Response to Stephanie A. Duriez, Francis T. Cullen, and Sarah M. Manchak:
Theory and Evidence on the Swift-Certain-Fair Approach to Enforcing Conditions of Community Supervision
by Mark A.R. Kleiman, University of California, Los Angeles (UCLA), Beau Kilmer, RAND Corporation, Daniel T. Fisher, John F. Kennedy School of Government, Harvard University
Swift-certain-fair (SCF) sanctioning improves on conventional practice in enforcing the conditions of community corrections both by substituting swiftness and certainty for severity and by increasing the predictability, and thus the perceived fairness, of the process from the offender’s viewpoint. SCF can also complement, or substitute for, the expensive and laborious process of formal risk-needs assessments in the process of allocating scarce supervisory and service capacity across offenders. SCF has both firm theoretical grounding and a growing body of empirical support as a means of reducing reoffending and the time participants spend behind bars.
EVERY COMMUNITY SUPERVISION program (probation, parole, pretrial release) has rules, and potential sanctions for breaking those rules. Most have loose monitoring and sporadic, unpredictable, but occasionally severe sanctions, including revocation of community supervision leading to sustained incarceration. Theory (Beccaria, 1764; Bentham, 1789; Schelling, 1960; Kleiman & Kilmer, 2009) and evidence (Bryjak & Grasmack, 1980, pp. 471-491; Paternoster, 1989; Nichols & Ross, 1990; Gendreau, 1996, pp. 144-161; Taxman, 1999) strongly support the idea that close monitoring with consistent and proportionate sanctioning works better than random severity. Stephanie A. Duriez, Francis T. Cullen, and Sarah M. Manchak (2014) criticize that idea, which they conflate with a single implementation of it: Hawaii’s HOPE probation.
Swift-certain-fair (SCF) is a set of operating principles for community supervision. HOPE is one instantiation of those principles: neither the first nor the largest, though so far the most intensively studied. The question facing other jurisdictions is not whether to replicate HOPE, but whether and how to implement swift-certain-fair principles in specific community-supervision agencies (Hawken & Kleiman, 2009, p. 49; Pearsall, 2014).
Another approach to allocating attention and services among clients is assess-and-treat (A&T). A&T programs, such as ORAS, conduct an elaborate, time-intensive, and therefore expensive risk-needs (RN) evaluation using self-report and official records in order to identify the overall risk and need levels and specific needs of each subject, and on that basis assign a supervision level and treatment plan (Latessa, Lemke, Lowenkamp, Makarios, & Smith, 2010).
SCF programs also incorporate official-records data in their decision-making, but rely primarily on the offender’s actual behavior under close monitoring, on the principle that the best predictor of future behavior is current behavior. This “behavioral triage” (BT) approach can be a substitute for or a complement to A&T in identifying offenders for intensive supervision and services (Hawken, 2010).
The use of drug-testing with swift and certain sanctions did not start in Honolulu. Operation Tripwire in Washington, DC, reduced rearrests and failure-to-appear rates among pretrial releasees starting in the early 1970s (Crosby, 1971; Dupont & Wish, 1992; Carver, 1993). Project Sentry was supervising probationers in Lansing, Michigan, from the early 1980s (Gallegher, 1996, 1997). In a head-to-head RCT against an A&T-based program in the DC Drug Court experiment run by Adele Harrell, an SCF program demonstrated much better outcomes at much lower cost (Cavanagh & Harrell, 1997; Cavanagh, Harrell, & Roman, 2000). The contemporaneous Project SWIFT, evaluated by Snell in Texas, has had results comparable to those of Hawaii’s HOPE, though it was implemented county-wide and therefore could not be studied using RCT methods (Snell, 2007).
Examples of SCF success are not limited to illegal drugs. South Dakota’s 24/7 Sobriety Program, which uses frequent alcohol monitoring with SCF sanctions, has demonstrated impressive results. Since 2005, more than 25,000 unique individuals have participated in the program, a large number for a state with roughly 650,000 adults (Kilmer & Humphreys, 2013). 24/7 Sobriety has been so successful in South Dakota that it is possible to detect effects of the program at the county level. In a paper published in the American Journal of Public Health, Kilmer et al. (2013) found that after counties adopted 24/7 Sobriety there was a 12 percent reduction in repeat drunk-driving arrests and a 9 percent reduction in domestic violence arrests. The latter result is especially noteworthy, since most participants are not in the program for a domestic violence charge.
Thus the claim that SCF lacks empirical support (Duriez et al., this issue) does not withstand scrutiny. Rather, given the weight of evidence for the idea that properly-executed SCF programs outperform the usual system of sporadic and delayed severity, it seems reasonable to ask both what standard of empirical and theoretical support Duriez et al. think necessary before a program can be called “evidence-based” and how many currently accepted programs, including the assess-and-treat approach based on risk-needs assessment, could actually satisfy that standard.
Neither was the success of properly-implemented SCF programs as surprising as Duriez et al. make it sound; it draws theoretical support not only from the Beccaria-Bentham tradition but also from the research tradition of operant conditioning (Bryjak & Grasmack, 1980; Nichols & Ross, 1990; Rhine, 1993; Gendreau, 1996; Tonry, 1996; Bachman, Brame, Paternoster, & Sherman, 1997; Taxman, 1999; Farabee, 2005). Kleiman and colleagues, following Robert DuPont, Eric Wish, and John Kaplan (Kaplan, 1983; DuPont & Wish, 1992), had laid out the principles of SCF with respect to drug use (under the unfortunate label “coerced abstinence”) in a number of publications years before the launch of the Hawaii program (Kleiman, 1992, pp. 192-1998; Kleiman & Rudolph, 1995, pp. 5-10; Kleiman, 1997; Kleiman, 2003). Duriez et al., portraying HOPE as an outlier without adequate theoretical foundations, cite none of that pre-HOPE work, nor do they mention SWIFT or 24/7.
As the article notes, Angela Hawken evaluated the A&T-based program under California Proposition 36 (Longshore et al., 2006). She reported her initial positive findings as evidence of the program’s apparent success (Magruder, 2007). When data from later years demonstrated the program’s failure, both operationally and in terms of outcomes, she reported those negative findings with equal vigor, much to the distress of the program operators. Duriez et al. also note that the data from Hawaii overcame Hawken’s initial skepticism and convinced her that HOPE was successful. But they are simply wrong to report that Hawken is “uncritical.” She has identified and reported problems with decreasing program fidelity in Honolulu (Hawken & Kleiman, 2012). In public presentations and in discussions with officials, Hawken has actively resisted attempts to label SCF an “evidence-based” program, insisting that more replication research is needed, even though the evidence of efficacy for SCF is much stronger than that for many programs that call themselves “evidence-based.”
More empirical work remains to be done about the psychological mechanisms underlying the demonstrated efficacy of SCF when implemented with fidelity to its underlying principles. Although the early work by Kaplan, DuPont, Wish, and Kleiman (Kaplan, 1983; DuPont & Wish, 1992; Kleiman, 1992) stressed deterrence, interviews with SCF subjects make it clear that the programs also benefit from fairness and transparent goodwill (Hawken & Kleiman, 2009, p. 37), and such procedural-justice effects on offenders’ attitudes and responses to rules should be included in any analysis, as should the therapeutic benefits of sustained desistance from drug-taking in the subject’s normal environment, as opposed to the artificial environment of a prison or residential treatment program. The claim that SCF does not change antisocial attitudes (and therefore must be ineffective) is unsupported by evidence.
Contrary to the assertion in the article by Duriez et al., SCF does not attribute to its subjects rationality in the economic sense of that term. In truth, perfectly rational offenders would respond strongly to the high-severity, low-certainty, deferred threats currently delivered by the criminal justice system in the United States. SCF is offered as an alternative to that “brute-force” approach, and is designed to manage the behavior of individuals who are strongly (irrationally, in economic terms) present-oriented, impulsive, and risk-acceptant (Kleiman, 2009). The consistent and dramatic decreases in violation rates among probationers subject to SCF suggest that Beccaria and Bentham were right where Becker (1968) was wrong.
The relationship between compliance with conditions of community supervision and re-offending need not, as Duriez et al. note, be a direct one, any more than the relationship between receipt of services and re-offending. But routine, unsanctioned violation of conditions—the status quo in much of the system—tends to discredit community corrections and thus encourage legislators, prosecutors, and judges to over-incarcerate (Hawken & Kleiman, 2009, p. 6).
Specifically in the case of drug-taking, logic and empirical studies agree: People who support drug habits by committing crimes do not stop committing crimes while they keep buying expensive drugs (Gropper, 1985, p. 2).
Duriez et al. are correct when they write that the empirical results from SCF programs would not have been predicted by the “correctional treatment” paradigm. It does not follow that those results are incorrect; perhaps the “correctional treatment” approach, with its heavy reliance on risk-needs assessment, requires modification in the light of new evidence.
Duriez et al. are also correct to ask about the post-supervision effects of these programs. Long-term follow-up data from the HOPE RCT and South Dakota’s 24/7 Sobriety Program will soon be released. But if, as appears to be the case, SCF programs reduce drug use, crime, and incarceration while offenders are subject to supervision, that alone can justify their use while the long-run data accumulate.
Duriez et al. cite evidence that the unsuccessful attempt to implement SCF in Delaware’s “Decide Your Time” program (DYT) did not lead to improvements in outcomes (O’Connell et al., 2013). DYT was not the first implementation failure of SCF. Multnomah County’s Structured Sanctions Program and Maryland’s Break the Cycle also had unsatisfactory results, Multnomah County in part because the program design provided for mere warnings as the “sanctions” for the first five violations (Cavanagh & Godfrey, 1995), Maryland because the lack of judicial “buy-in” led to the threatened sanctions not being consistently delivered (Kleiman, 2009, p. 36).
Hawken and Kleiman, both in the publications cited by Duriez et al. and elsewhere, have cited those examples as illustrating the importance of program design and fidelity in generating success or failure for SCF attempts. It is reasonable to argue that swiftness, certainty, and fairness may be beyond the operational reach of some agencies; it is not reasonable to argue that poor outcomes where swiftness, certainty, and fairness are not achieved cast doubt on the validity of the program design. Very few ideas work when not properly implemented.
It is possible that swift-certain-fair is more demanding in practice than assess-and-treat, and that fidelity to plan will be correspondingly lower. It is also possible that SCF, where fairness as reflected in consistency is central, may be more vulnerable to imperfect implementation than is A&T. If so, that suggests the importance—which Hawken has repeatedly stressed—of creating a strong fidelity-assurance component, so that every instance of deviation on the part of officials is evident to supervisors.
Duriez et al. overstate the severity of sanctions under SCF programs. In Hawaii and most other places, a first-time violator who comes in voluntarily and admits responsibility will be confined for as little as two days (Hawken & Kleiman, 2011). The sanction for a positive alcohol test in South Dakota’s Sobriety 24/7 is typically a night in jail (Dupont, Long, & Talpins, 2010, p. 2). In Washington State, the first “sanction” is acceptance of a performance contract. The search for the minimum effective dose of sanctions continues, and Hawken and Kleiman have both emphasized that, when it comes to punishment, “less is more.”
As Duriez et al. note, many studies of sanctions that are not swift, certain, or fair have demonstrated that such sanctions have only limited efficacy in changing behavior. From this they conclude that deterrent threats do not work. But the results from SCF programs suggest instead that badly-designed deterrents do not work, while well-designed deterrents do. The National Research Council report on incarceration (Travis, Western, & Redburn, 2014) notes the “strong evaluation design” of the Hawaii trial (p. 136). In a long footnote, the report points out that “the interpretation that certain but non-draconian punishment can be an effective deterrent is consistent with decades of research on deterrence.” Furthermore, “That such an effect seems to have been found in a population in which deterrence has previously been ineffective in averting crime makes the finding potentially very important.” Therefore, “Research on the deterrent effectiveness of short sentences with high celerity and certainty should be a priority, particularly among crime-prone populations” (Travis, Western, & Redburn, 2014, Fn. 8, pp. 136-7).
The treatment of probation-as-usual in Hawaii by Duriez et al. seems inconsistent. On the one hand, they criticize the HOPE evaluation for comparing HOPE results to probation-as-usual rather than to a program including more of what they consider evidence-based practices. Whether a specific SCF intervention is useful in any specific jurisdiction and institutional setting can be determined by an RCT comparing it with usual practices in that setting. How it would do compared with some other program is a different question.
On the other hand, Duriez et al. point out that probation-as-usual in Hawaii incorporates the evidence-based practices of cognitive-behavioral therapy and motivational interviewing, and speculate that the success of HOPE results from that fact. But the control group in the RCT showing the superior results of HOPE all benefited from CBT- and MI-trained probation officers (Hawken & Kleiman, 2009, p.11).
In all properly-conducted SCF programs so far studied, about half of all subjects never test positive when supervised under SCF (even when the program targets a heavy-using, high-risk caseload), while about one-fifth have three or more instances of detected use, thus demonstrating their need for greater attention. BT can, under some circumstances, be faster and more accurate than A&T in identifying not only high-risk subjects requiring great attention but also low-risk subjects who can safely be given minimal supervision or released from supervision entirely (Hawken, 2010).
The appropriate mix of A&T and BT in any specific circumstance is an empirical issue, not one that can be resolved by abstract reasoning alone. Hawken and Kleiman have been trying for most of a decade now to mount a trial of SCF against an A&T-based drug-court model. An experiment comparing a well-designed SCF program to a high-quality A&T program is long overdue.
Criminal justice practice generally, and correctional practice specifically, have indeed suffered from a series of theoretically unjustified and empirically unsupported fads. They have also suffered from the failure to swiftly recognize and adopt valuable innovations. Given the enormous burden created by over-incarceration, the potential contribution of SCF principles to reducing that burden should not be ignored.
Note: This article is #2 in a chain of academic debate. Follow the debate with the other texts (1, 3, 4) linked below:
1: Is Project HOPE Creating a False Sense of Hope?
3: Before Adopting Project HOPE, Read the Warning Label: A Rejoinder to Kleiman, Kilmer, and Fisher’s Comment
4: “They have learned nothing, and forgotten nothing”