Focused Deterrence Interventions: A Critical Review
Joseph Dule University of New Haven
Boston pioneered a focused deterrence strategy in the 1990s, to reduce high rates of youth and gang violence. Initially, Boston’s “Operation Ceasefire” working group conducted problem analysis. This research revealed approximately 60% of Boston homicides were gang related, and that both the perpetrators and victims of violence were highly criminal—i.e., they possessed criminal records and were typically known (often by-name) by the Boston Police Department (BPD). Accordingly, the intervention aimed to “communicate incentives and disincentives directly to known high risk chronic offenders” (Corsaro & Engel, 2015). To accomplish this, BPD conducted “call in” sessions with chronic offenders—often those who were out on parole—to make them aware that violence would not be tolerated (Kennedy et al., 2001). The BPD message was simple: if you commit violence, then every legal lever possible will be pulled to ensure that you will be held accountable. Further, not only will violence not be tolerated, but if one person in the group commits violence, then “heat” from the BPD will be brought on to the entire group (Kennedy, 2011).
This focused deterrence strategy also included parallel efforts to match the target population of high-risk violent offenders to the moral voices of the community and to social services. It was believed that community members who “bought into” the intervention could positively influence and/or morally sanction the intervention, while social services increased at-risk youths’ access to education, employment opportunities, housing, etc. (Braga, & Weisburd, 2012) Thus, it is important to stress that focused deterrence is ultimately a combination of push (demand a cease of violence) and pull (encouragement from the community, and redirecting youth to alternate pathways through education and employment opportunities). A successful implementation of the focused deterrence strategy therefore will not only require law enforcement to allocate sufficient personnel and resources to focus on the most at-risk groups/individuals, but also may require ongoing efforts to enhance broader police-community relations within targeted areas, with an expansion of social services. Depending on the severity of the problem and the existing policing and social service capacities, additional law enforcement and social service workers and resources may be needed.
This paper critically assesses the effectiveness of focused deterrence strategies implemented within the U.S. criminal justice system. Focused deterrence has shown to be a promising intervention strategy in reducing inner-city individual and group violence, illicit drug dealing, and improving parolee compliance amongst chronic substance abusers. However, much of the research on focused deterrence suffers from a weak research design. Therefore, additional research utilizing randomized controlled trials—the “gold standard” in experimental research—or more rigorous quasi-experimental designs should be commissioned to increase the validity of these findings.
Theoretical Framework and Prior Research
A focused deterrence strategy was pioneered in Boston in the 1990’s, to reduce high rates of youth and gang violence. From a theoretical perspective, focused deterrence stresses all three core tenets of Beccaria’s (2009) general deterrence theory: severity, certainty, and swiftness. However, a greater emphasis is placed on certainty of punishment. This is understandable considering evidence in support of deterrence theory is most consistent with the “certainty” effect (Nagin, 2013). Addressing “certainty” of punishment is accomplished through the “call in” sessions, whereby law enforcement assures at-risk offenders that if their behavior does not change, law enforcement will expend every ounce of their legal authority to make the violence or drug dealing stop (Kennedy, 2011). Put another way, the social contract between at risk offenders and police is made explicit: any deviation from acceptable behavior will result in severe and persistent police scrutiny. Separately, other focused deterrence strategies aiming to increase chronic substance abuser parolee compliance emphasize “certainty” through enforcement of strict drug testing, but also the deterrence tenet of “swiftness,” as violation of parole conditions result in an immediate sanction—i.e., a brief jail stay (Hawken, & Kleiman, 2009).
Results from focused deterrence strategies yield promising results. For example, a meta-analysis completed by Braga and Weisburd (2012) showed that of the 10 focused deterrence strategies examining area outcomes (changes at the neighborhood /city level), the mean effect size was .604. This is consistent with a medium or moderate standardized effect size (Cohen, 1988). However, it is important to emphasize that the research findings showing stronger effect sizes tend to suffer from weaker research designs. Specifically, the five interventions with the highest scoring standardized effect size (Lowell, MA; Indianapolis, IN; Nashville, NT; Stockton, CA; Boston, MA) had scores ranging .645 to 1.186, and all utilized non-equivalent quasi-experimental research designs. Four out of these five were citywide interventions. Nashville, in contrast, compared a neighborhood to the rest of the county, still making it a non-equivalent comparison group (Braga, & Weisburd, 2012).
Citywide interventions may be ideal if the program already has been demonstrably shown to be effective, as it increases the potential for greater crime reduction benefits. There is less need to incorporate comparison groups who will not receive treatment. However, absent improved research designs utilizing equivalent or near equivalent comparison groups, it remains unclear as to whether it was the intervention itself or other factors that generated the reported effect sizes. For instance, a city already may be on trajectory for less violence; there may be other effective preventative policing strategies underway; or there may be potential incarceration effects or other ongoing interventions (i.e. social programs, public health initiatives) which may provide predisposed youth with alternative pathways that move them away from a criminal path.
A common theme regarding why citywide interventions are the norm in focused deterrence interventions pertains to the difficulty in finding “counter-factual” groups (Braga, & Weisburd, 2014). Simply put, groups—like individuals—are unique, and they often engage in nefarious activities to serve different means. Further, there is often not enough of them to do a randomized control trial or quasi-experimental design at the neighborhood or city level, and instead most of the research consistently compares official area crime rates (typically at the city level) with other cities.
Arguably, an even greater problem with conducting a focused deterrence experiment within a city (as opposed to a city-to-city comparison), is what appears to be a lack of support from practitioners and policymakers. Randomized controlled trials are often advocated for by researchers, as in Boston’s seminal “Operation Ceasefire” in the 1990’s (Braga, 2013), and they were recommended in subsequent replications in other cities. For instance, the Rand Corporation was designated as the evaluation agency for a recent NIJ sponsored 12-site drug market intervention, which utilized a focused deterrence framework. The Rand Corporation specifically requested randomization procedures, but none of the twelve sites complied with Rand’s request (Braga, & Weisburd, 2014).
Randomization appears to be often an after-thought, as interventions tend to occur in places that are experiencing severe crime. Law enforcement and policymakers’ simply want to fix this problem and are less concerned with improving the fidelity of a research design. Ultimately, randomization would require that only certain groups get treatment while other groups are left out. The direct benefits to the city then would likely be significantly less, as fewer units would be assigned to receive the treatment intervention. As an elected official, policymaker’s have both a political and ethical obligation to do whatever they can to improve the welfare and safety of all community members residing within their district—particularly if they hope to win reelection in the future. By randomly selecting who receives an intervention and who does not creates a dilemma, which may even lead to public backlash when aggrieved citizens do not understand why their neighborhood is being left behind.
Advocates of problem oriented policing also contend that randomization is unrealistic or even inappropriate for focused deterrence strategies given that the results are generally no more generalizable than with quasi-experimentation. They may have a point, as Weisburd and Braga saliently point out (2014): “we know of no study to date that has actually shown a relationship between study design and external validity.” Put another way, the problems that a community or neighborhood faces are so unique and dependent on local conditions, that the effects of any intervention may not replicate well in other cities because the conditions will be too dissimilar. Problem oriented policing advocates generally argue that individual, group, organizational, community, and social structural conditions in community X are distinct from community Y. Therefore, the tailored intervention for community X will not produce a cookie-cutter type model that can be successfully applied in neighborhood Y. The question then becomes: why go through efforts to randomize if it will not sufficiently improve the external validity of the study, while also affording the city who is implementing the strategy less tangible crime reduction benefits? Critically important here is the fact that randomization would likely lessen overall crime reduction within the city—an effect that is fundamentally at odds with law enforcement and policymaker’s primary goals.
Yet another problematic issue with conducting focused deterrence experimentation is the issue of preserving the Stable Unit Treatment Value Assumption (SUTVA). The SUTVA requires that the treatment or control condition to which a unit is assigned has no impact on the response of another unit (Braga & Weisburd, 2014). Focused deterrence interventions on gangs can easily violate the SUTVA, as gangs tend to be socially connected to one another and could therefore experience vicarious treatment effects without ever sending their members to a “call in” session with law enforcement. In fact, this is exactly what the research suggests. For instance, Braga, Hureau, and Papachristos (2013) used social network analysis data to examine the social connections among the 123 gangs that participated in at least one shooting between 2006 and 2010 in Boston. The post-2007 implementation of Operation Ceasefire in Boston applied treatment to 19 of these gangs. Those 19 gangs were shown to be socially connected to 22 of the remaining 104 gangs. The results of the intervention showed that treated gangs experienced a 36% reduction in gang shootings, while the socially connected (or vicariously treated) gangs experienced a 27.4% reduction. Citywide interventions, of course, do not attempt to preserve the SUTVA, as treatment is directly applied to all relevant at risk offenders.
Within Braga and Weisburd’s (2012) meta-analysis, two of the three interventions utilizing a quasi-experimental design with near-equivalent treatment and control groups (Newark and Los Angeles) did not produce statistically significant effect sizes. The third intervention using such a design, Chicago, did produce a statistically significant effect size; however, it also scored the lowest mean effect size for area outcomes (of all ten interventions) with a score of .181. This is far below the average moderate effect size of .604. Chicago’s “Project Safe Neighborhoods” initiative involved individual repeat offenders who were recently out on parole and moving back into dangerous neighborhoods. These individuals went through “offender notification forums” that stressed increased federal prosecution for gun carrying and lengthy prison sentences for future violations. This study suggests that offender focused deterrence may prove an effective strategy, but such findings cannot be extrapolated to support focused deterrence strategies targeting violent groups, as individual and group behaviors remain mutually distinct.
The singular randomized controlled trial included within Braga and Weisburd’s (2012) meta-analysis was Hawaii Opportunity with Probation Enforcement (HOPE), Honolulu, Hawaii. This initiative was a community supervision strategy for substance-abusing probationers featuring mandatory drug tests and swift and certain sanctions. It is worth noting that HOPE was excluded from generating a mean effect size in the meta-analysis because the outcome measures utilized were not area outcomes like the 10 other studies; instead, HOPE relied on individual outcome measures for its dependent variables. Ultimately, the HOPE program was deemed effective, showing statistically significant reductions in missed probation appointments, positive urine tests for illicit substances, new arrests, probation revocations, and prison-days sentenced for HOPE probationers relative to control probationers (Hawken, & Kleiman, 2009).
Due to HOPE’s reported success, Oleson (2016) reports that at least 160 HOPE-like replications already have taken place within the U.S., and the concept is spreading internationally. A recent quasi-experimental study that examines a Project Hope-like implementation in Washington state finds the strategy to be deemed “effective” (Hamilton, et al., 2016). However, two recent randomized controlled experiments, conducted by O’Connell et al. (2016) in Delaware and Latimore et al. (2016) in four U.S. communities across the country, do not find evidence in support of HOPE’s effectiveness. Thus, HOPE may not be the paradigmatic model for focused deterrence that many have believed it to be, as support for its effectiveness remains empirically divided.
Conclusions and Recommendations
This paper principally recommends three ways to improve the quality of focused deterrence research: 1) randomized controlled trials (to include multi-site clustering); 2) stronger quasi-experimental research designs; and 3) measurements of police legitimacy and community “collective efficacy.” As discussed, there is an absence of randomized controlled trials for focused deterrence research regarding interventions targeting violent groups or offenders, or interventions that aim to eliminate drug markets. Urban drug markets are likely more amenable to randomization than violent gang focused deterrence, as drug markets tend to be concentrated in very small places (Braga, & Weisburd, 2014). In contrast, gangs are more likely to control larger swaths of territory (restricting the total number of control/treatments groups), and may be highly interconnected with other gangs (Braga, Hureau, and Papachristos, 2013) creating challenges for maintaining the integrity of the SUTVA. Similarly, individual focused deterrence in general is more amenable to randomization protocols than group-focused deterrence, as there are more potential treatment and control groups, and the SUTVA is easier to meet. Randomization for group-focused deterrence may benefit from multi-site clustered randomized control trials. This would increase the number of control and treatment groups available for study, and may reduce SUTVA violations as a lower number of groups per site could minimize inter-group social connectedness within a target city (Braga, & Weisburd, 2014). However, this would not likely resolve the previously discussed political and ethical concerns. In fact, such a process may only compound these concerns, as implementation would require more stakeholders to agree to randomization protocols.
If randomization is not feasible, future research also would benefit from improved quasi-experimental designs, whereby treatment and control groups are more closely matched. For example, the Chicago study compared targeted and non-targeted policing districts using propensity score matching, while the Newark study utilized propensity score matching to compare untreated census blocks with targeted census blocks (Braga & Weisburd, 2012). Propensity score analysis approximates a randomized controlled trial by matching subjects on a single score that represents all known factors that may have been associated with being assigned to the program being evaluated (Welsh & Harris, 2016). Propensity score matching (or other advanced statistical matching techniques) should be adopted in future research to better identify equivalent control and treatment groups.
Aside from measuring solely the overall crime reduction effects of focused deterrence interventions, it is also important that future research better explain why such a strategy is working in the first place. Scholars contend that other factors aside from deterrent effects may be equally important. For example, qualitative interviews from two Drug Market Interventions (Nashville and Rockford) showed that police legitimacy increased in response to the interventions (Brunson, 2015). An essential element of the focused deterrence strategy is that at-risk offenders are treated with dignity and respect from law enforcement (Kennedy, 2011). It is therefore not clear that any changes in behavior are ultimately a response to a change in perception from general deterrence, or a belief that they are being treated fairly and procedurally just, which may increase their likelihood of compliance. Focused deterrence also involves creating alternative pathways for at-risk offenders through greater access to social services, and increasing the collective efficacy of communities (Braga & Weisburd, 2012). To what degree do each of these factors interact with each other, and which aspects are more important to the others? This question has not been critically examined. Future research should aim to measure perceptions of police legitimacy, procedural justice, and the collective efficacy of communities. Qualitative interviews (individual and/or community forums) and surveying of targeted neighborhoods may help elucidate how influential these other factors are.
In summary, focused deterrence interventions have shown to be a promising form of crime prevention, although expectations of future success need to be tempered given the overall quality of research designs (whereby the strongest designs tend to net the weakest effect) coupled with contradictory research findings. Accordingly, future research should be commissioned that utilizes either randomization procedures or advanced statistical modeling techniques to create near-equivalent control and treatment groups. The causal mechanisms that might explain why focused deterrence works also remain underexplored, and should therefore be better captured in future research efforts.
Beccaria, C. (2009). On crimes and punishments and other writings. University of Toronto Press.
Braga, Anthony; Kennedy, David; Piehl, Anne. (2001). Reducing Gun Violence: The Boston Gun Project’s Operation Ceasefire. National Institute of Justice Research Report, September, NCJ 188741
Braga, A. A. (2013). Quasi-experimentation when random assignment is not possible: observations from practical experiences in the field. In B. C. Welsh, A. A. Braga, & G.
Braga, A. A., Apel, R., & Welsh, B. C. (2013). The spillover effects of focused deterrence on gang violence. Evaluation review, 37(3-4), 314-342.
Braga, A. A., & Weisburd, D. L. (2012). The effects of focused deterrence strategies on crime: A systematic review and meta-analysis of the empirical evidence. Journal of Research in Crime and Delinquency, 49(3), 323-358.
Braga, A. A., & Weisburd, D. L. (2014). Must we settle for less rigorous evaluations in large area-based crime prevention programs? Lessons from a Campbell review of focused deterrence. Journal of Experimental Criminology, 10(4), 573-597.
Braga, A. A., Kennedy, D. M., Waring, E. J., & Piehl, A. M. (2001). Problem-oriented policing, deterrence, and youth violence: An evaluation of Boston's Operation Ceasefire. Journal of research in crime and delinquency, 38(3), 195-225.
Bruinsma (Eds.), Experimental criminology: prospects for improving science and public policy (pp. 223–252). New York: Cambridge University Press.
Brunson, R. K. (2015). Focused deterrence and improved police–community relations. Criminology & Public Policy, 14(3), 507-514.
Cohen, J. (1988). Statistical Power Analysis for the Behavioral Sciences. 2nd ed. Hillsdale, NJ: Lawrence Erlbaum.
Corsaro, N., & Engel, R. S. (2015). Most challenging of contexts. Criminology & Public Policy, 14(3), 471-505.
Hamilton, Zachary; Campbell, Christopher; Van Wormer, Jaqueline; Kigerl, Alex; & Posey, Brianne. (2016). Impact of swift and certain sanctions: Evaluation of
Washington State’s policy for offenders on community supervision. Criminology & Public Policy, 15: 1009–1072.
Hawken, A., & M. A. R. Kleiman. (2009). Managing Drug Involved Probationers with Swift and Certain Sanctions: Evaluating Hawaii’s HOPE. Final report submitted to the National Institute of Justice, U.S. Department of Justice.
Kennedy, D. M. (2011). Don't shoot: one man, a street fellowship, and the end of violence in inner-city America. Bloomsbury Publishing USA.
Oleson, James C. 2016. HOPE springs eternal: New evaluations of correctional deterrence. Criminology & Public Policy, 15: 1163–1183.
O’Connell, Daniel, John J. Brent, and Christy A. Visher. 2016. Decide Your Time: A randomized trial of a drug testing and graduated sanctions program for probationers. Criminology & Public Policy, 15: 1073–1102.
Welsh, W. N., & Harris, P. W. (2016). Criminal Justice Policy and Planning: Planned Change. Routledge.