“You Know Baseball? 3 Strikes”: Understanding Racial Disparity with Mixed Methods for Probation Review Hearings

Abstract

Prior literature on judicial decision-making post-sentencing is relatively scarce, yet with the growth of problem-solving courts and offenders placed on probation, judges are responsible for overseeing compliance of offenders beyond traditional decision-making points. More recently, scholars have called for more nuanced methods of examining judicial decision-making, disparity, and attribution than traditional quantitative methods. This study examines the factors that influence judicial sanctioning of probationers for non-compliance in a domestic violence court. The following research questions are examined: Which factors predict whether a probationer is sanctioned for non-compliance? What are the discourses utilized to frame these violations? Are there differences in discourses utilized based upon a probationer’s race? This study combines participant observation of probation review hearings with agency records for a mixed-methods examination of which factors influence the decision to sanction non-compliant probationers, and whether differences emerge based on race. The sample included 350 cases of probation review hearings with 100 cases selected for critical discourse analysis. Results demonstrated that drug use, missed treatment sessions, gender, race, and family status influenced sanctioning decisions. Qualitative results demonstrated that judges evaluate probationers based upon contextual information, which at times relies on racial discourses of drug use and responsibility.

1. Introduction

Much of the research on racial disparity in the court process has focused on sentencing (e.g., Feldmeyer and Ulmer 2011; Ulmer et al. 2016) and pretrial release (e.g., Freiburger and Hilinski 2010; Wooldredge 2012). Fewer studies have examined decisions made post-conviction at probation review hearings (e.g., Tapia and Harris 2006) or in non-traditional courts, which have grown in popularity in the recent past (e.g., Arabia et al. 2008; Ray and Dollar 2013; Snedker et al. 2017). In these settings, judges often have more information provided to them about the defendant, few restrictions curbing their discretion, and a variety of possible sanctions (i.e., punishment) for non-compliant behavior (e.g., verbal reprimand, increased supervision, jail).

The number of individuals under community supervision continues to dominate corrections. Probationers continue to make up the majority of the correctional population, with 58% of offenders under supervision serving probation sentences in 2016 in the United States (Kaeble and Cowhig 2018). Although most probationers successfully complete their sentence, a sizeable proportion (42%) have had their probation revoked for reasons other than committing a new offense (Kaeble 2018). Judges and probation agents have vast discretion in whether and when to sanction for technical violations and seek revocation, which often reflects the goal of offender accountability (Matejkowski et al. 2011; Rodriguez and Webb 2007). Racial disparity in probation decision-making has been a growing concern in the literature. Some studies have examined whether there are racial differences in recidivism and technical violations (e.g., Iratzoqui and Metcalfe 2017; Rodriguez and Webb 2007), while fewer studies have examined racial disparity in revocation decisions (e.g., Gould et al. 2011; Steinmetz and Henderson 2016; Tapia and Harris 2006), and fewer still examined disparity in termination (i.e., failure to successfully complete the program, e.g., Ray and Dollar 2013) or sanctioning for non-compliance (i.e., technical violations) within problem-solving courts. Yet, these decisions involve more information provided to judges and great discretion.

The prior literature typically examines racial disparity though employing attribution-based theories (e.g., focal concerns) in quantitative analysis of existing datasets from secondary sources and agency records, adding race into models after controlling for legally relevant factors (see Baumer 2013; Ulmer 2012 for discussion). These approaches, however, do not adequately test the theories they purport to test, as the “black box” of attributions and cognitive processes are unspecified, nor do they take into account how so-called legally relevant factors are framed within context, or viewed differently depending on race and gender of offender (Mears 1998; Ulmer 2012; see Bridges and Steen 1998 for exception). This study expands on the prior literature by including both qualitative and quantitative analyses of decision-making in probation review hearings in a domestic violence court. Observational court data from 350 hearings in a large, Midwestern county in 2016 were examined to determine how aspects of non-compliance (i.e., technical violations) are evaluated, or framed, by judges and other court actors, whether there are differences in the framing of these factors dependent upon a probationer’s race, and ultimately whether racial disparity exists in sanctioning for non-compliance. Throughout this paper, I argue that utilizing mixed methods and multiple sources of data produces a more robust picture of what and how factors influence judicial decision-making than the current methods of quantitative analysis with agency datasets. Theories of decision-making can be better tested and applied to particular contexts by relying upon multiple data sources and actually examining the process of decision-making as interaction within hearings. Further, if racial disparity exists in the everyday talk of courtroom actors, policies can be drafted to bring awareness to implicit and explicit biases to reduce these disparities.

2. Review of the Literature

2.1. Decision-Making as Interaction among Court Agents

This paper utilizes symbolic interactionism and critical discourse theories from the sociological literature to more fully make sense of how judges make decisions. Symbolic interactionism states that individuals attribute meaning to others’ actions and demeanor through transformation of a common set of symbols into meaning (i.e., framing). Negotiation, persuasion, joint construction, and contestation of meaning occur in social interactions and thus affect decision-making in institutional settings (Goffman 1981; Feeley 1979). These attributions of meaning involve ongoing interactions, developing patterned meanings based upon knowledge of others’ assessments (Blumer 1969, 1994).

A similar line of sociological theory relies on Foucault’s (1980) concept of discourse, which reproduces social structure at the site of interaction based upon cultural reference frames. Discourse is the language used to interpret the world and is rooted in a particular cultural and historical context (see also Dittmer 2010; Fairclough 1993; van Dijk 1996; Waitt 2010). Critical discourse analysis is concerned with examinations of power and knowledge, which are mutually reinforcing (Foucault 1980). This type of discourse analysis examines power asymmetry in speaker and institutional roles, such as the ability to speak, ask questions, and manage topics. It allows the researcher to interrogate how “language […] creates categorical reality rather than the other way around”, thus discovering how people and their actions are constructed in light of dominant worldviews and competing discourses through social interactions (McCall 2005, p. 1777; Snow 2001).

Within the court context, the social worlds theory applies some of the tenets of symbolic interactionism. This theory states that informal social court communities develop norms and patterned practices through ongoing interactions amongst the same court actors (i.e., judge, defense attorney, prosecutor; Eisenstein and Jacob 1977; Feeley 1979; Heumann 1977; Eisenstein et al. 1988; Ulmer 1997). These court actors attribute meaning to (or socially construct) defendants’ actions, demeanor, and the nature of the offense or infraction, with these meanings rooted in previous experiences and cultural references, or toolkits of meanings, including primary frames of race, gender, age, and class (Blumer 1969, 1994; Goffman 1981). More specifically, any “fact” presented in court, such as a statement or test result, can be subject to multiple interpretations (Feeley 1979). Hearings, then, are disputes around the framing of facts, with each party mobilizing, manufacturing, or reframing particular facts to negotiate issues of culpability, responsibility, or assessments of dangerousness. Thus, the ‘fact’ of only one positive drug test may be mobilized to demonstrate a rare exception to a normally compliant probationer or be constructed as a personal, moral failing requiring correction.

Scholars have used the concept of discourse to examine court practices in juvenile courts (Travers 2007), sexual assault cases (Conley and O’Barr 2005; Frohmann 1991), and probation (Worrall 1990). Asymmetry in the distribution of power vis-à-vis ability to speak, ask questions, manage topics, and ultimately access discourses in the framing of offenses and offenders is inherent in legal settings (Thornborrow 2002; Conley and O’Barr 2005; Fairclough 1993; Merry 1990; Worrall 1990). Judges command power through their institutional roles; however, probation agents may also command power through their depth of knowledge on the probationer’s background and progress. The discourses utilized in courts may reflect focal concerns of judges, or factors specific to problem-solving courts.

2.2. Focal Concerns and Racialized Images of Culpability and Dangerousness

One of the dominant theoretical perspectives utilized to explain racial disparity in courts and sentencing is the focal concerns perspective. Developed by Steffensmeier et al. (1998), this framework states that judges are guided by three overarching concerns when making decisions in criminal cases—the blameworthiness of the defendant, dangerousness to the community (or potential for reform), and practical constraints. Although initially developed to explain decision-making in traditional courts, this theory has been utilized to explain decision-making in problem-solving courts (e.g., Jeffries and Bond 2013; Ray and Dollar 2013) and in probation/parole (Buglar 2016; Huebner and Bynum 2006), albeit with differing emphasis on factors influencing these concerns.

Central to this theory, judges take into account information about the defendant’s background in assessing dangerousness, culpability, and amenability to treatment (see Albonetti 1986; Steffensmeier et al. 1998). Within the realm of problem-solving courts, defendants are considered more blameworthy for non-compliant acts when there is a pattern of non-compliance, refusal to accept responsibility for these acts, and overt refusal to comply with conditions imposed (Ray and Dollar 2013). Judges also assess probationers’ potential and progress with rehabilitation based upon compliance (i.e., abiding by conditions of probation)—continued non-compliance is likely to be viewed as lacking progress, which may increase the likelihood of termination. Further, judges may view risk/needs scores, substance use, and employment history on whether defendants or probationers are likely to recidivate while under supervision (Hartley et al. 2007; Ray and Dollar 2013). Practical constraints may include whether the defendant has minor children, steady employment, health problems, and mental health issues that make sanctioning via incarceration a less attractive option (Freiburger 2011; Steffensmeier et al. 1993, 1998). Given the tenets of focal concerns, judges examine non-compliant behavior for aspects of culpability, such as purposeful disregard for conditions of probation, irresponsibility, and mitigating factors (e.g., illness, caring for a sick child, addiction or mental health issues) that could explain why a probationer tested positive for drugs or missed appointments.

Judges may rely on racialized attributions of culpability and amenability to treatment when examining aspects of non-compliance, specifically drawing on cultural scripts (i.e., discourses) of Black men as dangerous and irresponsible (Russell 1998; Steffensmeier et al. 1998; Travers 2007; van Cleve 2016; Wacquant 2010). Black probationers’ non-compliance may be attributed to irresponsibility and personal failings, while White probationer’s non-compliance may be viewed as caused by factors outside of their control (e.g., mental illness, competing work or family demands; see Albonetti 1991; Bridges and Steen 1998; van Cleve 2016).

Few studies have utilized a mixed methods approaches to examine racial disparity in court processing, yet this avenue of research has the strength of understanding the process qualitatively and outcomes quantitatively. Examples of this approach can be found through using case notes (e.g., Bridges and Steen 1998), court observations (e.g., Ray and Dollar 2013) and interviews (e.g., Steffensmeier et al. 1993; Steffensmeier et al. 1998) to understand what factors judges consider important and how these factors may reflect racialized attributions of culpability, dangerousness, or irresponsibility. In particular, Bridges and Steen (1998) conducted a content analysis of probation officer reports submitted to judges in juvenile courts, finding that probation officers typically referenced internal attributes when reporting on Black youth (e.g., dress, attitude, lack of remorse), while White youth reports mostly included external attributes (e.g., family disruption, delinquent peers). These raced attributions influenced probation officer recommendations, and subsequently judges’ decisions for community placement in their quantitative analysis. In contrast, Ray and Dollar (2013) conducted a sequential mixed methods approach, conducting participant observation of mental health court staffing to arrive at their findings that judges were guided by focal concerns that were gendered, with judges framing women’s non-compliance as mitigated due to bad male influence and time-management issues. They subsequently quantitatively examined the factors that predicted successful graduation and termination from the court based on agency records, finding that White women were least likely to be terminated. The statistical analysis found racial and gender disparity, while the qualitative analysis provided contextual information on why and how this disparity arose. These examples of mixed methods designs highlight the ability to examine disparities in outcomes as well as the processes of decision-making in which context and discourses influence perceptions of legally relevant factors, most notably whether court actors utilize different discourses to frame non-compliance.

2.3. Differences in Processes between Traditional and Problem-Solving Courts

Most of the court processing literature focuses on traditional courts, in which judges are faced with limited time and resources to tackle high caseloads. Central to theories of disparity in these courts is the idea that judges make decisions quickly and with imperfect information (Albonetti 1991; Steffensmeier et al. 1998). By contrast, problem-solving courts typically have small caseloads, more frequent contact with offenders, and vast information on a defendant’s progress (Jeffries and Bond 2013; Ray and Dollar 2013). Additionally, traditional court philosophies of punishment vary by judge, while problem-solving courts emphasize rehabilitation and therapeutic jurisprudence (Worrall 2008). Domestic violence courts are somewhat unique among problem-solving courts, however, in that some process defendants similar to traditional courts with an adversarial approach, yet also emphasize offender accountability and victim safety through reliance on probation and mandated batter intervention programing (BIP) (Healey et al. 1998; Worrall 2008). Problem-solving courts tend to have less guidelines or restrictions on sanctions for non-compliance, which means wide variability in the frequency of jail sanctions applied across different courts (Griffin et al. 2002) and in domestic violence courts in particular (Labriola et al. 2012).

Although there are differences between these courts in terms of process and philosophy, they may both suffer from reliance on racialized stereotypes and attributions of non-compliance. Prior research indicates that judges in problem-solving courts are guided by the same focal concerns as those operating in traditional courts, albeit with distinct differences in factors considered for culpability and suitability for rehabilitation (e.g., Ray and Dollar 2013). Other research finds related focal concerns that are specific to these specialty courts, such as discourses of responsibility and social work in juvenile courts (e.g., Travers 2007).

2.4. Factors Associated with Sanctioning and Revocation for Non-Compliance

The literature on probation decision-making is limited and primarily utilizes quantitative methods. Most studies examine broad types of issues that influence revocation (i.e., cancellation of probation and imposition of original incarceration sentence), such as any technical violation compared to new crime, or comparing “minor” (e.g., failure to update address), technical violations to “moderate” (e.g., positive drug tests), and “major” violations (e.g., new crimes, absconder; see Rodriguez and Webb 2007). These studies generally find that more serious violations and new crimes increase the likelihood of revocation (Rodriguez and Webb 2007; Steinmetz and Henderson 2016). Yet, judges may view different acts of non-compliance (e.g., drug use versus missed treatment sessions) differently and therefore take different actions when these issues are presented in a compliance or review hearing.

In contrast, studies on factors associated with termination and sanctioning in problem-solving courts tend to examine individual aspects of non-compliance and the timing of these acts. For example, Gallagher et al. (2015) examined factors that influenced drug court termination from an Indiana court, finding that unemployment, having any technical violations in the first month of the program, and the number of positive drug tests all increased the likelihood of revocation. In evaluating factors associated with termination from a mental health court, Ray and Dollar (2013) found that defendants who had been enrolled for longer periods were more likely to be terminated, which they noted stemmed from concerns of ability to be reformed given the amount of time and resources spent as well as opening spots for other potential participants. Unfortunately, they did not have measures of non-compliance in their study, which undoubtedly influence termination decisions.

2.5. Racial Disparity in Probation and Problem-Solving Court Decisions

Few studies have examined racial disparity in probation and revocation decisions, and the findings are mixed. Schulenberg (2007) is one of the few studies to examine probation agent sanction decisions—namely, whether a probationer was verbally admonished or required to attend a compliance hearing for committing technical violations. She found that probationers of color and those who had used alcohol or drugs during the offense were more likely to be verbally scolded by one’s agent; however, there were no racial differences in likelihood of attending a compliance hearing. Examinations of revocation decisions are more common within the literature. Most studies have found that Black probationers were more likely to be revoked, in part due to more frequent technical violations and new crimes (Gould et al. 2011; Steinmetz and Henderson 2016; Tapia and Harris 2006). Some studies have also found Hispanic probationers were more likely to be revoked (Steinmetz and Henderson 2016; Tapia and Harris 2006) and more likely to be adjudicated (i.e., unsuccessful deferred adjudication—having one’s judgment withheld until demonstrates serving probation satisfactorily) than successfully complete probation compared to their White counterparts (Steinmetz and Henderson 2016).

Racial disparity has also been examined in various problem-solving courts. Some studies have found that Black defendants are less likely to graduate from drug treatment court than Whites (e.g., Gallagher 2013; Ho et al. 2018). Similarly, Ray and Dollar (2013) found that White females were least likely to be terminated from a mental health court, while Black and Hispanic males had the highest odds of termination for non-compliance. As previously stated, this study did not include measures of non-compliance, which could potentially explain the disparities found.

2.6. The Current Study

Scholars have argued for refinement in research on court decision-making, particularly in testing theories of racial and ethnic disparity to capture the processes and conditions in which disparities are produced (e.g., Baumer 2013; Ulmer 2012). This paper responds to these critiques by utilizing a mixed methods approach to examining decision-making in probation review hearings. It is one of the first to link qualitative data gathered from participant observation of court hearings with quantitative analysis of relevant case, probationer, and processing factors to examine whether and how the everyday talk in courtrooms influences judges’ assessments of probationers, and ultimately their decisions to sanction for non-compliance. More specifically, by drawing on symbolic interactionism, this paper examines what factors are discussed, how they are framed, and ultimately what factors are highlighted in justifying a decision. To that end, four research questions are examined:

• Which factors predict whether a probationer is sanctioned for non-compliance?
• Are Black and Hispanic probationers more likely to be sanctioned for non-compliance?
• What are the discourses utilized to frame these violations?
• Are there differences in discourses utilized based upon a probationer’s race?

3. Methods

3.1. Setting

The study took place in three domestic violence courts in an urban Midwestern county in the United States, which process misdemeanor and felony domestic violence cases, as well as family violence cases involving child neglect and abuse. Participation in the domestic violence court is not voluntary, and all defendants sentenced to probation receive at least one routine review hearing approximately 60 days after sentencing as a measure of accountability. All probation review hearings for family violence offenders were observed from February through September 2016; only cases involving non-compliance with conditions of probation were included for analysis. The final quantitative sample included 347 cases, excluding other races (i.e., Asian, Native American, n = 3). In addition, I purposively sampled 100 cases for qualitative analysis. Cases were selected based on being information rich (i.e., filling 1/3 or more single-spaced pages), while purposively oversampling cases with women probationers and those in which mental health was discussed. Women probationers were oversampled in order to draw substantive conclusions on gender differences, given that they were less common in domestic violence courts.1

Judges in the county work on three-year judicial rotations and hold probation review hearings for all probationers sentenced in their court every 60 days, with additional hearings scheduled at the discretion of the judge. Prior to hearings, judges receive status reports on each probationer and review them before court. At these hearings, probationers are present in the gallery, with probation agents filling the jury box. Defense attorneys are a rare sighting, yet there are often 3–5 attorneys present on most days of hearings. Prosecutors are there simply to record notes on case files, and rarely ever speak. Bailiffs also line the court, often 6–15 present, ready to take a probationer into custody for non-compliance.

The structure of the hearings involves court clerks calling the case number, followed by the judge giving the probation agent an open invitation to speak, clarify, or amend their report previously filed with the judge, as well as make a recommendation on sanctioning. Questions then continue between judge and agent, judge and attorney, or judge and probationer with some interjections from the parties to clarify, contest, or support a statement. Sanction options include verbal praise (i.e., for improving since last hearing) or reprimand (i.e., warning to improve or face jail time) and a jail stay. Judges are able to set condition time and modify the time at any point, which is utilized for jail sanctions. In this jurisdiction judges have vast discretion on which option to impose for non-compliance and no restrictions on the length of a jail stay imposed. Judges may require a second or third review; some of the hearings observed were in this category. Hearings were often under 10 min, with the clerk calling the next case on the docket.

3.2. Research Design

This mixed-methods design utilized a QUAL + quan parallel design with conversion of qualitative findings into quantifiable measures (Creswell and Plan Clark 2011). This design prioritizes qualitative methods, which are used to inform quantitative analyses that will confirm emerging themes. Although the data were collected simultaneously, I conducted the qualitative analysis first, which was used to inform some of the modeling decisions in the quantitative component. After each portion of the research was analyzed, findings were integrated through meta-inference, in which the findings were examined in comparison to one another on how qualitative methods can help inform quantitative modeling and interpretation of decision-making in probation review hearings (Teddlie and Tashakkori 2009).

3.3. Quantitative Sources and Analysis

I conducted quantitative coding from three data sources for each case. The first was taken from field notes for information relevant to the hearing (e.g., non-compliance, judge, attorney present). In addition, I collected information from a publicly available court records website for probationer demographics, prior record, and case-processing factors. Finally, I obtained police reports for these cases by matching probationer name, date of birth, and offense date from court records for incident and offense-specific information (e.g., weapon use).

3.3.1. Dependent Variables

I created two variables from court observations that capture the sanctioning outcome. The first is a dichotomous measure of whether the defendant was sanctioned to jail with yes coded as 1, with those not ordered to jail coded as 0. Those that were not sanctioned for non-compliance were typically verbally reprimanded to “do better” or warned that anything less than perfect at a future hearing would lead to jail time. By contrast, judges congratulated some for improvement from a previous hearing, even if there were continued issues with attendance, unemployment, or substance use. The second was a count measure of the number of days judges decided to sanction a probationer for non-compliance. The range for jail time was 2–60 days, yet most probationers received less than one week of jail time. Similar to the sentencing literature, I conceived of sanctioning as a two-step process (Steffensmeier et al. 1998; Franklin 2017; Jordan and McNeal 2016).2

3.3.2. Independent Variables

As this study is exploratory, all probationer background and hearing-specific variables were treated as independent variables. The first set of variables captured probationer social status. I measured race as a trichotomous indicator with Hispanic coded as 2, Black coded as 1, and White as the reference category. Gender was a dichotomous indicator with 1 representing men, and 0 representing women. Probationer age was a continuous measure and captured the age of the probationer at the date of the hearing. I created a variable for family status based on whether they had minor children or were pregnant during the offense, with 1 representing parents, 0 for non-parents, and 2 for missing information on parenting status. Finally, employment status was captured as a trichotomous indicator, with 2 as no mention of employment at the hearing, 1 for employed individuals, and unemployed as the reference category, as prior literature has shown this influences termination and revocation decisions (Gallagher et al. 2015; Tapia and Harris 2006).

The second set of variables capture prior record and offense severity. Prior research has found that prior record and offense severity increased the likelihood of revocation (Lin et al. 2010; Rodriguez and Webb 2007), as well as resulted in longer terms of incarceration at sentencing (Steffensmeier et al. 1998; Freiburger and Romain 2018). Prior record was measured with two indicators of the nature of prior convictions. The first was a continuous measure of the number of prior violent convictions; the second was the number of prior felony convictions. Offense severity was taken from statute severity and class as a continuous measure, with 1 coded as an Unclassified Misdemeanor, and 8 as Class F Felony.3

The third set of variables include information specific to the hearing, capturing non-compliance issues as well as court actors present. I controlled for which judge presided over the hearing as a categorical variable for each of the five judges who presided over at least one date of hearings, with Judge A as the reference category.4 I also included a dichotomous indicator of whether a defense attorney was present at the hearing (1 coded as yes), as probationers with attorneys may have greater access to the law vis-à-vis legal and institutional knowledge from their lawyers in how to frame issues (Conley and O’Barr 2005; Gathings and Parrotta 2013; Worrall 1990).

The last set of variables captured common types of non-compliance issues that occurred in the domestic violence courts, for which probationers are commonly required to complete or abide by as a condition of their sentence. The first measured whether the probationer had any attendance issues with batterers’ intervention programming (yes = 1, no = 0). The second measured whether the probationer tested positive for alcohol or drugs (yes = 1, no = 0). The third measured whether the probationer had any attendance or interaction issues with their probation agent (yes = 1, no = 0). The last measured other issues that were raised for offender non-compliance, which varied depending on the conditions imposed on an individual probationer. These issues often included parenting class or mental health counseling attendance, or failure to pay court costs, and were commonly mentioned by court actors during the hearings. The last measure of probationer compliance was whether the probationer had contact with the victim, coded 1 for yes and 0 for no. In addition to coding for non-compliance issues presented at hearings, I also quantified the framing of said issues. Each variable was measured as positively framed (=0) or negatively framed (=1), or no issue stated (=2). Positive framing of issues was defined as framing the non-compliance as “doing better” than in past reviews with the act and an excuse offered considered “valid” such as missed appointments due to work conflict. By contrast, negative framing offered no mitigating view of the act, and often included statements by the judge or probation agent that the act was irresponsible or purposeful.5 The set of measures addresses issues found from the qualitative analysis—namely that facts are often framed in a particular light and attempts to capture the nuance of how judges view particular acts of non-compliance in deciding to sanction (see Feeley 1979).

3.3.3. Logistic and Negative Binomial Models

As there were two dependent variables (i.e., any jail sanction, length of jail sanction), two sets of analytical models were built. As the first dependent variable was dichotomous, a binomial logistic regression was performed predicting the likelihood of receiving a sanction for non-compliance. The second dependent variable, as a count of days jailed, required the use of a model in the Poisson family, as the assumptions of ordinary least squares regression would be violated (Heeringa et al. 2010; Long and Freese 2006). The ‘countfit’ command in Stata demonstrated that the zero-truncated negative binomial regression model was preferred based upon model fit and AIC and BIC estimates (Long and Freese 2006). Prior to analyses, multicollinearity was assessed; there were no issues when pair-wise correlations, the variance inflation factor, and tolerance were examined and were within thresholds recommended by Allison (2012).

3.4. Qualitative Sources and Analysis

I spent eight months conducting participant observation in the three domestic violence courts during scheduled probation review hearings. I took extensive field notes, capturing what each person (e.g., probationer, judge, attorney, probation agent) said, as well as tone, pauses, and interruptions. A total of 27 hours of observation were conducted for all hearings that occurred, producing 2.5 notebooks filled with handwritten records of what were essentially court transcripts. After hearings, I transcribed these field notes into a word processing document before transferring to NVivo for qualitative analysis.

Critical Discourse Analysis

Critical discourse analysis was used for the qualitative analysis, treating courtroom talk as text. Discourse is power-laden in regard to what can be accepted as fact, who can render this decision, and who is able to speak (Schram 2006). The analysis consisted of two stages: descriptive and interpretive (Jackson 2001). Based upon the recommendation of Jackson (2001), descriptive coding began with reading the hearings, looking for common phrases and words (i.e., keywords), as well as affect toward probationers (e.g., terse tone, sarcasm, disbelief, sympathy; see also Waitt 2010). Further, descriptive coding for each hearing contained codes for context (e.g., judge, whether a defense attorney was present, who was involved in exchanges), turn-taking (e.g., question, response, interruption, diatribe, silence), and knowledge references (e.g., bench experience, scientific studies) (Waitt 2010).

I identified initial patterns in phrases and terms used through an inductive approach, which required me to think critically about deeper meanings, and positing the ways in which factors may be inter-related in the framing of the defendant’s actions. The process of analytical coding also included determining which key discourses were present as discursive repertoires. Discursive repertoires are “cultural resources everyday speakers may use” (e.g., rhetoric) when talking about court cases to attribute meaning to past and future behaviors (Jackson 2001, p. 208). This inductive analytical approach generated discursive themes, repertoires, and dispositions. The framing of defendants’ actions positively or negatively and rhetoric employed by court actors in making sense of their actions reflect the discursive dispositions and repertoires that will likely influence practices (i.e., sanctions) during probation review hearings.

4. Results

The quantitative results are presented first to address the first two research questions and to demonstrate how mixed-methods approaches that utilize court observations can illustrate nuance and context in how racial differences emerge. The qualitative results address the third and fourth research questions—namely discourses used in framing violations and differences in discourses based upon a probationer’s race. The discourses in the qualitative section are presented because they are the dominant discourses utilized by court actors, as well as to highlight the need for contextualizing non-compliance issues when examining quantitative data.

4.1. Factors Associated with Sanctioning for Non-Compliance

4.1.1. Descriptive Statistics

Table 1. Descriptive Statistics on Dependent, Independent, and Control Variables (n = 350) *.

Variable	Frequency	Percent
Sanction
Verbal (=0)	266	76.0
Jail	84	24.0
Days Sanctioned
(Range 1–60)	M = 9.57	s = 13.03
Gender
Female (=0)	60	17.14
Male	290	82.86
Race/Ethnicity
White (=0)	72	20.57
Black	230	65.71
Hispanic	45	12.86
Other	3	0.86
Age
Range (18–62)	M = 32.32	s = 9.67
Employment Status
Unemployed (=0)	78	22.29
Employed	172	49.14
Not Mentioned	100	28.57
Minor Children
No Children (=0)	53	15.14
Children	262	74.86
Missing Information	35	10.00
Prior Violent Convictions
Range (0–4)	M = 0.23	s = 0.60
Prior Felony Convictions
Range (0–7)	M = 0.52	s = 1.20
Severity at Conviction
Range (1–8)	M = 2.85	s = 1.05
Judge
Judge A (=0)	99	28.29
Judge B	147	42.00
Judge C	64	18.29
Judge D	18	5.14
Judge E	22	6.29
Attorney Present at Hearing
No (=0)	220	62.86
Yes	130	37.14
Missed BIP
No (=0)	162	46.69
Yes	185	53.31
UA Issue
No (=0)	196	56.48
Yes	151	43.52
Missed PO Visits
No (=0)	287	82.71
Yes	60	17.29
Other Issue
No (=0)	165	47.55
Yes	182	52.45
Contacted Victim
No (=0)	314	89.71
Yes	36	10.29
BIP Framing
Positive (=0)	201	57.43
Negative	88	25.14
No issue	61	17.43
UA Framing
Positive (=0)	164	46.86
Negative	100	28.57
No issue	86	24.57
PO Visits Framing
Positive (=0)	45	12.86
Negative	55	15.71
No issue	250	71.73
Other Issue Framing
Positive (=0)	159	45.43
Negative	113	32.29
No Issue	78	22.29

* Note: for continuous variables, means and standard deviations are reported.

displays the descriptive statistics for the sample of probation review hearings from February through September 2016. As can be seen, most probationers were not sanctioned to a jail stay (76%), and of those sanctioned, the average number of jail days was 9.57 (SD = 13.03). Most probationers were jailed for 1–3 days (50.6%) and 71.08% were jailed for one week or less, demonstrating the variable is skewed. Most probationers were male (82.86%) and almost two-thirds were Black (65.71%), followed by White (20.57%), and Hispanic (12.86%). The average age of probationers at the time of the hearing was 32.32 years (SD = 9.67). Almost half of probationers were employed at the time of the hearing (49.14%), most had had children (74.86%), did not have a prior criminal history (violent convictions M = 0.23, SD = 0.60; felony convictions M = 0.52, SD = 1.20, and were convicted of relatively minor offenses (M = 2.85, SD = 1.05), with the average severity as a Class A Misdemeanor.

In terms of non-compliance issues, it was most common for probationers to miss at least one Batterers Intervention Programming (BIP) session (53.14%), have some other issue with complying with the conditions of probation (52.29%), test positive for drugs or alcohol (43.43%), and miss appointments with their probation agent (17.14%). Finally, only 10.29% of probationers contacted the victim, violating the no contact order condition. Although issues with non-compliance were common, there was variation in how these issues were framed. Other issues with non-compliance had the highest frequency of being framed negatively (32.29%), followed by testing positive for drugs or alcohol (28.57%), missing sessions for BIP (25.14%), and missing probation agent visits (15.71%).

4.1.2. The “In/Out” Decision of Probation Reviews

Table 2. Logistic Regression Results Predicting Whether a Probationer was Sanctioned to Jail (n = 347).

Variables	Model 1			Model 2
	Coeff.	S.E.	Exp(B)	Coeff.	S.E.	Exp(B)
Age	−0.020	0.019	---	−0.016	0.022	---
Male	1.04 *	0.519	2.820	0.990	0.614	---
Black	−0.399	0.417	---	−0.529	0.502	---
Hispanic	−0.377	0.308	---	−0.277	0.339	---
Children ^	0.987 ^	0.510	2.683	1.289 *	0.570	3.628
Children-Missing	0.344	0.730	---	1.491 ^	0.801	4.439
Employed	−0.290	0.424	---	−0.275	0.475	---
Employment Not Mentioned	0.476	0.472	---	0.370	0.519	---
Prior Violent Convictions	−0.288	0.353	---	−0.195	0.418	---
Prior Felony Convictions	−0.243	0.166	---	−0.285	0.185	---
Severity of Offense	−0.097	0.157	---	0.076	0.193	---
Judge B	−0.673 ^	0.401	0.510	−0.699	0.470	---
Judge C	−1.297 *	0.523	0.273	−1.674 **	0.623	0.187
Judge D	−1.243 ^	0.737	0.288	−0.865	0.889	---
Judge E	0.157	0.629	---	−0.064	0.708	---
Attorney Present	0.541	0.332	---	0.842*	0.388	2.320
Missed BIP	1.095 **	0.342	2.990	---	---	---
Positive UA Test	1.853 ***	0.345	6.331	---	---	---
Missed PO Visits	1.843 ***	0.412	6.314	---	---	---
Other Issue	1.274 ***	0.344	3.575	---	---	---
Contacted Victim	0.731	0.503	---	0.975 ^	0.558	2.652
BIP Negative Rating	---	---	---	2.026 ***	0.438	7.586
BIP No Issue	---	---	---	0.026	0.621	---
UA Test Negative Rating	---	---	---	2.193 ***	0.495	8.963
UA Test No Issue	---	---	---	0.283	0.602	---
PO Visits Negative Rating	---	---	---	0.782	0.749	---
PO Visits No Issue	---	---	---	−1.075	0.688	---
Other Issue Negative Rating	---	---	---	2.457 ***	0.496	11.667
Other Issue No Issue	---	---	---	1.202 *	0.527	3.327
Constant	−3.951 **	1.217	---	−4.886 **	1.673	---
	Log Likelihood = −131.33, chi-square = 119.15, p < 0.001, Pseudo R2 = 0.31			Log Likelihood = −104.614, chi-square = 172.57, p < 0.001, Pseudo R2 = 0.452

Note: ^{^} p < 0.10, * p < 0.05, ** p < 0.01, *** p< 0.001. Odds ratios are reported only for significant variables.

presents the binary logistic regression models of the decision on whether to sanction for non-compliance. Model 1 (Chi-square = 119.15, p < 0.001, Pseudo R² = 0.31) includes the indicator of whether there was an issue for the various conditions of probation. Race was not statistically significant, however, there were gender differences (OR = 2.82). All four aspects of non-compliance influenced the likelihood of a jail sanction: missing at least one BIP session (OR = 2.99), at least one positive drug test (OR = 6.38), missing a PO visit (OR = 6.31), and having any other issue of non-compliance (OR = 3.57). Model 2 includes the framing of particular issues of non-compliance, with positively framed excluded as the reference category (Chi-square = 172.57, p < 0.001, Pseudo R² = 0.45). There were no significant racial differences in the likelihood of a jail sanction, however, family status was significant (OR = 3.65). With respect to issues of non-compliance, having a negative framing of missing BIP (OR = 7.59), positive drug test (OR = 8.96), and other issue of non-compliance (OR = 11.67) increased the odds of jail compared to those who’s issues were positively framed. Interestingly, probationers who had no other issues of non-compliance had greater odds of a jail sanction compared to positively framed issues (OR = 3.33).

4.1.3. Length of Jail Stays for Non-Compliance

Table 3. Zero-Truncated Negative Binomial Regression Results Predicting the Length of a Jail Sanction (n = 76).

Variables	Model 1			Model 2
	Coeff.	S.E.	Exp(B)	Coeff.	S.E.	Exp(B)
Age	−0.067 ***	0.019	0.935	−0.067 ***	0.017	0.936
Male	−0.304	0.462	---	1.238 *	0.548	3.448
Black	−0.860 **	0.318	0.423	−0.601 *	0.294	0.548
Hispanic	−0.816 ***	0.229	0.442	−0.566 **	0.218	0.568
Children	−0.555	0.407	---	−0.751 ^	0.414	0.472
Employed	0.287	0.328	---	0.361	0.295	---
Employment Not Mentioned	0.375	0.455	---	−0.059	0.376	---
Prior Violent Convictions	−0.405	0.296	---	−0.531 ^	0.271	0.588
Prior Felony Convictions	0.215	0.147	---	0.149	0.131	---
Severity of Offense	−0.379 *	0.152	0.685	−0.162	0.133	---
Judge B	0.755 *	0.320	2.128	0.965 **	0.308	2.626
Judge C	1.308 **	0.430	3.700	0.822	0.576	---
Judge D	1.554 *	0.629	4.732	1.532 **	0.589	4.625
Judge E	0.522	0.471	---	0.703 ^	0.406	2.020
Attorney Present	−0.367	0.281	---	−0.015	0.318	---
Missed BIP	0.405	0.292	---	---	---	---
Positive UA Test	0.419	0.299	---	---	---	---
Missed PO Visits	0.522 ^	0.277	1.685	---	---	---
Other Issue	0.602 *	0.285	1.825	---	---	---
Contacted Victim	−0.571 ^	0.333	0.565	−0.195	0.302	---
BIP Negative Rating	---	---	---	0.404	0.280	---
BIP No Issue	---	---	---	1.303 **	0.481	3.680
UA Test Negative Rating	---	---	---	2.373 **	0.764	10.729
UA Test No Issue	---	---	---	2.309 ***	0.546	10.065
PO Visits Negative Rating	---	---	---	0.157	0.414	---
PO Visits No Issue	---	---	---	−0.435	0.424	---
Other Issue Negative Rating	---	---	---	1.726 ^	0.935	5.620
Other Issue No Issue	---	---	---	1.102 ^	0.626	3.011
Hazard	0.617	0.157	---	0.917	0.848	---
Constant	−4.717 *	1.172	---	−0.807	2.680	---
	Log likelihood = −216.30, chi-square = 49.68, p < 0.001. Pseudo R2 = 0.103.			Log Likelihood = −206.92, chi-square = 68.44, p < 0.001, Pseudo R2 = 0.141

Note: ^{^} p < 0.10, * p < 0.05, ** p < 0.01, *** p < 0.001. Missing children category was omitted due to collinearity with the variable children. Odds ratios are reported only for significant variables.

presents the results of the zero-truncated negative binomial regression models examining the count of days jailed for non-compliance. Again, Model 1 (Chi-square = 49.68, p < 0.001, Pseudo R² = 0.10) includes whether there were any issues with compliance. Black and Hispanic probationers had decreased expected days jailed for non-compliance than White probationers (OR = 0.42, OR = 0.44, respectively). While there were no gender differences, age was a significant negative predictor (OR = 0.94) of day jailed. With respect to non-compliance factors, only other issues of non-compliance was a significant predictor of increased days jailed (OR = 1.83), although missing probation agent visits and contact with victim were marginally significant, with the former increasing days jailed and the latter decreasing days jailed (OR = 1.70, p = 0.06; OR = 0.57, p = 0.08, respectively). Model 2 presents the results when framing of non-compliance issues are modeled (Chi-square = 68.44, p < 0.001, Pseudo R² = 0.14). Black and Hispanic probationers had decreased expected days jailed compared to White probationers (OR = 0.55, OR = 0.57, respectively). Additionally, gender and age were significant predictors, with men having greater days jailed and older probationers with decreased days jailed (OR = 3.45, OR = 0.94, respectively). Finally, two aspects of non-compliance were significant: negatively framed positive drug tests (OR = 10.73) and no issue with BIP attendance (OR = 3.68) increased the expected days jailed compared to probationers with positively framed compliance issues for these factors.

4.2. Discourses Utilized to Contextualize Non-Compliance Issues

Issues of non-compliance relied on judges’, probation agents’, and defense attorneys’ discursive understandings about behavior more broadly and compliance with the law in particular. The main discourses that judges, probation agents and defense attorneys utilized in framing non-compliance included responsibility, mental health, and therapeutic benefit. At times, these discourses were applied differently across probationers of color and White probationers, often reflecting stereotypical views of men of color in particular. Finally, I examine the ways in which contextual information and discourses affect sanctions.

During the observations, some of the themes and main ‘issues’ for judges became apparent through seeing repeated assertions, rhetorical questions, and extensive diatribes. Although the judges differed in their approaches and main concerns, the most common issues highlighted in hearings were drug and alcohol use, attendance at BIP and additional programming (e.g., AODA, parenting classes), contact with probation agents, and employment status. These issues elicited different explanations offered by probation agents, judges, defense attorneys, and probationers, rooted in underlying themes of personal responsibility, therapeutic benefit, and mental health. These dominant discourses reflect the main purposes of the problem-solving courts, notably offender responsibility and accountability while addressing underlying biological, social, and psychological needs of offenders (e.g., Winick 2002).

Discussions within hearings largely centered on attendance issues and drug use, yet these concerns were not discussed individually, but were rather related to other factors in probationers’ lives. Similar to what qualitative research on pretrial release and sentencing has found (e.g., Gathings and Parrotta 2013; Feeley 1979), “facts” about probationer progress and status did not simply exist, but were evaluated by probation agents, judges, defense attorneys, and probationers themselves based upon context and how these “facts” were framed by court actors. One act of missing a BIP session or testing positive for drugs did not automatically render the probationer sanctioned; the number of times, context of consistency or “getting better” with violations, and reasons cited for violations often carried more weight to judges than a simple indicator of whether there were any violations.

What contextual factors were drawn upon in framing, or making sense, of probationer behavior varied by who was speaking (e.g., role) and by the judge. Not surprisingly, probationers and, when present, defense attorneys often cited employment, financial, and caregiving concerns when constructing why programming was missed, as well as stress and life strains for why drug tests were positive. Probation agents and judges commonly discussed these issues with respect to parenting responsibility, employment, need for AODA treatment, and irresponsibility. Whether an issue with compliance was ultimately framed by the judge as mitigated or aggravated depended on the interaction of court actors and probationers highlighting and connecting issues, as well as judges’ worldviews and bench experience.

Further, whether discourses of responsibility and personal choice, as compared to mental health, were used to frame a probationer’s non-compliance depended on the probationer’s social location (i.e., race). In these hearings, racial discourses were not made explicit, but rather differences emerged in the discourses used to make sense of non-compliance between probationers of color and White probationers. As will be demonstrated, White probationers often had their non-compliance framed as mitigated, with court actors often drawing upon mental health discourses in emphasizing the need for help. Black and Hispanic probationers, by contrast had their non-compliance framed as blameworthy, with court actors drawing upon responsibility discourses in emphasizing personal choice, failings, and laziness as reasons for issues. The following sections highlight these differences in discourses used based upon racial background for two common issues discussed at length by judges: drug use and programming attendance issues.

4.2.1. Drug Use as Framed within Responsibility and Addiction Discourses

Drug and alcohol use was one of the most frequently discussed aspects of probation compliance by judges and probation agents, who commonly connected use with employment and caregiving responsibilities, framed within broader responsibility discourse. Although some judges would state that drug use is illegal, particularly for probationers who continually used marijuana and cocaine, they would more frequently assert that positive drug tests would create problems with obtaining or maintaining employment. This concern about employability reflected a larger discourse of responsibility, connecting maintaining employment and not using illicit drugs as part of being a responsible, successful adult. Thus, judges often viewed continued drug use as problematic beyond the concern about violating conditions of probation: it affected prospects of rehabilitation by interfering with aspects of daily life. For unemployed probationers, however, continued drug use was framed as part of a drug dealing lifestyle, thus suggesting that unemployed probationers who had money to afford drugs must be “employed” within the drug trade. In one hearing, the judge makes this connection with the question, “So you don’t have a job? How did you get the cocaine?” With these connections, judges drew on cultural scripts of unemployed people living in poverty as drug dealers, particularly for Black men (Beckett et al. 2006; Ramasubramanian 2011).

Probationers with children, who engaged in drug use often were framed as irresponsible parents. For some, judges asked pointed questions about when drug use occurred, attempting to discern whether probationers were watching their children while high and thus irresponsible due to a diminished cognitive state, risking the safety of their children. More commonly, judges would assert that familied probationers using drugs were irresponsible providers. As with unemployed probationers, an underlying class and race assumption is present when judges frame drug use by parents as being irresponsible. In one hearing, the judge made connections between drug use and capability to provide as a parent:

Judge A:

You used cocaine and marijuana? You tested positive two times, you used more than once. I’m not saying you spent a fortune. How many kids do you have?

Probationer:

Seven.

Judge A:

That’s a lot of kids. I want you to keep a picture in your head—a quantity of marijuana and something for any one of your seven kids. Every time you pick up the marijuana, that’s one thing your kid doesn’t get. It’s not like you had all the money in the world—then you could do both. Most people have to make the choice between those two things.

Within this hearing, the judge assumes that the probationer is acting selfishly, placing his own wants for a high over the needs of his children. Thus, drug use is framed with the lens of responsibility discourses vis-à-vis providing for children.

In addition to drug use discussed in connection with employment and parenting status, drug use was commonly framed as either an addiction requiring help (i.e., mental health) or a personal choice (i.e., responsibility). Which discourse emerged as dominant depended largely on drug type, frequency or consistency of use, and racialized assumptions of responsibility or mental health. Marijuana and cocaine were commonly framed under responsibility discourses, particularly for probationers of color, thus rendering probationers who used these drugs as irresponsible, making bad choices, and having personal weaknesses. Judges and probation agents often cited scientific studies on the limited addiction potential for these drugs, as well as how long drugs stay in one’s system when probationers claimed they had used a while ago or had a false positive test. For example, Judge A often interrupted probationers with repeated questions on the last time smoked, such as “You have one more chance—you know baseball? Three strikes—when was the last time?” when met with responses that they had smoked a month ago. Alcohol and opioids, however, were mainly framed under addiction discourses, particularly for White probationers, with continued positive drug tests as indicative of relapse or substance dependence. Judges’ and probation agents’ framing of these drugs as requiring AODA inpatient or intensive treatment reflects the current national public health epidemic with heroin and other opioid addictions and overdoses (CDC 2017).

One hearing offers a prime example of the diverging discourses used for both drugs for a Black man probationer:

PO:

Before court, he had a UA that was positive for marijuana. He says benzos on 9/4 and marijuana use was in July. I did get a letter that he is starting counseling. […] I’m concerned about him—he’s got some addiction issues, he minimizes—I’m trying to get him to accept responsibility.

Judge A:

Mr. ____—what’s going on?

Probationer:

It’s stressful, I miss my babies.

Judge A:

How old are they?

Probationer:

Four—well I take care of her, and one. I just don’t see them on a daily basis.

Judge A:

So because you miss them you take opiates and benzos?

Probationer:

I hurt my back, I talked to my grandma, my auntie overheard and gave me some pills—I felt better.

[…]

PO:

He said he had a Percocet issue since he was 22, this is what I’m talking about.

Judge A:

What about the THC?

Probationer:

I’m a chronic user, I’m working on stopping.

Judge A:

What? Marijuana doesn’t have a physiological addiction—you can stop any time you want.

At first, the probation agent utilized both addiction and responsibility discourses to frame the probationer’s drug use, as taking responsibility is connected with acknowledging a substance abuse problem. After the judge denies the veracity of the probationer’s response of drug use as a way to cope with missing his children by posing a rhetorical question, he asks about the continued marijuana use. Again, when the probationer tries to frame his use as a “chronic user” via dependence and addiction discourse, the judge challenges this statement, asserting that because he “can stop any time,” marijuana use is an irresponsible personal choice, while opiate use is framed as addiction stemming from pain management that led to dependence.

The above example is also indicative of the common concern by judges on the consistency of drug use. Probationers who had multiple, consistent, positive drug tests for marijuana were often framed as chronic users, while those testing positive for opiates were framed as dealing with addiction. Those who had inconsistent histories of drug tests, particularly for marijuana, were framed as making poor choices, with judges often utilizing the rhetoric of “you showed me you can stop” as evidence that marijuana and cocaine use is an irresponsible act, yet a personal choice. Probationers who had a series of positive drug tests, followed by several clean tests, were framed more positively than their counterparts who had consistent positive or inconsistent histories. Judges often emphasized the newly clean history as a “step in the right direction”, whether for marijuana or opiate use. Interestingly, probationers who tested positive only once or twice for opiates were often constructed as making poor choices and were often sanctioned to jail, yet these individuals tended to cite pain management as reasons for engaging in illicit drug use. Thus, both the nature of the drug and history of drug tests mattered in how judges evaluated drug use. Ultimately, Black and Hispanic men probationers were more frequently (n = 24) found to have their drug use framed under personal choice and responsibility discourses, which often translated into short jail sanctions (i.e., 2–7 days).

By contrast, White probationers who tested positive for drugs often were more likely to have their drug use framed under addiction and mental health discourses (n = 3), which translated into verbal warnings and urging to get treatment. In one example hearing, the White man probationer’s non-compliance included failure to find employment and violating the sobriety order by continuing to consume alcohol. At first, the judge’s reaction is to sanction for non-compliance:

Judge B:

I remember the report thinking I don’t understand why he’s not being revoked. It’s like he can find any reason to drink: “it’s Friday—have a drink” “My birthday—have a drink” “a funeral—have a drink” always a reason to drink [in this state], always a reason to drink.

After this assertion, the probation agent informs that the probationer needs inpatient to truly address the underlying addiction needs, yet insurance and access issues are an issue. After some back and forth between the defense attorney, probation agent, and judge determining logistics between applying for inpatient and day treatment, which is viewed by all as a less effective alternative to address alcohol use. The judge ends the exchange by framing his alcohol use as addiction, under mental health discourse, which is better treated in day treatment programming than a jail sanction or potential revocation:

Judge B:

When’s the last time you’ve been to alcoholics anonymous?

Probationer:

Never.

Judge B:

When’s the last time you went to a bar?

Probationer:

Two weeks ago.

Judge B:

So if I send to jail, maybe he gets treatment, or day treatment which will probably fail.

Defense Attorney:

Judge B required mental health assessment—he never had one. I’m concerned it may be mental health, self-medication.

Judge B:

20 days stayed conditioned on day treatment—cost be damned get in into inpatient. If day treatment doesn’t work there’s two options: he needs inpatient treatment, or revoke and send him to prison. Seems the first option makes more sense, the second doesn’t. If you show up drunk, I suppose they’ll put you in jail.

In this instance, the probationer avoids a jail sanction, much like other White probationers who engaged in alcohol and opiate use, as their drug use was framed under the mitigating discourse of mental health.

4.2.2. Missing Programming as Framed by Competing Responsibilities or Laziness

The second most common concern highlighted in probation review hearings was attendance at programming and agent meetings, including arriving late, leaving early, scattered absences, or a delay in starting programming. Judges also were concerned about the number of times missed—simply missing one session did not inherently render a probationer sanctioned—and they often asked for reasons why programing was missed. For those with continued absences from programming, probationers and their attorneys often framed absence within the context of balancing work and programming schedules, and childcare issues, such as having no babysitter or a sick child to stay home with, reflecting responsibility discourses. For many probationers, these contextual details framed the “fact” of missed sessions as mitigated due to competing demands of a responsible adult, which Feeley (1979) called the “malleability” and “mobilization” of facts in framing behavior (p. 176). One example highlights this mobilization of additional information to mitigate missed sessions within responsibility discourses for a review hearing of a Black woman probationer:

Judge B:

Good, she’s also compliant with office visits. Why did you miss two visits there?

Probationer:

I had parenting class and my son had an appointment.

Judge B:

Make sure you make your appointments.

PO:

The programs will work with each other; the program is very important for her son.

Judge B:

You get candy.

In this exchange, the probation agent actively frames the missing of office visits as mitigated due to the scheduling conflict with programming described as “very important” for the child. The judge accepts this contextual information as a valid excuse from a responsible mother. When competing responsibilities were cited by probationers as reasons for missing programming, judges often reminded them to call ahead in order to get the absence excused, while accepting these reasons as valid for responsible adults, and jail sanctions were less common.

While judges viewed competing priorities as mitigating factors for missing programing, other excuses such as oversleeping, forgetting, or not having transportation were seen as illegitimate excuses. Indeed, discourses on probationer progress often centered on responsibility—whether one was acting as a responsible adult balancing multiple priorities, or if one was “doing nothing” all day. Probationers who gave answers of being vaguely “busy” but did not offer excuses linked to employment or childcare responsibilities were often framed as not taking probation seriously and were often sanctioned to jail. Twenty-two probationers of color were framed as lazy, irresponsible, or doing nothing, compared to one White probationer. One judge stated to a Black man probationer with delays in programming, attendance issues, and failure to obtain employment, “I recommend you revoke yourself. You don’t want to comply. Why not serve March in jail? You’re trying to game it—it won’t work”. This was particularly true for probationers of color who had delays in beginning programming. In one such case, the judge homes in on missing BIP, asserting her desire to jail the probationer by beginning the hearing with increasing the condition time available for sanctioning:

Judge C:

I am amending to 45 days condition time for use in the future for, say, you don’t make it to the Green Place. […] When you go, you gotta be there the whole time. Why didn’t you go on 5/20?

Probationer:

I tried to reschedule…

Judge C:

Where is it?

Probationer:

15th and Main.

Judge C:

Does a bus go there? You gotta learn to take a bus. […] Not having a ride to chauffer you to the Green Place is not an excuse—you’re a healthy young man. You can walk, ride, roller skate, take the bus, there’s lots of ways to get there.

In this exchange, the judge challenges the Black man probationer’s offering of trying to reschedule BIP, asserting that there are multiple ways to arrive at a central location within the city. For probationers of color who were not actively seeking employment, or who had delayed starting programming, they were chastised for not taking initiative, framed as lazy and irresponsible.

In summary, judges mainly utilized responsibility and mental health discourses to frame probationer’s non-compliance with drug use and program attendance, which were the main concerns identified in these hearings. These factors, however, were not simply listed as issues, but were framed within context—the timing of drug use and missed appointments, discursive worldviews of addiction versus personal choice based upon drug type and racial understandings of responsibility versus mental health. Thus, non-compliance issues were evaluated in context, and were not seen as inherently “good” or “bad” but were framed as aggravating or mitigating based upon what relevant information was proffered by probation agents, probationers, and defense attorneys, similar to the findings of Feeley (1979). These discourses typically influenced sanctioning decisions, as probationers of color often received short jail sanctions for one or two positive drug tests, missed programming that was not due to competing responsibilities, and delays in starting programing due to not taking initiative in calling to schedule. These acts of non-compliance were typically framed under responsibility discourses as being a personal choice (e.g., drug use) or laziness (e.g., missed or not started programming). White probationers, by contrast tended to receive longer jail sanctions if and only if they had a multitude of non-compliance issues, which were framed under mental health discourses of addiction (e.g., continued positive tests for drugs) or depression (e.g., failure to comply with any programming requirements).

5. Discussion

This study sought to determine how everyday talk by judges, probation agents, and defense attorneys during review hearings influence judges’ assessments of probationers, and how the discourses used in these interactions influence sanctioning decisions. More specifically, it examined whether racial disparity exists in jail sanctions for non-compliance and whether the discourses drawn upon to make sense of non-compliant acts are different depending on probationer race. The findings of racial disparity and difference in discourse were mixed. No differences were found between White, Black, and Hispanic probationers in the likelihood of receiving a sanction, yet Black and Hispanic probationers received shorter jail sanctions compared to their White counterparts. This finding is contrary to much of the literature that tests racial disparity in sentencing (e.g., Bridges and Steen 1998; Steffensmeier et al. 1998) and revocation decisions (Gould et al. 2011; Steinmetz and Henderson 2016; Tapia and Harris 2006).

The qualitative findings present a more complete picture on why Black and Hispanic probationers received shorter jail sanctions. Probationers of color tended to be sanctioned for minimal amounts of drug use (e.g., testing positive once for marijuana), yet they also tended to receive short sanctions for such violations (i.e., 2–7 days). White probationers received longer sanctions if and only if they had extensive multiple issues (e.g., continued positive drug tests, extensive attendance issues, failure to start programming, contact with victim). The differences in factors contributing to length of a jail sanction lie in racialized discursive understandings of responsibility and mental health. The starkest example of this is framing drug use as personal choice for probationers of color and addiction for White probationers. Therefore, the perceived ‘leniency’ toward Black and Hispanic probationers that was observed in the quantitative results is actually due to the differential treatment of their non-compliance, which re-inscribes assumptions of Black and Hispanic probationers as irresponsible and lazy (Cesario et al. 2010; Todd et al. 2016; van Cleve 2016). Relying solely on quantitative methods of examining case records misses this nuanced understanding of decision-making, as factors are made sense of based upon contextual information, such as why programming was missed and how often someone tested positive for drugs. Quantitative methods alone also fail to capture the attributions that court actors make for non-compliant acts, which reflect underlying discourses about behavior.

The findings also demonstrated that attendance issues and drug use mattered in sanctioning decisions, both in modeling indicators of issues of non-compliance, as well as the framing of these issues. Drug use, when framed negatively, increased the likelihood of a sanction as well as the length of jail stays. Missed BIP or other required courses, when framed negatively, increased the likelihood of sanctions, yet did not result in an increase in days jailed, which is consistent with previous research (Rodriguez and Webb 2007; Tapia and Harris 2006). Quantitative analysis tells us that drug tests, BIP attendance, visits with probation agents, and other issues such as missing parenting class or therapy matter to judges, while the qualitative results give nuanced details in how much these factors matter, why they are viewed as important to judges, and the discourses and contextual factors that judges use when framing a factor as ‘positive’ or ‘negative’. Missing BIP, as a ‘fact’ can influence a sanction, but it is the context of the miss (e.g., only missed once, attendance has improved recently, missed due to work) that judges take into account when ascribing responsibility to probationers and determining whether to sanction. Judges consider other factors, such as missing BIP due to childcare issues, connecting reasons for missing or testing dirty with other factors present in the person’s life—and even assumptions about parenting and responsibility when framing a probationer’s missteps. Often two or more issues were tied together interjections and changing topics, with probationers and court actors drawing connections between seemingly separate issues. Therefore, issues exist not simply as ‘facts’ existing outside of context but are imbued with meaning by court actors to frame non-compliance as aggravated or mitigated depending on what additional information is supplied by probation agents and probationers (Feeley 1979). Quantitative analysis typically fails to address these processual issues of the interconnection of issues and the framing of a ‘fact.’

5.1. Limitations

Several limitations existed in both qualitative and quantitative data collection. The sample size and limited variation on the dependent variable may have led to reduced statistical power. This also extended to the selection of 100 cases for qualitative analysis. It could be that cases excluded may be different than those selected. Second, models would have ideally included both the indicator of a compliance issue and the framing of said issue. A high degree of multicollinearity existed between these variables, however, precluding the modeling of both simultaneously. Contingency tables for each set of variables demonstrated variability between issues with non-compliance and how they were evaluated. Future research could benefit by examining interactions between non-compliance issues as well as whether race and gender moderate the impact of particular violations. Third, relying on hearing notes to determine issues of non-compliance led to issues with precision in the measure of how many instances occurred (i.e., number of positive drug tests). Qualitative analysis demonstrated that judges do rely on timing and number of instances when evaluating probationers; future research could benefit from utilizing multiple data sources (e.g., probation reports) for better measures of non-compliance.

Additionally, what is left unsaid by court actors is also a limitation qualitatively. At times judges would make vague references to an “event” in the case file, chastising probationers to “not let it happen again”. Thus, what judges and probation agents choose to disclose during hearings may not be the only factors that judges consider when deciding to sanction for non-compliance. Finally, the current study examined domestic violence courts in only one jurisdiction in a large, Midwestern county. The factors that emerged as important to judges in evaluating whether to sanction may not generalize to domestic violence courts in other jurisdictions, or even to other judges. Future research should examine whether these same factors are present in other types of specialized courts, as well as in other aspects of probation decision-making found in traditional courts (e.g., revocation). Future research also should examine other locations to determine if the factors that were important to judges in this jurisdiction may be transferable.

5.2. Research Implications

Ulmer (2012) has argued that existing methods of examining racial disparity (i.e., quantitative analysis of secondary data) cannot fully test theories without utilizing additional methods and sources of data to uncover the processes through which attributions of responsibility, dangerousness, or culpability emerge (see also Baumer 2013). These approaches also miss the fact that these “facts” are not created in a vacuum, but are framed, or made sense of, within the context of other factors about the person or offense, such as his or her background, prior experience with the person, demeanor of the person, and discourses, or worldviews that judges bring with them. I took one approach of modeling the framing of issues, which attempts to capture what qualitative methods found—namely that facts are framed in context to other information, and this framing is ultimately what influences decisions. Future research could build upon this idea, refining methods of measuring framing and contextual information for quantitative methods.

The current study utilized participant observation of probation review hearings to capture, both qualitatively and quantitatively, how probationers are talked to and talked about with respect to issues of non-compliance. When studying decision-making in courts, it is imperative to examine not only the outcomes (e.g., sanctions, bail) and factors (e.g., drug use, prior record) but the framing of these issues as well. Without examining the everyday language court actors use in the courtroom, tests of attribution theories will be incomplete (see Mears 1998; Ulmer 2012). Research that incorporates participant observation or additional data sources (e.g., probation or attorney memos) blending quantitative and qualitative methods may bet better suited to test and refine existing sentencing theories, and to determine whether and where disparity exists.