Taking Stock of Drug Courts: Do They Work?

ABSTRACT

Although researchers, policymakers, and practitioners alike have long known about the established link between substance abuse and criminal behavior, criminal justice agencies in the United States are still tasked with managing an influx of individuals who display symptoms of abuse and dependence. By the late 1980s, the drug court model emerged as an innovative response to this problem, and this reform has since proliferated to such an extent that it is the most common type of problem-solving court in America. Still, there remains much variation in how drug courts are implemented across jurisdictions, which can have strong implications for the outcomes among the courts’ participants. In this review, we summarize the key research on drug court implementation, followed by an assessment of whether they can be said to “work” in terms of reducing criminal behavior and relapse among adults. We conclude that the model remains an evidence-based practice and suggest some directions for future work, including increased emphasis on theory and causal dynamics and key measurement issues.

KEYWORDS:

• drug courts
• therapeutic jurisprudence
• problem-solving courts
• addiction substance use disorder
• evidence-based practice

How have specialized criminal courts impacted crime and justice in the United States? Over the past three decades, a precipitous shift has occurred in the use of empirical research to inform both criminal justice policy and practice, with a specific focus on “what works” to reduce recidivism and improve offender outcomes (Andrews & Bonta, 2014; Cullen, Myer, & Latessa, 2009; Lowenkamp, Latessa, & Smith, 2006). More so than any other type of specialized court, the creation and implementation of drug courts (hereafter referred to as “DCs”), which were designed to accommodate those with substance abuse issues and ultimately divert them from custodial to community-based supervision, have seen dramatic growth (Green & Winik, 2010; Spohn & Holleran, 2002). Although its implementation varies greatly across jurisdictions (and perhaps even within), the hallmark of the drug court is that it requires the participation of staff members from both the criminal justice system and treatment and social service agencies—all of whom are unified by the goal of improving clinical and criminal outcomes for DC participants. Rooted in a “carrots-and-sticks” approach, the model places greater emphasis on principles of rehabilitation compared to traditional courts, although elements of deterrence, such as sanctions for a failed drug test, are still routinely employed. As such, the American drug court is considered by many to be the paragon of “therapeutic jurisprudence” in action (Winnick, 1997).

The DC model and its promise to fundamentally alter how justice operates has generated national interest since its inception in Dade County, Florida in 1989 (Belenko, 2001). Over the past three decades, the model has evolved and spread such that virtually every state uses DCs in some capacity. Recent reports suggest that there are currently more than 3,100 in operation across the country (Lowenkamp, Holsinger, & Latessa, 2005; Office of Justice Programs, 2018). To a large extent, the proliferation of the DC model can be attributed to the confluence during the latter part of the 20th century of three interrelated events: (a) the enhanced enforcement on drug users, as part of the “War on Drugs,” during a time when criminal justice was already punitive across the board; (b) a renewed and sustained interest in treatment- and community-based correctional philosophies; and (c) the inability of correctional facilities to house and properly treat individuals incarcerated for substance-related offenses (Boldt, 1998; Lowenkamp et al., 2005; Maguire & Pastore, 1999). At the same time, there was widespread public frustration with the way traditional criminal courts processed those with substance abuse issues in a revolving door fashion (Goldkamp, White, & Robinson, 2002). “Business-as-usual” was not working, and thus the time was ripe for a paradigm shift.

Since then, a spate of empirical research has been conducted regarding the effectiveness of DCs—with a focus on whether they work to reduce rates of recidivism and substance abuse (Goldkamp et al., 2002; Gottfredson, Najaka, & Kearley, 2006; Peters & Murrin, 2000) and on which practices or characteristics produce the most beneficial outcomes for offenders and community as a whole (Marlowe, 2010; Marlowe, Festinger, Foltz, Lee, & Patapis, 2005). Given decades of research and the scholarship it has produced, the goal of this article is to “take stock” of the research on drug courts—from their conceptual and theoretical underpinnings to their empirical status—with a specific focus on assessments of adult offender populations. In doing so, we ask the following questions: (a) What does science tell us about the DC model as a process and as a way to decrease offending and substance abuse? And (b) in which directions should they move in the future? As we demonstrate, the future of the DC model is positive: The majority of the literature suggests that they are generally succeeding in their goals to process those with substance abuse issues in ways that may disrupt the cycle of relapse, crime, and reincarceration (Latimer, Morton-Bourgon, & Chrétien, 2006; Marlowe, DeMatteo, & Festinger, 2003; Mitchell, Wilson, Eggers, & MacKenzie, 2012; Shaffer, 2011).

In the following sections, we review the conceptual framework of the DC model, including the setting, its key characters, and critical issues of implementation that directly affect the model’s success. We then take stock of its empirical status by reviewing key studies of the impact of drug courts on adults with substance use disorders, with a focus on knowledge generated through meta-analyses and systematic reviews. We conclude with a discussion on gaps in the literature with respect to research on the fundamental model. Particular attention is given to issues regarding proposed causal mechanisms embedded within the model, including potential issues with measurement and program evaluation.

Conceptual framework and implementation

Prior to reviewing the empirical literature, we review the research on the conceptual foundation and ideal implementation of the model. This discussion is important because the drug court is a complicated system comprised of various actors across several criminal justice and treatment agencies. Moreover, the structure of drug courts varies greatly from courtroom to courtroom. This variation is not without its consequences. The evaluation literature, as a result, is replete with numerous versions of drug courts—some of which, because of their design, may decrease substance use and crime for some but not for others. Goldkamp, White, and Robinson (2002, p. 28) elucidate the problem this poses for scholars and practitioners succinctly:

Without such a framework to isolate the critical instrumental elements of the approach, findings from scattered evaluations will accumulate like apples and oranges and other ingredients for a mixed fruit salad of research. The result is that the practice-oriented consumer of the research is left to weed through diverse findings from disparate studies to identify directions or themes relating to drug court effectiveness.

Variation in the implementation of the DC model can therefore affect the extent to which it can be deemed “effective” (National Association of Drug Court Professionals, 2015). Congruent with the central tenets of other community-based programs in the criminal justice system, an adherence to best practices (i.e., the fidelity principle) is strongly correlated with the model’s success (Latessa & Smith, 2011). We review some of the key points from this literature n the following.

In general, there are three phases of the drug court process: the stabilization phase, the intensive treatment phase, and the transition phase (National Association of Drug Court Professionals, 1997). Stabilization refers to immediate alcohol or drug (AOD) detoxification processes and initial assessment screens for substance dependence disorder and risk of criminal behavior. When a person with AOD issues is arrested, it is often a time of serious crisis. As such, stabilization efforts and entry into the DC must occur quickly (National Association of Drug Court Professionals, 1997). Once stabilization is achieved and admittance to the DC has been secured, the intensive treatment phase begins—the likes of which primarily entail AOD treatment services in conjunction with ancillary services for particular social- or health-related needs (National Association of Drug Court Professionals, 2013). Finally, the transition phase marks the nearing of the completion of the DC, where the participant is prepared for independence from the DC. Efforts to maximize the likelihood of social integration are performed as the participant is untethered from the supervision of the drug court judge and other rehabilitative workers (National Association of Drug Court Professionals, 2015). The following reviews the essential evidence-based components within the drug court phases.

Assessment and treatment provision

Perhaps the most critical element of the stabilization phase (and maybe the DC model as a whole) is assessing individuals and identifying who is appropriate and eligible for this type of therapeutic jurisprudence. Participants should fit the criteria for substance addition/dependence and pose a high risk of recidivism, which is usually determined by their criminal history. Another factor DCs often consider is the likelihood that a less rehabilitative sanction will fail to produce a positive outcome, which again can be assessed based on past sanction failures. Drug courts that can successfully identify these candidates and admit them into the DC process see crime reduction benefits approximately twice as large as those programs that enroll less serious offenders (Cissner et al., 2013; Lowenkamp et al., 2005). This assessment works to enhance public safety while also yielding financial benefits: Programs that are successful in selecting high-risk individuals see about 50% greater cost savings in the public coffers (Bhati, Roman, & Chalfin, 2008; Downey & Roman, 2010). In short, getting the initial selection process “right” is integral to the program’s success.

It is also imperative that DCs have written, objective, and defensible criteria that delineate how screening and eligibility determinations are to be made. Due to substantial variation in the implementation of DCs across jurisdictions, research indicates that several DCs have informal or subjective selection criteria (Belenko, Fabrikant, & Wolff, 2011). Such informal determinations of whether someone is fit for a DC (usually made by a judge or another court administrator, which is also known as a “suitability” assessment) are highly problematic, as research demonstrates that they do not predict graduation from drug courts or recidivism (Rossman et al., 2011). When non-standardized types of filters are used, certain subjects may be improperly excluded from DC treatment. Moreover, exposing other people to intensive treatment may even exacerbate situations—or cause “iatrogenic effects”—for individuals who turn out to be low on risk and need (Lovins, Lowenkamp, Latessa, & Smith, 2007; Lowenkamp & Latessa, 2005; Wexler, Melnick, & Cao, 2004). When subjective decision-making leads to DC environments that inappropriately contain a mixture of individuals with low, medium, and high risk and needs, the lowest risk participants can suffer worsened substance abuse and criminal outcomes. This negative effect may occur because of their exposure to antisocial peers or because the rigorous demands of the program can absorb time that would otherwise be spent doing productive activities such as working, going to school, or engaging with family (DeMatteo, Marlowe, & Festinger, 2006; Lowenkamp & Latessa, 2004; McCord, 2003).

Given that selection into the program is so critical for several outcomes, it is essential to ask: How is it to be achieved? Research consistently demonstrates that standardized assessment instruments have superior reliability and validity compared to the “clinical” judgment of criminal justice professionals (Andrews, Bonta, & Wormith, 2006; Miller & Shutt, 2001) and the determination of who should be placed into a DC is no different. Drug courts that rely on such validated and standardized tools show improved substance abuse and criminal recidivism outcomes compared to those that do not take advantage of these instruments (Shaffer, 2011).

Apart from the use of standardized tools, DCs are defined by rules, policies, or laws that dictate who is qualified for participation in a DC, even if results from assessment instruments indicate otherwise. In general, there are offense type disqualifications and health-related disqualifications (National Association of Drug Court Professionals, 2013). Some DCs disqualify people who have been charged with or have a record of felony convictions, and thus they solely serve those who have been charged with drug use crimes. Those charged with drug dealing offenses can be disqualified as well, even if their dealing is motivated by the need to support a use habit (National Association of Drug Court Professionals, 2013). These exclusions are sometimes the result of stipulations set forth by an agency that may be subsidizing the DC, such as the federal government (Saum & Hiller, 2008). This type of blanket exclusion, however, may not produce the best possible outcomes, as research has shown that courts can reap double the financial benefit when they cater to individuals charged with felony theft and property crimes (Carey, Mackin, & Finigan, 2012). In fact, DCs that exclusively provide treatment to those charged with drug offenses may not see financial benefits in the form of reduced recidivism that can offset the cost of the DC program itself (Downey & Roman, 2010). Regarding violent offenders, the literature is mixed, as some studies have shown that they perform well (Saum & Hiller, 2008; Saum, Scarpitti, & Robbins, 2001) whereas others have shown that they do not (Mitchell et al., 2012; Shaffer, 2011). As in other areas of the model, there exists wide variation in the constraints put on DCs in terms of who has access, thus potentially limiting some of their beneficial impacts.

In the treatment phase, and based on the assessment of the participant’s risk and need, various DC services are available, ideally, and fall within a continuum of care that ranges from outpatient and intensive outpatient therapy to detoxification and residential treatment. Importantly, programs that offer more intensive services such as residential treatment have shown better outcomes for participants whose needs require a certain level of care (Koob, Brocato, & Kleinpeter, 2011). Perhaps the two most common types of substance abuse treatment found in DCs are individual sessions with a therapist or case manager and group sessions. Treatment outcomes have shown to be enhanced when the participant is able to meet one-on-one with a therapist at least once per week, especially during the beginning stages of treatment, when therapeutic engagement is critical (Rossman et al., 2011). Group therapeutic sessions have also been shown to be beneficial and are most effective when they are small (< 12 patients) and have two facilitators (Sobell & Sobell, 2011; Yalom, 2005). An important exception here is that some participants diagnosed with other conditions may not perform well in a group milieu, such as those with serious cognitive impairments, a history of exposure to trauma, or certain mental health conditions (Yalom, 2005). As such, persons who present these issues may be more suitable for specialized treatment. Finally, medication-assisted treatment (MAT) is an evidence-based practice offered in DC programs, and it has been shown to improve conditions for those with opioid addition or alcoholism (Connery, 2015; Roman, Abraham, & Knudsen, 2011). Common medications include: agonists such as methadone, partial agonists such as buprenorphine, and antagonists such as naltrexone. Among offenders recently released from incarceration, those who receive MAT services show increased engagement in treatment and reduced drug use and criminal recidivism (Coviello et al., 2012; Havnes et al., 2012; Kinlock, Gordon, Schwartz, & O’Grady, 2008; Magura et al., 2009). Unfortunately, likely due to a variety of biases and barriers, recent research shows that almost half of American drug courts do not take advantage of MAT therapies (Matusow et al., 2013).

Generally speaking, the longer participants remain in these services, the better their outcomes (Gottfredson, Kearley, & Bushway, 2008; Taxman & Bouffard, 2005). Some research suggests that optimal outcomes are achieved when participants remain in DC treatment for up to one year (Huebner & Cobbina, 2007). In combination with length of treatment, DC outcomes are maximized when, toward the conclusion of the program (during the transition phase), plans vis-à-vis continuing or aftercare services are offered (Carey et al., 2012). These include assistance finding employment and housing, furthering the participant’s education, and services for family or chronic medical issues. Although DCs are primarily concerned with chemical relapse and recidivism, attention should be paid to the participant as a whole, including catering to the pressing needs and stressors in the individual’s life that are important in and of themselves, and are essential for the pursuit of sobriety.

The critical role of judges

Research demonstrates that judges have a large impact on how their drug court operates, and thus service delivery can be punctuated or hampered based on judicial knowledge of drug courts and substance abuse treatment generally (Carey et al., 2012; Jones, 2013; Zweig, Lindquist, Downey, Roman, & Rossman, 2012). This effect is not surprising, given the judge is the most powerful legal actor in the courtroom. Outcomes are influenced by the judicial demeanor toward the participant and the drug court process (Goldkamp et al., 2002; Jones & Kemp, 2014). Akin to the concept of procedural justice and system legitimacy (Tyler, 2003), research indicates when judges are perceived as being fairer and more respectful toward participants, the DCs they manage show primarily positive outcomes (Farole & Cissner, 2007; Senjo & Leip, 2001; Zweig et al., 2012). Conversely, therapeutic outcomes can be worsened when judges display less-than-sympathetic attitudes toward the subjects (Farole & Cissner, 2007; Miethe, Lu, & Reese, 2000). The research is clear: The judge is a powerful moderator of drug court outcomes, and they impact the goals of the DC in critical ways.

Best practices suggest that judges should support ongoing trainings and evaluations of drug courts within their jurisdictions. Data on treatment, relapse, criminal recidivism, and other psycho-social variables should be collected on a continuous basis. Often, this task is most appropriately performed by outside evaluators who can bring objectivity to the situation (National Association of Drug Court Professionals, 2015). Equally important, however, is that drug courts ought to assess themselves on a regular basis, so as not to deviate from ideal implementation over time (Carey et al., 2012). Fortunately, continuous self-monitoring and training can increase the likelihood that fidelity to the model can be maintained over time so as to maximize effectiveness of the model (Taxman & Belenko, 2013). We now turn to these evaluation studies that assess how effective drug courts are in reducing relapse and recidivism.

Key empirical evaluations

Since their inception, the causal mechanisms of DCs thought to influence offender-based outcomes have generated a substantial amount of empirical interest. In fact, more research has been published on the effects of DCs than any other criminal justice program combined (Marlowe, 2010). Over the past three decades, academics and practitioners alike have embraced a scientific approach rooted in evidence-based practice to closely examine the DC model. The following section reviews this body of research, paying special attention not only to the results generated by individual empirical assessments but also to the average effect sizes generated by a host of meta-analyses. Much of the research on DCs focuses on outcomes related to future drug use and crime (i.e., recidivism) as well as cost on savings to the criminal justice system. In general, these studies indicate that DCs produce more favorable outcomes for offenders relative to traditional criminal case processing or probation. As we discuss, however, substantial variability exists in the performance of individual DCs (Listwan, Sundt, Holsinger, & Latessa, 2003; Marlowe, 2010). Moreover, the degree to which DCs are considered effective in achieving their goals is contingent on how various outcomes are operationalized and statistically analyzed.

Although the implementation of DCs began in 1989, the corresponding empirical research regarding their effectiveness did not fully appear for nearly another decade. In 1997, the Government Accounting Office (GAO) released a summary statement concluding that DCs were successful at reducing recidivism—a report that was subsequently corroborated by extensive reviews conducted by Belenko (2001). Indeed, the evaluations of DCs across the country—from Florida (Peters, Haas, & Murrin, 1999) to Pennsylvania (Brewster, 2001) to Nebraska (Spohn, Piper, Martin, & Frenzel, 2001) to Ohio (Listwan et al., 2003) to California (Wolf & Colyer, 2001)—have been extensive, and they indicate that participants tend to have lower rates of recidivism, based on rates of re-arrest, than traditionally adjudicated offenders. Studies also suggest that individuals who successfully graduate from DC programs are significantly less likely to be re-arrested (Dynia & Sung, 2000; Peters et al., 1999), re-convicted (Vito & Tewksbury, 1998) and are more likely to be self-supporting (Sechrest & Shicor, 2001) compared to those who do not successfully complete their DC-based programming.

Perhaps the greatest amount of support for the utility of DCs comes not from individual studies but from the comprehensive meta-analytic reviews that have been conducted. As previously mentioned, substantial variability exists between and among individual DCs, thus necessitating a quantitative synthesis of empirical research to assess the average size of the relationship between (1) the DC model and its proposed mechanisms and (2) future rates of recidivism. As Cullen and Jonson (2016) point out, meta-analyses are akin to “batting averages” in professional baseball where each study is similar to a time up at bat: If a study shows that a variable – such as a measure of drug court graduation – is found to influence recidivism in its analysis, then it is like the batter getting a hit; on the other hand, if a variable does not influence future rates of recidivism, then it is like making an out. Keeping with this analogy, meta-analytic reviews of the DC model essentially determine the batting average for particular variables across all individual studies, where higher averages (“effect sizes”) are indicative of support for the relationship between predictor (i.e., DC interventions) and outcome variables (i.e., recidivism).

To date, several meta-analyses have examined the link between DC interventions and recidivism – most of which support the notion that DCs are effective in reducing future crime and drug use (see Table 1). For example, Mitchell et al. (2012) reviewed 92 independent evaluations of adult drug courts and reported an average drop in recidivism from 50% to 38%, based on a three-year follow-up period. Likewise, Wilson, Mitchell, and Mackenzie’s (2006) analyses of 55 independent drug court-comparison samples, which included both experimental and quasi-experimental designs, yielded effect sizes corresponding with a 14–26% decrease in average rates of recidivism (see also Drake, Aos, & Miller, 2009; Lowenkamp et al., 2005; Shaffer, 2011). There is also evidence to suggest that the relationship between DC interventions and recidivism extends beyond the United States. In a meta-analytic review conducted by the Canadian Department of Justice, which included evaluations based on Canadian, Australian, and American data (n = 66), researchers found that DC participation was associated with a 14% reduction in crime and future drug use (Latimer et al., 2006).

Table 1. Results from meta-analyses of drug courts.

Authors	Number of Evaluations	Recidivism Outcome(s)	Percent Decreased Recidivism
Sevigny et al., 2013	19	Jail; incarceration	8–12
Mitchell et al., 2012	92	Re-arrest; future drug use	12
Shaffer, 2011	60	Re-arrest; re-conviction	9
Latimer et al., 2006	66	Charged with a new criminal offense	14
Wilson et al., 2006	55	Re-arrest; re-conviction	14–26
Drake et al., 2006	57	Re-arrest; re-conviction	8
Lowenkamp et al., 2005	22	Re-arrest; re-conviction	8

Despite the general consensus regarding the effectiveness of DCs on recidivism, empirical findings are not completely uniform. For example, while the focus of this review has been on adult drug courts, the literature on the effectiveness of juvenile drug courts indicates that they have produced unfavorable results (see Stein, Homan, & DeBerard, 2015; Sullivan, Blair, Latessa, & Sullivan, 2016; Tanner-Smith, Lipsey, & Wilson, 2016). Even among adult populations, some research reports null findings regarding differences between DC and non-DC participants. In one of the first empirical DC evaluations ever conducted, Belenko, Fagan, and Dumanovsky (1994) examined official arrest data to compare drug offenders adjudicated in New York City’s fast-track drug courts to those processed through standard means (e.g., prison or probation) over a two-year period (1989–1991). Results from their logistic regression analyses indicated no difference in the likelihood of arrest between drug court participants and comparison group members. Similar patterns were observed by Deschenes and Greenwood (1994), who found no difference in arrest rates between DC and non-DC participants in Maricopa County, California (see also Granfield, Eby, & Brewster, 1998; Listwan et al., 2003). Most recently, a meta-analysis by Sevigny, Fuleihan, and Ferdik (2013) found that while DCs significantly reduced the likelihood of incarceration – corresponding with an 8% reduction for confinement and a 12% reduction for incarceration, respectively – they did not reduce the actual amount of average time spent incarcerated. This finding indicates that the potential benefits of a lower incarceration rate may be offset by longer sentences for offenders who fail their programs.

Finally, some empirical research suggests that DCs can have unintended consequences – if implemented in a certain way – that might actually work to increase recidivism among its participants. A study by Miethe et al. (2000) employed Braithewaite’s (1989) theory of reintegrative shaming as an interpretive framework to examine a sample of felony cases within and outside the DC system in Las Vegas (Clark County District) to compare the relative risk of recidivism across courts. Logistic regression models were estimated to determine magnitude and significance of the net impact of type of court on recidivism and revealed findings contrary to the logic and goals of the DC model. Specifically, overall recidivism rates were about 10% higher for drug-court participants (26%) than the control sample (16%), and overall recidivism risks were about 1.8 times higher for drug court participants than non-drug court participants – net of relevant control variables regarding demographic characteristics and criminal history. Why these untoward outcomes may occur, as we elaborate below, may be related to a general lack of theory-informed practice within drug courts.

Future directions

This project sought to assess the current status of American drug courts. In taking stock of the literature, the evidence shows that the model remains an important evidence-based practice for addressing the challenging cycle of substance use and crime. Of course, outcome and process evaluations will still be needed on a continual basis in order to enhance the model’s crime- and drug use-reducing potential. Although outcome studies on drug courts are abundant and generally suggest that the model works with adult offenders, we contend that, moving forward, research in this area would benefit from two things: (1) leveraging more theory when assessing which causal dynamics of DCs should be credited with bringing about change, and (2) paying attention to some key measurement issues.

Criminological theory and drug court research

Although the evaluation literature on drug courts has received a significant amount of attention since its implementation in Miami three decades ago, many of the studies can be characterized as atheoretical assessments of whether recidivism (or drug use) did or did not happen. Despite some key exceptions, most of this literature describes core drug court functions (i.e., treatment and sanctions) and assesses whether those in the treatment group showed improved outcomes. Often absent from work in this area is a detailed accounting of the mechanisms that may bring about some of the aggregate benefits seen in meta-analytic reviews. Ironically, criminological theory and the insights it brings to the study of offending and drug use have rarely been put to use in the study of drug courts (see, however, Gottfredson et al., 2006; Miethe et al., 2000). And while deterrence theory has been implicated in the model’s use of monitoring and sanctions (Marlowe et al., 2005), numerous other perspectives in the discipline have yet to be fully utilized. Below, we discuss the few studies that have centered on key elements and what causal dynamics they suggest in the context of drug courts, and then we suggest additional, potentially promising theoretical avenues. Given that the drug court literature is defined by variation in implementation and outcomes, more thought in this area is warranted.

In one of the earliest pieces exploring the “black box” of drug courts and which components may be responsible for beneficial outcomes, Goldkamp et al. (2002) examined both treatment factors and sanctions among drug courts in Las Vegas and Portland. They showed that, after accounting for outside factors and attributes of the offender, treatment services mattered for some participants – albeit in an inconsistent manner. Jail sanctions were also significantly related to recidivism. However, these were in the opposite direction as would be anticipated by deterrence theory, with greater sanctions increasing the chances of further crime (Goldkamp et al., 2002). Perhaps this is not surprising in the broader context of treatment alternatives, which generally show more efficacy than deterrence-based measures (Petersilia, 2003).

In a similar study that focused on juvenile drug courts in particular, Long and Sullivan (2017) found that referrals to substance abuse treatment was inversely related to recidivism, and – congruent with prior literature (Gendreau, 1996) – that a higher ratio of incentives or treatment services to sanctions was similarly linked with lower reoffending. In terms of sanctions, they found that drug screens were positively related to recidivism (Long & Sullivan, 2017). Taken together, their findings are similar to those of Goldkamp et al. (2002) in that they underscore the important role of treatment factors in fostering positive behaviors and how deterrence-based measures can have unintended consequences.

Two studies stand out as clear exceptions to the assertion that criminological theory has not been leveraged in the study of drug courts. Miethe, Lu, and Reese’s (2000) evaluation suggested that certain core elements of the DC model – including increasing social embeddedness, (de)certification of behavior, and involvement with family – would bring about decreases in further criminal involvement, which did not happen. Drawing on the insights of reintegrative shaming, they found that many of the DCs in their study employed harsh and stigmatizing tactics such as blaming and embarrassment that brought about further labeling and exclusion – results that highlight the importance of fidelity and implementation in criminal justice settings.

Shortly after, Gottfredson and colleagues (2008) expanded on the literature by examining which theoretical mediators may be responsible for observed declines in recidivism within DCs. They posited that age-graded life course theory (Laub & Sampson, 2003; Sampson & Laub, 2003), with its emphasis on informal social control via bonds to family and to the labor force, would be linked to improvements in behavior among DC participants. In addition, they suggested that principles of fairness, such as those captured in the procedural justice and legitimacy literature (Tyler, 2003), would be associated with decreased rates of recidivism. Despite being theoretically distinct concepts, both perspectives were partially supported: Increases in social control through treatment programs and heightened perceptions of procedural justice through judicial hearings were related to lowered substance use and crime.

In light of these studies, the importance of the work focusing on theory in the evaluation of DCs cannot be overstated. Still, there may be a few untapped perspectives that may shed additional light on why DCs may work, including Cullen’s (1994) theory of social support, the Risk/Needs/Responsivity paradigm (Andrews et al., 2006), and perhaps, given the rigorous time demands of drug court treatment, routine activities approaches to crime (Cohen & Felson, 1979; see also Schaefer, Cullen, & Eck, 2016). The DC scholarship may wish to pivot toward these or other criminological frameworks and integrate them by examining clinical and criminal outcomes as the result of interactions between background, treatment, and environmental factors. During this process, it may be possible to better understand “the black box” and to establish a clearer and more nuanced model that can explain why the intervention works for some people in some places and under some conditions.

Measurement issues

Scientific support for the effectiveness of DCs in reducing recidivism and future drug use is robust and firmly rooted in evidence-based practice. Notwithstanding the small number of studies reporting mixed or contrary results, the DC model appears to significantly reduce recidivism while providing substantial cost savings to the criminal justice system (Marlowe, 2010). Despite the strength and breadth of empirical research, however, it is important at this point to consider the extent to which positive findings may be influenced, at least in part, by the ways in which study outcomes are operationalized – specifically, how recidivism, subsequent drug use, and other important concepts are quantified by researchers.

This issue stems from the fact that, although every DC evaluation is based on some metric of recidivism, the concept itself is very broad. As presented in Table 1, we show that recidivism measures range from technical violations and future drug use (Deschenes & Greenwood, 1994; Mitchell et al., 2012) to re-arrest and re-conviction (Drake et al., 2009; Lowenkamp et al., 2005; Shaffer, 2011; Wilson et al., 2006) to serving time in either jail or prisons (Sevigny et al., 2013), among others. On the one hand, this is not inherently problematic, as the literature clearly indicates that DCs are effective at reducing recidivism regardless of its definition; on the other hand, marked differences exist – both in frequency and severity – in the nature of recidivism.

This begs the question: Are some recidivism outcomes more important or germane to the study of the DC model than others? For instance, Sevigny et al. (2013) note that drugs courts reduce the likelihood of initial incarceration but that “any benefits realized from a lower incarceration rate are offset by the long sentences imposed on participants when they fail the program” (p. 1). Which of the two outcomes is more preferable for proponents of the DC model in this instance: lower rates of incarceration for some or longer and more punitive sentences for others? In the same way, are technical violations and re-arrests synonymous with re-conviction or incarceration? One outcome might result in appearing before a judge before being released back into the community; the other outcome could entail an extensive prison sentence – the likes of which could negatively impact and stigmatize DC participants, especially offenders who are otherwise deemed “low risk” (Lowenkamp & Latessa, 2004). As Miethe et al. (2000) note, certain measures of recidivism – such as court appearance – are conservative measures compared to arrests because parole and probation violations are not considered in court records, and that recidivism rates would therefore be far higher if arrests were used as a measure of repeat offending and the scope of repeat offending was extended to other jurisdictions. Thus, while the title of our paper implies that we are interested in whether or not DCs “work,” we must acknowledge – as others have recently pointed out – that the conclusions drawn about DCs are invariably influenced by how recidivism is operationalized (Ostermann, Salerno, & Hyatt, 2015).

Conclusion

Research on the DC model is both extensive and promising and indicates that, compared to traditional measures such as intensive probation, they are effective in reducing recidivism and drug use. Support for the DC model has come in the form of extensive individual evaluations as well as a number of meta-analyses whose effect sizes are robust, notwithstanding a smaller number of studies that show null or mixed findings. The implications that DC research has for criminal justice policy and reform are therefore tangible, practical, and – most importantly – quantifiable.

Yet despite all that is known about the effectiveness of DCs on reducing future recidivism and drug use, more research is needed. Indeed, the most recent meta-analytic review of which we are currently aware was conducted 5 years ago by Sevigny and colleagues (2013). As Marlowe (2010) points out, the legitimacy of the DC model is in large part due to its explicit endorsement of the scientific method, including best practices and evidence-based practices. If DCs truly “work” to reduce future drug use and recidivism, then continued support for the model should be contingent upon replication studies. In other words, it behooves academics and practitioners alike to know whether the causal mechanisms responsible for reducing recidivism operate similarly in the Northwestern and Southeastern regions of the United States – or whether the results are generalizable to other countries such as Canada.

Unfortunately, this is a practice that occurs all too infrequently in criminal justice research, and social science research more generally. For example, McNeeley and Warner (2015) content analysis of the five most influential journals in criminology shows that replication research is typically devalued in comparison to studies with nuanced findings, as evidenced by the fact that they constitute approximately 2% of the total manuscripts accepted for publication. We are not suggesting that empirical assessments of the DC model are lacking in this regard. Rather, we simply emphasize the need for ongoing research in this area, especially given the fact that various criminal justice policies and practices evolve over time and across different social and political contexts. In doing so, anyone with a vested interested in criminological theory or policy can effectively “take stock” of the merit regarding the DC model – or any topic of interest in criminal justice research.

Disclosure statement

No potential conflict of interest was reported by the authors.