Examining the Parole Officer as a Mechanism of Social Support During Reentry From Prison

Abstract

Emerging research has shown that the parole officer, much like friends and family, can be an important source of social support for returning persons. While this body of literature is growing, existing research provides little insight into understanding how specific types (e.g., interpersonal and/or professional) of parole officer support matter. Using panel data from the Serious and Violent Offender Reentry Initiative, results of mixed-effects models demonstrate that greater levels of parole officer support are associated with decreased odds of reincarceration. Furthermore, parole officer professional support (e.g., providing correct information) exerts a more robust effect than interpersonal support (e.g., listening and caring). Findings suggest policy makers should consider programming to strengthen the professional relationship between the parole officer and returning person.

Introduction

Standing at a rate of just under 600 per 100,000 residents by the end of 2016 (Carson, 2018), the United States incarcerates more individuals than any other country in the world. At the same time, nearly every person who is incarcerated will eventually be released (Travis, 2005) and undergo the process of reentry into society. The process of reentry is tumultuous and fraught with challenges, and many individuals become reincarcerated (Bahr, Harris, Fisher, & Armstrong, 2010; Durose, Cooper, & Snyder, 2014). In fact, estimates suggest that about one third of all individuals released will become reincarcerated within the first year of release (Durose et al., 2014). As a result of this relatively high rate of failure, a wide range of research has sought to understand factors that promote reentry success. One of the most widely researched and supported factors related to reentry success is social support (Bahr et al., 2010; Berg & Huebner, 2011; Martinez, 2008; Meyers, Wright, Young, & Tasca, 2017; Naser & La Vigne, 2006; Pleggenkuhle, Huebner, & Kras, 2016; Visher, 2004).

Recently released individuals tend to have very few resources (Petersilia, 2003; Visher & Courtney, 2007) and often rely on others for social support. Largely defined as the receipt of tangible or intangible resources from social connections (Colvin, Cullen, & Vander Ven, 2002), social support is central to the reentry process. Existing research tends to highlight the importance of families (Bahr et al., 2010; Berg & Huebner, 2011; Mitchell, Spooner, Jia, & Zhang, 2016; Visher, 2004; Western, Braga, Davis, & Sirois, 2015) and peers (Bahr et al., 2010; Martinez & Abrams, 2013; Mowen & Boman, 2019) as sources of social support. These respective groups can provide important mechanisms of support including emotional (e.g., love and care) and instrumental (e.g., physical assistance such as housing and financial) support. Perhaps unsurprisingly, individuals with social support tend to report more prosocial reentry outcomes than those who lack support (e.g., Boman & Mowen, 2018; Visher & Courtney, 2007). Although studies have tended to focus almost exclusively on family (e.g., Berg & Huebner, 2010; Naser & La Vigne, 2006; Pleggenkuhle et al., 2015), and to a lesser extent peers (e.g., Martinez & Abrams, 2013), other significant individuals/groups can also offer social support to the returning individual. In particular, an emerging line of research has shifted focus onto the parole officer.

Research has recently highlighted the importance of the parole officer for returning persons (e.g., Blasko, Friedmann, Rhodes, & Taxman, 2015; Chamberlain, Gricius, Wallace, Borjas, & Ware, 2018; Vidal, Oudekerk, Reppucci, & Woolard, 2015). As just one example, Chamberlain et al. (2018) found that positive parole officer rapport was associated with reduced recidivism, while negative parole officer rapport was associated with increased recidivism. Although this small group of studies has established that parole officer support matters, it is likely that the link between the parole officer and returning person is more than just “positive” or “negative.” When placed within the framework of social support—again, the receipt of resources through social connections (Colvin et al., 2002)—the parole officer is an important mechanism of social support as they provide the returning person with a wide range of resources such as assisting in acquiring and maintaining employment (Rakis, 2005), counseling the individual by talking and listening to them about their problems and hardships (Seiter, 2002), and connecting them with resources that target their reentry needs (Lowenkamp, Holsinger, Flores, & Koutsenok, 2013; Paparozzi & Caplan, 2009). In fact, the parole officer is often one of only a few social connections and sources of social capital for returning persons as they transition back into society (Wolff & Draine, 2004), and are likely to be a particularly important mechanism of social support (see also Seiter, 2002). However, beyond understanding the parole officer simply as another source of social support, when placed within the broader research on peers and families as mechanisms of support during reentry (e.g., Mowen, Stansfield, & Boman, 2018), there is ample rationale to believe that parole officer social support is multidimensional and comprised of interpersonal (e.g., listening and caring) and professional (e.g., providing important resources and acting professionally) social support components. Yet, existing research provides little in understanding how these specific types (e.g., interpersonal and/or professional) of parole officer support matters. As the vast majority of individuals reintegrating back into society will be placed on parole, and about 4.5 million individuals—one out of every 55 adults in the United States—were under some form of community supervision by the end of 2016 (Kaeble, 2018), a greater focus on the salience of parole officer social support is sorely needed.

Overall, understanding the capacity in which parole officer social support relates to reentry outcomes is pressing in terms of both theory as well as policy. Using four waves of data from the Serious and Violent Offender Reentry Initiative (SVORI), the goal of this project is to examine the influence of parole officer social support on reincarceration. Specifically, we first explore how overall levels of social support derived from the parole officer relates to reincarceration. We then examine the extent to which parole officer support may represent fundamentally unique theoretical and empirical dimensions of social support (e.g., interpersonal and/or professional), and how each factor relates to reincarceration. Below, we review the theoretical underpinnings highlighting why social support matters for the process of prison reentry and then overview the research on parole officer support and reentry.

Social Support and Prison Reentry

As Colvin et al. (2002) note, a wide range of theoretical traditions in the field of criminology focus on the notion that “social support prevents crime” (p. 19). In fact, in his 1994 address to the Academy of Criminal Justice Sciences, Cullen (1994) called upon researchers to recognize the “social support paradigm” (p. 527) of criminology, thus highlighting the utility of understanding that social support reduces criminal involvement. With ties to the concepts of social and cultural capital (Cullen, Wright, & Chamlin, 1999), social support is generally defined as the receipt of assistance from social connections (Colvin et al., 2002). Social support can take both material (e.g., tangible support such as financial assistance, transportation, and housing) as well as immaterial (e.g., less tangible support such as a sense of belongingness, love, and emotional support) forms. Social support can also be delivered from a wide range of entities including family, peers, social networks, governmental agencies, and communities (Colvin et al., 2002; Cullen, 1994).

Although Cullen (1994) and others (e.g., Colvin et al., 2002) have argued that social support must constitute its own focus of study, the concept of social support shares strong epistemological roots with other criminological traditions such as social control (Hirschi, 1969), strain theory (Agnew, 1992), and differential association (Sutherland, 1947). For example, strong attachment to others (social control) can be a form of social support whereby the emotional bonds one shares with others restrain criminogenic behaviors. Likewise, a wide range of studies has shown that coping mechanisms—such as financial assistance—provided by significant others (Bahr et al., 2010; Petersilia, 2003; Western et al., 2015)—can reduce the effect of strain on criminal behavior. Furthermore, elements of social support can also be found in differential association/social learning theory whereby individuals are likely to learn—and model—behaviors after significant others with whom they share close ties (e.g., peers), which may also be a form of social support.

When applied to the process of prison reentry, the importance of social support both as an empirical reality and as a theoretical perspective becomes immediately apparent. Individuals going through the process of reentry often have a difficult time transitioning back into society (Travis, 2005). Some of their difficulties arise when individuals have problems acquiring housing and employment, avoiding substances, encountering financial difficulties, and experiencing a lack of interpersonal relationships (Bahr et al., 2010; Martinez & Christian, 2009; Petersilia, 2003; Phillips, 2010; Travis, 2005; Visher & Courtney, 2007; Visher & Travis, 2003; Western et al., 2015). Perhaps unsurprisingly, research on prison reentry has shown that individuals with greater levels of social support tend to fare better than individuals who lack social support (Cochran, 2014; Hochstetler, DeLisi, & Pratt, 2010; Mowen & Boman, 2019; Visher, 2004). Existing research has tended to focus on how family, peers, and broader social networks can provide social support for returning individuals.

Berg and Huebner (2011) examined the link between social ties, employment, and recidivism among returning individuals. Findings highlighted that family support was a particularly important factor in assisting returning men in securing employment. Likewise, a wide range of studies have demonstrated that emotional family support is robustly related to prosocial reentry outcomes (Visher & Courtney, 2007). Similar findings exist for peers and social networks. For example, Cochran (2014) found that individuals who were able to maintain connections with their social networks outside of prison reported having lower rates of offending than individuals whose social ties were cut during incarceration. In this case, individuals who left prison received social support via their social network, which aided in the reentry process. Likewise, Mowen and Boman (2019) found that peers can offer instrumental forms of support (e.g., transportation, housing, and help desisting from substance use) to returning persons, and individuals with greater levels of peer support tended to report lower odds of recidivism than those with lower support. Although findings were mixed, Martinez and Abrams (2013) also found that, generally speaking, informal peer support helped young individuals navigate the reentry process.

Overall, research has clearly demonstrated the importance of social support within the reentry process. Yet, existing work has tended to focus on family and/or friends as mechanisms of social support. However, social support can also be drawn from others—including members of the criminal justice system—such as the parole officer, which we turn to next.

Research on the Parole Officer

Most individuals released from prison are placed on parole and assigned to a parole officer (Kennealy, Skeem, Manchak, & Eno Louden, 2012). The purpose of the parole officer during the reentry process is twofold. First, the parole officer serves as an important mechanism of control—to ensure that individuals do not pose public risk. The second role of the parole officer is to supply aid and support to the returning individual (Glaser, 1969; Kennealy et al., 2012). In this role, parole officers may engage in a wide range of supportive functions such as assisting the individual in finding a job or ensuring the returning person is receiving appropriate services. Functions such as assisting with housing needs, providing services and treatment, and facilitating connections with the broader community are also examples of social support. We now turn to a discussion of the extant research on the link between the parole officer and the returning person and tie these findings to a social support framework.

In a sample of 412 individuals mandated to psychiatric treatment, Skeem, Louden, Polaschek, and Camp (2007) found that individuals who perceived their supervising officer as caring, accommodating, and fair were far less likely to experience a parole violation than those who did not perceive their supervising officer as caring, accommodating, and fair. Likewise, increased trust between the individual and the supervision officer was also related to lower odds of a parole violation (Skeem et al., 2007). While the authors do not directly conceptualize the parole officer’s care, fairness, or accommodating actions as social support, these characteristics and actions could be considered intangible resources as they benefit the returning individual. Furthermore, as the resources are being provided from a member of the individual’s social network, these experiences likely represent parole officer social support (Colvin et al., 2002).

Producing similar findings, using a sample of 109 men and women on parole, Kennealy et al. (2012) found that individuals who reported a supportive relationship with their parole officer—characterized by being caring and fair—reported far less odds of arrest than those who did not perceive having a caring and fair relationship with their parole officer. Again, this finding suggests—at least implicitly—that parole-returning person relationships characterized by high levels of social support (re: supportive relationships) benefit from the social support they derive from the parole officer. This finding has been mirrored in more recent work; for example, in a sample of 480 returning men, Blasko et al. (2015) found that individuals with a more positive and supportive relationship (a form of social support) with their parole officer reported lower odds of parole violations and drug use than those with less supportive relationships, or lower parole officer social support.

Findings have also been mirrored in research using samples of women and youth. In a sample of 226 women spanning 2 years, Morash, Kashy, Smith, and Cobbina et al. (2015) found that dimensions of social support derived from the parole officer did not produce direct effects on recidivism. However, more supportive and less punitive relationships were associated with lower levels of anxiety, which was highly predictive of recidivism (Morash et al., 2016). The study by Morash and colleagues (2015) demonstrates that supportive social relationships between the returning person and the parole officer are vital during the reentry process as parole officer social support indirectly promoted prosocial reentry outcomes. In a sample of 140 youth, Vidal et al. (2015) found that a strong bond between the youth and the supervising officer was related to less violent recidivism. Specifically, the authors found that parole officer support exerted a significant effect on recidivism in the presence of low family support. Findings suggest that even in the absence of other forms of social support (e.g., family), social support from the parole officer can help promote prosocial outcomes.

In a more recent study, Chamberlain et al. (2018) recognized the parole officer relationship was potentially multifaceted. Using data from the SVORI, and the same seven measures as the current study (explored in greater detail in the next section), Chamberlain et al. (2018) divided individuals into having either (a) positive rapport assessed using five positively orientated statements; or (b) negative rapport, assessed using two negatively oriented statements. Findings demonstrated positive rapport was associated with decreased recidivism, while negative rapport was associated with increased recidivism. While this study is important as it demonstrates that the parole officer relationship can take on more than one form, the a priori classification of “negative” and “positive” parole officer rapport is an important limitation. The parole officer serves an important role as a distributor of social support instead of simply being negative or positive; therefore, it is likely that a more rigorous statistical analysis of these items will reveal theoretically and methodologically distinct components of the parole officer relationship.

While past research has shown that the parole officer can benefit individuals as they transition back into society (Chamberlain et al., 2018; Morash et al., 2015), very little, if any, research has conceptualized the supportive functions of the parole officer as social support. Furthermore, there is potential that the support from the parole officer mimics that of social support from family and friends in that it is multifaceted and comprised of both interpersonal and professional components. For example, when the parole officer offers social support through listening and providing a trustworthy and caring environment—similar to a counselor or family member (see Mowen, Stansfield, & Boman, 2019)—they are offering an interpersonal form of social support. Research in other fields exploring the benefits of receiving interpersonal support—feeling supported or being listened to—has found that this type of support can improve individuals’ social-emotional wellbeing (Suldo et al., 2009), improve social integration (Nils & Rimé, 2012), and in some populations of former offenders can reduce alcohol use (Rowe et al., 2007). Findings on interpersonal support in other fields, such as teaching and nursing, suggest interpersonal support is related to lower levels of emotional exhaustion (Kinman, Wray, & Strange, 2011) and to increases in ability to cope with stress (Teasdale, Brocklehurst, & Thom, 2001). This form of social support is focused on the emotional side of reentry, and provides an atmosphere whereby the returning person experiences interpersonal social support (e.g., Mowen et al., 2019) during the difficult, and often emotionally taxing, process of reentry (Travis, 2005). While having someone to listen to and talk to about problems may benefit returning individuals and help avoid reincarceration, other forms of support, such as professional support, may influence reentry success independent of the interpersonal support an individual is receiving, which we turn to now.

Past research has often identified professional support as receiving information, access to resources and other tangible opportunities, and being treated with respect (Fordham, Gibson, & Bowes, 2012; Singh & Billingsley, 1998). While professional support specifically from a parole officer has yet to be examined by scholars, research in nursing and teaching has also found that when individuals feel professionally supported (e.g., feel respected, are provided with professional resources and information), they report more positive outcomes such as lower rates of burnout and also report more commitment to their occupation (Antonopoulou, Killian, & Forrester, 2017; Hong, Song, Liu, Wang, & Wang, 2014; Singh & Billingsley, 1998). If the parole officer provides professionally supportive functions such as information on employment, treatment, and housing opportunities or access to them, it is likely the returning person will benefit from these supportive functions. In addition, parole officers also provide professional support when they act professionally and treat individuals on parole with respect. Professional support in the form of respect should increase compliance to parole conditions (Ireland & Berg, 2007) and, ultimately, a successful transition in society.

To date, no research that we are aware of has examined the interplay between interpersonal and professional parole officer social support during the process of reentry. Research has shown that both interpersonal and professional social support can be beneficial in respective ways, and it is possible that different dimensions of parole officer support may promote differing reentry outcomes. This discussion on the potential multidimensional parole officer relationship, consisting of both interpersonal and professional support, raises attention to the goals of the current study.

Current Study

Although prior research has shown that social support promotes prosocial reentry outcomes, much less remains known about how—and in what capacity—social support derived from the parole officer relates to reentry success. Using four waves of data from the SVORI, we address two specific research questions. Our first research question asks, Does parole officer support relate to reentry success? Like prior work on social support (e.g., Berg & Huebner, 2011; Pleggenkuhle et al., 2015; Visher, 2004), we expect (Hypothesis 1) that parole officer support will relate to decreased odds of reincarceration.

Second, again mirroring prior work on social support derived from other entities, our second research question asks, How do specific forms of parole officer social support relate to reincarceration? We expect that parole officer social support will be comprised of multiple factors including a professional factor (e.g., capturing instrumental forms of social support), and an interpersonal factor (e.g., capturing emotional forms of social support). Accordingly, we believe (Hypothesis 2) that parole officer professional support and interpersonal social support will relate to reduced odds of reincarceration.

Method

Data

In the current study, we use data from SVORI. SVORI was a federally funded reentry initiative focused on improving reentry success among individuals following release from prison. The overall goal was to examine the extent to which enhanced reentry programming related to improved housing, familial relations, mental/physical health, and decreased substance use/criminal behavior (Lattimore & Visher, 2009). Males, females, and juveniles participated in SVORI programming at 69 different locations across 14 states (Lattimore & Steffey, 2010; Lattimore & Visher, 2013). To examine the effectiveness of SVORI programs, roughly half of the participants were assigned to the treatment programs that were designed for SVORI, while the other half received normal treatment programs. Due to smaller sample sizes and high attrition rates within the female and juvenile samples, we only use the sample of returning males.

SVORI contains a total sample of 1,697 men at Wave 1, approximately half of whom were identified as SVORI program participants. To collect data, the SVORI research team used in-person interviews at four different time periods. Data for Wave 1 were collected while the respondent was still incarcerated, but within about 30 days of their scheduled date of release. Subsequent data collection occurred around 3, 9, and 15 months after the individual was released. Overall, 79.3% of respondents participated in the first wave and at least one other wave, with 58.0% responding at Wave 2, 61.0% at Wave 3, and 65.6% at Wave 4 (see Lattimore & Steffey, 2010, pp. 72–74 for a discussion of attrition). Because the overall goal of SVORI was to examine which, and to what extent, reentry programs and treatment services were effective in improving post-release success, data were collected on a variety of different topics including demographic characteristics, family relations, housing, employment, education, physical and mental health, and crime and substance use (Lattimore & Visher, 2009).

Dependent Measure

Given concern over the number of individuals who return to prison following release (Durose et al., 2014), the dependent variable in the current study is reincarceration indicating that the respondent experienced reentry “failure.” Descriptive statistics demonstrate that around 23% of the sample was reincarcerated during the 15-month sampling time frame (overall SD = .42). However, individuals could—and did—experience changes in incarceration status across the post-release waves. As a result, this measure is time variant and the within-person SD is .24. Descriptive statistics for reincarceration and all other variables included in the models can be found below in Table 1.

Table 1.

SVORI Sample Descriptive Statistics (n = 1,697).

Variable	M	SD	Range	SD Between	SD Within
Dependent Measure
Reincarceration	0.234	0.424	0, 1	0.375	0.246
Time Variant Measures
Parole Officer Support
Overall	20.814	4.062	7–28	3.795	1.912
Professional	9.260	1.793	3–12	1.657	0.920
Interpersonal	11.475	2.525	4–16	2.371	1.192
Family Support	9.739	1.980	3–12	1.762	1.079
Substance Use (logged)	0.695	0.554	0–2.485	0.503	0.284
Depression	8.038	3.889	5–25	3.290	2.293
Marital Status
Married	0.104	0.306	0, 1	0.260	0.161
Single	0.735	0.442	0, 1	0.416	0.153
Other	0.161	0.368	0, 1	0.341	0.148
Children	0.636	0.481	0, 1	0.463	0.149
Post-Release Treatment	0.300	0.458	0, 1	0.361	0.321
Failed Parole Conditions	0.290	0.540	0, 1	0.476	0.304
Post-Release Criminal Activity	0.339	0.776	0–5	0.672	0.455
Employment	0.776	0.417	0, 1	0.377	0.230
Time Invariant Measures
Age	29.090	7.306	18–69	—	—
Race
White	0.341	0.474	0, 1	—	—
Black	0.534	0.499	0, 1	—	—
Other	0.125	0.331	0, 1	—	—
Legal Cynicism	11.893	2.309	4–16	—	—
Length of Incarcerationa	918.294	932.349	43.8–9,486.35	—	—
Education	0.595	0.491	0, 1	—	—
Number of Convictions	5.870	7.529	1–50	—	—
Offense Type
Property	0.226	0.418	0, 1	—	—
Drug	0.193	0.395	0, 1	—	—
Violent	0.135	0.345	0, 1	—	—
Other	0.446	0.497	0, 1	—	—
SVORI Participant	0.509	0.499	0, 1	—	—
Reentry Preparation	0.484	0.499	0, 1	—	—

Note. SVORI = Serious and Violent Offender Reentry Initiative.

^aLength of incarceration in days.

Focal Independent Measure: Parole Officer Relationship

The focal independent measure in the current study is parole officer support. To capture parole officer support, we first identified seven questions from the SVORI data set that asked respondents about relationship dynamics with their parole officer. Specifically, respondents were asked about the extent to which their parole officer (a) provided them with correct information, (b) acted in a professional manner, (c) treated them with respect, (d) was helpful in their transition, (e) appeared trustworthy, (f) did not listen to them, and (g) was too busy to help them.¹ Possible responses ranged along a 4-point Likert-type scale (1 = strongly disagree, 2 = disagree, 3 = agree, 4 = strongly agree). We use these items in two specific ways. First, we combine all seven to create an overall measure of parole officer support. This measures has a mean of 20.81, overall SD of 4.06, and ranges from 7 (very low support) to 28 (very high support). The within-person SD of 1.91 captures the extent to which this measure varies across time.

Second, we examined the extent to which these items constituted multidimensional constructs. Results of confirmatory principal components factor analyses (Thompson, 2007) demonstrated two factors. The first factor is comprised of the items asking the respondent the extent to which the parole officer acts professionally, treats them with respect, and provides correct information (Eigenvalue = 1.89, all factor loadings exceed .7). We label this factor professional support. This measure has a mean of 9.26, overall SD of 1.79, and ranges from 3 (very low professional support) to 12 (very high professionals support). The second factor is comprised of items that asked if the parole officer was helpful with the transition, appeared trustworthy, listened to them, and made time for them (Eigenvalue = 2.16, all factor loadings exceed .7). We label this factor interpersonal support. This measure has an overall mean of 11.47, a SD of 2.52, and ranges from 4 (low interpersonal support) to 16 (high interpersonal support).

Time Variant Controls

Based on prior literature, we control for a wide range of measures. All time variant controls are drawn from all three post-release waves. First, we include a measure capturing post-release family support (Visher, 2004). To create this measure, we draw data from three items that asked if they had someone in their family to talk to, to turn to for suggestions, and/or who understands their problems. Response categories ranged on a 4-point Likert-type scale (1 = strongly disagree, 2 = disagree, 3 = agree, 4 = strongly agree). To create a measure capturing family support, we summed the items together (averaged α across all waves = .89). Overall, family support has a mean of 9.73, an SD of 1.98, and a range from 3 (low support) to 12 (high support). The within-individual standard deviation of 1.07 demonstrates that levels of family support vary across time.

We also include a measure capturing substance use as past research has shown that substance use during the reentry process can contribute to reincarceration (Phillips, 2010). The measure for substance use was created by drawing data from the post-release waves that asked respondents if they used (1 = yes, 0 = no) (a) alcohol, (b) tranquilizers, (c) stimulants, (d) steroids, (e) marijuana, (f) hallucinogens, (g) cocaine, (h) heroin, (i) amphetamines, (j) inhalants, (k) sedatives, (l) pain relievers without prescription, and/or (m) methadone. These measures were summed to create an index of substance use. The mean for substance use is 1.35 (overall SD = 1.46) and ranges from 0 (no substance used) to 13 (all substances use). As substance use varies across time, the within-person SD = .72. To correct for a significant positive skew in the distribution, we used the natural logarithm of the variable in the analyses.

We control for marital status as research has shown that individuals who are married tend to have better reentry outcomes compared with those who are not married (Bahr, Armstrong, Gibbs, Harris, & Fisher, 2005). At each wave, respondents were asked about their marital status. Descriptive statistics demonstrate that about 12.3% of the sample was married, 71.0% of the sample was single, and 16.7% had a marital status other than married or single. Married serves as the reference group in the analyses. In addition, we control for having children. At each wave, respondents were asked if they currently had any children (1 = yes). Descriptive statistics demonstrate that about 63.6% of the sample reported having children (within-individual SD = .14).

We also control for post-release substance abuse treatment. At each post-release wave, respondents were asked if they had received any drug or alcohol treatment since their prior interview (yes = 1). Overall, around 30% of the sample reported receiving treatment during at least one wave. The within-individual SD is .32, highlighting that individuals tend to move in and out of receiving treatment.

A key factor of the parole officer relationship is parole compliance (Werth, 2012). To account for this, we include a measure indicating that the respondent failed to comply with parole (1 = failed to comply, 0 = complied). Descriptive statistics demonstrate that around 29% of respondents reported failing to comply with their parole conditions during at least one post-release wave. The within-individual SD of .30 suggests parole compliance varies across time. In addition to parole compliance, we also include a measure capturing post-release criminal activity. To create this measure, we used five binary measures that asked respondents if they had committed the following: (a) violent crime, (b) weapon crime, (c) drug sales, (d) driving under the influence (DUI)/driving while intoxicated (DWI), and/or (e) property offense. We then summed these five measures to create an index that captured the respondent’s total criminal offenses at each wave. The mean for post-release crime is .33, an SD of .77 (within-individual SD = .45), and a range from 0 (no offending reported) to 5 (all type of offending reported).

The final time-variant measure is employment status as past research has shown that individuals who secure employment are less likely to be reincarcerated than those who are not employed (Visher, Debus-Sherrill, & Yahner, 2011). At each wave, respondents were asked if they currently held legal employment (1 = yes). Overall, 77.6% of the sample reported being employed. The within-individual SD of .230 suggests that individuals vary in their employment status over time.

Time-Invariant Controls

In addition to the time-variant measures, we also include a number of time-invariant measures collected from Wave 1. First, we include age in the models as research has shown that older and younger individuals differ in their reentry outcomes (Severson, Veeh, Bruns, & Lee, 2012). The average age at Wave 1 is 29.09 years, SD of 7.30, and a range from 18 to 69 years. We include a series of binary measures capturing race/ethnicity (Marbley & Ferguson, 2005). Overall, 34.1% of the sample identified as White, 53.4% of the sample as Black, and 12.5% as Other race. For each analysis, White serves as the reference group.

We also account for legal cynicism, assessed at Wave 1. To capture legal cynicism, we drew data from four questions derived from Sampson’s (Sampson & Bartusch, 1998 [Project on Human Development in Chicago Neighborhoods]) legal cynicism scale. These items asked the respondent along a 4-point Likert-type scale (1 = strongly agree, 2 = agree, 3 = disagree, and 4 = strongly disagree) how much they agreed with the following statements: (a) laws are made to be broken; (b) it’s okay to do anything you want as long as you don’t hurt anyone; (c) to make money, there are no right and wrong ways, only easy and hard ways; and (d) fighting with friends and family is nobody else’s business. The α for these items is .702 indicating acceptable levels of inter-item reliability. To construct a scale of legal cynicism, these items were summed, and this measure has an overall mean of 11.89, an SD of 2.31, and ranges from 4 (low legal cynicism) to 16 (high legal cynicism).

Past research has shown that individuals who spend more time incarcerated may differ in their experiences once released from incarceration compared with those who spend less amount of time incarcerated (Wolff, Shi, & Schumann, 2012). For this reason (see also Rydberg & Clark, 2016), we control for length of incarceration. The mean for length of incarceration was 918.29 days with a standard deviation of 932.34 (range = 43.8 to 9,486.35 days).

In addition to age, race/ethnicity, and length of incarceration, we include a measure for the total number of prior convictions the respondent reported (Visher & Courtney, 2007). Due to extreme outliers and less than 1% of the sample reporting more than 50 convictions, any response over 50 was recoded to represent the maximum value of previous convictions. The mean for this variable is 5.87 convictions with an SD of 7.52. We also control for respondent’s offense type for their previous conviction using four binary measures from Wave 1. The four types of offenses that were included in the analyses were property (22.6% of the sample); drug (19.3%), or violent (13.5%), in contrast to any other offense; or multiple offenses (44.6% of the sample).

We also control for levels of education as research shows education may influence important aspects of the reentry process (Travis, 2005). To control for education, we use a binary measure from Wave 1, which asked respondents if they completed high school or received a general education development (GED). Overall, 59.5% of the sample reported completing high school or their GED.

Although research on the efficacy of SVORI programming is mixed (see Visher, Lattimore, Barrick, & Tueller, 2017), we include a control for SVORI participation due to potential between SVORI participants and nonparticipant differences. Overall, 50.9% of the sample reported being a SVORI participant in the sample. In addition, we include a separate control that asked the respondent—at Wave 1 prior to release—if they had worked with anyone to plan for release. Overall, 49.4% of respondents reported working with someone to plan for release. The correlation between SVORI participants and working with someone to plan for release was minimal (r = .34).

Missing Data

Missing data have been well documented in SVORI (Lattimore, Steffey, & Visher, 2010). Of the 1,697 respondents interviewed at Wave 1, our analytical sample encompasses 894 individuals. While this attrition rate is considerable, prior research using SVORI has shown that missing data occur at random (Mowen et al., 2018; Stansfield, Mowen, O’Connor, & Boman, 2017), and individuals present at Wave 1 do not significantly differ from those at Wave 4 (Lattimore & Steffey, 2010). However, to confirm that missing data did not significantly bias the results of the forthcoming analysis, we performed multiple imputation using chained equation (White, Royston, & Wood, 2011) and estimated each of our models with the imputed dataset. Results were substantively the same as the models we present, suggesting that missing data do not significantly bias the substantive results of the forthcoming models.

Analytic Strategy

Because SVORI is a longitudinal panel data set, a model is required that accounts for the lack of independence across time. A multilevel, or hierarchical, model is commonly used in the case of panel data and nests time within the individual through the introduction of a random intercept (Rabe-Hesketh & Skrondal, 2012). The use of this model also allows us to examine the effects of time-invariant measures (e.g., race, length of incarceration) as well as time-variant measures (e.g., post-release substance use, parole officer social support) on reincarceration. As our outcome is binary, we use a generalized linear mixed-effects model in the forthcoming analyses (Rabe-Hesketh & Skrondal, 2012). First, we examine the extent to which the overall measure of parole officer support relates to reincarceration. Then, following the results of the factor analysis, we examine the extent to which each specific type (professional and instrumental) of parole officer support relates to reincarceration.

Results

Prior to presenting the results, we examined the correlations and variance inflation factors across all measures used in each analysis. No variance inflation factor exceeded 2 in any model suggesting that multicollinearity does not significantly bias results (O’Brien, 2007). Furthermore, one of the major assumptions of the mixed-effects model is that the between-person (time invariant) estimate is approximately equal to the within-person (time variant) measure. This assumption, often called the assumption of equality (Rabe-Hesketh & Skrondal, 2012) can be tested using a Hausman test. We invoked the Hausman test, which compares the between- and within-person effects for each of the forthcoming models. Results yielded nonsignificant results for each of the forthcoming models, demonstrating that the assumption of equality was met.

Results of the mixed-effects generalized linear model examining overall parole officer support are shown in Table 2. Results from this model demonstrate that overall parole officer support relates to significantly lower odds of reincarceration. Converting the logged-odds coefficient to an odds ratio through exponentiation can help provide more parsimonious interpretations (University of California, Los Angeles Statistical Consulting Group, 2016). Specifically, a 1-unit increase in parole officer support is associated with an 18.5% reduction in the logged odds of reincarceration. Results from the control measures largely support prior research as substance use, criminal offending, violating parole, and depression are significantly associated with increased odds of reincarceration. Similarly, Black respondents, relative to White respondents, report increased odds of being reincarcerated.

Table 2.

Mixed Effects Models Assessing Reincarceration (n = 894).

Variables	Model 1
	Coef.	SE
Time-Variant Measures
Parole Officer Support
Overall	−0.189	0.036***
Family Support	0.045	0.065
Substance Use (logged)	0.833	0.309**
Depression	0.241	0.041***
Marital Status
Single	0.047	0.462
Other	0.370	0.546
Children	−0.074	0.297
Post-Release Treatment	−0.242	0.297
Failed Parole Conditions	1.940	0.315***
Post-Release Crime	0.717	0.175***
Employment	0.333	0.326
Time-Invariant Measures
Age	−0.029	0.025
Race
Black	1.310	0.369***
Other	0.578	0.457
Length of Incarceration	0.001	0.001
Legal Cynicism	0.016	0.062
Education	0.229	0.308
Number of Convictions	0.009	0.020
Offense Type
Property	0.150	0.346
Drug	−0399	0.415
Violent	−0.128	0.432
SVORI Participant	0.067	0.300
Reentry Preparation	0.379	0.300
χ2	73.10***
Rho (ICC)	0.504

Note. Coef. = unstandardized regression coefficient; SVORI = Serious and Violent Offender Reentry Initiative; ICC = intraclass correlation coefficient.

^*p ≤ .05, where * refers to the significance of the coefficients.

^**p ≤ .01.

^***p ≤ .001.

In the next series of models, shown in Table 3, we use the result from the factor analysis described above to examine the respective effects of professional and interpersonal support on reincarceration. In Model 1, we explore the relationship between parole officer professional support and reincarceration. Results demonstrate that parole officer professional support is associated with decreased odds of reincarceration. Specifically, a 1-unit increase in parole officer professional support is associated with a 34.7% reduction in the logged odds of reincarceration. Results of the control measures are substantively the same as the prior model as substance use, depression, failure to comply with parole, and criminal offending are all associated with increased odds of reincarceration. Like the prior model, too, Black respondents—relative to White respondents—report significantly elevated odds of reincarceration.

Table 3.

Mixed Effects Models Assessing Reincarceration (n = 894).

	Model 1		Model 2		Model 3
Variables	Coef.	SE	Coef.	SE	Coef.	SE
Time-Variant Measures
Parole Officer Support
Professional	−0.409	0.080***	—	—	−0.280	0.107**
Interpersonal	—	—	−0.279	0.055***	−0.132	0.077
Family Support	0.039	0.065	0.040	0.065	0.044	0.066
Substance Use (logged)	0.843	0.317**	0.801	0.306**	0.841	0.310**
Depression	0.246	0.041***	0.243	0.040	0.241	0.041***
Marital Status
Single	0.085	0.467	0.004	0.456	0.062	0.464
Other	0.407	0.552	0.305	0.541	0.390	0.548
Children	−0.110	0.292	−0.051	0.287	−0.087	0.290
Post-Release Treatment	−0.233	0.297	−0.258	0.295	−0.237	0.297
Failed Parole Conditions	1.969	0.320***	1.944	0.312***	1.945	0.317***
Post-Release Crime	0.767	0.177***	0.698	0.174***	0.731	0.176***
Employment	0.342	0.327	0.311	0.324	0.339	0.327
Time-Invariant Measures
Age	−0.028	0.025	−0.031	0.025	−0.028	0.025
Race
Black	1.334	0.374***	1.290	0.365***	1.319	0.371***
Other	0.612	0.462	0.568	0.454	0.587	0.459
Legal Cynicism	0.011	0.063	0.016	0.617	0.015	0.062
Length of Incarceration	0.001	0.001	0.001	0.001	0.001	0.001
Education	0.174	0.310	0.235	0.307	0.207	0.310
Number of Convictions	0.008	0.019	0.011	0.019	0.009	0.019
Offense Type
Property	0.122	0.349	0.150	0.344	0.144	0.348
Drug	−0.420	0.419	−0.386	0.411	−0.405	0.417
Violent	−0.088	0.434	−0.161	0.430	−0.114	0.433
SVORI Participant	0.052	0.302	0.091	0.228	0.060	0.301
Reentry Preparation	0.346	0.302	0.384	0.298	0.371	0.300
χ2	71.04***		74.06***		72.49***
Rho (ICC)	0.513		0.513		0.506

Note. Coef. = unstandardized regression coefficient; SVORI = Serious and Violent Offender Reentry Initiative; ICC = intraclass correlation coefficient.

^*p ≤ .05, where * refers to the significance of the coefficients.

^**p ≤ .01.

^***p ≤ .001.

In Model 2, we explore the relationship between interpersonal parole officer support and reincarceration. Results of the mixed-effects model show that interpersonal support is significantly associated with lower odds of reincarceration. Specifically, a 1-unit increase in parole officer interpersonal support is associated with a 24.3% reduction in the logged odds of reincarceration. Concerning the control measures, findings from this model are substantively the same as the prior models.

Finally, in Model 3, we include both parole officer interpersonal support and parole officer professional support together to examine their respective effects on reincarceration. Interestingly, the results of this model demonstrate that professional parole officer support is significantly—and robustly—related to decreased odds of reincarceration. Yet, interpersonal parole officer support fails to reach significance once accounting for the effect of parole officer professional support. The results of the control variable are substantively the same as the prior models. Overall, results of these analyses demonstrate that broad parole officer support is significantly associated with decreased odds of reincarceration. Yet, and moving this body of literature forward, parole officer support is multidimensional and, as highlighted by the analysis above, certain dimensions of parole officer support (i.e., professional) appear to be significantly more impactful than other dimensions (i.e., interpersonal) for individuals undergoing the process of reentry from prison.

Discussion

The goal of this study was to examine the influence of parole officer support on reincarceration with attention given to specific forms of parole officer support. Using four waves of panel data from the SVORI, results of the analysis support a body of limited, but growing, research that highlights the importance of the social ties between returning individuals and their supervising officers. We now return to our hypothesis and prior literature to unpack these findings.

Our first hypothesis premised that overall parole officer support would relate to decreased odds of reincarceration. This hypothesis is supported as the results of the mixed-effects regression highlighted that general parole officer support was significantly—and robustly—related to decreased odds of reincarceration. The second hypothesis premised that each factor of parole officer support—professional and interpersonal—would relate to reincarceration. This hypothesis is partially supported. While analyses demonstrated that both professional and interpersonal support related to decreased odds of reincarceration on their own, once both were included in the model together, only parole officer professional support significantly related to decreased odds of reincarceration. At present, we offer two potential explanations for this set of findings.

First, the parole officer serves in a dual role as both an agent of the criminal justice system with the ability to surveil and punish, and also as a mechanism of support whereby they provide resources to the returning individual. It could be that professional support derived from the parole officer more closely captures the official function of the parole officer. That is, professional support—comprised of being respectful, providing the individual with correct information, and acting professionally—may be indicators of a parole officer who is providing the returning person with more formal social support in contrast with the interpersonal support, which captures dimensions more aligned with less formal types of emotional support (e.g., being trustworthy and helpful). Consequently, professional support may exert a more robust effect on recidivism than interpersonal support because it represents a more formal component of support. Thus, while interpersonal support does appear to make a difference, professional support may be more indicative of social support receipt in an official capacity.

A second explanation, largely complementary to the first, could be that individuals are able to derive interpersonal support from other sources—such as family and peers—from whom they would not typically be able to draw professional support. As a consequence, the effect of interpersonal support from the parole officer may not be as robustly related to recidivism as professional support because the returning person has ample sources to also draw interpersonal support from. Drawing interpersonal support from others may dampen the effect of parole officer interpersonal support on reincarceration. Yet, parole officer interpersonal social support may be particularly important for individuals who lack a broader social network, and/or for those who are unable to derive social support for other critical sources such as family and friends. Future research should consider, and investigate, this possibility. Overall, our findings mesh with research in other arenas including nursing and teaching (Hong et al., 2014; Kinman et al., 2011) demonstrating the importance of differing forms of social support.

Findings from this study carry implications for theory. Although research on the parole officer is expanding (e.g., Blasko et al., 2015; Chamberlain et al., 2018; Kennealy et al., 2012; Morash et al., 2015; Skeem et al., 2007; Vidal et al., 2015), findings from this study suggest that the parole officer, like family and peers (e.g., Berg & Huebner, 2011; Cochran, 2014; Mowen & Boman, 2019), has the clear potential to be an important agent of social support for returning persons. Placed within other theoretical traditions, the parole officer may function as an important bond (e.g., social control), coping mechanism against forms of strain (e.g., general strain theory), or be an important model for prosocial behavior (e.g., differential association). Future research should investigate the importance of parole officer support from additional theoretical perspectives. For example, from a procedural justice perspective (Sunshine & Tyler, 2003; Tyler, 2003), professional support may represent the procedurally just treatment from the parole officer to the individual on parole. There is potential that procedurally just treatment from a parole officer to an individual on parole may lead to more prosocial reentry outcomes, which future research should examine. Likewise, existing studies have shown that changes in family social support may function as an important turning point during the reentry process (e.g., Mowen & Visher, 2015). It is possible that changes in social support from the parole officer may operate similarly, and future research should examine this possibility.

Our findings both echo, and diverge, from prior work. For example, Chamberlain et al. (2018) used the same data to examine how negative and positive parole officer relationships related to reincarceration. In our study, we considered how the questions about the parole officer captured distinct theoretical constructs (e.g., professional support and interpersonal support) as opposed to negative or positive orientations. Thus, our findings support the conclusions of Chamberlain et al. (2018) as we find that, overall, the parole officer relationship is generally related to recidivism. However, we move beyond their findings by demonstrating that the primary motivator for reentry success is not simply having a positive or negative orientation toward the parole officer. Instead, our findings establish that the presence of professional support—irrespective of interpersonal support—promotes reentry success.

In addition to the theoretical implications, findings from this study also carry considerations for policy. Although our findings from Hypothesis 1 demonstrate that the overall level of parole officer support relates to prosocial reentry outcomes, our subsequent analysis suggests that professional support is largely the primary motivator for decreased odds of reincarceration. As a consequence, policy makers should consider placing an emphasis on building professional rapport between the parole officer and the individual under supervision. This policy recommendation mirrors the findings of Johnson (2015) who, in a qualitative study of 60 previously incarcerated individuals, found that “Positive rapport between parole officers and parolees is important in women parolees’ ability to access treatment, community resources, and to foster desires to fulfill their parole obligations” (p. 806). Yet, individuals with negative perceptions of the parole officer tended to highlight that they did not receive enough professional support (e.g., help finding a job or employment). Our quantitative analysis echoes these findings and suggests that policy makers should consider the essential role that parole officer professional support plays in the process of reentry.

In addition to the findings discussed above, there are additional findings worthy of consideration as they pertain to the parole officer support and reincarceration. Specifically, in our analysis, one of the most robust covariates of reincarceration was race, with Black respondents significantly more likely than White respondents to be reincarcerated. To examine whether parole officer support moderated this relationship, we performed a series of interactions between all types of parole officer support and race. None of these interactions reached significance, suggesting that parole officer social support does not moderate the relationship between race and reincarceration. On one hand, this speaks to the strength and importance of the parole officer relationship in that it appears to relate to lower reincarceration irrespective of race/ethnicity. On the other hand, it also demonstrates that parole officer social support does mitigate—or reduce—inequality in reincarceration. Future research should investigate the extent to which specific characteristics of the parole officer may moderate the relationship between race and reincarceration.

Outside of the contributions of this study, there are also important limitations and possibilities for future research. First, the SVORI sample is comprised of designated serious and violent individuals. Although the sample is similar in demographic composition to the broader incarcerated population, findings from this study may not be generalizable to other populations, including less serious offenders, juveniles, and females. Future work should investigate how specific types of parole officer support relate to reentry success in diverse samples. Second, differences in experiences prior to, and during, incarceration may result in individuals differing in their ability to succeed—or be prepared to succeed—on parole. For example, an individual who was previously on parole or had received some form of reentry programming while incarcerated may be more likely to succeed on parole compared with an individual without these experiences. In our analysis, we accounted for whether or not the individual worked with someone to prepare for release, but this limited measure fails to account for the wide range of experiences and preparation an individual may experience before release. Future research should consider triangulating differing levels of parole preparedness to more comprehensively examine its effect on the parole officer-returning person relationship and reentry outcomes. In addition, we are also limited by our inability to identify within the data if the respondent was assigned to a parole officer they had worked with on a previous parole experience. Prior experience with a parole officer may alter the transfer of social support. Future research should examine the role of parole officer familiarity and its influence on the transfer of social support to returning individuals.

A third limitation concerns the trend that far more individuals are currently on probation than parole (Kaeble, 2018), and our results may not speak to the broader experience of all individuals under the umbrella of community supervision. It is possible that characteristics of support that matter for individuals on parole are not the same characteristics that matter for individuals on probation. In addition, our results speak only to reincarceration as an outcome and not other important reentry outcomes such as employment and mental health. While future research should examine alternative outcomes, the focus on reincarceration is also a strength of this project given the relatively high rate at which individuals return to prison (e.g., Durose et al., 2014). In addition, our models account for measures of recidivism (substance use and criminal offending) as well as parole compliance, suggesting that parole officer professional support exerts a significant effect on reincarceration independent of parole violations and recidivism. Finally, SVORI only contains four waves of data spanning 15 months, and longer term studies on SVORI programming have revealed that SVORI programming had little-tono effect on recidivism (Visher et al., 2017). However, while the 15-month time frame is a limitation to the current analysis, given the research finding null results on the effectiveness of SVORI programming, our research suggests that other mechanisms outside of programming—such as professional support for the parole officer—provide scholars a fruitful avenue in which to explore “what works” in promoting prosocial reentry outcomes. Unfortunately, we are unable to determine the longer term implications of parole officer support on reentry success, however, but would encourage future research projects to exam this question.

In addition to limitations concerning these data, there are important theoretical limitations. There are likely additional domains of the parole officer relationship beyond the two factors examined in the present study. Martinez (2008), for example, highlights that family support can take on numerous forms beyond just the emotional or instrumental. It is likely that parole officer support can take a multitude of forms as well, which future research should seek to investigate. However, a multilevel analysis conducted by Orrick and colleagues (2011) found mixed support for the importance of social support on recidivism among returning persons. This finding suggests, among other things, that community-level factors outside of the scope of individual-level social support may be important factors in understanding reentry outcomes. Unfortunately, we are unable to account for neighborhood- or community-level conditions in the current study. Future research should examine the role of both micro- and macro-sources of social support on reentry success. Finally, there are a wide range of outcomes that likely define reentry “success” beyond reincarceration. Employment, desistance from substance use, reconnecting with family and friends, and education are just some additional measures of reentry success. Future research should examine how parole officer social support relates to a more diverse range of reentry outcomes.

Overall, findings from this study highlight the importance of parole officer social support during the reentry process. Social support generally, and professional support specifically, derived from the parole officer significantly contributed to lower odds of reincarceration. These findings also highlight the utility of understanding the parole officer as an important mechanism of social support. Yet, as we note above, high caseloads and limited direct contact between the supervising officer and the returning person may decrease perceptions of social support. Future research must continue to investigate these dynamics and researchers should recognize that the parole officer—much like family and peers—can provide important ties that bind through social support.

Author Biographies

Kyle J. Bares is a PhD student in the department of Sociology at Bowling Green State University. With concentrations in criminology and quantitative methods, Kyle’s research is primarly focused on understanding the influence of social support during the process of reentry from prison, applying life-course perspectives to crime, and understanding the correlates of juvenile delinquency. His recent research has appeared in Youth Violence and Juvenile Justice and Criminal Justice Studies.

Thomas J. Mowen is in the department of Sociology at Bowling Green State University. His research examines the effect of punishment on youth and family outcomes and the role of family within the process of reentry. His recent research has appeared in Criminology, Justice Quarterly, and Journal of Research in Crime and Delinquency.