Building Evidence for Grassroots Community Violence Prevention
Katheryne Pugliese, Paul Odér, Talib Hudson, and Jeffrey A. Butts
Research and Evaluation Center
John Jay College of Criminal Justice
City University of New York
The crime and justice field recently started to label a wide array of violence prevention strategies as Community Violence Interventions (or CVI). Many of these strategies depend on law enforcement and social services, but the most innovative approaches are community-centered and community-sourced. They are grassroots efforts that rely on the resources of neighborhoods and residents themselves, operating separately from law enforcement and traditional human services. These strategies could be called Community Violence Interventions at the Roots (or CVI-R). The most established CVI-R programs are Cure Violence and Advance Peace. They offer highly localized and potentially cost-effective approaches to public safety, but do they work? Evaluation evidence is recent and not yet consistent, but the grassroots approach to community violence prevention is highly promising. To build sustainable CVI-R models, communities and researchers must collaborate in designing rigorous evaluations to produce reliable and actionable evidence.
The violence prevention field embraced a new name in recent years. Community Violence Interventions (or CVIs) are strategies designed to prevent and reduce community violence( 1 ) using varying approaches. The CVI name is used increasingly by philanthropies, nonprofit organizations, state and local governments, federal agencies, and even the White House.
1. “Community violence” refers to the type of violence neighborhood residents may experience while going about daily activities. It does not include domestic, intimate partner, or intra-family violence. It also does not include all forms of gun violence. Mass shootings and suicides are usually not included in the definition of community violence.
In 2021, the Biden Administration announced the White House Community Violence Intervention Collaborative, a group of sixteen jurisdictions using federal funds to build or strengthen their local “CVI infrastructure.” The White House previewed the investment by issuing a statement describing CVI programs as “effective” because they:
“leverage trusted messengers who work directly with individuals most likely to commit gun violence, intervene in conflicts, and connect people to social, health and wellness, and economic services to reduce the likelihood of violence as an answer to conflict”
(White House Fact Sheet 2021).
Attention from the White House helps to advance discussions of violence prevention. However, the new CVI label covers a very broad range of strategies. Are some of these strategies more effective than others? Is police involvement essential for CVI? Some CVI programs rely on traditional social services and therapeutic interventions. Is that required? Is direct interruption of ongoing conflicts fundamental? Is street outreach to connect with participants a key ingredient? What is a credible messenger and why does this come up so often in conversations about CVI?
Research has begun to answer some of these questions, but evidence is still developing. If one of these components turns out to be indispensable, do we even know how to measure whether programs are doing it correctly? If the best answers to such questions are, “you’ll know it when you see it” or “we can’t explain it, but we can show you,” the field of CVI is a long way from becoming “evidence-based.”
Do CVIs have to provide concrete assistance, such as employment and housing, or is it sufficient to form strong relationships with residents and use those ties to help them rethink their attitudes about violence? Perhaps community organizing is the essential component in CVI strategies.
The level of intervention is another central issue. Do CVI programs achieve more when they intervene at the level of individuals? In other words, is community violence prevented most effectively by working with one person at a time or can programs also intervene at the level of neighborhoods and communities with less attention to individual residents?
Researchers are just starting to investigate these issues. Studies demonstrate that some CVI strategies offer promising results, but even the most celebrated programs do not yet qualify as “evidence-based.” In other words, it is not yet possible to use the findings of research to identify and implement the most effective strategies while rejecting or reforming ineffective approaches. The nascent quality of research evidence is at least partly due to the unique challenges CVI strategies present for evaluation research.
Building strong research evidence for the CVI approach is difficult because many different strategies and outcomes are included. Evaluating CVI is also politically complicated. Unlike the consistent support provided to law enforcement, elected officials must be convinced that non-policing approaches to violence prevention are effective. Common-sense appeals or political rhetoric are not enough. Advocates for CVI must be willing to answer tough questions about their methods and results. Communities must invest in rigorous research to identify CVI effects if the strategies are to become key elements of public safety.
Building evidence for CVI must be done systematically by asking key questions. Do CVIs reduce violence and improve safety? Can researchers say precisely how they work so that effective practices can be expanded and replicated? How much do CVI efforts cost? Do the benefits outweigh costs? What are taxpayers getting for their money?
Evaluations may eventually identify some CVI programs as effective and worthwhile components of public safety, but developing effective programs and practices requires informed debate and evidence. Research about CVI strategies is still in a developmental phase. Even the meaning of CVI is not settled.
What are CVI programs? President Biden signed legislation to expand the use of CVI programs in 2021, but the Administration’s definitions were very inclusive. The U.S. Department of the Treasury provided written guidance for using CVI funding provided by the American Rescue Plan Act. The Department suggested local authorities use federal resources to “reduce and respond to violent crime” and “particularly gun violence” by investing in strategies that have been “proven to reduce crime.” The text from Treasury then defined Community Violence Interventions as:
“evidence-based strategies including focused deterrence, street outreach, and hospital-based violence intervention models, complete with wraparound services such as behavioral therapy, trauma recovery, job training, education, housing and relocation services, and financial assistance.”
Using that definition, CVI includes many strategies used for decades by social services, healthcare providers, and law enforcement agencies. The guidance from federal agencies provides minimal information about CVI methods. Some interventions labeled as CVI concentrate on groups and entire neighborhoods, but most focus on changing individual behavior. Some attempt to reset broad social norms and attitudes toward violence, but most use social and therapeutic services to alter individual perceptions. A few strategies focus on community organizations not identified with law enforcement, but many work closely with police.
Some of the most well-known programs now called CVI focus on individual-level behavior change and professionalized client services. Roca, for example, is a successful violence prevention program launched initially in several locations around the Eastern and Northeastern United States (Roca 2022). The program does street outreach to connect individuals with supportive services, cognitive therapy, and employment readiness and partners with law enforcement and probation offices to reduce negative interactions with justice institutions. Another popular program, READI Chicago, identifies those residents most at risk of violence and provides them with social services, employment assistance, and cognitive-behavioral therapy to improve their decision-making skills and readiness for work.
The most celebrated CVI programs focus on individualized social services and therapeutic interventions. In this way, they embrace a traditional theory of violence prevention that has existed for centuries. They address violence one person at a time with professional services. Such an approach will always be an important component of public safety, but a full CVI menu must include other tools.
Comprehensive violence prevention should include community-sourced and community-empowered grassroots strategies. Besides policing and social services, sustainable public safety requires action at the neighborhood level, rooted in the power and resources of communities, and directed at community mobilized social change.
Community-centered strategies that operate at the grassroots could be called CVI-R, or Community Violence Intervention at the Roots. These strategies represent an important and innovative subset of CVI approaches.
Because they operate autonomously and not as an arm of social services or police, community-resourced interventions struggle for recognition, funding, and acceptance. Clearly, CVI-R cannot replace law enforcement and human services. Both sectors will always be a part of public safety, but introducing the label CVI-R could inspire broader investment in violence prevention strategies that 1) rely on the talents and power of communities and residents and 2) leverage social change in neighborhoods rather than individuals.
The concepts behind CVI-R are hardly new. During most of the 20th Century, public safety scholars portrayed the origins of crime as a constellation of social factors and not simply individual culpability. Social science theory and research suggested true public safety would require measures to mitigate poverty, improve housing, correct educational deficits, and reduce unemployment (Glaser 1958). However, turning these insights into sustainable policies and programs was always challenging.
Reducing crime and violence at the community level is frustrating when political officials and financial supporters require rapid results. Policymakers and researchers alike tend to conceptualize crime as a problem of individual behavior, at least in part because individual-level interventions are easier to design and manage. Individual outcomes can be measured using shorter time frames.
In the early 1900s, public officials started America’s juvenile courts because they viewed the illegal behaviors of young people not as individual pathologies but as indicators of family problems aggravated by community stressors (Platt 1977). The celebrated Chicago Area Project (CAP) started in the 1930s to prevent crime by supporting neighborhoods and residents’ inherent strengths and capacities (Schlossman and Sedlak 1983; Snodgrass 1976). Community-based approaches were used extensively during the 1950s and 1960s to address serious youth crime and gang-related violence (Cloward and Ohlin 1960; Spergel 1965; Spergel and Grossman 1997).
The community-centered approach to crime prevention lost ground as the 1970s approached (Hinton 2016). The decline was represented by the passage of the 1968 Omnibus Crime Control and Safe Streets Act which provided more funding for local policing and strengthened state and federal justice systems. Federal lawmakers added billions of dollars to police, courts, and prisons. However, that same year, the Johnson Administration received the final Report of the National Advisory Commission on Civil Disorders, known as the Kerner Commission. The Commission warned that aggressive policing combined with the nation’s ongoing failure to address economic inequality would worsen social conditions, especially in racially and ethnically diverse communities. Nevertheless, officials pressed ahead with stronger and more punitive systems of crime control.
Of course, community-focused approaches were never completely out of the picture. National interest in Community Violence Interventions was particularly strong following the disruption of the COVID-19 pandemic and widespread protests reacting to police violence, including the on-camera suffocation of Minneapolis resident George Floyd in 2020. Political support for CVI strategies surged from 2020 to 2022, encouraging states and municipalities to enhance existing efforts and establish new programs.
Emboldened by the growing support for community-based violence prevention, progressive activists and social reformers even argued policing could be de-funded or abolished. The debate changed the tone of public safety policy. Support for CVIs turned into a social movement, with growing numbers of public events, national conferences, and publications. Community Violence Intervention became a litmus test of sorts. When someone aligned with progressive policies looks for effective crime prevention strategies today, CVI is their natural focus.
In many ways, these developments are welcome. Communities should never depend solely on law enforcement and criminal justice for safety and security. A strong system of community-based violence prevention would be popular with public officials — if it works and if research proves that it works. Building a strong system of prevention requires evidence from credible evaluation research.
Elected officials and political leaders frequently rely on instinct and rhetorical arguments to justify crime prevention policies, and their natural preference tends to be enforcement. To get a fair shot in policy debates, CVI-R must be grounded in evidence.
A small number of community-centered strategies grew in prominence recently. Cure Violence, for example, became one of the most well-known violence prevention strategies in the United States and internationally. It inspired many programs, including Safe Streets Baltimore and the Crisis Management System of New York City.
Another popular brand in the violence prevention orbit is Advance Peace, first piloted in Sacramento, Stockton, and Richmond, California. Both models use community-centered and community-sourced strategies to reduce violence. They could be called the leading examples of the CVI-R approach. Of course, policymakers and communities continue to ask the key question: do they work?
Research evidence for Advance Peace and Cure Violence continues to grow, but neither strategy has been proven effective at a level that would merit the label “evidence-based.” Researchers use that term to describe interventions that have been replicated with fidelity and evaluated multiple times using reasonably rigorous research designs. Both strategies seem to be effective enough to merit further investment, but research to date can only be called promising.
Advance Peace is a relatively new but prominent intervention approach that began in California in 2018, piloted in the cities of Sacramento, Stockton, and Richmond (Corburn and Fukutome 2019; Corburn and Fukutome-Lopez 2020). The Advance Peace approach includes individual-level supports, but it also focuses on community-based resources (Corburn et al. 2021).
The program provides individuals with Peacemaker Fellowships that include monthly stipends, educational opportunities, and work opportunities (i.e., internships, apprenticeships). Neighborhood Change Agents create life plans individually crafted for each person in the program. This plan, along with the monthly stipend, is designed to incentivize individuals away from violent activity.
Researchers studied Advance Peace in 2018 and 2019 in Sacramento, comparing homicides and assaults before and after the intervention. In addition, the evaluators compared police data in program areas with non-program areas having similar demographics (matched comparisons). The study also compared intervention areas with the entire city of Sacramento to observe the city-wide trends of violent assaults and homicides before and after implementation. Researchers hoped these double comparison groups would provide more reliable evaluation findings.
Findings compared the mean number of gun homicides and violent assaults from 2014 to 2017 with the number of gun homicides and violent assaults in the first 18 months of program implementation from July 2018 to December 2019. The findings show a decrease in homicides and violent assaults in all three Advance Peace zones. In non-program zones, there was a slight increase in homicides and assaults during the same period. However, data from the entire city also showed decreases in gun homicides and violent assaults.
These findings could suggest that Advance Peace had a positive impact on reducing violence in the areas where it has been implemented in Sacramento. The increases in non-program areas, however, could indicate violent activities simply moved. Evaluations of operational, placed-based interventions must consider that possibility.
Researchers tried to evaluate Advance Peace again in Stockton from 2018 to 2020 using interrupted time-series analyses to assess changes in homicides and violent assaults (Corburn et al. 2021). Unlike the Sacramento study, Stockton’s evaluation did not include genuine comparison areas, at least partly because the city’s relatively small population made it difficult to distinguish between intervention areas and non-intervention areas.
Like in Sacramento, evaluators used police data to examine violent crime rates across the community before and after the intervention. Researchers compared crime averages before implementation with averages in 12-month intervals after implementation. The study examined six districts of the city. When comparing pre- and post-intervention outcomes in each community, researchers found decreases in gun homicides and assaults. In total, the city experienced a 21 percent decrease in gun-related violence. It was not possible, however, to establish this as an effect of the intervention because the evaluation did not use comparison areas.
The most successful and well-known CVI-R strategy is Cure Violence. Often called the public health approach to violence prevention, Cure Violence is inspired by methods used to combat contagious disease outbreaks (Slutkin et al. 2015). It relies on community members utilizing trust and credibility to intercede with residents likely to be affected by interpersonal violence. The approach presumes that people with local credibility are best suited for influencing their community instead of outsiders.
Cure Violence staff members use several tactics to reduce community violence. First, Violence Interrupters (VIs) form trusting relationships with residents. They show up to mediate before interpersonal disputes escalate into violent confrontations. Second, Outreach Workers (OWs) link the same neighborhood residents with whatever social services, educational opportunities, and employment may help them avoid violence. Finally, the program works to shape community norms toward non-violence, disseminating materials through public messages and signage, meeting with community groups, and holding public events.
Cure Violence works on multiple fronts to aid communities most likely affected by violence. Researchers have tested Cure Violence in a wide array of communities. Many studies have produced promising results, and the stronger studies point to the most positive effects (Butts et al. 2015; Cure Violence Global 2021).
Safe Streets Baltimore
Baltimore implemented a strategy inspired by Cure Violence more than 15 years ago. Initially launched in five neighborhoods, the Safe Streets program was intended to reduce gun violence and non-fatal shootings while shaping community views on violence. The program largely followed the Cure Violence approach, but it also differed in some ways. Contrary to the recommended management structure for Cure Violence sites, Safe Streets sites sometimes shared office space and supervisory staff. At least one team of outreach workers was involved in several locations simultaneously. This could have diluted the program’s neighborhood-specific effect.
Webster et al. (2012) evaluated Safe Streets using negative binomial regression (NBRM) models that enabled researchers to track changes in gun violence while also accounting for police activity and other factors in each program area. The research design allowed researchers to determine if the predictor variable (intervention) was associated with changes in the dependent variable (violence) while controlling for other variables (e.g., police activity) that could affect violent crime outcomes.
Safe Streets had promising results in the Cherry Hill neighborhood. Homicides and non-fatal shootings each declined. Beyond the Cherry Hill neighborhood, however, other study sites returned mixed findings. McElderry Park experienced reduced homicides but an increase in non-fatal shootings. Ellwood Park and Madison-Eastend also experienced fewer homicides, but overall differences in homicides and non-fatal shootings failed to reach statistical significance.
In another Safe Streets study, Webster et al. (2018) combined NBRM models with multiple interrupted time-series (ITS) analyses to examine the effects of Safe Streets on homicide and non-fatal shootings while accounting for other violence prevention programs and policing strategies operating across the city. Using data from 2003 to 2017, the study replicated many of the same findings from Webster et al. (2012). Non-fatal shootings declined approximately 30 percent in Cherry Hill, but the effects were not significant in three other sites.
More recently, Buggs et al. (2022) analyzed Safe Streets using a synthetic control design, again finding mixed results across all sites. Synthetic control group studies employ weighted combinations of multiple control groups to compare outcomes with treatment groups (Robbins, Saunders and Kilmer 2017). The study found that three of seven Safe Street sites had increases in homicides and non-fatal shootings during the program’s operation. The remaining four had reductions in at least one outcome, but “none of the seven Safe Streets sites experienced violence reductions outside the norm when compared with placebo tests” (Buggs et al. 2022: 65).
On the other hand, Safe Streets may have had positive effects on participants. More than half of the participants surveyed by researchers indicated that a Safe Street outreach worker helped them to settle conflicts and avoid violence. Respondents in at least one site, McElderry Park, reported lower approval of firearm use to settle conflicts (Webster et al. 2012). Still, the impact of Safe Streets varied by location and seemed to erode after initially positive results. Buggs et al. (2022) hypothesized that factors such as differing drivers of violent conflicts at program sites, an absence of supportive services for participants, and civil unrest following the 2015 death of Freddie Gray in police custody may have combined to limit the program’s impact.
Save Our Streets NYC
Save Our Streets (SOS) was inspired by the Cure Violence approach and launched in New York City’s Crown Heights Community Mediation Center by the Center for Court Innovation (CCI) in 2010. The SOS program was a precursor to New York City’s subsequent implementation of two dozen other Cure Violence sites. SOS followed the Cure Violence approach with reasonable fidelity other than one significant staffing change: program managers combined the roles of Violence Interrupter and Outreach Worker.
Researchers evaluated SOS implementation in Crown Heights and the South Bronx. Both sites yielded promising results (Picard-Fritsche and Cerniglia 2013). Using an interrupted time series design to compare three years of SOS outcomes in Crown Heights against three neighboring areas, the study found that Crown Heights experienced a six percent reduction in average monthly shooting rates. On the other hand, comparison areas saw increases of 18 to 28 percent. The evaluation also surveyed the community, but the SOS program appeared to have little effect on residents’ sense of safety or opinions about using firearms for self-protection.
Crisis Management System NYC
The NYC Mayor’s Office of Criminal Justice (MOCJ) continued to fund and oversee two dozen Cure Violence sites collectively called the Crisis Management System. Program designs varied somewhat between neighborhoods, but core operations were intended to follow the principles of Cure Violence. In the early years of the effort, programs were piloted in one area of each New York City borough – East New York (Brooklyn), South Bronx, South Jamaica (Queens), North Harlem (Manhattan), and North Staten Island.
The Research and Evaluation Center at John Jay College estimated the program’s effectiveness using various measurement strategies, including surveys of young men in intervention neighborhoods compared with those in a matching set of non-program communities (Delgado, Alsabahi and Butts 2017). Gun violence trends in program catchment areas were compared with a set of matching non-program sites using reports from law enforcement and hospital records about shooting victims from each neighborhood (Delgado et al. 2017).
The study found promising evidence that Cure Violence may help to prevent violence. Compared with matched neighborhoods not hosting programs, gun violence declined significantly in areas with Cure Violence sites. In one neighborhood of Brooklyn, gun injuries reported by hospitals fell 50 percent following the implementation of Cure Violence, while injuries declined just 5 percent in the matched comparison area. Another Cure Violence neighborhood experienced 63 percent fewer shootings after the program opened, compared with a decline of just 17 percent in the comparison area.
The study used a quasi-experimental design, but it relied on a purposive, data-driven matching strategy rather than a purely statistical matching process. Treatment and comparison areas were well-matched on most socioeconomic and crime indicators, but with limited availability of suitable matches, some comparisons were only approximate. Ideally, studies of community-level interventions would use stringent matching procedures to detect differences between areas with and without interventions. The ARIMA models used in the study were also only able to detect breaks in a single time series. Regression-based tests, such as difference-in-difference or latent growth curve models, were not possible because of the small sample size of two to four communities. Stronger results would require a larger sample.
Abriendo Caminos implemented a version of the Cure Violence approach in Colombia. The program focused on two neighborhoods in Cali, a large city near the Pacific coast. Cali’s geographical location made it a hot spot for drug and gun trafficking, leading to sharp increases in violent activity (Mejia et al. 2015). Abriendo Caminos addressed spikes in violent activity in two city areas with high rates of homicides and assaults and a similar acting violence prevention program. The program integrated the core ideas of the original Cure Violence approach: interrupt the transmission of violence, identify, and change the way of thinking of individuals inclined to violence and change the norms around the community to be non-violent. In addition, Abriendo Caminos expanded from the original Cure Violence approach by focusing closely on addressing substance use in communities.
Abriendo Caminos used violence interruption strategies to guide individuals away from violent behavior, and OWs supported individuals by providing internship or apprenticeship opportunities while also focusing on reducing the consumption of toxic substances. The program also worked with neighborhoods to facilitate community change by organizing public events, workshops, and community gatherings.
Researchers examined changes in homicides, violent assault, and violent threats within areas where Abriendo Caminos operated during the 18-month evaluation period (Moreno et al. 2020). Matched comparison groups allowed the study to compare rates in program areas with non-program areas having similar demographics. The evaluation examined three dependent variables (homicide, violence-related injury, and violent threats) while accounting for other predictor variables, including policing and unrelated gang violence programs.
In addition, evaluators used negative binomial regression to assess changes in violent outcomes over seven-day increments in each neighborhood, those with and those without the program. In one of the two program neighborhoods, researchers found clear reductions in homicides, violent injuries, and threats compared with the matched non-program area. The other program area presented mixed results, showing an increase in homicides with a decrease in violent injuries relative to the comparison area, although consistent with citywide rates.
Researchers noted the program experienced several administrative complications during the study period. Staff members were often overworked, with some reporting they needed to work several jobs. In addition, evaluators noted that services designed for high-risk individuals were geographically inaccessible for some individuals. Finally, some participants reported a lack of trust in the program and staff.
Researchers measured the impact of the Cure Violence approach used by Philadelphia’s CeaseFire program (Roman et al. 2017). Impact evaluations are designed to create statistical evidence that changes in outcome variables may be attributed to the effects of an intervention rather than merely being associated with the intervention. The study tracked changes in firearm homicides and nonfatal shootings and measured the effect of the city’s implementation of a program based on the Cure Violence approach.
Using interrupted time series analyses, the evaluation team compared changes in program areas with a set of statistically matched comparison areas. Shooting data were examined before and after April 2013, the single point of interruption when the program was considered fully implemented. All program components were operational at that time, all staff had been formally trained, violence mediations had begun, and the number of participants nearly tripled from one month before in March 2013.
The results showed that shootings declined over the study period in both the intervention zones and the comparison areas, but the scale of the decline favored the program. The overall rate of gun violence fell by 2.4 shootings per month per 10,000 residents in neighborhoods served by the CeaseFire program, while it dropped by 1.24 shootings per month per 10,000 residents in the matched comparison areas.
Trinidad & Tobago implemented a program based on the Cure Violence approach, and researchers tracked its results from 2015 to 2017. Project REASON worked in 16 neighborhoods, focusing on preventing gun violence, improving public perceptions of safety, reducing the likelihood of antisocial behavior by high-risk individuals, and preventing conflicts from escalating into violence (Maguire et al. 2018).
Project REASON included five core components: street outreach for at-risk youth, public education, faith-leader involvement, community mobilization, and monitoring of local police reports. The program followed the Cure Violence approach in most ways. Still, it differed in requiring a higher threshold for identifying high-risk individuals, enrolling victims of violence and those in need of social support without a violence risk assessment, and using informal methods for detecting potential violence.
A 2018 evaluation of the project found positive effects on several outcomes (homicide, shooting with intent, etc.). Using difference-in-differences (DID) estimates, synthetic control methods, and an interrupted time series to compare outcomes with a synthetic comparison group, Maguire et al. (2018) found that violent crime at program sites was 45 percent lower than at the comparison site and emergency calls decreased while the comparison site saw an increase. The hospital closest to the intervention sites also reported fewer admissions for gunshot wounds. A cost-benefit analysis revealed that each prevented violent crime, service call, and hospital admission could have saved $3,500 – $4,300.
Although Project REASON was associated with a significant decline in fear of crime among residents, results from a local two-wave survey on self-reported violent victimization, perceived community mobilization, and perceived influence of gun/gang violence revealed no program impact. Publicly responding to violent incidents is a crucial part of the Cure Violence model but was absent at times for Project REASON. In conjunction with Project REASON’s targeted interventions with high-risk individuals, this shortcoming may have affected community awareness of the program.
Cure Violence and Advance Peace operate with resources derived from residents and neighborhoods. Each program shows promise, but neither can be described as “evidence-based.” Researchers call interventions evidence-based only after several studies have shown meaningful and predictable effects on designated outcomes (Farrington 2003).
It is certainly appropriate to refer to both programs as evidence-informed, or even evidence-backed, as they have shown promise in previous evaluations. To achieve the status of evidence-based, however, CVI-R strategies must continue to work with researchers to conduct high-quality outcome evaluations, and the quality of evaluations must be assessed using the standards of research science rather than political viability or journalistic interest.
Evaluations must compare the effects of interventions with estimates of the “counterfactual,” or what would have happened to key outcomes if an intervention had not been implemented (Roman, Klein and Wolff 2018). Simply measuring outcomes before and after an intervention, or what researchers call a pre-post design, is not reliable evidence. Many other factors may contribute to changes in outcomes. Studies that fail to estimate the counterfactual may be useful for exploring new innovations, but they cannot be considered evidence of effectiveness.
Of course, the classic method for estimating the counterfactual is the randomized controlled trial in which treatment and control groups are constituted at random, thus reducing selection bias and equalizing as many sources of change in key outcomes as possible. However, many programs and interventions in the public safety area are not amenable to randomized designs for a range of practical and ethical reasons (Butts and Roman 2018).
Fortunately, effective alternatives exist. The best evaluations of CVI-R programs employ some form of quasi-experimental design involving non-randomized but statistically generated comparison groups. They also use quantitative models to measure outcomes over time while accounting for factors potentially associated with the same outcomes. They do not simply report a change in one indicator between two points in time and then suggest such a finding is sufficient to prove an intervention effective.
It is also not sufficient to estimate the effects of a violence-prevention effort in one neighborhood by comparing changes in that neighborhood with all remaining areas of a city or other large area. Programs to prevent violence are placed in specific settings for a reason. Violence levels are often elevated before launching a new program. A statistical property known as “regression to the mean” predicts that extreme values tend to moderate over time. In other words, an area with a high rate of violence in one period will likely experience declines in subsequent periods, just as an area with low violence in one period can expect to see increases in a later period. For this reason, credible evaluation research involves multiple measures over time in various locations.
Measuring program outcomes against an estimate of the counterfactual is a key task for CVI-R evaluations, but it is also important to measure whatever process is used to achieve outcomes (Haynes et al. 2015). Until interventions can be replicated in detail with proven “fidelity” to an intentional process, even an apparently effective violence prevention strategy is assumed to be one-of-a-kind. A single success could be due to some unknown combination of circumstances, such as pure coincidence, historically lucky timing, or a uniquely charismatic leader.
Rigorous evaluations measure what programs do and what they produce in the short term just as carefully as they measure changes in long-term outcomes. In other words, high-quality studies measure “inputs” and “outputs” as well as “outcomes.” A community violence intervention evaluated with outcome measures alone may be admired and even imitated, but it will never be established as an evidence-based prevention model.
Before investing in evaluations, public officials should also consult with researchers to assess the “evaluability” of interventions. In other words, is an intervention being designed and managed in a way that will allow evaluation researchers to measure inputs, outputs, and outcomes? Evaluability describes whether a program operates with observable and measurable components that could feasibly achieve its intended goal(s). Interventions are not ready for evaluation if they are still innovating and discovering routine procedures. Launching evaluations prematurely may lead to mixed or even negative findings that aren’t indicative of a program’s quality.
In addition to assessing evaluability, community leaders should encourage researchers to begin studies of CVI-R strategies by exploring the potential uses of qualitative methods, including direct observation of program operations, unstructured interviews with staff and stakeholders, and focus groups of relevant participants. Qualitative investigations can highlight the internal practices and worker-participant relationships that affect a program’s overall impact. These methods are not always employed due to the cost of researcher labor and their lack of generalizability. Still, their inclusion in outcome evaluations may help uncover implementation issues and explain unexpected findings (Steckler and Linnan 2002).
Funding for community violence prevention is typically contingent on evidence, despite the complexities involved. Unlike law enforcement or therapeutic interventions, CVI-R strategies are detached from conventional bureaucracies in the criminal justice and social services sectors. Their independence may be part of their effectiveness, but it presents many political and technical challenges.
Finally, CVI-R programs are often confronted with operational issues and workplace culture problems associated with staff burnout from long hours, low pay, and the ongoing effects of trauma, both vicarious and experienced. Programs are often unable to safeguard the wellness of their most crucial staff members who work in the community to interrupt imminent conflicts and connect with residents. High stress and staff turnover have detrimental effects on programs (Fixsen 2005).
Conventional techniques from the organizational development and human resources field may not apply to programs in the CVI-R sector, but transparency and open communication between supervisors and staff are even essential (Weiss 2020). The stressors inherent in violence prevention work are hard to avoid, but mitigating staff burnout, properly compensating staff, and ensuring that staff voices are heard in practice and policy discussions should be considered critical features of CVI-R programs (Bocanegra et al. 2021).
Researchers must continue to study grassroots Community Violence Interventions to establish them as effective and worthwhile practices. Evaluations show promising results, but many lawmakers are skeptical of any model not directly associated with law enforcement or conventional social services. The violence prevention field must confront these challenges, and advocates need more than political rhetoric and ideology. Sustainable progress requires evidence. Researchers must work closely with policymakers and community groups to answer these important questions using rigorous evaluations of CVI-R processes and outcomes.
Bell, Holly, Shanti Kulkarni and Lisa Dalton (2003). Organizational prevention of vicarious trauma. Families in Society: The Journal of Contemporary Social Services, 84(4), 463–470.
Buggs, Shani A.L., Cassandra K. Crifasi and Daniel W. Webster (2018). Estimating the Effects of Law Enforcement and Public Health Interventions Intended to Reduce Gun Violence in Baltimore. Baltimore, MD: Johns Hopkins Center for Gun Policy and Research.
Buggs, Shani A., Daniel W. Webster and Cassandra K. Crifasi (2022). Using synthetic control methodology to estimate effects of a Cure Violence intervention in Baltimore, Maryland. Injury Prevention, 28(1), 61–67.
Butts, Jeffrey A., Caterina Gouvis Roman, Lindsay Bostwick and Jeremy Porter (2015). Cure Violence: A public health model to reduce gun violence. Annual Review of Public Health, 2015(36), 39–53.
Butts, Jeffrey A. and John K. Roman (2018). Good questions: Building evaluation evidence in a competitive policy environment. Justice Evaluation Journal, 1(1), 15–31.
Cloward, Richard A. and Lloyd E. Ohlin (1960). Delinquency and Opportunity: A Theory of Delinquent Gangs. Glencoe, IL: The Free Press.
Corburn, Jason, DeVone Boggan, Khaalid Muttaqi, Sam Vaughn, James Houston, Julius Thibodeaux and Brian Muhammad (2021). Advancing urban peace: Preventing gun violence and healing traumatized youth. Youth Justice, June 2021, 1–18.
Corburn, Jason and Amanda Fukutome (2019). Advance Peace Stockton: 2018-19 progress report. Berkeley, CA: Institute of Urban and Regional Development, University of California-Berkeley.
Corburn, Jason and Amanda Fukutome-Lopez (2020). Outcome evaluation of Advance Peace Sacramento, 2018-19. Berkeley, CA: Institute of Urban and Regional Development, University of California-Berkeley.
Delgado, Sheyla, Laila Alsabahi and Jeffrey A. Butts (2017). Young Men in Neighborhoods with Cure Violence Programs Adopt Attitudes Less Supportive of Violence. [JohnJayREC DataBits 2017-01]. New York, NY: Research and Evaluation Center, John Jay College of Criminal Justice, City University of New York.
Delgado, Sheyla A., Laila Alsabahi, Kevin Wolff, Nicole Alexander, Patricia Cobar and Jeffrey A. Butts (2017). The Effects of Cure Violence in the South Bronx and East New York, Brooklyn. In Denormalizing Violence: A Series of Reports from the John Jay College Evaluation of Cure Violence Programs in New York City. New York, NY: Research and Evaluation Center, John Jay College of Criminal Justice, City University of New York.
Farrington, David (2003). Methodological quality standards for evaluation research. Annals of the American Academy of Political and Social Science, 587(1), 49–68.
Fixsen, Dean L., Sandra F. Naoom, Karen A. Blase, Robert R. Friedman and Frances Wallace (2005). Implementation research: A synthesis of the literature. Tampa, FL: University of South Florida, Louis de la Parte Florida Mental Health Institute, The National Implementation Research Network (FMHI Publication #231).
Glaser, Daniel (1958). The sociological approach to crime and correction. Law and Contemporary Problems, 23(4), 683–702.
Haynes, Abby, Sue Brennan, Sally Redman, Anna Williamson, Gisselle Gallego, Phyllis Butow and the CIPHER team (2015). Figuring out fidelity: A worked example of the methods used to identify, critique and revise the essential elements of a contextualised intervention in health policy agencies. Implementation Science, 11(1), 23.
Himle, David P., Srinika Jayaratne and Paul Thyness (1986). Predictors of job satisfaction, burnout and turnover among social workers in Norway and the USA: A cross-cultural study. International Social Work, 29(4), 323–334.
Hinton, Elizabeth (2016). From the War on Poverty to the War on Crime: The Making of Mass Incarceration in America. Cambridge, MA: Harvard University Press.
Maguire, Edward R., Megan T. Oakley and Nicolas Corsaro (2018). Evaluating Cure Violence in Trinidad and Tobago. Washington, DC: Inter-American Development Bank.
Moreno, Carlos E., María I. Muñoz, and Juan Carlos G. Gómez (2020). Informe Final de la Evaluación de Impacto del Programa Abriendo Caminos de la Fundación Alvaralice. Cali, Colombia: Laboratorio de Medición de Impacto Social, Universidad Icesi.
Mejia, Daniel, Daniel Ortega and Karen Ortiz (2015). Un Análisis de la Criminalidad Urbana en Colombia. [Documento de Trabajo]. Bogotá, Colombia: CAF Banco Inter-Americano de Desarrollo de América Latina.
Phalen, Peter, Erricka Bridgeford, Latrice Gant, Aaron Kivisto, Brad Ray and Simon Fitzgerald (2020). Baltimore Ceasefire 365: Estimated impact of a recurring community-led ceasefire on gun violence. American Journal of Public Health, 110(4), 554–559.
Platt, Anthony M. (1977). The Child Savers: The Invention of Delinquency (2nd Edition). Chicago & London: University of Chicago Press.
Picard-Fritsche, Sarah and Lenor Cerniglia (2013). Testing a Public Health Approach to Gun Violence: An Evaluation of Crown Heights Save Our Streets, a Replication of the Cure Violence Model. New York, NY: Center for Court Innovation.
Robbins, Michael W., Jessica Saunders and Beau Kilmer (2017). A framework for synthetic control methods with high-dimensional, micro-level data: Evaluating a neighborhood-specific crime intervention. Journal of the American Statistical Association, 112(517), 109–126.
ROCA (2022). How we Do it: Our Intervention Model. Accessed May 10, 2022.
Schlossman, Steven and Michael Sedlak (1983). The Chicago Area Project revisited. Crime & Delinquency, 29(3), 398–462.
Slutkin, Gary, Charles Ransford and R. Brent Decker (2015). Cure Violence: Treating violence as a contagious disease (pp. 43-56). In Michael D. Maltz and Stephen K. Rice (Editors), Envisioning Criminology: Researchers on Research as a Process of Discovery. Switzerland: Springer International.
Snodgrass, Jon (1976). Clifford R. Shaw and Henry D. McKay: Chicago criminologists. British Journal of Criminology, 16(1), 1–19.
Spergel, Irving A. (1965). Selecting groups for street work service. Social Work, 10(2), 47–55.
Spergel, Irving A. and Susan F. Grossman. (1997). The Little Village Project: A community approach to the gang problem. Social Work, 42(5), 456–470.
Steckler, Allan and Laura Linnan (Editors) (2002). Process Evaluation for Public Health Interventions and Research. San Francisco, CA: Jossey-Bass: A John Wiley & Sons Imprint.
Webster, Daniel W., Jennifer Mendel Whitehill, Jon S. Vernick and Elizabeth M. Parke (2012). Evaluation of Baltimore’s Safe Streets Program: Effects on Attitudes, Participants’ Experiences, and Gun Violence. Baltimore, MD: Johns Hopkins Center for the Prevention of Youth Violence.
Weiss, Leah (2020). Burnout from an organizational perspective. Stanford Social Innovation Review, October 20, 2020.
The New York City Mayor’s Office of Criminal Justice (MOCJ) provided funding support for this analysis, but all conclusions are those of the authors. Funders and partners of the Research and Evaluation Center are not responsible for any findings presented in Center publications.
The authors appreciate the contributions of doctoral student Yu-Hsuan Liu who assisted with the initial collection and review of the studies discussed in this report. The authors are also indebted to the scholars who reviewed this report in draft form and provided feedback to shape the final product. They include Caterina Roman, Charles Elliott, and Sheyla Delgado.
Katheryne Pugliese, Paul Odér, Talib Hudson, and Jeffrey A. Butts (2022). Community Violence Intervention at the Roots (CVI–R) — Building Evidence for Grassroots Community Violence Prevention. New York, NY: Research and Evaluation Center, John Jay College of Criminal Justice, City University of New York.