control measures are the same as the prior model.

Finally, in Model 3, we include each measure of family and peer coercion and support in the same model. Like the prior two models, this model fits the data well (χ2 = 108.02, p < .001) with about 44.3% of the variation occurring between individuals. Findings demonstrate that once family and peer support and coercion are included, family conflict relates to significantly higher odds of reincarceration whereas peer support relates to significantly lower odds of reincarceration. Specifically, a standard deviation increase in family conflict is related to an 87.3% increase in the logged odds of reincarceration. The same increase in peer instrumental support is related to a 22.5% decrease in the logged odds of offending. Notably, family support and peer crime do not significantly relate to reincarceration in the full model. Overall, these findings suggest that family plays a strong coercive role in the reentry process while peers provide a strong protective effect as agents of social support.

To examine the interactive nature of these coercive and supportive mechanisms on reincarceration, we examine interaction terms among the key theoretical predictors. Due to the number of interaction terms, the measures are shown in matrix format in Table 3 for brevity. Each interaction term was stepped into the full model in separate, but unreported, models (the main effect results were identical to Model 3 in Table 2). As shown by this matrix, no interaction reaches significance, thereby suggesting that the supportive and coercive elements of family and peer relationships are independent—and not interdependent—in their effects of reincarceration.

Table 3.

Interaction Matrix From Six Generalized Mixed-Effects Models.a

Family support Family conflict Peer support Peer offending
95% CI
 95% CI
 95% CI
 95% CI
b Lower Upper B Lower Upper b Lower Upper b Lower Upper
Family support — —
Family conflict −0.005 −0.027 0.019 — — —
Peer support 0.004 −0.008 0.016 −0.002 −0.023 0.019 — —
Peer offending −0.004 −0.021 0.014 0.028 −0.059 0.002 −0.014 −0.028 0.001 — — —
Open in a new tab

Note. No asterisks noted = nonsignificant interaction; b = regression coefficient, CI = confidence interval.

a

Full models not tabled for brevity.

*

p ≤ .05.

**

p ≤ .01.

***

p ≤ .001.

Discussion

The goal of this study was to examine the influence of family conflict, peer offending, family support, and peer support on reincarceration among a large sample of returning men. Viewed through the lens of DCSS (Colvin et al., 2002) theory, we sought to examine the independent and interdependent relationships of mechanisms of coercion and support on reincarceration. In this concluding section, we review key findings, offer some thoughts on theory and policy alike, and discuss limitations of this study.

The current study had three research questions and adjoining hypotheses. The first hypothesis proposed that family and peer coercion would relate to reincarceration. This expectation is only partially supported. In support of this hypothesis, findings demonstrated that family conflict was significantly associated with reincarceration. However, peer offending behaviors were not associated with reincarceration. Adhering to the social support side of DCSS theory, the second hypothesis premised that family and peer social support would relate to lower odds of incarceration. Results for this hypothesis are also mixed. Although initial findings demonstrated that peer and family support related to lower incarceration likelihood, this finding disappeared after coercive measures were included. As a result, our findings demonstrate that only peer support—and not family support—relates to decreased levels of reincarceration. The third hypothesis expected the coercive elements to undermine the prosocial effects of support. Results from a series of interactions demonstrated that family conflict, peer crime, family support, and peer support are entirely independent—and not interdependent—in their effects on reincarceration. Thus, this hypothesis is not supported.

The results of these findings offer three specific contributions to the literature. First, our findings highlight the importance and salience of family conflict as a coercive force on returning individuals. This effect is so strong that peer criminality—an established predictor of recidivism (e.g., Boman & Mowen, 2017)—was not significantly related to reincarceration. However, results from our study demonstrate that peer criminality is largely spurious when family conflict is added to the equation. Thus, it is not that peer deviance and peer coercion do not affect the likelihood of reincarceration, but rather that family conflict appears to be the most significant and robust coercive force. Although a broad range of criminological research has demonstrated the significance of differential association on offending (e.g., Sutherland, 1947), our results suggest that family conflict—at least among returning individuals—may be a stronger source of coercion than peer crime. This finding on the importance of family coercion meshes with emerging research. Recently, Mowen and Visher (2015) found that family conflict was a significant predictor of substance use and offending among returning individuals. Our findings hone this result by emphasizing that negative family relationships may be more harmful for people who are reentering society than criminal peer influences. This finding not only extends the scope of reentry research, but also stresses the importance of further examining the roles of coercive elements of social relationships among a broader group of “differential associates” and crime than simply peers or friends.

The second major contribution of this work regards the finding that the presence of family conflict superseded the role of family social support— something highlighted in recent qualitative work. For example, Grieb and colleagues (2014) interviewed family members of returning individuals in West Baltimore, Maryland, to better understand their experiences. Grieb and colleagues note that every interviewed participant highlighted important challenges in providing support for a returning individual, and—as a result— all 39 interviewees said that providing support was “extremely stressful” (p. 1187). This stress had dramatic impacts on family members:

Participants discussed numerous sources of stress created through providing support of any kind; this stress, which was added to stress endured through other parts of their lives (chronic stress), was understood by the participants to have a negative impact on their health and well-being. (pp. 1187–1189)

Taken together, it is entirely possible that the stress and conflict that accompanies reentry may undermine the protective influence of family during this process.

In addition to the two important findings regarding family coercion, the third contribution is the importance of peer social support. Our findings revealed that even though family conflict “drowned out” the influence of family support and peer crime, peer social support remained a significant predictor of lower odds of incarceration. Two points are pertinent to this finding. First, some research has alluded to the possibility that peers can provide meaningful support during the reentry process. Using qualitative techniques, Western and colleagues (2015) found that returning people tend to evolve toward having less contact with family and more contact with friends the longer they are out. Building on this observation, Grieb and colleagues (2014) also found that friends are quite willing to provide support to the returning person. With these scholars’ research combined with these findings, the results at least imply that family conflict may drive returning people away from family and instead to supportive peers during the reentry process. This observation contributes to an important second observation: Based on these results, it appears that peers are protective factors—not risk factors—during the reentry process.

When placed within the framework of DCSS, our findings lead us to conclude that coercion is primarily a function of family relationships while support is primarily a mechanism derived from peers. Due to the established understanding that “family is good, peers are bad” (e.g., Andrews & Bonta, 2006; Boman & Mowen, 2017), this result paints a quite different picture of the roles of social relationships for those who are undergoing the reentry process. Moreover, the interaction terms demonstrated that these relationships were entirely independent—not interactive—in their effects. These findings further demonstrate the utility of DCSS as family coercion/support and peer coercion/support were derived from the exact same social group (families and peers, respectively).

In addition to the theoretical implications, this study also offers implications for policy and practice. Currently, a significant number of programs attempt to increase a person’s level of family support. A number of correctional institutes have programs in place to maintain family support (see Boman and Mowen, 2018, p, 214). Instead of these programs trying to increase family support, our results suggest another potential avenue that could prove to be fruitful: Refine these programs to try to provide people with the capability to reduce family conflict. This change could be relatively easily implemented and could dramatically reduce reincarceration rates.

Despite the contributions, this study is not without limitations. First, these findings speak only to the reentry experiences of men, and findings may not be applicable to women. Future research should examine how peer and family support and coercion relate to reentry outcomes among returning females. Second, although we focus on the pressing issue of reincarceration, there are other reentry outcomes and experiences that merit additional consideration such as employment, mental health and well-being, and substance use and abuse. Future research should examine how family and peer support and coercion relate to these other important reentry outcomes. In addition, the SVORI data end at 15 months postrelease. However, the reentry time frame may last many years. As a result, our findings are unable to speak to longer term outcomes within the reentry process. This is particularly important as social relationships with family and peers are likely to change well beyond the 15-month time frame. “Family” is also a broad term and may encompass parents and grandparents, children, spouses and extended family members for some individuals. As a result, it is not possible to specifically determine which individuals comprise the family members who were asked about in the SVORI instrumentation.

In addition, our measures of family and peer support capture instrumental support and not other forms of support such as emotional support (Martinez & Abrams, 2013). However, supplemental analyses (not shown) demonstrated that including measures of family emotional support produced substantively identical findings to family instrumental support. That is, after accounting for family and peer coercion as well as peer support, family support did not significantly relate to reincarceration irrespective of the specific measure of support used. Unfortunately, the SVORI data do not contain peer measures of emotional support, resulting in an important limitation. Future work should examine other forms of peer support on reentry outcomes. Finally, prior work has noted that the importance of housing during the reentry process (Fontaine, 2013). Due to the manner in which housing data were collected, we are unable to control for specific housing conditions, thus making this another notable limitation.

Even though there are some notable shortcomings, the take-home message of this study carries importance for researchers, practitioners, and policy makers alike. Our results clearly demonstrate that families may actually be “bad” during the reentry process whereas peers, perhaps paradoxically, may be “good.” Specific to our sample, our findings demonstrate that family conflict might provide a key causal mechanism that leads people to being reincarcerated. In the face of family conflict, people likely turn to peers for support. In turn, this peer support appears to protect people from reincarceration. Although these results apply to men reentering society after release from prison, researchers must explore the broader possibility that families may actually be criminogenic agents by introducing mechanisms of coercion into people’s lives. Equally strange, peers—who are so often viewed as catalysts of deviance—may instead provide people with forces and influences, which protect people from offending. Although the issue at hand is certainly relevant to reentry, the application of a broader sociological and criminological perspective to the traditional view that “families are good, peers are bad” makes one thing clear: Social relationships with families and peers—both during reentry and perhaps more generally—may be considerably more complex than previously realized.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This research was supported in part by the Center for Family and Demographic Research, Bowling Green State University, which has core funding from the Eunice Kennedy Shriver National Institute of Child Health and Human Development (P2CHD050959).

Author Biographies

Thomas J. Mowen is an assistant professor in the Department of Sociology at Bowling Green State University. His research focuses on the impact of punishment and incarceration on family and youth as well as life-course criminology. Tom’s recent work has appeared in Criminology, Justice Quarterly, and Journal of Research in Crime and Delinquency.

John H. Boman, IV is an assistant professor at Bowling Green State University in the Department of Sociology. His research is primarily focused on development, friendships, and social relationships through the life course. Recently, his work appears in Crime & Delinquency, Deviant Behavior, and Criminology.

Footnotes

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

1.

Families can provide both emotional and instrumental forms of support. Although the measure we use in this study represents instrumental family support, we do note that the SVORI data also contain measures of emotional family support. We explored differences in the analysis using measures of both family emotional support and family instrumental support. Substantive results were the same. As the SVORI data contain only instrumental measures of support for peers, we chose to use only instrumental family support measures for parsimony.

References

  1. Abrams LS (2007). From corrections to community: Youth offenders’ perceptions of the challenges of transition. Journal of Offender Rehabilitation, 44, 31–53. doi: 10.1300/J076v44n02_02 [DOI] [Google Scholar]
  2. Agnew R (1991). A longitudinal test of social control theory and delinquency. Journal of Research in Crime and Delinquency, 28, 126–156. doi: 10.1177/0022427891028002002 [DOI] [Google Scholar]
  3. Agnew R (1992). Foundation for a general strain theory of crime and delinquency. Criminology, 30, 47–88. doi: 10.1111/j.1745-9125.1992.tb01093.x [DOI] [Google Scholar]
  4. Agnew R (2006). Pressured into crime: An overview of general strain theory Los Angeles, CA: Roxbury. [Google Scholar]
  5. Allison PD (1991). Logistic regression using the SAS system: Theory and application Cary, NC: SAS Institute. [Google Scholar]
  6. Andrews DA, & Bonta J (2006). The psychology of criminal conduct (4th ed.). Newark, NJ: LexisNexis. [Google Scholar]
  7. Andrews DA, Bonta J, & Hoge RD (1990). Classification for effective rehabilitation: Rediscovering psychology. Criminal Justice and Behavior, 17, 19–52. doi: 10.1177/0093854890017001004 [DOI] [Google Scholar]
  8. Andrews DA, Bonta J, & Wormith JS (2011). The risk-need-responsivity (RNR) model: Does adding the good lives model contribute to effective crime prevention? Criminal Justice and Behavior, 38, 735–755. doi: 10.1177/0093854811406356 [DOI] [Google Scholar]
  9. Bahr S, Harris L, Fisher JK, & Armstrong AH (2010). Successful reentry: What differentiates successful and unsuccessful parolees? International Journal of Offender Therapy and Comparative Criminology, 54, 667–692. doi: 10.1177/0306624X09342435 [DOI] [PubMed] [Google Scholar]
  10. Berg MT, & Huebner BM (2011). Reentry and the ties that bind: An examination of social ties, employment, and recidivism. Justice Quarterly, 28, 382–410. doi: 10.1080/07418825.2010.498383 [DOI] [Google Scholar]
  11. Binswanger IA, Nowels C, Corsi KF, Glanz J, Long J, Booth RE, & Steiner JF (2012). Return to drug use and overdose after release from prison: A qualitative study of risk and protective factors. Addiction Science & Clinical Practice, 7, 1–9. doi: 10.1186/1940-0640-7-3 [DOI] [PMC free article] [PubMed] [Google Scholar]
  12. Boman IVJH, & Mowen TJ (2017). Building the ties that bind, breaking the ties that don’t: Family support, criminal peers, and reentry success. Criminology & Public Policy, 16, 753–744. 10.1111/1745-9133.12307 [DOI] [Google Scholar]
  13. Breese JR, Ra’el K, & Grant GK (2000). No place like home: A qualitative investigation of social support and its effects on recidivism. Sociological Practice: A Journal of Clinical and Applied Research, 2, 1–21. doi:1522-3442/00/0300-0001 [Google Scholar]
  14. Carson EA (2018). Prisoners in 2016 Washington, DC: U.S. Bureau of Justice Statistics; Retrieved from https://www.bjs.gov/index.cfm?ty=pbdetail&iid=6187 [Google Scholar]
  15. Cobbina JE, Huebner BM, & Berg MT (2012). Men, women, and postrelease offending: An examination of the nature of the link between relational ties and recidivism. Crime & Delinquency, 58, 331–361. doi: 10.1177/0011128710382348 [DOI] [Google Scholar]
  16. Colvin M (2000). Crime and coercion: An integrated theory of chronic criminality New York, NY: St. Martin’s Press. [Google Scholar]
  17. Colvin M (2007). Applying differential coercion and social support theory to prison organizations: The case of the penitentiary of New Mexico. The Prison Journal, 87, 367–387. doi: 10.1177/0032885507304774 [DOI] [Google Scholar]
  18. Colvin M, Cullen FT, & Vander Ven T (2002). Coercion, social support, and crime: An emerging theoretical consensus. Criminology, 40, 19–41. doi: 10.1111/j.1745-9125.2002.tb00948.x [DOI] [Google Scholar]
  19. Comfort M (2008). Doing time together: Love and family in the shadow of the prison Chicago, IL: The University of Chicago Press. [Google Scholar]
  20. Costello BJ, & Zozula C (2018). Peer influence: Mechanisms and motivations. Deviant Behavior, 39, 94–110. doi: 10.1080/01639625.2016.1260387 [DOI] [Google Scholar]
  21. Cronbach LJ (1951). Coefficient alpha and the internal structure of tests. Psychometrika, 16, 297–334. [Google Scholar]
  22. Cullen FT (1994). Social support as an organizing concept for criminology: Presidential address to the academy of criminal justice sciences. Justice Quarterly, 11, 527–559. [Google Scholar]
  23. Davis C, Bahr SJ, & Ward C (2012). The process of offender reintegration: Perceptions of what helps prisoners reentry society. Criminology & Criminal Justice, 13, 446–469. [Google Scholar]
  24. Day JC, Brauer JR, & Butler HD (2015). Coercion and social support behind bars: Testing an integrated theory of misconduct and resistance in U.S. prisons. Criminal Justice and Behavior, 42, 133–155. doi: 10.1177/0093854814546352 [DOI] [Google Scholar]
  25. Department of Justice. (2018). Prisoners and prisoner re-entry Retrieved from https://www.justice.gov/archive/fbci/progmenu_reentry.html
  26. Ekland-Olson S, Supancic M, Campbell J, & Lenihan KJ (1983). Postrelease depression and the importance of familial support. Criminology, 21, 253–275. doi: 10.1111/j.1745-9125.1983.tb00261.x [DOI] [Google Scholar]
  27. Ender P (2018). Applied categorical & nonnormal data analysis: Collinearity issues Retrieved from http://www.philender.com/courses/categorical/notes2/collin.html
  28. Fontaine J (2013). Examining housing as a pathway to successful reentry: A demonstration design process Washington, DC: Urban Institute. [Google Scholar]
  29. Grieb SM, Crawford A, Fields J, Smith H, Harris R, & Matson P (2014). “The stress will kill you”: Prisoner reentry as experienced by family members and the urgent need for support services. Journal of Health Care for the Poor and Underserved, 25, 1183–1200. doi: 10.1353/hpu.2014.0118 [DOI] [PubMed] [Google Scholar]
  30. Hlavka HR, Wheelock D, & Jones RS (2015). Exoffender accounts of successful reentry from prison. Journal of Offender Rehabilitation, 54, 406–428. doi: 10.1080/10509674.2015.1057630 [DOI] [Google Scholar]
  31. Hughes T, & Wilson DJ (2002). Reentry trends in the U.S Washington, DC: U.S. Bureau of Justice Statistics; Retrieved from https://www.bjs.gov/content/reentry/reentry.cfm [Google Scholar]
  32. Lattimore PK, & Steffey DM (2009). The multi-site evaluation of SVORI: Methodology and analytic approach (Document 230424). Washington, DC: U.S. Department of Justice. [Google Scholar]
  33. Lattimore PK, & Visher CA (2009). Multi-site evaluation of SVORI: Summary and synthesis Raleigh, NC: RTI International. [Google Scholar]
  34. Listwan SJ, Colvin M, Hanley D, & Flannery D (2010). Victimization, social support, and psychological well-being: A study of recently released prisoners. Criminal Justice and Behavior, 37, 1140–1159. doi: 10.1177/0093854810376338 [DOI] [Google Scholar]
  35. Mallik-Kane K, & Visher CA (2008). Health and prisoner reentry: How physical, mental, and substance abuse conditions shape the process of reintegration Washington, DC: Urban Institute. [Google Scholar]
  36. Martinez DJ, & Abrams LS (2013). Informal social support among returning young offenders: A meta-synthesis of the literature. International Journal of Offender Therapy and Comparative Criminology, 57, 169–190. doi: 10.1177/0306624X11428203 [DOI] [PubMed] [Google Scholar]
  37. Martinez DJ, & Christian J (2009). The familial relationships of former prisoners: Examining the link between residence and informal support mechanisms. Journal of Contemporary Ethnography, 38, 201–224. doi: 10.1177/0891241608316875 [DOI] [Google Scholar]
  38. Mowen TJ, & Boman JH IV. (2018a). A developmental perspective on reentry: Understanding the causes and consequences of family conflict and peer delinquency during adolescence and emerging adulthood. Journal of Youth and Adolescence, 47, 275–289. doi: 10.1007/s1096 [DOI] [PMC free article] [PubMed] [Google Scholar]
  39. Mowen TJ, & Boman JH IV. (2018b). The duality of the peer effect: The interplay between peer support and peer criminality on offending and substance use during reentry. Crime & Delinquency, 64, 1094–1116. doi: 10.1177/0011128717740529 [DOI] [PMC free article] [PubMed] [Google Scholar]
  40. Mowen TJ, & Visher CA (2015). Drug use and crime after incarceration: The role of family support and family conflict. Justice Quarterly, 32, 337–359. doi: 10.1080/07418825.2013.771207 [DOI] [Google Scholar]
  41. Naser RL, & La Vigne NG (2006). Family support in the prisoner reentry process: Expectations and realities. Journal of Offender Rehabilitation, 43, 93–106. doi: 10.1300/J076v43n01_05 [DOI] [Google Scholar]
  42. National Institute of Justice. (2018). About the serious and violent offender reentry initiative Retrieved from https://www.nij.gov/topics/corrections/reentry/pages/about-svori.aspx
  43. Paccagnella O (2006). Centering or not centering in multilevel models? The role of the group mean and the assessment of group effects. Evaluation Review, 30, 66–85. doi: 10.1177/0193841X05275649 [DOI] [PubMed] [Google Scholar]
  44. Petersilia J (2003). When prisoners come home: Parole and prisoner reentry Oxford, UK: Oxford University Press. [Google Scholar]
  45. Rabe-Hesketh S, & Skrondal A (2012). Multilevel and longitudinal modeling using Stata College Station, TX: Stata Press. [Google Scholar]
  46. Shapiro C, & Schwartz M (2001). Coming home: Building on family connections. Corrections Management Quarterly, 5, 52–61. [Google Scholar]
  47. Solomon AL, Gouvis C, & Waul M (2001). Summary of focus group with exprisoners in the district: Ingredients for successful reintegration Washington, DC: Urban Institute. [Google Scholar]
  48. Sutherland EH (1947). Principles of criminology (4th ed.). Philadelphia, PA: Lippincott. [Google Scholar]
  49. Travis J (2005). But they all come back: Facing the challenges of prisoner reentry Washington, DC: Urban Institute Press. [Google Scholar]
  50. UCLA Statistical Consulting Group. (2018). How do I interpret odds ratios in logistic regression? Retrieved from https://stats.idre.ucla.edu/other/mult-pkg/faq/general/faq-how-do-i-interpret-odds-ratios-in-logistic-regression/
  51. Unnever JD, Colvin M, and Cullen FT (2004). Crime and coercion: A test of core theoretical propositions. Journal of Research in Crime and Delinquency, 41, 224–268. [Google Scholar]
  52. Visher CA, & Courtney SME (2007). One year out: Experiences of prisoners returning to Cleveland Washington, DC: The Urban Institute. [Google Scholar]
  53. Visher CA, Lattimore PM, Barrick K, & Tuller S (2017). Evaluating the longterm effects of prisoner reentry services on recidivism: What types of services matter. Justice Quarterly, 34, 136–165. doi: 10.1080/07418825.2015.1115539 [DOI] [Google Scholar]
  54. Visher CA, & Travis J (2003). Transitions from prison to community: Understanding individual pathways. Annual Review of Sociology, 29, 89–113. doi: 10.1146/annurev.soc.29.010202.095931 [DOI] [Google Scholar]
  55. Wallace D, Fahmy C, Cotton L, Jimmons C, McKay R, Stoffer S, & Syed S (2016). Examining the role of familial support during prison and after release on post-incarceration mental health. International Journal of Offender Therapy & Comparative Criminology, 60, 3–20. doi: 10.1177/0306624X14548023 [DOI] [PubMed] [Google Scholar]
  56. Western B, Braga AA, Davis J, & Sirois C (2015). Stress and hardship after prison. American Journal of Sociology, 120, 1512–1547. doi: 10.1086/681301 [DOI] [PubMed] [Google Scholar]
  57. Yahner J, & Visher CA (2008). Illinois prisoners’ reentry success three years after release Washington, DC: The Urban Institute. [Google Scholar]

RESOURCES