The Effect of Individual Characteristics and Supervision Experiences on the Perceived Quality of the Supervision Relationship

Abstract

Research has shown that high-quality relationships between individuals on probation/parole and their supervising officers can reduce recidivism and increase compliance. Although this relationship clearly matters, little attention has been given to understanding the factors that influence this relationship. Drawing on research in psychology and counseling, this study explores how both individual characteristics and supervision experiences affect the perceived quality of the supervision relationship. Results from the Serious and Violent Offender Reentry Initiative (SVORI) reveal that both individual characteristics—such as mental health and family support—and supervision experiences—such as the use of sanctions and incentives—exert significant effects on the supervision relationship. Yet, the effects of supervision experiences were substantially more robust than the individual characteristics. Findings suggest community supervision agencies should prioritize positive supervision experiences to build positive relationships between the returning person and supervising officer.

Over the last two decades, the correctional field has experienced a fundamental transformation in which the get-tough approach of the last century has been gradually supplanted with a smart on crime mentality. This transformation has been especially evident in community corrections. Motivated by the realities and consequences of the deterrence- and incapacitation-based policies of the 1980s and 1990s, community supervision agencies have adopted a diverse set of policies and practices loosely organized under the mantras of what works or evidence-based practices. Risk/needs assessments, cognitive-behavioral programming, and the adoption of incentives and sanctions are just a few examples of evidenced-based policies aimed at improving success rates and promoting long-term behavioral change that have been implemented in jurisdictions across the country (Latessa et al., 2014).

The reorientation of values in community corrections has been accompanied by a pragmatic shift in the role of probation and parole officers in many locales. While community supervision has always involved a balance of treatment and surveillance (J. Miller, 2015), in this modern, evidence-based practices era, the therapeutic role of the community supervision officer has been redefined. Officers, who traditionally acted as service brokers by connecting individuals under their supervision to treatment resources in the community, are now called upon to take an active role in the behavioral change process (Bourgon et al., 2012; Taxman, 2008). Community supervision agencies, for example, routinely incorporate line-level officers into behavioral change initiatives such as motivational interviewing, contingency management, drug-related counseling, and cognitive-behavioral therapy.

As the therapeutic role of the supervision officer has evolved from service broker to change agent, there has been a heightened emphasis on the importance of the supervision relationship. A high-quality interpersonal relationship between the supervision officer and client, for example, is a key component of the well-established core correctional practices (CCP) literature first articulated by Andrews and Kiessling (1980). The responsivity principle within the Risk–Need–Responsivity (RNR) framework has also been utilized to theorize the importance of the supervision relationship. This principle holds that better results will be gained when treatment approaches and components are designed to suit the “learning style, motivation, abilities, and strengths” of the person on supervision (Latessa et al., 2014, p. 103). As such, correctional staff who are better able to engage individuals through interactional styles that are open, motivational, nonjudgmental, flexible, and respectful will achieve greater success in promoting long-term, prosocial change (Andrews, 2011; Andrews et al., 1990).

Evidence is accumulating to show that high-quality supervision relationships contribute to beneficial outcomes such as reduced violations, revocations, and reincarcerations (see next section), yet little attention has been given to understanding the factors that influence an individual’s perception of his or her relationship with the supervising officer. Instead, most discussion on this issue has focused on the role of the supervision officer’s demeanor and style of interaction. It should also be recognized that other factors, many of which are outside the officer’s immediate control, may also impact the perceived quality of the supervision relationship. Research in the psychology and counseling fields, for example, has consistently revealed that therapeutic relationships are influenced by the characteristics of the clients themselves, such as their optimism in achieving treatment goals and their level of interpersonal functioning (Horvath & Bedi, 2002; Horvath & Luborsky, 1993; Sass-Stanczak & Czabala, 2015). In addition, certain functions associated with the supervision of individuals on probation or parole might also affect the supervision relationship¹ such as the enforcement of punitive supervision conditions or the imposition of sanctions and incentives.

This study seeks to broaden our understanding of the factors that influence the supervision relationship. Using data from the Serious and Violent Offender Reentry Initiative (SVORI), the study begins with a preliminary inquiry into the importance of the quality of the supervision relationship on reentry outcomes, followed by a more in-depth examination of the factors that influence the quality of this relationship. More specifically, we draw on prior research and what is known about the dual nature of community supervision to examine the degree to which the perceived quality of the supervision relationship is influenced by individual characteristics and supervision experiences.

Prior Research on Supervision Relationship Quality and Supervision Outcomes

The value of service providers developing meaningful relationships with clients has been promoted in the psychology literature since Greenson (1967) coined the term “working alliance” (Horvath & Symonds, 1991). A working alliance emphasizes the necessity of caregivers and recipients establishing a collaborative approach to treatment engagement as a unique and indispensable determinant to therapeutic success. Meta-analytic reviews in the psychotherapy field demonstrate that a strong working alliance is a robust predictor of positive therapeutic outcomes above and beyond the impact of specific treatment modalities (Horvath & Symonds, 1991). A critical review of drug treatment studies similarly concluded that positive therapist–client relationships were predictive of greater program engagement and retention (Meier et al., 2005).

By the nature of their jobs, community supervision officers are challenged to form affective bonds, trust, and the collaborative spirit essential to creating a healthy working alliance. Unlike therapists and counselors operating in pure treatment roles, probation and parole officers are positioned in a dual-role occupation where they are tasked with facilitating and sometimes directly delivering services while simultaneously surveilling clients and enforcing criminal justice sanctions (Taxman & Ainsworth, 2009). As a result, individuals under community supervision may be more suspicious and offer greater resistance when officers make efforts to connect and build rapport (Kennealy et al., 2012; Trotter, 1999).

Despite the potential obstacles, existing studies tend to confirm the importance of high-quality supervision relationships for generating successful outcomes during the community supervision period. Skeem et al. (2007) examined probationers with mental disorders in both a southern and western U.S. city. They found that probationers who characterized their supervising officers as less demanding and more caring, fair, and accommodating were less likely to receive probation violations and revocation; those who expressed greater trust in officers were less apt to violate supervision conditions. In a study of parolees with no mental health problems residing in the western United States, Kennealy and colleagues (2012) reported that more favorable scores on a relationship measure integrating dimensions of officer caring/fairness, trust, and toughness predicted rate of rearrest, with the result holding even when personality and risk level were considered. Blasko et al. (2015) found that stronger supervision relationships were linked with less drug use and fewer violations for moderate-to-high risk parolees participating in a multisite drug treatment study. Using data from the SVORI project, Chamberlain et al. (2018) observed that parolees who described their parole officers as more supportive, such as more helpful and trustworthy, were less likely to be reincarcerated. Finally, a Michigan study of female probationers and parolees reported that more positive supervision relationships reduced the anxiety and psychological reactance experienced by the study sample, which, in turn, were associated with fewer arrests and convictions during supervision (Morash et al., 2016).

Factors That Influence the Supervision Relationship

To date, most attention on the supervision relationship has centered on the contributions of the supervising officer, such as her demeanor or communication style (Dowden & Andrews, 2004). In some respects, this narrow focus on officer contributions seems justified and consistent with prior research. Substantial research in the psychotherapy and counseling fields reveals that patient–client relationships are strongly influenced by certain qualities of the therapist (for reviews of this research, see Ackerman & Hilsenroth, 2003; Horvath & Bedi, 2002; Horvath & Luborsky, 1993; Meier et al., 2005; Sass-Stanczak & Czabala, 2015). Ackerman and Hilsenroth’s (2003) often cited review of research on therapist contributions to the therapeutic alliance, for instance, identified a number of therapist attributes and techniques that positively influence the working alliance, such as being supportive, honest, understanding, flexible, friendly, confident, and trustworthy. These findings parallel a recent study by Epperson et al. (2017) which used qualitative methods to better understand how adult probationers with serious mental illness perceived their relationships with supervision officers. Individuals in the study identified qualities such as kindness and empathy as important in developing positive relationships. In addition, behaviors such as being treated fairly and offering help and support were also recognized by probationers as critical for the development of good relationships. By contrast, probationers had less favorable views of supervision officers who were authoritarian and inflexible in their interactions.

Beyond the established importance of the officer’s contribution to the supervision relationship, the quality of this relationship may be influenced by other factors, many of which may be outside the supervising officer’s immediate control. The characteristics of individuals on supervision may be important to consider. Research in the counseling and clinical psychology fields reveals that the quality of the working alliance is influenced by various patient characteristics (Horvath & Luborsky, 1991; Horvath & Bedi, 2002; Sass-Stanczak & Czabala, 2015). As noted by Ross et al. (2008), “a client is not a blank slate or passive receiver of the therapeutic process” (p. 467).

There is some research to suggest that the quality of the therapeutic alliance may vary based on certain demographic characteristics of the client such as age, gender, race, and educational attainment (Barrowclough et al., 2010; Connors et al., 2000; Hersoug et al., 2009; Taft et al., 2004; Urbanoski et al., 2012). Among these variables, age has been one of the most consistent predictors with older clients generally reporting higher quality relationships with their counselor or therapist (Barrowclough et al., 2010; Connors et al., 2000; Sass-Stanczak & Czabala, 2015; Taft et al., 2004; Urbanoski et al., 2012). Only one study was located that looked specifically at the association between demographic characteristics and the quality of the supervision relationship. Springer et al. (2009) explored the degree to which characteristics such as age, race, education level, and employment status influenced individual perceptions of a supervising officer among a sample of probationers in Florida. Their findings revealed that only race/ethnicity was significantly related to the quality of the supervision relationship.

Moving beyond demographic characteristics, research has found that the quality of the therapeutic alliance is affected more consistently and profoundly by a variety of other patient characteristics and attributes, including interpersonal functioning, mental health, level of social support, and attitudes toward treatment and outcomes. Studies have found, for example, that individuals who struggle in building healthy interpersonal relationships, especially those with hostile and dominant interpersonal functioning problems, are less likely to form positive therapeutic connections with counselors and other treatment providers (Connolly Gibbons et al., 2003; Constantino et al., 2005; Kokotivic & Tracey, 1990). In addition, mental health functioning, to include the presence and severity of depressive symptoms, has been shown to be associated with lower quality patient–therapist relationships (Barrowclough et al., 2010; Sass-Stanczak & Czabala, 2015).

Research has also found a positive association between perceptions of the therapeutic alliance and patient levels of social support from family and other sources (Connors et al., 2000; Kokotivic & Tracey, 1990). Finally, attitudinal factors, to include the individual’s motivation and readiness to achieve treatment goals, have been found to predict the quality of the patient–therapist alliance. More specifically, individuals with higher levels of optimism and readiness to change are more likely to form positive patient–therapist bonds (Connors et al., 2000; Fitzpatrick & Irannejad, 2008; Mander et al., 2013; Taft et al., 2004). It should be noted, however, that the research described above has focused exclusively on the patent–therapist relationship. The degree to which these same factors might influence the supervision relationship remains unclear.

In addition to patient characteristics, another largely unexamined dimension of community supervision that has the potential to shape the perceived quality of the supervision relationship is the dual role aspect of the job. As Skeem and colleagues (2007) observed, unlike the traditional patient–therapist relationship, probation and parole officers are expected to simultaneously be both social workers and law enforcers. This dual role does more than just create an authoritarian dimension to the relationship that is absent from the traditional treatment provider–patient relationship; it also imposes a unique set of interactions and requirements that have the potential to influence the supervision relationship in meaningful ways.

One aspect of community supervision that differs from the traditional patient– therapist interactions is the modes and frequency of interactions between individuals and their supervising officers. Probationers and parolees, for example, are often required to report to their supervising officer multiple times per month or week depending on their level of supervision. In addition to office visits, officers frequently engage in other types of contacts with their caseloads, including home visits, phone calls, and collateral contacts with third-party sources such as employers (Petersilia, 1997). There are a number of potential ways in which the modes and frequency of supervision might influence the supervision relationship. More frequent and longer lasting meetings, for example, may provide greater opportunity for a positive supervision relationship to form. On the contrary, certain types of contacts, such as office visits, may cause disruption in work schedules and can present difficulties related to travel, especially if the individual has to rely on public transportation or family and friends for rides to the office. Thus, the burden of frequent office visits might dampen probationer or parolee attitudes toward supervision officers. This is consistent with Springer et al. (2009) who found that probation meetings that caused work disruptions and longer travel time and distance to the probation office resulted in less favorable ratings of the supervision relationship.

A second aspect of probation supervision that sets the supervision relationship apart from the patient–therapist relationship is the presence and enforcement of supervision conditions. Individuals under probation and parole often have a diverse range of conditions that they have to follow to remain in the community (Petersilia, 1997). Even though these conditions are normally determined by the judge or parole granting authority, it is the role of the supervision officer to communicate and enforce these conditions in their daily interactions with individuals on their caseloads. As such, a reasonable expectation is that when supervision officers are required to enforce certain conditions, especially those that are inherently punitive or place a substantial burden on their time or other resources, the quality of the supervision relationship might suffer.

Finally, the nature of interactions between the individual and his or her supervising officer brought on by the dual role nature of community supervision has the capacity to shape the supervision relationship. In their day-to-day dealings with individuals they supervise, officers are required at times to be social workers, talking through problems and offering encouragement, and at others to be rule enforcers, expressing disapproval and sanctioning noncompliance (Wodahl & Garland, 2018). The officer’s engagement in these different types of interactions likely influences the supervision relationship. More specifically, the expectation is that interactions emphasizing the enforcement role, such as discussions related to the collection of court-ordered fees and costs, reprimanding individuals for their transgressions, and issuing sanctions for noncompliance will negatively influence the supervision relationship. At the same time, interactions that emphasize the social work role, such as talking with individuals about potential problem areas (e.g., substance use, housing, family) and praising or rewarding prosocial behavior, will likely strengthen the individual’s view of his or her relationship with the officer.

Method

Data

Data for this project come from the SVORI (for an overview, see Lattimore & Visher, 2009). In light of increasing incarceration rates in the 1990s and early 2000s, the U.S. Departments of Justice, Labor, Education, Housing, and Urban Development provided funding to support programs to facilitate the reentry process for returning individuals. As a result, SVORI, a federally funded initiative, examined whether enhanced reentry programming—such as participation in anger management classes, substance abuse treatment, and reentry planning—resulted in prosocial reentry outcomes in housing, education, criminal justice, health, and employment (Lattimore & Steffey, 2009). Data were collected between 2005 and 2007 from individuals who were all incarcerated at the time of initial data collection. Although SVORI contained subsamples of youth and females, given the small sizes of those subsamples, we use the male sample which encompasses a total of 1,697 individuals across 12 different sites with approximately half randomly assigned to the SVORI program.

As longitudinal panel data, SVORI contains four distinct waves of data. Wave 1 data were collected while the respondent was still incarcerated, approximately 30 days prior to the scheduled release date. Wave 2 data were collected about 3 months post-release, Wave 3 data about 9 months post-release, and Wave 4 data collected 15 months post-release. A total of 79.3% of all respondents participated in Wave 1 (pre-release) and at least one other post-release wave. Specifically, 58.0% participated at Wave 2, 61.0% of respondents participated at Wave 3, and a total of 65.6% at Wave 4 (Lattimore & Steffey, 2009). At each wave, respondents were asked questions across a number of domains including substance use and criminal offending behaviors, family dynamics, housing/educational outcomes, mental health, and programming in addition to others (see Lattimore & Steffey, 2009). In the post-release waves, respondents were asked about a variety of experiences including questions about their relationship with their supervision officer, questions about their parole experiences, family relationships, and employment in addition to other measures. For this project, we draw data from all four waves. Descriptive statistics for all measures are shown in Table 1.

Table 1.

Descriptive Statistics for the SVORI Sample (n = 778).

Variable	M	SD	Range	SD between	SD within
Dependent measure
Relationship with parole officer	20.860	4.225	7–28	3.949	1.961
Individual characteristic variables
Race
Black	0.508	0.500	0, 1	—	—
Other	0.121	0.326	0, 1	—	—
Age	29.450	7.047	18–69	—	—
Married	0.111	0.314	0, 1	—	—
Less than high school education	0.320	0.467	0, 1	—	—
Employment	0.703	0.457	0, 1	0.424	0.223
Interpersonal violence	4.457	3.673	1–24	2.967	2.375
Depressive symptoms	7.334	3.556	5–25	3.323	1.578
Family support	13.788	2.135	4–16	1.975	0.930
Readiness for change	10.363	2.366	0–18	2.097	1.216
Prior arrests	14.515	20.626	1–300	—	—
Primary conviction
Sex offense	0.060	0.237	0, 1	—	—
Violent offense	0.273	0.445	0, 1	—	—
Length of incarceration	918.297	932.354	44–9,486	—	—
Parole revocation	0.241	0.427	0, 1	—	—
SVORI participant	0.551	0.497	0, 1	—	—
Supervision experiences variables
OV frequency	2.446	0.976	1–6	0.892	0.479
PO meeting length	1.979	0.524	1–4	0.476	0.271
HV frequency	0.871	0.995	0–6	0.898	0.488
PO phone calls	1.265	1.282	0–6	1.155	0.666
PO contact with employer	0.322	0.467	0, 1	0.423	0.254
PO assistance topics	2.405	1.315	0–4	1.244	0.550
Assistance topic need	5.641	1.571	3–9	—	—
PO health topics	0.374	0.685	0–2	0.635	0.317
Health topic need	3.144	1.083	2–6	—	—
PO payment topic	0.722	0.448	0, 1	0.401	0.232
Drug testing required	0.824	0.380	0, 1	0.361	0.168
Treatment programs required	0.316	0.465	0, 1	0.425	0.230
Punitive requirements	0.099	0.298	0, 1	0.279	0.153
PO sanction	0.220	0.414	0, 1	0.358	0.234
PO incentive	0.333	0.472	0, 1	0.417	0.262
PO praise	0.462	0.498	0, 1	0.452	0.253
PO reprimand	0.264	0.441	0, 1	0.383	0.241

Note. SVORI = Serious and Violent Offender Reentry Initiative; SD = standard deviation; PO = parole officer; HV = home visit, OV = office visit.

Dependent Measure

The dependent measure in this study is a scale variable aimed at capturing the perceived quality of the respondent’s working relationship with his parole officer. To create this measure, we draw data from seven questions asking respondents about their attitude toward the supervision officer measured along a 4-point Likert-type scale (strongly agree, agree, disagree, strongly disagree) and collected at each post-release wave. The questions asked whether the officer has been helpful with the transition back to the community, seems trustworthy, gives you correct information, acts too busy to help you, treats you with respect, acts in a professional way, and doesn’t listen to you. All items were coded such that higher values indicated more positive orientations toward the parole officer. The averaged Cronbach’s alpha (Cronbach, 1951) across each wave was .897, indicating high inter-item reliability. Items were summed to create a scale capturing the respondent’s Relationship with his Parole Officer (M = 20.860, SD = 4.225) with a range 7 (a very negative relationship) to 28 (a very positive relationship, within-individual SD = 1.957).

Independent Variables: Individual Characteristic Variables

The first set of independent measures used in the analysis encompasses individual characteristics aimed at exploring the influence of demographics, interpersonal functioning, mental health, family support, and motivation to change on the perceived quality of the supervision relationship. Demographic variables in our analyses include Race, measured as Black, White, and other race; Age, which is a continuous variable capturing the respondent’s age at Wave 1; Married, which is a dichotomous measure capturing whether the respondent was married or in a stable partnership at Wave 1; Less than a High School Education, which captures whether the respondent has less than a high school education (coded as 1) or a high school education (including a GED) or higher (coded as 0) at Wave 1; and Employment, which is a time variant measure capturing whether an individual was employed (coded as 1) or unemployed (coded as 0) at each wave of data collections.

To explore the relationship between interpersonal functioning and quality of the supervision relationship, we include the variable Interpersonal Violence, which is a time variant scale that measures the frequency in which an individual had threatened to hit, throw, push, slap, or use a weapon on a family member (0 = never, 1 = once, 2 = a few times, 3 = once a month, 4 = a couple of times a month, 5 = once a week, and 6 = several times a week). Items were summed together to create an Interpersonal Violence scale.

To examine the potential influence of mental health functioning on the perceived quality of the supervision relationship, we include a Depressive Symptoms measure. At each wave, respondents were asked along a 5-point scale (1 = not at all, 2 = a little bit, 3 = moderately, 4 = quite a bit, 5 = extremely) whether they felt lonely, blue, hopeless, worthless, or if they had no interest in anything. This summed scale of Depressive Symptoms has a mean of 7.334, SD of 3.556, and ranges from 5 (no depression) to 25 (high levels of depression, alpha = .845).

To capture Family Support, we draw from four items measured along a 4-point scale (strongly agree, agree, disagree, strongly disagree) asking respondents how much they have someone in their family who loves them, someone in their family to talk to, whether they feel close to their family, and if they want family involved in their life. This measure has an overall mean of 13.79 (alpha = .823), SD of 2.14, and ranges from 4 (low support) to 16 (high support).

To explore the relationship between motivation to change and the supervision relationship, we include a Readiness for Change scale created by SVORI researchers. This scale captures the extent to which the individual is motivated and optimistic about desistance and comprises six items asking the respondent if: (1) you are tired of problems caused by crimes you committed; (2) you want to get your life straightened out; (3) you think you will need help staying straight; (4) you will give up friends that get you in trouble; (5) it’s urgent you find help not to commit crimes after release; and (6) you think you will be able to stop committing crimes when released. Items were coded such that higher values represent greater levels of readiness for change (0 = strongly disagree, 1 = disagree, 2 = agree, 3 = strongly agree) and summed. This measure has an overall mean of 10.363, SD of 2.366, and ranges from 0 (low levels of readiness for change) to 18 (high levels of readiness).

Finally, consistent with prior research utilizing the SVORI data, we include a series of variables related to the individual’s criminal and correctional history (see, for example, Alward et al., 2020; Mowen et al., 2020; Stansfield et al., 2020). Given that previous research has found criminal and correctional history variables to be predictive of reentry outcomes (Makarios et al., 2010; Spivak & Damphousse, 2006), we include these measures to explore their potential influence on the supervision relationship. These variables include Prior Arrests (a continuous variable which captures the total number of lifetime arrests), Sex Offense and Violent Offense (dichotomous measures indicating whether the individual’s current offense is a sex or violent crime), Parole Revocation (dichotomous variable measuring whether the individual’s most recent incarceration was the result of a parole revocation), Length of Incarceration (number of days spent in prison for current offense), and SVORI Participant (binary measure indicating whether or not the respondent was identified as a SVORI participant).

Independent Variables: Supervision Experience Variables

The first set of supervision experience variables explores how the supervision relationship may be influenced by the frequency and mode of supervision contacts. We first include a measure of Office Visit Frequency which captures how frequently the individual is required to report in person to the supervision office along a 6-point scale (0 = not at all, 1 = once or twice, 2 = about once a month, 3 = two or three times a month, 4 = once a week, 5 = several times a week, and 6 = every day or almost every day, M = 2.446, SD = 0.976). Second, we include a measure of Parole Officer (PO) Meeting Length capturing the total time the respondent spent with the parole officer during their meetings (0 = none/do not meet, 1 = less than 5 min, 2 = 5–30 min, 3 = 31 min to an hour, 4 = more than an hour, M = 1.979, SD = 0.524). Third, we include Home Visit (HV) Frequency captured along a 6-point scale (0 = not at all, 1 = once or twice, 2 = about once a month, 3 = two or three times a month, 4 = once a week, 5 = several times a week, and 6 = every day or almost every day, M = 0.871, SD = 0.995). We also include a measure of PO Phone Calls capturing how often the respondent reported speaking with a supervision officer on the phone (0 = not at all, 1 = once or twice, 2 = about once a month, 3 = two or three times a month, 4 = once a week, 5 = several times a week, and 6 = every day or almost every day, M = 1.265, SD = 1.282). Finally, we include a measure encompassing PO Contact with Employer which indicates whether the individual was aware of any contact between the supervising officer and his employer (1 = yes, 0 = no).

The second category of supervision experience variables relates to the individual’s specific conditions of supervision. First, we include a binary measure (1 = yes, 0 = no) indicating whether or not Drug Testing was a requirement of supervision. Second, we include a measure that captures whether attending Treatment Programs was a condition of parole (1 = yes, 0 = no). Finally, we draw from questions asking whether the use of electronic monitoring, house arrest, or the requirement of community service was a condition of supervision. Respondents answering yes to any of these questions were coded “1” to indicate having a Punitive Requirement in contrast to those who answered no who were coded as “0.” All of these measures are time variant.

The next category of supervision experience variables focuses on the content of the interactions between the individual and supervising officer. Respondents were asked if they discussed employment, housing, substance use, substance use treatment, mental health, physical health, payment of legal fees, payment of child support, or issues with the family (1 = yes, 0 = no). Results of an exploratory analysis (Gorsuch, 1988) demonstrated two factors encompassing topics related to assistance and topics related to health. The first summative scale capturing PO Assistance Topics includes a discussion of substance use, substance treatment, employment, and housing (M = 2.405, SD = 1.315). The second factor is PO Health Topics which comprises items related to mental and physical health (M = 0.374, SD = 0.685). Given that the receipt of these items could be due to individual needs, we also include Assistant Topic Need which captures how much respondents believed they needed help with finding employment, housing, and substance use treatment (1 = not at all, 2 = a little, and 3 = a great deal). Items were summed (M = 5.641). We also include Health Topics Needs which, measured along the same scale, asked the respondent if he needed mental health treatment or medical treatment (M = 3.144). We also include a stand-alone measure capturing PO Payment Topic which asked respondents if they discussed their progress in paying fees when they meet with their parole officer (1 = yes, 0 = no).

The final set of supervision experience variables looks at officers’ use of sanctions and incentives. To capture sanctions, we include a measure examining whether the respondent had received enhanced drug testing, jail time, required substance use meetings, or increased Alcoholics Anonymous meetings. Due to the low frequency of any specific type of sanction, we combined all yes responses to the individual items into a binary measure capturing PO sanction (1 = any sanction received). To measure supervision officer use of incentives, we draw data from the respondent asking if he received decreased mandatory meetings with a PO, decreased drug testing, decreased drug treatment, or decreased Alcoholics Anonymous meetings. To measure PO Incentive, respondents answering yes to any of these questions were coded 1 in contrast to those who responded no (coded “0”). We also included a binary measure indicating whether individuals had received praise (PO Praise) from their parole officer (1 = yes), and a separate measure capturing whether a verbal reprimand was received (PO Reprimand) from their parole office (1 = yes). All four measures are time variant.

Sample and Missing Data

Missing data within the SVORI sample is well documented (Lattimore & Steffey, 2009). Of the 1,697 men in the SVORI sample, 817 were identified as being placed on parole following release. Missing data on some of the control measures resulted in a total sample of 778 individuals. Prior methodological reports regarding SVORI as well as a variety of research using this sample have demonstrated that individuals present at Wave 1 do not significantly vary from individuals at Wave 4 across a variety of measures (e.g., Lattimore & Steffey, 2009). Following Mowen and Culhane’s (2017) suggestion regarding the SVORI sample, we used multiple imputations to examine robustness of results (see also Royston & White, 2011, on imputation). Results (not shown) were substantively similar to the results we report below. Given the replication using multiple imputation, and prior reports on patterns of missing data, we cautiously conclude that sample attrition does not significantly bias the results we present below.

Analytic Strategy

As the SVORI sample comprises four waves of panel data whereby data are collected from the same individual over time, a method must be used that accounts for this lack of independence across time. As we are interested in both between- and within-person estimates, we use a mixed-effects regression model which introduces a random intercept to account for a lack of independence (Rabe-Hesketh & Skrondal, 2012). Although mixed-models have become the most popular form of panel data analysis in the social sciences (Allison, 2015), there are limitations. Perhaps the most important limitation is the treatment of the between- and within-individual estimators. Mixed-effects models assume the between-individual effect exerts approximately the same magnitude on the outcome as the within-individual effect. Referred to as the assumption of equality (Rabe-Hesketh & Skrondal, 2012), this assumption can be difficult to meet. Consequently, prior to presenting the results, we invoke the Hausman test which compares a fixed-effects model to the random-effects model to examine how between-individual and within-individual estimators differ, and whether these differences bias the estimates.

To analyze the data, we examine how individual characteristics and supervision experiences relate to parole officer relationship quality. Specifically, we present three models that examine how: (a) only individual characteristics relate to perceived relationship quality, (b) only supervision experiences relate to relationship quality, and (c) how both individual characteristics and supervision experiences relate to perceived relationship quality in a full model.

Results

In Model 1, we present the results of a mixed-effects model exploring how individual characteristics relate to the parole officer relationship. Prior to examining the specifics, we note that the Hausman test comparing fixed and random effects was significant in this model (p =.03). Results tended to demonstrate stronger between-person effects than within-person effects of the time variant estimators on relationship quality. For example, the effect of employment tended to be stronger between-people (comparing those employed with those unemployed) than within-people (the effect of an individual losing or gaining employment across time). However, differences in these estimates tended to be small and separating the between-person effect from the within-person effect and including both in the model resulted in similar substantive conclusions. As a result, we present the results of the mixed-effects model below.

The significant chi-square value indicates appropriate model fit to the data and the intra-class correlation show that about 46% of the variability in the parole officer relationship is within-persons across time. Substantive results demonstrate that age and family support are both significantly related to better perceived relationship quality with the PO whereby older respondents and those with greater levels of family support report higher relationship quality. On the contrary, individuals with higher levels of interpersonal violence and depressive symptoms report lower levels of relationship quality than their PO.

In Model 2, we examine how supervision experiences relate to parole officer relationship quality. Like the prior model, the Hausman test was significant (p < .05); however, the differences in the between- and within-person effects did not appear to bias the substantive findings. Overall, the fit statistic indicates strong fit to the data. Substantive results show that individuals who report more phone calls with their supervising officer, those whose supervision officer contacts their employer more frequently, those who report having more conversations with their parole officer regarding assistance, and individuals who receive praise from their parole officer report higher levels of relationship quality. On the contrary, sanctions and reprimands both relate to lower relationship quality between the returning individuals and their supervising agent.

Finally, in Model 3, we examine the relationships among both individual characteristics and supervision experiences on perceived relationship quality. Once all measures are included in the model, age no longer significantly relates to relationship quality with the PO; however, results are otherwise the same as the prior models. In terms of individual characteristics, family support is significantly related to higher levels of relationship quality while interpersonal violence and depressive symptoms relate to lower levels of relationship quality. The measures encompassing supervision experiences once again indicate that phone calls, contact with the employer, coverage of assistance topics, praise, sanctions, and reprimands are significantly associated with relationship quality.

To gain a more comprehensive understanding of the robustness through which these factors are associated with relationship quality, we standardized the coefficients from Model 3, which we present in the final column of Table 2. Interestingly, what these standardized coefficients reveal is that the supervision experience measures tend to be more robust in their effects on perceived relationship quality. Specifically, phone calls (.430), coverage of assistance topics (.388), sanctions (−.380), praise (1.101), and reprimands (−.405) are all more robust in their effect on relationship quality than any of the individual characteristic measures. Only contact with employer (.309) is weaker in its substantive effect than the significant individual characteristics (family support, depressive symptoms, and interpersonal violence). Thus, the results demonstrate that, generally speaking, supervision experiences appear to be far more predictive of the parole officer relationship than characteristics of the individual under supervision.

Table 2.

Mixed-Effects Regression Models Assessing Parole Officer Relationship Quality (n = 778).

	Model 1		Model 2		Model 3
Variable	B	SE	B	SE	B	SE	Z score
Individual characteristic variables
Race
Black	−0.282	0.315	—	—	−0.023	0.288	−0.012
Other	−0.763	0.455	—	—	−0.547	0.410	−0.179
Age	0.046	0.021*	—	—	0.024	0.019	0.165
Married	−0.447	0.444	—	—	−0.336	0.405	−0.106
Less than high school education	−0.591	0.305	—	—	−0.411	0.277	−0.192
Employment	0.108	0.254	—	—	0.147	0.239	−0.067
Interpersonal violence	−0.099	0.030***	—	—	−0.069	0.029*	−0.216
Depressive symptoms	−0.109	0.036**	—	—	−0.112	0.035**	−0.397
Family support	0.226	0.057***	—	—	0.174	0.053***	0.371
Readiness for change	0.597	0.545	—	—	1.062	0.518	0.227
Prior arrests	−0.036	0.177	—	—	−0.018	0.162	−0.015
Primary convictions
Sex offense	0.559	0.611	—	—	0.751	0.553	0.178
Violent offense	−0.603	0.318	—	—	−0.489	0.288	−0.218
Length of incarceration	−0.124	0.173	—	—	−0.331	0.156	−0.283
Parole revocation	−0.366	0.320	—	—	−0.179	0.291	−0.077
SVORI participant	−0.054	0.275	—	—	−0.283	0.248	−0.141
Supervision experience variables
OV frequency	—	—	−0.018	0.116	−0.026	0.116	−0.027
PO meeting length	—	—	0.326	0.204	0.361	0.203	0.190
HV frequency	—	—	−0.082	0.110	−0.100	0.109	−0.100
PO phone calls	—	—	0.327	0.088***	0.335	0.087***	0.430
PO contact with employer	—	—	0.717	0.228**	0.662	0.227**	0.309
PO assistance topics	—	—	0.315	0.094***	0.295	0.092***	0.388
Assistance topics need	—	—	−0.083	0.081	−0.093	0.081	−0.147
PO health topics	—	—	0.261	0.171	0.282	0.169	0.193
Health topics need	—	—	0.135	0.120	0.212	0.124	0.230
PO payment topic	—	—	−0.106	0.240	−0.090	0.240	−0.040
Drug testing required	—	—	−0.074	0.299	−0.079	0.298	−0.030
Treatment programs	—	—	0.118	0.244	0.181	0.243	0.084
Punitive requirements	—	—	0.052	0.352	0.003	0.351	0.001
PO sanction	—	—	−1.126	0.275***	−0.917	0.281***	−0.380
PO incentive	—	—	0.295	0.228	0.378	0.228	0.179
PO praise	—	—	2.301	0 224***	2.208	0 222***	1.101
PO reprimand	—	—	−1.067	0.261***	−0.919	0.260***	−0.405
Intercept	17.731	1.824***	18.299	0.767***	16.351	1.766
Σ2	86.18***		275.22***		356.23***
Intra-class correlation	0.458		0.425		0.405
r2	.076		.195		.245

Note. SVORI = Serious and Violent Offender Reentry Initiative; PO = parole officer; HV = home visit.

^*p ≤ .05.

^**p ≤ .01.

^***p ≤ .001.

Discussion and Implications

While prior research has established that that the relationship between the supervising officer and returning person is an important factor in reentry outcomes, few—if any—studies have examined how a variety of individual characteristics and supervision experiences relates to this relationship. Consequently, our goal was to explore this relationship as an outcome so that supervision agencies can develop and implement policies and practices that promote the development of high-quality relationships between supervision officers and the individuals they supervise.

Focusing first on individual characteristics, our findings revealed that individuals with greater levels of family support reported more positive relationships with their supervising officers than those with lower levels of family support. Although future research should more closely examine why family support relates to the parole officer relationship, it could be that social support in one context—the family—contributes to, or helps to create, social support in other contexts such as with the parole officer. In any case, these findings provide further support for the need for correctional agencies to promote policies and practices that preserve and foster family relationships such as prison visitation and family-based reentry programs (J. M. Miller et al., 2015).

Results also showed that individuals with greater levels of depressive symptoms were less likely to report positive relationships with their supervising officer. This finding, which is consistent with research in the psychology and counseling fields (e.g., Barrowclough et al., 2010; Sass-Stanczak & Czabala, 2015), highlights a potentially important way in which mental illness might indirectly influence supervision and reentry outcomes. It has long been recognized that mentally ill individuals fail community supervision at disproportionately high rates; yet, there is little empirical support to contend that mental illness itself has a direct criminogenic effect (Skeem et al., 2011). In trying to reconcile these seemingly contradictory findings in the literature, it has been suggested that while mental illness may not directly cause recidivism, it may influence supervision outcomes in other indirect ways (Skeem et al., 2011; Skeem & Peterson, 2012).

One likely explanation for the observed relationship between depression and the supervision relationship is that the depressive symptoms interfere with an individual’s interpersonal functioning and ultimately his capacity to form positive relationships. Another potential interpretation is that mentally ill individuals are subject to more intense law enforcement-oriented styles of supervision as a result of supervision officers’ biased perceptions toward those with mental illness as being more dangerous and in need of close monitoring, which impacts the capacity to develop high-quality supervision relationships (Skeem et al., 2011). In either case, it is imperative for individuals with mental illness to be paired with supervision officers who possess the understanding and skills necessary to effectively work with this population and build positive relationships (Skeem et al., 2011). In addition, these findings underscore the need to consider ways in which the prison environment might cause or exacerbate problems with mental health functioning and thereby hamper successful post-release reintegration (Haney, 2002). One attractive possibility is to actively promote programs and activities that have counteracted the negative mental health effects in prisons, such as art therapy programs and providing inmates access to income-generating activities (Bedaso et al., 2018; Gussak, 2007).

The final individual characteristic variable that was found to influence the supervision relationship was a measure of interpersonal violence. Perhaps not surprisingly, individuals who were more likely to resort to interpersonal violence in their personal relationships were less likely to report having high-quality relationships with their supervising officer. It is highly likely that these individuals struggle in many aspects of their interpersonal relations and could benefit greatly from programming to address these shortcomings.

Our second set of findings highlighted the importance of a variety of supervision experiences that influenced the quality of supervision relationship. Before discussing these specific variables, it is important to acknowledge that overall the effects of supervision experience variables on the supervision relationship were substantially more robust than the individual characteristic variables. The r² value for the model that examined the effect of individual characteristics on the PO relationship only was nearly a third (7.5%) than that of the model examining the effect of supervision experiences only on the PO relationship (19.3%). This finding lends support to the notion that supervision experiences are more important in understanding the supervision relationship than individual characteristics. Furthermore, through standardizing the coefficients in the full model, we found that supervision experiences—overall—exerted much more robust effects on the PO relationship relative to the effect sizes of the individual characteristics. Although more research is needed, our findings suggest that supervision experiences may be more salient than characteristics of the individual under supervision in promoting positive—or negative—relationships with the supervising officer. From a policy standpoint, these findings seem to offer reason for optimism as supervision experiences appear to be more easily modified through policy changes.

Looking more closely at the specific findings related to supervision experiences, we see that two contact-related variables were positively associated with the perceived quality of the supervision relationship. Increased contact via the phone and increased contact with the employer were associated with more positive relationships. The reasoning behind these findings is not entirely clear, nor was it expected. Collateral contacts with employers, for example, are generally considered a law enforcement function of supervision; thus, it is not clear why an increase in these types of contacts is related to more positive relationships. It might be that individuals on parole perceive officers who utilize these types of contacts as more invested in their success. It might also be that less formalized contact, such as through the phone calls, provide a better medium for building positive relationships. In any case, these results question the notion that supervision intensity is inherently detrimental to supervision outcomes (e.g., Morash et al., 2019). Moreover, it is clear that additional research in this area is needed, to include qualitative research, to better understand these findings. At the practical level, supervision agencies should look at avenues for increasing the use of these types of contacts as a normal part of the overall supervision strategy.

The remaining significant supervision experience variables, including PO Assistance Topics, PO Sanction, PO Praise, and PO Reprimand, are similar in that they are all interactional in nature.² It was projected that a number of supervision experiences that are commonly outside the control of the supervising officer, such as having to enforce punitive supervision conditions, would affect the perceived quality of the supervision relationship. This, however, does not seem to be the case; rather, the supervision relationship is most strongly affected by the nature of the interactions between the individual and officer. For example, officers who utilized more praise and focused on topics related to an individual’s successful reentry in their communications with individuals on their caseloads developed more positive supervision relationships. A recent study suggests that a more carefully calibrated officer approach to communicating with probationers and parolees has the potential to further enhance the interactions (Smith et al., 2019). On the contrary, officer use of sanctions and reprimands led to more negative perceptions of the supervision relationship. The implications stemming from these findings suggest that officer training should emphasize the importance of positive interactions that focus on assisting individuals to achieve successful reentry and praising success. At the same time, these findings raise questions about the long-term efficacy of supervision policies that emphasize the sanctioning of transgressions as the primary means to promote compliance with supervision conditions, as the damage these approaches cause to the supervision relationship might outweigh their short-term deterrent effect.

Limitations and Conclusion

Despite the contributions of this project, it is not without limitations. First, given our use of secondary data to carry out this study, we had to rely on certain measures that are likely limited in their capacity to capture the full depth and complexity of the constructs they are meant to measure. For example, the seven items used to create our relationship scale falls short of the current “gold standard” for measuring the supervision relationship—the dual role inventory revised (DRI-R) developed by Skeem and associates (2007). While our relationship measure falls short of the DRI-R standard, many of the questions that make up our scale are similar to questions found in the DRI-R. For instance, both measures ask respondents to assess certain interactional qualities of the supervision officer, such as treating the individual with respect, being trustworthy, and a willingness to listen (Skeem et al., 2007). At the same time, future research should examine the supervising officer-returning person relationship with additional measures.³

Although one of the strengths of the SVORI data is that it is longitudinal panel data, the final wave of data collection occurred 15 months post-release. The process of reintegrating following release from prison may continue well past this time and, consequently, we are unable to determine the long-term factors that relate to the parole officer relationship. Likewise, although we used a longitudinal modeling technique, it is possible that the initial quality of the supervising officer relationship plays a key role in understanding changes over time. Unfortunately, mixed (and fixed) effects models cannot include a lagged measure of the dependent variable, and thus, we are unable to examine how “baseline” levels of this relationship matter for future changes. In addition, our analysis is limited to examining males returning from prison and findings may not be applicable to females. Future research should examine factors that relate to the supervision relationship among women. Relatedly, while the SVORI sample is generally similar to the incarcerated population in the United States, it comprises individuals convicted of more serious offenses and may not be generalizable to the broader population. Our sample is also limited to individuals on parole; yet, given the expansion of those placed on community supervision, future research should examine whether these findings are consistent among individuals placed on probation. Finally, prior research suggests that officer-related factors such as demeanor and interaction style are likely important predictors of the supervision relationship; however, this information was not available in the SVORI data. Similarly, all our data were collected from the returning person, which raises the possibility of a single source bias. Future research should examine both the role of demeanor and interactions of the supervising officer in this relationship and collect data about the relationship reported by the supervising officer.

Over the last two decades, the corrections field has sought to distance itself from unproven and often counterproductive supervision practices and focus more closely on evidence-based approaches that enhance community safety and promote behavioral change. Findings from this study add to the growing body of research which demonstrates that developing and maintaining high-quality supervision relationships can contribute to prosocial reentry outcomes and decreased reentry failure. Perhaps more importantly, our findings suggest that while characteristics of the returning individual—like family support and mental health—relate significantly to one’s relationship with their supervising officer, supervision experiences appear to be far more robust in forming and maintaining this relationship. While correctional institutions may be limited in their ability to change the characteristics of the returning individual, probation/parole officers do have the ability to dictate—at least on some level—how they interact with, and supervise, the individuals under their supervision. Consequently, these findings have the potential to shape future policy to increase supervision practices that can enhance the quality of the supervision relationship, while also recognizing the need to reexamine existing practices that may be counterproductive to developing high-quality supervision relationships.

Author Biographies

Eric J. Wodahl is an associate professor of criminal justice at the University of Wyoming. He received his PhD in 2007 from the School of Criminology and Criminal Justice at the University of Nebraska, Omaha. His research interests include prisoner reentry, alternatives to revocation in the supervision of individuals in the community, and correctional policy issues.

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

Brett E. Garland is a professor and head of the Department of Criminology and Criminal Justice at Missouri State University. He received his PhD in 2007 from the School of Criminology and Criminal Justice at the University of Nebraska, Omaha. His research interests include criminal justice organization and management, prisoner reentry, and public opinion on justice-related topics.