Sentencing Drug Court Failures: Judicial Considerations With Increased Offender Information

Abstract

This study explores the influential predictors of sentence severity within a sample of drug court failures. This sample is unique, in that, judges possess greater amounts of offender information at the time of sentencing, relative to conventional adjudications. Due to defendants’ participation in a drug court program, judges possessed offender program performance information in addition to traditionally assessed criminogenic and offender indicators. Data were collected from 320 individuals who participated, yet failed, in Midwestern adult felony drug court program. Results suggest that, in these instances, judicial sentencing decisions were influenced by case and program performance characteristics. Moreover, judges who oversaw defendants in the program sentenced participant failures more harshly than judges who had no affiliation with the drug court or offender prior to sentencing.

Keywords

sentencing drug courts drug offenders punishment

Introduction

In judicial sentencing decisions, judges consider several factors when assessing offender culpability and likelihood of offender recidivism (Steffensmeier, Ulmer, & Kramer, 1998).

The existing research in criminal sentencing demonstrates judges predominantly take into account legally relevant variables, such as offense severity and criminal history in their decision-making process (Bushway & Piehl, 2001; Koons-Witt, Sevigny, Burrow, & Hester, 2014; Steffensmeier & Demuth, 2006; Ulmer & Johnson, 2004), while extralegal factors have also shown to be influential in final determinations of criminal sanctions (Koons-Witt et al., 2014; Spohn, 2009; Steffensmeier & Demuth, 2006). The consideration of legally irrelevant defendant characteristics, such as age, race, and gender, are thought to be the latent result of limited offender information judges possess at the time of sentencing (Steffensmeier et al., 1998). Couched in causal attribution (Albonetti, 1991) and the focal concerns’ perspective (Steffensmeier et al., 1998), scholars suggest that judges, in part, unintentionally assess defendants’ threat to the community based on these extralegal factors (Doerner & Demuth, 2010; Koons-Witt et al., 2014; Rodriguez, Curry, & Lee, 2006).

Prior research on focal concerns have found that age (Ulmer & Johnson, 2004), gender (Koons-Witt et al., 2014; Spohn, 2009; Steffensmeier & Demuth, 2006; Ulmer & Johnson, 2004), and race/ethnicity (Koons-Witt et al., 2014; Steffensmeier & Demuth, 2006; Ulmer & Johnson, 2004) are often predictors in the sentencing decisions, yet the samples used in these studies do not vary in their posited interaction between judge and defendant. Judges usually have little interaction with defendants prior to sentencing and typically have incomplete information regarding offender backgrounds with the exception of presentence investigation reports (Hoffman, 2000; Wheeler, Weisburd, & Bode, 1982). As a result, judges are forced to make assessments based on limited information regarding defendants’ likelihood of recidivism or threat to the community, leading to illegitimate considerations of offender demographics.

An alternative to traditional models of criminal justice processing are specialty courts, such as drug courts, which have taken a more interactive approach to judicial participation and adjudication. Under this model, offenders are supervised in the community, frequently interact with judges, and in instances of failure, are subject to the full statutory menu of sentencing options (Hora, Schma, & Rosenthal, 1999). An essential component of the drug court model is judicial status hearings, where offenders detail their progress through interactions with drug court judges (Senjo & Leip, 2001). These hearings, along with participant program performance, provide judges with greater information and understanding of offenders and their life circumstances (Hoffman, 2000; Hora et al., 1999). From a theoretical standpoint, judges may be less reliant on a stereotype to assess offenders’ threat to the community, thus limiting the consideration of extralegal factors in sentencing decisions. An increase in offender information available to sentencing judges is not without its potential drawbacks. Critics argue drug court performance information may unjustly lessen the likelihood or disqualify a defendant from traditional probation sentences upon program failure (Hoffman, 2000; Hora et al., 1999). Moreover, drug court judges who also sentence program failures may hold biases against defendants based on information collected during the normal course of drug court operations (Satal, 1998).

To date, there has been no research focusing on sentencing outcomes of drug court participants; therefore, the purpose of this study is to examine the factors influencing judicial decision-making in criminal sentencing with a sample of drug court participant failures. I analyzed sentencing outcomes, both the decision to incarcerate and length of sentence decision, of 320 drug court participants who failed the program to determine the direct influence legal, extralegal, and program performance characteristics have on criminal sentencing. In addition, a measure to capture increased interaction and familiarity a judge has with an offender was included in an attempt to determine its effect on the sentencing decision. Specifically, I focused on (1) what are the individual direct effects legal, extralegal, and program performance characteristics have on sentencing drug court failures? (2) What is the effect increased interaction between a judge and offender has on criminal sentencing?

The current research contributes to the existing literature on criminal sentencing and drug courts in a number of ways. First, this research addresses the gap in the existing literature by analyzing sentencing outcomes among drug court participants. I am not aware of any existing studies focusing on criminal sentencing that uses such a sample. Second, program performance measures were analyzed to assess their impact on judicial decision-making at the sentencing phase. In addition, this research addresses the criticism that drug court judges sentence their failed participants more severely than judges not intimately involved with the drug court program. This study sample includes observations who differed in their sentencing experiences. Two-thirds of the sample were sentenced by their drug court judge, while the remaining observations faced final adjudication from an outside judge who lacked familiarity with offenders being sentenced relative to the drug court judges. This split sample allowed me to examine the effect judicial–offender familiarity had on sentence severity. Finally, this study adds to the research by Holleran and Spohn (2004) by including jail as one of the sentencing categories rather than the dichotomous in/out outcomes often analyzed in prior sentencing studies.

Theoretical Framework

This research is guided by the focal concerns perspective, which posits that judges have three primary concerns when sentencing defendants—blameworthiness, protection of the community, and practical constraints and consequences (Steffensmeier et al., 1998). Unlike prior perspectives that attempt to explain sentencing disparities solely based on offender demographics (Anderson, 1976; Bobo & Hutchings, 1996; Chesney-Lind, 1986; Ulmer & Johnson, 2004), focal concerns posit that judges consider a host of crime-specific and offender characteristics when assessing individual culpability and threat to the community. Yet, this assessment is handicapped due to the limited offender information judges possess at the time of sentencing. Subsequently, judges develop a “perceptual shorthand,” using stereotypes and generalizations to assess individuals’ likelihood of recidivism. Albonetti (1991) described this evaluative process as uncertainty avoidance/causal attribution. The stereotypes considered by judges are causal attributes, linking extralegal offender characteristics to offenders’ inclination toward future criminality. These attributes are assessed within a bounded rationality, where judges base sentencing decisions on prior experiences to generate satisfactory contemporary results, but not necessarily ideal ones. The focal concerns’ perspective suggests that judges will engage in this bounded rationality to best address their concerns of blameworthiness, protections of the community, and practicality of their resource constraints.

The judicial concern of blameworthiness centers on legal elements, such as seriousness of offense and harm caused by the offender. Those convicted of serious crimes can expect severe sentences. Protection of the community involves a judge’s assessment of offender dangerousness and the likelihood of repeat criminality. As there is no way to accurately predict the future, judges consider an individual’s prior behavior (criminal history) in their assessment of dangerousness. Moreover, judge’s may invoke stereotypes that serve to reduce the uncertainty of who is dangerous and who is not. It is the use of stereotypes that lead to disparate sentences based on race, gender, and age. Finally, Steffensmeier and colleagues (1998) posit that judges consider practical constraints and consequences when sentencing offenders. It is within this concern judges may consider particular resources, such as financial expenditure when sentencing offenders. The focal concerns perspective is supported in past research, which has found all three concerns playing a role in the discretionary decision-making of judges (Doerner & Demuth, 2010; Ulmer & Bradley, 2006).

Review of the Literature

Much of the extant literature finds that sentencing decisions are influenced by a combination of legally relevant and extralegal factors (Koons-Witt et al., 2014; Spohn, 2009; Warren, Chiricos, & Bales, 2012). Within traditional adjudication, judges are limited in the offender information they possess. Subsequently, these decisions are reduced to a consideration of offense-specific characteristics, while simultaneously factoring irrelevant offender attributes. However, criminal sentencing of drug court failures provide judges and defendants a unique situation in which judges maintain access to offense and offender characteristics (similar to traditional adjudications), but also possess greater amounts of legitimate and illegitimate offender information which may negatively impact the defendant at sentencing. In these situations, judges possess defendants’ program performance information, which may disqualify (formally or informally) offenders from less onerous community-based supervision programs (Fluellen & Trone, 2000; Hora et al., 1999). Moreover, judicial bias may play a role in cases where the offenders have the same judge in drug court and during the sentencing phase. The relationship between judges and offenders within the drug court model is an integral component but may also present an opportunity for unconscious consideration of additional extralegal factors in the sentencing decision. Thus, in the sentencing of drug court failures, it is important to consider the conventional predictors of sentence severity, along with the additional information garnered through an offender’s drug court participation.

Conventionally Sentenced Offenders

In traditional sentencing situations, judges are often left with only case-specific and offender characteristics to consider, which may result in disparate sentences based on extralegal factors (Steffensmeier et al., 1998). Concerns of the legally irrelevant characteristics are not unwarranted, but much of the existing literature suggests that criminogenic characteristics are often the strongest predictors of sentence severity (Steffensmeier & Demuth, 2001; Steffensmeier et al., 1998; Ulmer & Bradley, 2006). Prior research has repeatedly found a positive relationship between offense type and offense seriousness as predictors of judicial decisions to incarcerate and length of sentence (Brennan & Spohn, 2008; Embry & Lyons, 2012; Ulmer & Bradley, 2006; Warren et al., 2012). Offenders convicted of multiple charges are also sentenced more harshly than a single-count conviction (Rodriguez et al., 2006; Steffensmeier & Demuth, 2006). The influence of an offender’s criminal history is well-founded and generally seen as a legitimate consideration in the sentencing decision (Albonetti, 1991; Engen & Gainey, 2000; Ulmer & Kramer, 1996). As a primary concern of judges, offenders’ criminal history is the best indicator of future criminality (Steffensmeier et al., 1998). Not surprisingly, past findings report the greater the criminal past, the more severe the current sentence (Bushway & Piehl, 2001; Holleran & Spohn, 2004; Steffensmeier & Demuth, 2006; Ulmer & Johnson, 2004). The influence of these criminogenic characteristics in sentencing decisions is expected, thus these factors are often positioned as control variables rather than the primary focus of research. It is the illegitimate influence of offender demographics that is often the centerpiece in sentencing research (Doerner & Demuth, 2010; Koons-Witt et al., 2014).

Legally irrelevant, or extralegal, predictors in sentencing research have been attributed to judicial use of stereotypes in their assessment of offenders’ dangerousness (Steffensmeier et al., 1998). Offender demographic information—gender, race, and age—has frequently been found to play an influential role in the administration of sentences (Brennan & Spohn, 2008; Bushway & Piehl, 2001; Doerner & Demuth, 2010; Engen & Gainey, 2000; Koons-Witt et al., 2014; Rodriguez et al., 2006; Spohn & Beichner, 2000; Spohn & Sample, 2013). Whether females are perceived as less dangerous than males (Steffensmeier et al., 1998) or the criminal justice system is acting paternalistically (Chesney-Lind, 1986), females are overwhelmingly found to be sentenced less severely than similarly situated males (Koons-Witt et al., 2014; Rodriguez et al., 2006; Spohn & Beichner, 2000). For example, Doerner and Demuth (2010) found that females were 42% less likely to be sentenced to prison, and in instances where they were sentenced to prison, they served an average of 25% less time than males. Past findings are less definitive in regard to race and age (Engen & Gainey, 2000; Spohn, 2009; Wu & Spohn, 2009).

Research on the direct effects of race and age on sentencing has been less consistent than findings concerning gender (Brennan, 2006; Bushway & Piehl, 2001; Spohn, 2009; Spohn & Sample, 2013). Prior works have found that minority offenders are sentenced more harshly than Whites. Studies show that minorities are more likely to be incarcerated (Brennan & Spohn, 2008; Koons-Witt et al., 2014; Spohn & Beichner, 2000; Steffensmeier & Demuth, 2006; Ulmer & Johnson, 2004; Warren et al., 2012) and are sentenced to longer periods of time (Doerner & Demuth, 2010; Rodriguez et al., 2006; Steffensmeier et al., 1998). But, however, some research finds that the effect of race disappears when controlling for other variables (Brennan, 2006; Engen & Gainey, 2000; Spohn, 2009). Similarly, inconsistent results are also found with age and sentencing outcomes (Bushway & Piehl, 2001; Doerner & Demuth, 2010; Koons-Witt et al., 2014; Rodriguez et al., 2006; Steffensmeier et al., 1998; Wu & Spohn, 2009). Steffensmeier and colleagues describe an inverted “U” effect, where the youngest and oldest offenders are treated more leniently than offenders in their 20s and 30s (Steffensmeier & Demuth, 2000; Steffensmeier et al., 1998), yet a meta-analytical review of 60 studies concluded that age had no effect on sentence severity (Wu & Spohn, 2009). Moreover, prior research has found that the effect of gender, race, and age is often conditioned by one another, resulting in more nuanced conclusions regarding the effects of extralegal characteristics on sentencing (Doerner & Demuth, 2010; Holleran & Spohn, 2004; Koons-Witt et al., 2014; Spohn, 2009; Spohn & Beichner, 2000; Steffensmeier & Demuth, 2006; Steffensmeier et al., 1998).

Sentencing Drug Court Failures

Judges possess a great deal of more information on offenders who participated, but failed, in a drug court program. Primarily, these judges are made aware of the offender’s program performance beyond the fact that these individuals were unsuccessful. Although program performance measures have not been examined in previous sentencing literature, these factors are prevalent in drug court research. Previous drug court literature suggests program performance measures are legitimate sources of information for judges to consider at sentencing (Banks & Gottfredson, 2003; Goldkamp, White, & Robinson, 2001; Krebs, Lindquist, Koetse, & Lattimore, 2007; Shaffer, 2011). Prior recidivism studies have found that certain drug court performance measures are predictive of future criminality (Banks & Gottfredson, 2003; Goldkamp et al., 2001; Krebs et al., 2007; Shaffer, 2011). For example, offenders who incur a program violation within the first 30 days of the program are significantly more likely to recidivate than those who remain compliant in the program (Krebs et al., 2007). Moreover, Goldkamp and colleagues (2001) found that the number of violations resulting in jail sanctions was predictive of future participant criminality (Goldkamp et al., 2001). From a focal concerns perspective, the consideration of one’s drug court performance appears to be legally relevant and may be a way for judges to protect the community. Critics, however, contend that those who are identified as drug court failures are unfairly sentenced more harshly than conventionally adjudicated offenders and that the severity of sentences are an inherent product of the additional information judges possess (Boldt, 2010; Hoffman, 2000).

Judges may be reluctant to sentence these offenders to probation, as failing a drug court program may be viewed as a strong predictor of future performance in alternative community-based supervision programs (Fluellen & Trone, 2000; O’Keefe & Rempel, 2006). However, some argue that drug court performance is an unfair indicator of future community-programming outcomes because drug court programs are far more cumbersome than traditional forms of community-based corrections (Hora et al., 1999). Judicial considerations may extend beyond these points of criticism, especially in situations where the sentencing and drug court judges are the same.

Drug court judges glean insight into the offender through drug court status hearings. Through progress reports and one-on-one interactions, these hearings allow judges to become more familiar with offenders than traditional courts (Burns & Peyrot, 2003; Hoffman, 2000; Hora et al., 1999; Satal, 1998). This judicial interaction has generally been perceived as beneficial to participants (Burns & Peyrot, 2003; Goldkamp, 2002; McIvor, 2009; Satal, 1998). Judges are considered influential in offenders maintaining compliance of program rules (McIvor, 2009) and offenders’ continued involvement in substance abuse treatment (Goldkamp, 2002). However, some suggest that increased judicial insight may cause biases that can negatively impact sentencing upon program failure (Hoffman, 2000; Orr et al., 2009). Judges may unconsciously hold negative feelings when faced with participants’ failed treatment opportunities (Satal, 1998), or more simply, be offended that individuals failed drug court programming (Orr et al., 2009).

Critics’ contention that drug court failures are sentenced more harshly than conventionally adjudicated offenders is not unfounded (Fluellen & Trone, 2000; O’Keefe & Rempel, 2006). An evaluation of the Staten Island, NY, drug court reported that drug court failures were significantly more likely to be incarcerated and serve nearly 6 months longer sentences than traditionally sentenced offenders (O’Keefe & Rempel, 2006). Unfortunately, research is limited on this topic beyond this particular study. The influence program performance measures have on sentencing is unknown, in addition to the potential biases of drug court judges when sentencing program failures.

In sum, previous sentencing literature has provided significant insight into the predictive influence legal and extralegal characteristics have on judicial sentencing decisions. Moreover, past drug court studies have found relationships between program performance measures and future criminality. The current study looks to build upon previous research by examining offenders’ legal, extralegal, and performance characteristics and their effects on sentence severity within a drug court population, while also exploring potential bias that may exist in the sentencing of drug court failures. The review of past research above informs the current study and its four hypotheses.

Research Expectations

The review of the existing literature suggests that offenders’ criminogenic characteristics, such as charge of conviction, the number of counts convicted, and criminal history, will be influential in the severity of sentence judges are likely to levy. Thus,

Hypothesis 1 (H1): Offenders convicted of an income-generating crime, with a greater number of charges, and a more extensive criminal history, will possess a greater probability of being sentenced more harshly than their counterparts.

Specifically, those with greater criminogenic characteristics will possess a greater probability of being sentenced to prison than either probation or jail and serve longer periods of time when sentenced to a period of incarceration. Moreover, those with greater criminogenic characteristics will possess an increased probability of being sentenced to jail relative to probation.

Hypothesis 2 (H2): counter to the existing literature, extralegal offender characteristics, such as race, age, and gender, will play no influential role in the sentencing decision.

Judges will be less reliant on a perceptual shorthand using stereotypes due to their past interaction with these offenders and/or information detailing offenders’ drug court performance. As a result, it is expected that offenders’ previous drug court performance will influence judges sentencing decision.

Hypothesis 3 (H3): offenders who demonstrated a lack of compliance with the drug court program will be sentenced more harshly than their counterparts.

Those with a greater number of positive alcohol and/or illicit drug screens and total program violations will see an increased probability of being sentenced to prison relative to jail and probation, and when incarcerated, will serve longer periods of time. Moreover, those failures who have a greater noncompliance record will have an increased probability of being sentenced to jail over probation. The probability of being sentenced to prison will also be increased for individuals who are convicted of a crime during their participation in the drug court program and/or are absconders at the time of their termination. Finally,

Hypothesis 4 (H4): drug court failures who are sentenced by their drug court judge will experience an increased probability of a prison or jail sentence and will be sentenced for longer periods of incarceration than those sentenced by a non-affiliated judge.

Data and Methods

The data for this study derive from a Midwestern metropolitan adult felony drug court serving a county of approximately 555,000 residents. One of the requirements of the program is that eligible offenders who want to participate in the program are required to plead guilty to their charge(s) of record before entering the program. Their plea is held in abeyance until offenders complete their program, either successfully or unsuccessfully. In instances where an offender graduates the program, their charge(s) is dismissed. Conversely, those who fail the program are immediately convicted of their charge(s) and sentenced by their original judge of record. The state in which this drug court operates maintains an indeterminate sentencing framework, leaving the judges with a great deal of discretion in their sentencing decisions.

Sample

This study examined judicial sentencing patterns of individuals who were unsuccessful in an adult felony drug court from 2008 to 2013. A total of 824 individuals participated in the program during this period. Approximately 61% (n = 504) successfully completed the requirements of the drug court, leaving 320 observations for this study. Sample demographics and program performance information were collected from the state’s web-based management information system and hard copy files archived within the drug court agency. Criminogenic information including sentencing information was provided by the program administrator.

Dependent Variables

Criminal sentence

Criminal sentencing is often examined as a two-stage process, where a judge first makes the decision to incarcerate or not to incarcerate (in/out decision) criminal defendants (Wheeler et al., 1982) and secondarily decides on the length of sentence for those incarcerated. To examine the in/out decision, the categorical variable—criminal sentence—was positioned as the dependent variable. This measure includes three possible dispositions, prison (0 = reference category), jail (1), and probation (2) (see Table 1).¹ The available data allowed for a more nuanced examination of the sentencing decision. In lieu of measuring the sentencing decision dichotomously, which has often been used as the dependent variable in previous research, jail was included as a separate sentence category. As Holleran and Spohn (2004) argued, there exists a difference between jail and prison and should remain distinct within sentence categories. Practically, jail represents an incarceration sentence less severe than prison, as it is often used for shorter periods of confinement. The statutory maximum jail sentences that could be served by the sample is 12 months. A little over half of the observations (n = 165) were sentenced to prison for their original charge (see Table 1), while the remaining offenders were split between serving time in jail (n = 88) and probation (n = 67). The categorical nature of the dependent variable and the use of the two-part model (TPM) strategy necessitates analyzing the first stage of the sentencing process with a probit model (Bushway, Johnson, & Slocum, 2007; Koons-Witt et al., 2014). In instances of more than two categories within the dependent variable, as the current study possesses, a multinomial probit regression is appropriate (Bushway et al., 2007; Long & Freese, 2006).

Table 1.

Sample Characteristics.

N = 320	Frequency	%	M	SD
Dependent variable
Disposition
Prison	165	51.56	—	—
Jail	88	27.50
Probation	67	20.94
Sentence length (in months)			15.29	15.89
Independent variables: offender/criminogenic characteristics
Age of offender	—	—	30.09	9.28
Sex
Male	214	66.88	—	—
Female	106	33.13	—	—
Race
White	204	63.75	—	—
Non-White	116	36.25	—	—
Convicted charge
Possession	211	65.94	—	—
Property	50	15.63
Distribution	59	18.44
Convicted counts			1.225	0.5306
Criminal history	—	—	2.572	2.814
Independent variables: program performance measures
Violations
Positive tests	—	—	1.71	1.55
Total violations	—	—	6.81	5.45
Crime conviction during participation
Yes	93	70.94	—	—
No	227	29.06	—	—
Absconder at time of termination
Yes	165	51.56	—	—
No	155	48.44	—	—
Judicial familiarity
Yes	225	70.31
No	95	29.69

Sentence length

The second dependent variable was a count variable calculated in months. The indeterminate sentencing framework of this state allows judges to sentence defendants to a range (e.g., 12-24 months) of incarceration or a precise time period. Sentence length was measured as the minimum sentence within the time range levied. In instances where a defendant was convicted of two or more charges, a judge had the discretion to run the sentences concurrently or consecutively. To calculate the sentence length for consecutive sentences, the sum of the minimum time levied for each convicted charge was calculated as the length of sentence. In addition, I subtracted jail time credit earned by the defendant from the official minimum sentence to measure sentence length. For those sentenced to a period of confinement (n = 253)—jail or prison—the average length of incarceration was just more than 15 months (SD = 15.89).

In 11 instances, a judge sentenced an offender to jail in accordance with the individual’s earned jail time credit, leaving an observed sentence length of zero days. For example, one observation was sentenced to 63 days in jail and credited with 63 days, which prompted their immediate release from all criminal justice supervision. To maintain the integrity of these decisions, an ordinary least squares regression with a logged dependent variable could not be completed (Doerner & Demuth, 2010; Koons-Witt et al., 2014). Subsequently, an analysis of the subsample of those sentenced to jail and prison was undertaken using negative binomial regression. This method is appropriate as the outcome variable—sentence length—possesses a Poisson distribution but demonstrates over dispersion between the mean and the variance (Long & Freese, 2006).

Independent Variables

Offender and criminogenic characteristics

The mean age of the sample was 30 years (SD = 9.28 years), with one-third being female (0 = male; 1 = female; N = 106; Table 1). Due to the relative homogeneity of offenders, the race variable was dichotomized—White and non-White (0 = White; 1 = non-White)—with nearly 64% (n = 204) of the sample categorized as Caucasian (Table 1).

The criminogenic characteristics included participants’ charge of conviction, number of counts convicted, and criminal history. Like most drug courts, this program had eligibility requirements such as the charge of record must be alcohol- or drug-related. Moreover, persons charged with a violent offense are ineligible to participate in the program. These requirements resulted in a homogeny of charges across participants. The convicted charge was categorized into three attributes—possession offense, property offense, and drug distribution offense.² The majority of the offenders were convicted for a felonious possession charge (n = 211, 66%), where only 59 (18%) participants were charged with distribution and the remaining 50 (15%) offenders were charged with a property offense. Due to a lack of variance across the three criminal sentence categories, property offense and drug distribution offense were combined creating a dichotomous variable, Possession (0) and Income Crime (1).

A portion of offenders in the sample pled guilty to more than one charge before their participation in drug court, resulting in being sentenced for multiple charges upon their program failure. The sample was convicted and sentenced for an average of 1.23 offenses (SD = .5306). Finally, criminal history was measured as a continuous variable. Criminally sentenced participants averaged 2.6 convictions—misdemeanors and felonies—before and during their drug court participation.³

Program performance measures

As addressed above, judges are able to rely on a greater amount of information when sentencing former drug court participants. With this consideration, it is believed program performance measures, beyond the fact that one failed the program, would be particularly salient to judges. Four performance measures (Positive alcohol/drug screens, total program violations, criminal convictions during participation, and absconder status) were included to capture the variation in program noncompliance of individual participants. Crime conviction during participation was measured dichotomously (yes = 1; no = 0) with 29% (n = 93) of offenders being convicted of a crime, either misdemeanor or felony during their participation with drug court.⁴ In addition, more than half of those who failed the program were considered “absconders” (yes = 1; no = 0) at the time or leading up to their termination from the program. An absconder is an individual who failed to report to the drug court or status hearings, resulting in a warrant being issued for their arrest. These measures represent an offender’s unwillingness to continue their participation and compliance with the program. In this particular program, not all absconders are terminated. Some were given an additional opportunity to comply with the program. Performance or program noncompliance was additionally measured by two continuous variables. On average, the sample yielded 1.71 (SD = 1.55) positive alcohol or other drug (AOD) use screens during their participation. Program failures were also sanctioned (which includes positive drug screens), on average, for 6.81 (SD = 5.45) violations throughout their participation.

Judicial familiarity

A measure indicating the increased interaction between judge and offender was included in this analysis. This particular variable attempts to capture any potential sentence disparities between those offenders sentenced by their drug court judge and those who were sentenced by an unaffiliated judge. Judicial familiarity is a dichotomous categorical predictor (yes = 0; no = 1) measuring whether or not the sentencing judge also acted as the defendant’s drug court judge. The district (county-level) court in which this program is based has 16 judges serving on the bench. Felony cases are randomly assigned to these judges in a rotation. Of the 16 judges, 4 of them also serve the drug court program. Judges who officiate the drug court program and sentence program failures assigned to their court, presumably, possess greater information or familiarity with these offenders than a judge who did not serve within the drug court program. Seventy percent (n = 225) of drug court failures were sentenced by their drug court judge, while 30% (n = 95) were sentenced by a judge not involved with the drug court program.⁵

Statistical Diagnostics

Potential collinearity issues were examined through Pearson’s correlation matrix and variance inflation factors (VIFs). The highest VIF was 1.37 with an average of 1.17. Particular attention was paid to statistical power and cell population or events per predictive variable (EPV). A series of post hoc power analyses were run using G*Power 3.1. All models were found to possess adequate power exceeding .80 at an alpha of .05. Moreover, there existed sufficient numbers within each cell to satisfy suggested EPVs (Peduzzi, Concato, Kemper, Holford, & Feinstein, 1996; Vittinghoff & McCulloch, 2006).

Results

The Two-Stage Sentencing Process

The two-stage analyses using multinomial probit and negative binomial regressions sought to answer the study’s primary research question, what are the individual effects legal, extralegal, and program performance characteristics have on sentencing drug court failures (Table 2). The multinomial probit regression addresses the judges’ discretionary decision to sentence individuals to prison, jail, or probation. The negative binomial regression addresses the second stage of sentencing process, the length of sentence imposed. Cumulatively, both models test H1–H3. Results from the multinomial probit model are reported in average marginal effects for greater clarity in interpretation. Unlike reported odds ratios in logit models, results from probit models cannot be converted into odds and the coefficients are not directly interpretable. Similarly, coefficients produced by a negative binomial regression are not clearly interpretable; thus, the direct effects on length of sentence are reported in incidence-rate ratios (IRR). Finally, results from a propensity score matching (PSM) analysis are reported in an effort to determine the difference in sentence severity for those sentenced by their drug court judge and those who were not (Table 4).

Table 2.

Multinomial Probit and Negative Binomial Regression Results.

Dependent variable	Multinomial probit			Negative binomial
	Sentence type			Sentence length
	Prison	Jail	Probation	β	IRR
	dy/dx	dy/dx	dy/dx	β	IRR
Offender age	−.0055827(.0030082)	.0055839* (.0025751)	−1.17e–06(.0024703)	−0.0035107(.0069506)	0.9964954(.0069263)
Female	−.0389122(.0556425)	−.0486024(.0486456)	0.0875145(.0451315)	0.0863383(.1233115)	1.90175(.1344311)
Non-White	.0261398(.0573302)	.007682(.0513087)	−0.0338218(.048524)	0.0265326(.121996)	1.026888(.1252762)
Income crime conviction	.2077335** (.0551948)	−.2790528** (.051182)	0.0713193(.0444788)	0.6203456** (.1218185)	1.858571** (.2265301)
Convicted counts	.0598597(.0569612)	−.1538569* (.0613939)	0.0939972* (.0398635)	0.2872785* (.1086891)	1.332795* (.1448603)
Criminal history	.0304582* (.0102425)	−.0122616(.0094177)	−0.0181966(.0097804)	−0.0032803(.01931)	0.9967251(.0192468)
Crime during participation	.1605676* (.0576242)	−.1059435* (.0541142)	−0.0546242(.0512709)	0.4304858** (.1216643)	1.538005** (.1871203)
Absconder	.1410942* (.0521636)	−.0605078(.0470007)	−0.0805864(.0443389)	0.1721691(.1179544)	1.187879(.1401156)
Positive drug screens	.038042* (.0193697)	−.0509196* (.0182119)	0.0128776(.0159553)	0.0812152(.0427437)	1.084604(.0463599)
Total violations	−.0014488(.0056566)	.0100974* (.0048117)	−0.0086485(.0048537)	−0.0120093(.0126752)	0.9880625(.0125239)
Judicial unfamiliarity	NR	NR	NR	NR	NR
Constant	—	—	—	1.869925** (.2897346)	6.487813** (1.879744)

Note. Standard errors are in parentheses. NR = not reported.

p < .05. **p < .001.

Sentence type

H1 posits offenders convicted of an income-generating crime, with a greater number of charges, and a more extensive criminal history, will possess a greater probability of being sentenced more harshly than their counterparts. Results partially support this hypothesis. The probability for being sentenced to prison is 21 percentage points higher for those who were convicted of income crimes than those convicted of possession crimes (Table 2). This finding is intuitive, as income crimes are the more serious of the two charge categories. With this in mind, there was an additional expectation that those convicted of an income crime would possess a greater probability of receiving a jail sentence or a significantly decreased probability of a probation sentence. Yet, this was not the case. Results suggest that those convicted of an income crime possess a decreased probability of receiving a jail sentence (−.279), where the convicted charge had no effect on whether or not a participant was sentenced to probation. A similar pattern emerged within the direct effect of convicted counts. Unexpectedly, for every additional convicted charge, the probability of a probation sentence increased by nearly 10%, while the probability of a jail term decreased by 15%. If one were to consider a jail sentence a more severe punishment than probation, then opposite results would be expected. Moreover, the number of charges one was convicted had no predictive effect on prison sentences. Participants’ criminal history played little influential role in judges’ sentencing decision. Aligned with H1, for every one prior criminal conviction an offender increased their probability of being sentenced to prison by 3%.

H2 posits extralegal offender characteristics, such as race, age, and gender will play no influential role in the sentencing decision. The reported average marginal effects are supportive of this hypothesis. Neither gender nor race played a role in the type of sentence levied. Age of participants was significantly related to a jail sentence, but only marginally so. As participant failures increased in age, their probability of receiving a jail sentence increased by .05%. Participation in the drug court model may lessen the impact extralegal factors have on sentencing, but one’s involvement in the program provides judges with greater amounts of legally relevant information.

At the time of sentencing, if not before, sentencing judges are made aware of the offenders’ performance while in the drug court program. Offenders’ level of compliance (beyond failure) or type of noncompliance may influence Judges’ sentencing decisions. It is for this reason it was expected (H3) offenders who demonstrated greater noncompliance with the drug court program will be sentenced more harshly than their counterparts experience a greater probability of being sentenced to either jail or prison. All four measures—crime during participation, absconder at the time of termination, the number of positive AOD screens, and the total number of program violations—were significantly influential in the sentencing decision, but not necessarily in the expected direction. An increased probability of a prison sentence is exhibited in instances where a participant was convicted of crime during their participation (.161) or was an absconder at the time of their termination (.141). In addition, a 3% increase in the probability of a prison term was experienced for every positive AOD screen. However, predictors of a local confinement punishment act against expectations. The probability of a jail sentence decreased by 11 percentage points in instances when a participant was convicted of crime during program participation; yet had no significant impact on receiving probation. In addition, for every positive AOD screen, the probability of jail decreased by 5%, while the probability of a local confinement sentence increased by just 1% for every violation accumulated by the participant. In regard to judges’ decision to sentence individuals to one of the three sentence types, expectations of H1-H3 were met within the prison category. Greater criminogenic factors, along with less program compliance, increased the probability would receive a prison term. The more nuanced and unexpected results were found between the jail and probation categories. The findings may suggest that a jail sentence, rather than a probation sentence, was considered to be the more lenient punishment.

Length of sentence

Predictors of sentence length were examined through a negative binomial regression and were reported in IRR for the purposes of interpretation (Table 2). Beyond sentence-type comparisons, H1 and H3 predict that those convicted of more serious charges, with a greater number of charges, with a more extensive criminal history, and possess a greater record of program noncompliance will be sentenced to longer periods of incarceration. As predicted, offenders convicted of an income crime and with multiple counts were sentenced more harshly than their counterparts. Program failures convicted of an income crime were sentenced to 86% longer periods of incarceration than offenders convicted of a possession crime. Moreover, for each additional convicted count, individuals’ sentence length increased by 33%. These findings align themselves with previous research and the focal concerns’ perspective, where greater perceived culpability increases the severity of punishment. Against expectations, criminal history had no impact on the length of sentence. Of the four program performance measures, only one measure played an influential role in the sentence length decision. Offenders convicted of a crime during their program participation were sentenced to a period of incarceration 53% longer than those who remained crime free.

Counter to the existing literature, there was no expectation that offender demographics would play a role in the sentencing decision (H2). Again, it was believed that the frequent judicial–offender interaction through status hearings would nullify the use of stereotypes in assessing one’s danger to the community. As a whole, this hypothesis was confirmed in the multinomial probit analysis. This expectation is more completely realized in terms of the sentence length decision. None of the extralegal variables—age, gender, and race—demonstrated a significant influence in judges’ determination of length of sentence.

The Impact of Judicial Familiarity—PSM

H4 was tested using PSM. PSM was used for this analysis as it is reported to provide greater sensitivity in capturing an isolated treatment effect than multivariate analyses (Guo & Fraser, 2010; Harding, 2003). This approach calculates propensity scores by predicting an observation’s inclusion in the treatment group using relevant covariates in a multivariate probit model (Guo & Fraser, 2010; Rosenbaum & Rubin, 1983). Observations are then matched according to their calculated propensity scores between the treatment and control groups. The determination of an observed treatment effect is then calculated through an appropriate post-matching analysis.

The 10 independent variables included in the multivariate analyses above were chosen to fit the propensity score model. These variables included the three criminogenic characteristics, three personal characteristics, and the four program performance measures. The 11th predictor—judicial familiarity—was positioned as the treatment variable. Due to the disparate sizes between the two groups (judicial familiarity = 225; no judicial familiarity = 95), the less populated group—no judicial familiarity—was positioned as the treatment variable, while the opposite was considered the control. The—psmatch2—command in STATA 15.1 was used to carry out nearest-neighbor, one-to-one matching (Leuven & Sianesi, 2003). Calipers were set at the suggested distance of .20 of the standard deviation of the propensity score (SD=.1,293,878; Austin, 2011; Rubin, 2001). Balance between the two groups was assessed through a series of t-tests generated through the—pstest—command in Stata 15.1. As reported in Table 3, the treatment and control groups were imbalanced within the non-White, crime during participation, and total violations covariates. Similar imbalance was found when only considering the incarcerated subsample. Balance was achieved between the two groups upon the PSM process.

Table 3.

Group Comparisons Between Unmatched versus Matched Samples.

Criminal sentence	Unmatched samples (n = 320)			Matched samples (n = 169)			% reduction bias
	M		t-test	M		t-test
	Treated	Control	t-test	Treated	Control	t-test
Age	28.642	30.702	−1.82	29.373	28.974	0.24	80.6
Female	0.32632	0.33333	−0.12	0.33735	0.34211	−0.05	32.2
Non-White	0.47368	0.31556	2.71*	0.42169	0.47368	−0.53	67.1
Income crime convicted	0.32632	0.34667	0.35	0.33735	0.36842	0.33	−52.7
Convicted counts	1.1895	1.24	−0.78	1.2048	1.3158	−1.01	−119.6
Criminal history	2.8211	2.4667	1.03	2.8193	2.8947	−0.11	78.7
Positive drug screens	1.8526	1.6578	1.02	1.7229	1.3684	1.24	−81.9
Crime during participation	0.37895	0.25333	2.27*	0.31325	0.34211	−0.31	77.0
Absconder	0.53684	0.50667	0.49	0.54217	0.57895	−0.37	−21.9
Total violations	8.3684	6.1511	3.38*	7.1325	7.8421	−0.60	68.0
Sentence length	Unmatched samples (n = 253)			Matched samples (n = 125)			% reduction bias
	M		t-test	M		t-test
	Treated	Control	t-test	Treated	Control	t-test
Age	29.397	30.422	−0.83	30.016	29.148	0.41	15.3
Female	0.31507	0.28889	0.41	0.30645	0.40741	−0.92	−285.6
Non-White	0.47945	0.33333	2.19*	0.45161	0.44444	0.06	95.1
Income crime convicted	0.31507	0.32778	−0.19	0.32258	0.18519	1.32	−981.1
Convicted counts	1.1644	1.944	−0.43	1.1774	1.2593	−0.70	−172.2
Criminal history	3.1644	2.5944	1.39	3.0968	3.3333	−0.27	58.5
Positive drug screens	1.9315	1.65	1.30	1.7258	1.9259	−0.54	28.9
Crime during Participation	0.39726	0.27778	1.86	0.32258	0.48148	−1.43	−33.0
Absconder	0.60274	0.53889	0.92	0.58065	0.62963	−0.43	23.3
Total violations	9.0959	6.3	3.69*	7.4032	9.8148	−1.78	13.7

Denotes significant difference between means at p < .05.

A multinomial probit regression analysis was, again, used to examine the effect judicial familiarity had on sentence type. The treatment group, no judicial familiarity, was positioned as the sole predictor in the model, regressed upon the categorical dependent variable sentence type. The results indicate that offenders sentenced by a judge who was not their drug court judge possessed a decreased probability of receiving a prison sentence, but an increased probability of being sentenced to local confinement (Table 4). The average marginal effect (AME) suggests that no judicial familiarity results in a 19 percentage point decrease in the probability that an offender will be sentenced to prison. The AME also reports that offenders not sentenced by their drug court judge experienced a 12 percentage point increase in the probability of a jail sentence. Judicial assignment did not increase or decrease the probability a failed participant would receive a probation sentence.

Table 4.

Propensity Score Matching Analysis.

Multinomial probit model (criminal sentence)
N = 169	Prison	Jail	Probation
N = 169	dy/dx	dy/dx	dy/dx
No judicial familiarity	−.187414* (.0706079)	.1243936* (.0626945)	.0653478(.0618621)
Difference in mean (sentence length)
N = 125	No judicial continuity	Judicial continuity	Difference	Standard error	T-stat
Average treatment effect	15.8774193	17.1548385	−1.27741919	2.65571126	−.48

Note. Standard errors are in parentheses.

p < .05. **p < .001.

Similar procedures were undertaken to determine an existing difference in sentence length between the treatment and control groups. A subsample of all observations sentenced to either prison or jail was included, leaving a subsample size of 253. A t-test was completed to determine the difference in mean sentence lengths across the matched groups. The PSM analysis yielded no difference in sentence lengths between those who were sentenced by their drug court judge and those who were not. Overall, H4 was only partially confirmed. Results of the PSM suggest drug court judges sentenced program failures more harshly than unaffiliated judges. Unaffiliated judges decreased the probability of a prison sentence and were more inclined to sentence offenders to local confinement. However, judicial assignment played no influential role in the length of incarceration.

Discussion and Conclusion

The current study sought to fill in a gap in the drug court and sentencing literature by examining the criminal sentences levied against drug court failures. Specifically, conventional legal and extralegal variables, along with program participant measures were analyzed to determine their predictive roles in sentencing decisions. Moreover, the effect of judicial–offender familiarity was explored in an effort to discover how judges’ increased knowledge of offenders may impact the severity of sentence levied. The sample was made up of 320 observations, all of which failed an adult felony drug court program. The overwhelming majority of defendants were sentenced to a period of incarceration. Just over half of the sample was remanded to a state penal institution, while 28% were incarcerated in the local county jail. Due to their failure in drug court, it is unsurprising that only 21% of the observed defendants received probation as their criminal sentence.

Separate multivariate analyses were used to examine the effect demographic, criminogenic, and program performance measures had across sentence type (multinomial probit regression) and sentence length (negative binomial regression). The results were assessed through the prism of the focal concerns perspective, where judges are thought to be primarily concerned with offenders’ blameworthiness, protection of the community, and practical constraints and consequences. Consistent with this perspective, it was hypothesized that offenders convicted of the more serious charge category (income crimes) with a greater number of counts would experience a greater probability of a prison or jail sentence, relative to probation. Moreover, those failed participants who had a greater record of program noncompliance would be less likely to be sentenced to probation. Contrary to previous tests of focal concerns, it was expected that offender demographics would have no influential impact on the sentencing decision due to the unconventional adjudication of the sample examined. As a whole, these expectations were realized through the analysis.

The guiding theory suggests that blameworthiness is akin to offender culpability, measured in the present study through the criminogenic characteristics offense type, number of counts, and criminal history. The results were consistent with prior research, the focal concerns perspective, and confirmed the first hypothesis. The more serious offense type, income crime, and increased convicted counts resulted in a greater probability of receiving a prison sentence, while also predicting significantly longer periods of incarceration. Despite the non-traditional adjudication, these judges still based much of their decision-making on legally relevant characteristics and conventional measures of blameworthiness. Those offenders assessed to be most culpable in their offending received the most severe punishments. Yet, the findings were less predictable in the assessment of jail and probation sentences.

Prior research informed current research expectations that a prison sentence was the most severe punishment followed by jail and probation. Evidence supporting this assumption can be found in the coding decisions of past studies, where prison and jail dispositions have been combined to create a total “incarceration” measure (Doerner & Demuth, 2010; Koons-Witt et al., 2014; Steffensmeier & Demuth, 2006). Moreover, studies that separated jail and prison sentences have found that the greater culpability measures were predictive of a jail sanction over probation sanction (Holleran & Spohn, 2004; Warren et al., 2012). Yet, the current findings, when limiting the focus to jail and probation, suggest that jail was treated as the least severe sanction. Those convicted of income crimes experienced a decreased probability of 28 percentage points in receiving a jail sentence. In addition, for every additional count conviction, the probability of a local confinement sentence decreased by 15%. Conversely, a null effect was found regarding the relationship between the charge of conviction and a probation sentence, while every additional convicted count increased the probability of a probation sentence by 9%. These results suggest that the greater culpability measures, the less likely an offender would be sentenced to jail. The same pattern did not hold true for probation sentences, as an increase in counts convicted predicted an increased probability one would be administered a term of probation. There existed an expectation the opposite would be true, assuming a jail sentence is a more severe sanction than a term of probation. This notion is further bolstered by the raw data. In 11 instances, judges levied jail sentences that corresponded to the jail time credit earned, immediately freeing the offender from all criminal justice supervision.

Judicial use of local confinement, as the least severe consequence, may be a product of the uniquely situated sample and alternative concerns of protection of the community and practical constraints and consequences. These judges were provided program performance information that could be used to evaluate one’s perceived danger to the community. Despite the fact all current observations failed the drug court program, their performance was not uniform. This study sought to determine how participants’ performance ultimately factored into judges sentencing decision. Crime during participation, absconder status, and positive AOD (alcohol and other drugs) screens were all positively associated with a prison sentence. Similarly, these performance attributes were significantly related to a decreased probability of a jail term. Program performance, as an indicator of offender dangerousness, appears to have provided judges legally relevant information in which to assess offenders’ threat to the community. These offenders failed to comply with a community-based supervision program, continued their criminal activity, and maintained their substance use. For these reasons, judges may have felt that incapacitation was the most effective way in which to protect the community. However, a secondary explanation may focus on judges’ consideration of resource expenditure.

When sentencing drug court failures, judges may consider the significant resources the court and county already expended toward the rehabilitation of these offenders. Judicial concerns of resource allocation may have led them to believe that additional spent resources would not further benefit these offenders or the community as a whole. Although beyond the scope of this study, this particular concern may explain the relative sentence severity of drug court failures compared to conventionally adjudicated defendants. Many of these judges knew how much time and money were spent on individual participants. In instances of failure, judges may have felt that there was limited utility in expending more community resources. As a result, drug court failures were more likely to be sentenced to prison than receive another period of community-based supervision. At the same time, judges may have believed that certain offenders and offenses were not deserving of a prison term; thus, a period of local confinement was considered more appropriate. In addition, these judges may have felt that, in certain instances, some offenders would not benefit further from probation, while a prison term was not the most effective expenditure of resources.

In addition, the program performance of offenders may have provided judges with adequate amounts of information to significantly reduce their reliance on legally irrelevant offender characteristics. As hypothesized (H3), extralegal offender demographic characteristics would be less influential than has been found in past sentencing studies (Brennan & Spohn, 2008; Koons-Witt et al., 2014). The results largely confirmed this assertion. Neither offenders’ sex nor race was significantly predictive of the judges’ sentencing decision, and age was found to only be minimally influential of a jail sentence.

The final analysis addressed the criticism that judicial familiarity of an offender can bias judges at the sentencing phase (Hoffman, 2000; Satal, 1998)—a criticism supported by the current findings. The PSM analysis demonstrated that participants who were not sentenced by their drug court judge experienced a 19 percentage point decreased probability of receiving a prison term; yet, a 12 percentage point increased probability of a jail sentence. There are several potential explanations for this finding. Satal (1998) suggests a phenomenon of judicial counter-transference, where personal interaction between judge and offender may trigger unconscious negative reactions of participant failure. Judicial status hearings, an integral component of the drug court model, provide judges an opportunity to personally interact with the participant (Hora et al., 1999). The judges are keenly aware of defendants’ transgressions and the resources spent for rehabilitation and understand the time and effort they have been personally spent in the process. They have made encouragements, behavioral modification efforts through intermittent sanctioning, and other accommodations to motivate offenders’ success. In instances of failure, it is not surprising that drug court judges were less amenable to the idea that further supervision under probation may be more fruitful in its returns.

The findings also indicate that sentencing disparities do exist based on a legally irrelevant factor. The illegitimacy of inequitable punishment due to random judicial assignment is not a new complaint (Johnson, 2006). Judicial discretion in sentencing allows for arbiters to levy punishments they personally deem appropriate within the confines of a criminal code, yet the reasons for doing so may be problematic. Drug court judges’ personal involvement in the program may unconsciously bias them against the defendant, or perhaps, the explanation is as simplistic as judges being offended at the mere failure of the program (Burns & Peyrot, 2003). For example, the state of Oklahoma has legally recognized the possibility of bias in these situations by allowing defendants to request judicial recusal (Alexander v. State of Oklahoma, 2002). This appellate ruling potentially addresses the harshness in failures’ sentences, where policies in other jurisdictions may have eliminated the disparate punishments.

In certain jurisdictions, the sentence in instances of program failure is built into the process and known by the defendant (Boldt, 2010; Hoffman, 2000). For example, in the Bronx (New York) program, defendants agree to a specific sentence before entering the program (Quinn, 2001). In other jurisdictions, there are mandatory prison terms for program failures (Hoffman, 2000). In 2010, the Midwestern drug court used in the current study implemented a policy in which drug court judges are always the sentencing judges. Based on the current findings, it appears the policies adopted by these jurisdictions will assist in the alleviation of sentence disparity based on judicial assignment.

Although it is believed the findings of this study make a valuable contribution to the existing literature, there are some limitations that should be noted. First, I caution against the generalization of these results to a larger population of drug court participants as the data are derived from one Midwestern drug court and may not be representative of other programs across the country. Second, there must be an acknowledgment of the limitations of agency data (Maxfield & Babbie, 2014). Much of the data collected for this study was done so by drug court professionals and not done with research as its primary purpose. Third, the sample size is small relative to other sentencing studies and is an inevitable by-product of using a subsample of drug court participants. As noted previously, there exists substantial statistical power for the examination of direct effects, but inadequate numbers for interaction terms. Much of the existing sentencing research focuses attention on moderating effects, combining offender demographic characteristics (Koons-Witt et al., 2014; Steffensmeier & Demuth, 2006; Steffensmeier et al., 1998). For example, Steffensmeier and colleagues (1998) analyzed the interaction between race, gender, and age and concluded that Black male youths are perceived to be the most dangerous class of offenders. In the current study, however, there exist insufficient EPVs in cells which prohibited an analysis of subpopulations. Finally, the data lack a sample of defendants who did not participate in the drug court program (control group). A subsample of conventionally adjudicated offenders would have allowed for counterfactual analysis comparing the sentences of drug court failures and traditionally sentenced offenders.

Overall, this study supports the focal concerns perspective, in that judicial considerations of offender culpability, protection of the community, and practical constraints and consequences remain primary concerns when sentencing drug court failures. The current study supports previous research, in that criminogenic characteristics remain the strongest predictors of sentence severity. Unlike studies that analyzed conventionally adjudicated offenders, this study found that extralegal offender characteristics had little influence in the sentencing decision. Judges sentencing drug court failures considered participants’ program performance, rather than relying on stereotypes to assess offenders’ threat to the community. Finally, the findings lend some credence to critics who argue that judicial roles within the drug court model may bias the same judges in their role at sentencing. Future research should include qualitative studies which explore the disparate sentences and assess the potential sentence severity differences between drug court failures and conventionally adjudicated offenders.

Footnotes

Declaration of Conflicting Interests

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

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

Notes

Author Biography

Benjamin R. Gibbs is an assistant professor in the Department of Criminal Justice and Criminology at Ball State University. His current research includes sentencing, institutional and community corrections, and sex offenders.

References

Albonetti

(1991). An integration of theories to explain judicial discretion. Social Problems, 38, 247-266.

Alexander v. State of Oklahoma, 48 P. 3d 110 (No. F-2000-472; Ok. Ct. App. 2002)

Anderson

E. A.

(1976). The “chivalrous” treatment of the female offender in the arms of the criminal justice system: A review of the literature. Social Problems, 23, 350-357.

Austin

P. C.

(2011). Optimal caliper widths for propensity-score matching when estimating differences in means and differences in proportions in observational studies. Pharmaceutical Statistics, 10, 150-161.

Banks

Gottfredson

D. C.

(2003). The effects of drug treatment and supervision on time to rearrest among drug treatment court participants. Journal of Drug Issues, 33, 385-412.

Bobo

Hutchings

V. L.

(1996). Perceptions of racial group competition: Extending Blumer’s theory of group position to a multiracial social context. American Sociological Review, 61, 951-972.

Boldt

R. C.

(2010). The “Tomahawk” and the “Healing Balm”: Drug treatment courts in theory and practice. University of Maryland Law Journal of Race, Religion, Gender, and Class, 10(1), 45-70.

Brennan

P. K.

(2006). Sentencing female misdemeanants: An examination of the direct and indirect effects of race/ethnicity. Justice Quarterly, 23, 60-95.

Brennan

P. K.

Spohn

(2008). Race/ethnicity and sentencing outcomes among drug offenders in North Carolina. Journal of Contemporary Criminal Justice, 24, 371-398.

10.

Burns

S. L.

Peyrot

(2003). Tough love: Nurturing and coercing responsibility and recovery in California drug courts. Social Problems, 50, 416-438.

11.

Bushway

S. D.

Johnson

B. D.

Slocum

L. A.

(2007). Is the magic still there? The use of the Heckman two-step correction for selection bias in criminology. Journal of Quantitative Criminology, 23, 151-178.

12.

Bushway

S. D.

Piehl

A. M.

(2001). Judging judicial discretion: Legal factors and racial discrimination in sentencing. Law & Society Review, 35, 733-764.

13.

Chesney-Lind

(1986). “Women and crime”: The female offender. Signs, 12, 78-96.

14.

Doerner

J. K.

Demuth

(2010). The independent and joint effects of race/ethnicity, gender, and age on sentencing outcomes in U.S. Federal Courts. Justice Quarterly, 27, 1-27.

15.

Embry

Lyons

P. M.

(2012). Sex-based sentencing: Sentencing discrepancies between male and female sex offenders. Feminist Criminology, 7, 146-162.

16.

Engen

R. L.

Gainey

R. R.

(2000). Modeling the effects of legally relevant and extralegal factors under sentencing guidelines: The rules have changed. Criminology, 38, 1207-1230.

17.

Fluellen

Trone

(2000). Do drug courts save jail and prison beds? New York, NY: The Vera Institute.

18.

Goldkamp

J. S.

(2002). An honest chance: Perspectives on drug courts. Federal Sentencing Report, 14, 369-372.

19.

Goldkamp

J. S.

White

M. D.

Robinson

J. B.

(2001). Do drug courts work? Getting inside the drug court black box. Journal of Drug Issues, 31, 27-72.

20.

Guo

Fraser

M. W.

(2010). Propensity score analysis: Statistical methods and applications. Thousand Oaks, CA: Sage.

21.

Harding

D. J.

(2003). Counterfactuals models of neighborhood effects: The effect of neighborhood poverty on dropping out and teenage pregnancy. American Journal of Sociology, 109, 676-719.

22.

Hoffman

M. B.

(2000). The drug court scandal. North Carolina Law Review, 78, 1439-1534.

23.

Holleran

Spohn

(2004). On the use of the total incarceration variable in sentencing research. Criminology, 42, 211-239.

24.

Hora

P. F.

Schma

W. G.

Rosenthal

T. A.

(1999). Therapeutic jurisprudence and the drug treatment court movement: Revolutionizing the criminal justice system’s response to drug abuse and crime in America. Notre Dame Law Review, 74, 439-538.

25.

Johnson

B. D.

(2006). The multilevel context of criminal sentencing: Integrating judge-and county-level influences. Criminology, 44, 259-298.

26.

Koons-Witt

B. A.

Sevigny

E. L.

Burrow

J. D.

Hester

(2014). Gender and sentencing outcomes in South Carolina: Examining the interactions with race, age, and offender type. Criminal Justice Policy Review, 25, 299-324.

27.

Krebs

Lindquist

Koetse

Lattimore

P. K.

(2007). Assessing the long-term impact of drug court participation on recidivism with generalized estimating equations. Drug & Alcohol Dependence, 91(1), 57-68.

28.

Leuven

Sianesi

(2003). Stata module to perform full Mahalanobis and propensity score matching, common support graphing, and covariate imbalance testing (Statistical Software Components S432001). Chestnut Hill, MA: Department of Economics, Boston College.

29.

Long

J. S.

Freese

(2006). Regression models for categorical dependent variables using Stata. College Station, TX: Stata Press.

30.

Maxfield

M. G.

Babbie

E. R.

(2014). Research methods for criminal justice and criminology. Stamford, CT: Cengage Learning.

31.

McIvor

(2009). Therapeutic jurisprudence and procedural justice in Scottish drug courts. Criminology & Criminal Justice, 9, 29-49.

32.

O’Keefe

Rempel

(2006). The Staten Island treatment court evaluation: Planning, implementation, and impacts. New York, NY: Center of Court Innovation.

33.

Orr

C. H.

Hall

J. W.

Reimer

N. L.

Mallett

E. A.

O’Dowd

Frazer

A. C.

(2009). American’s problem solving courts: The criminal costs of treatment and the case for reform. Washington, DC: National Association of Criminal Defense Lawyers.

34.

Peduzzi

Concato

Kemper

Holford

T. R.

Feinstein

A. R.

(1996). A simulation study of the number of events per variable in logistic regression analysis. Journal of Clinical Epidemiology, 49, 1373-1379.

35.

Quinn

M. C.

(2001). Whose team am I on anyway? Musings of a public defender about drug treatment court practice. New York University Review of Law & Social Change, 26(37), 37-75.

36.

Rodriguez

S. F.

Curry

T. R.

Lee

(2006). Gender differences in criminal sentencing: Do effects vary across violent, property, and drug offenses. Social Science Quarterly, 87, 318-339.

37.

Rosenbaum

P. R.

Rubin

D. B.

(1983). The central role of the propensity score in observational studies for causal effects. Biometrika, 70(1), 41-55.

38.

Rubin

D. B.

(2001). Using propensity scores to help design observational studies: Application to the tobacco litigation. Health Services and Outcomes Research Methodology, 2, 169-188.

39.

Satal

S. L.

(1998). Observational study of courtroom dynamics in selected drug courts. National Drug Court Institute, 1(1), 56-87.

40.

Senjo

S. R.

Leip

L. A.

(2001). Testing and developing theory in drug court: A four-part logit model to predict program completion. Criminal Justice Policy Review, 12, 66-87.

41.

Shaffer

D. K.

(2011). Looking inside the black box of drug courts: A meta-analytic review. Justice Quarterly, 28, 493-519.

42.

Spohn

(2009). Race, sex, and pretrial detention in federal court: Indirect effects and cumulative disadvantage. Kansas Law Review, 57, 879-901.

43.

Spohn

Beichner

(2000). Is preferential treatment of female offenders a thing of the past? A multisite study of gender, race, and imprisonment. Criminal Justice Policy Review, 11, 149-184.

44.

Spohn

Sample

(2013). The dangerous drug offender in federal court: Intersections of race, ethnicity, and culpability. Crime & Delinquency, 59, 3-31.

45.

Steffensmeier

Demuth

(2000). Ethnicity and sentencing outcomes in U.S. Federal Courts: Who is punished more harshly? American Sociological Review, 65, 705-729.

46.

Steffensmeier

Demuth

(2001). Ethnicity and judges’ sentencing decisions: Hispanic-Black-White comparisons. Criminology, 39, 145-178.

47.

Steffensmeier

Demuth

(2006). Does gender modify the effects of race-ethnicity on criminal sanctioning? Sentences for male and female, White, Black, and Hispanic defendants. Journal of Quantitative Criminology, 22, 241-261.

48.

Steffensmeier

Ulmer

Kramer

(1998). The interaction of race, gender, and age in criminal sentencing: The punishment cost of being young, Black, and Male. Criminology, 36, 763-797.

49.

Ulmer

J. T.

Bradley

M. S.

(2006). Variation in trail penalties among serious violent offenses. Criminology, 44, 631-670.

50.

Ulmer

J. T.

Johnson

(2004). Sentencing in context: A multilevel analysis. Criminology, 42, 137-177.

51.

Ulmer

J. T.

Kramer

J. H.

(1996). Court communities under sentencing guidelines: Dilemmas of formal rationality and sentencing disparity. Criminology, 34, 383-408.

52.

Vittinghoff

McCulloch

C. E.

(2006). Relaxing the rule of ten events per variable in logistic and Cox regression. American Journal of Epidemiology, 165, 710-718.

53.

Warren

Chiricos

Bales

(2012). The imprisonment penalty for young Black and Hispanic males: A crime-specific analysis. Journal of Research in Crime & Delinquency, 49, 56-80.

54.

Wheeler

Weisburd

Bode

(1982). Sentencing the white-collar offender: Rhetoric and reality. American Sociological Review, 47, 641-659.

55.

Spohn

(2009). Does an offender’s age have an effect on sentence length? Criminal Justice Policy Review, 20, 379-413.