Special Collection on Community-Led Strategies to Reduce Gun Violence: Developing Evidence for Policy and Practice: An Editorial

Introduction

In recent years, the United States has invested considerable funding in a national network of community violence intervention (CVI) programs. ¹ Many types of violence reduction efforts are now called CVI. They share the goal of reducing shootings and other serious violence, but CVI approaches vary. Models in the CVI domain often provide direct services and resources to individuals at the highest risk of violence. Many focus on the neighborhoods and geographic “hot spots” most affected by violence. Others even attempt to mitigate basic social norms and community conditions that engender violence.

There is no single, widely accepted definition of community violence intervention. Theories of change differ, operating at either the individual level through direct engagement and support or the aggregate level through community change and mobilization across geographic areas (neighborhoods, blocks, census tracts, police districts, etc.). A promising model of CVI activity focuses on community-resourced efforts rather than law-enforcement–driven approaches or those that rely entirely on professionalized social services. Community-resourced and community-led models emphasize relationship-building, the voices and efforts of “credible messengers,” and local ownership of safety strategies. Some leading CVI models derive from a public health framework, conceptualizing violence as a preventable condition that can be addressed through systematic, population-level strategies—identifying risk and protective factors, interrupting transmission of violence, and promoting resilience and social cohesion. Essentially, community-led public health approaches to CVI focus on the “roots” of violence, or CVI-R, and may involve law enforcement as secondary rather than leading partners. ² These models are important and potentially cost-effective, but they confront numerous evidentiary challenges.

Evidence Agenda

Efforts to address social problems may be categorized as primary prevention to impede the initial occurrence of problems, secondary prevention to reduce the severity and escalation of problems as they emerge, and tertiary intervention to apply consequences and remediations for fully developed problems. Research on violence reduction thus far has largely focused on secondary prevention and tertiary intervention.^2,3 Secondary prevention programs managed by social services and law enforcement work with individuals at risk of but not deeply involved in violence. Tertiary interventions are the primary business of bureaucracies comprising the criminal justice system, including law enforcement, courts, community supervision agencies, and correctional facilities.

Efforts to strengthen primary and secondary prevention of violence in the U.S. benefited significantly from increased federal funding after 2020. Researchers were involved in multiple efforts to evaluate various CVI initiatives. ⁴ In 2024, however, a national election in the U.S. produced dramatically different legislative and executive branches of the government. Newly empowered officials immediately canceled most CVI funding, including numerous research projects already underway. More than $100 million was eliminated from the resources available for CVI programs across the country. ⁵ The timing was unfortunate. Research on CVI was just beginning to show results. The funding setback delayed the emergence of much-needed evidence about the value and cost-effectiveness of community-led programs to reduce violence.

Community-led violence-reduction strategies are highly promising but not yet evidence-based. ⁴ Evaluation results are not yet strong enough to defend community-level CVI approaches from hostile government officials who instinctively prefer the more familiar methods of law enforcement and social services. It is especially challenging to separate the effects of violence-reduction efforts from general crime trends, as trends tend to fluctuate. Community violence was generally declining after 2023 as the crime surge associated with the COVID-19 pandemic began to wane. ⁶ Some of the decline may have been due to CVI investments, but program effects must be measured in detail to determine their contribution to improved safety trends. Identifying the role of CVI requires careful research using causal inference controls, implementation science to gauge program efforts, and theoretically appropriate outcomes measured with intent and precision.

Policymakers looking for reasons to support CVI programs now find a relatively small base of research about community-level violence prevention approaches, but far stronger evidence for individual-level interventions. Studies of community-led and community-resourced CVI programs inevitably require some outcomes to be measured at an aggregate neighborhood level, such as a police district or census tract, which involves smaller samples, less precise instrumentation, and more complex statistical methods. These complications affect the development of persuasive evidence.^7,8 Moreover, some advocates who value CVI as a social movement may disparage efforts to develop rigorous evidence, further limiting the growth of essential research.

Building the CVI evidence base for the future will require theoretically informed, intentionally causal evaluation studies. The first task in designing such evaluations is to articulate a research-oriented theory of change. How do program developers describe the operational theory behind CVI? Is it supported by previous research and scientific literature? Are evaluation designs properly aligned with program goals and operations? Both process and impact evaluations must complement one another and identify how intended changes occur. Research must identify which program components are most successful. When studies can isolate the effects of the multiple components of a complex program strategy and compare results across individuals or geographic areas, evaluation researchers can begin to pinpoint the key components of CVI. Are some specific components responsible for the changes attributed to an entire strategy? What is the minimum number of components needed to effect change? Could future efforts be effective using only a subset of known components?

Contributors to the Inquiry Special Collection focused on these issues to advance the development of research evidence for community violence prevention.

Contents of the Special Collection

Ross et al address a key blind spot in the evaluation of street-outreach CVI models (CVI-SO). ⁹ Studies rarely capture the work outreach workers do, for whom they do it, and at a specific level of intensity. The authors lay out why estimates of effectiveness are often mixed-documentation and privacy constraints, lengthy pre-enrollment relationship-building that goes unrecorded, selection into services based on “readiness,” and crude treatment metrics that ignore strategic variation by risk. Because CVI-SO typically operates simultaneously at community, group, and individual levels, outreach workers can have effects across an array of dimensions. Yet, evaluations often flatten the variation, making it difficult to detect “critical dosage.” The authors describe methods for measuring services and dosage at scale. Using a mixed-methods pipeline on administrative data from a Chicago CVI coalition (2017-2023), the team qualitatively coded and then programmatically grouped ~190 000 service entries for more than 4000 participants into 13 categories. Analyses examine patterns in demographics and network-based risk and use descriptive statistics and latent profile analysis to uncover distinct service-dosage groups. The authors note difficulties faced in measuring person-to-person interactions.

Ivey et al draw on interviews and focus groups with CVI practitioners in 3 cities to argue for a coherent training ecosystem build with a core curriculum—structural violence frameworks, trauma-informed practice, relational engagement and motivational interviewing, risk and needs assessment, conflict mediation, and safety and ethics—as well as policy literacy and advocacy so street-level insights can inform policy and funding debate. ¹⁰ The authors emphasize the need to build data and evaluation fluency, from distinguishing process and outcome measures to documenting services, using asset-based indicators, and employing dashboards for continuous quality improvement. Training should cultivate research partnerships and analytics that refine program theories, with progression pathways tailored by role and tenure (eg, outreach worker, case manager, supervisor, and data lead). Findings from interviews and focus groups underscore that CVI’s potential depends on professionalizing the workforce without losing its community roots. Investing in standardized, continuous training that converts lived experience and structural understanding into consistent practice, evidence generation, and policy influence will strengthen effectiveness, scalability, and legitimacy—equipping practitioners to address immediate risk while also confronting the upstream conditions that reproduce violence.

Ransford et al conduct a PRISMA-guided systematic review of the evidence base for outcomes that the Cure Violence (CV) approach is designed to affect—shootings and homicides—excluding 36 studies that focused on other, secondary outcomes. ¹¹ Searching PubMed, Google Scholar, and Cure Violence Global’s internal files, the authors identify 13 papers covering 27 program sites and reporting 83 outcome findings. Overall, 68.7% of findings show reductions in shootings or killings, with 32.5% reaching statistical significance. They find that effects vary by place. Baltimore alone accounts for most null or adverse findings, underscoring how local context and implementation conditions shape impact. Other than Baltimore, 95.8% of sites report reductions, and 54.2% are statistically significant. The authors, who acknowledge they are closely aligned to the model as the developers and implementers of Cure Violence, conclude that, given proper implementation, the program produces meaningful reductions in nonfatal shootings and homicides.

Simonsson et al zero in on the “engagement phase” of CVI programs that use street outreach—specifically with participants and staff from a Cure Violence program in Philadelphia. ¹² The study seeks to answer 2 practical questions: (1) how outreach workers (OWs) build relationships and motivate high-risk individuals to engage, and (2) how those individuals actually use OWs in the early stages of program participation. Framed by desistance theory but aimed at the day-to-day mechanics of CVI practice, the study argues that behavior change is fundamentally relational. Outreach workers leverage credibility, consistent presence, and empathic care to create early “hooks for change,” while also calibrating timing to participants’ vulnerabilities, readiness, and developmental stage. Participants, in turn, draw on OWs and the broader CVI community to step back from antisocial peer networks, replacing those ties with a pro-social “surrogate family” that models different norms, offers cover to reorganize time away from the street, and provides immediate problem-solving support. The findings bolster relational/collective accounts of desistance: change spreads through networks, not just individuals, and CVI can supply the social infrastructure—trust, belonging, and reciprocal obligations—that sustains movement away from violence.

Costa et al call for better alignment of research, policy, and practice. ¹³ They stress that clear, consistent definitions of foundational CVI constructs are needed but remain unavailable. Without shared terms and established metrics, policies tend to splinter, funding becomes inconsistent, and it’s hard to scale or evaluate approaches with integrity. The ongoing confusion over basic concepts can lead practitioners, institutions, and communities to unknowingly accept enforcement-centric models that dilute CVI’s public health roots. The authors call for a unified framework that honors CVI’s history, centers practitioner expertise, and uses credible evaluation methods to capture the full array of model variations, while also encouraging the field’s development. “Without a shared conceptual framework, the research community risks misaligning its inquiries by focusing on metrics and methodologies that may fail to capture the nuanced realities of CVI efforts.” In short, we must get the definitions and measures right, or risk blunting CVI’s transformative potential.

Girma et al take on key measurement challenges and conduct a scoping review of how CVI programs in the U.S. have been evaluated—what gets measured, at what level, and who is involved. ¹⁴ Analyzing 1763 records and 149 studies, the authors find that methods and measures are uneven. Most outcome evaluations lean heavily on deficit-based indicators (eg, arrests, injury recidivism). Many studies concentrate on individual-level outcomes, and few incorporate strengths-based or protective factors. Fewer than 4 in 10 studies integrate process and outcome measures, limiting findings about implementation mechanisms. Units of analysis are split. Process measures are mostly at the individual level, while outcome measures are roughly evenly divided between the individual and community levels, depending on the specific CVI approach. The authors note that community-level outcome measures may dilute program effects if researchers define the community to include large numbers of people who are not expected to be aware of or affected by the program.

Solomon et al bring an implementation science (IS) lens to Cure Violence, shifting the focus from whether the program “works” to what we know about how and why it works (or fails) across contexts. ¹⁵ Rather than treating implementation as a black box, the authors map determinants, strategies, mechanisms, and outcomes, clarifying the pathways through which Cure Violence activities plausibly generate change. Grounded in a scoping review, the authors catalog 42 strategies spanning the intervention lifecycle—from pre-planning and hiring through sustainment—linking these strategies to mechanisms of action and both proximal (implementation) and distal (effectiveness) outcomes. Fidelity, adaptation, and equity considerations move from afterthoughts to testable elements of program design. The authors examine a template for comparative evaluation, a hybrid trial design, and continuous quality improvement. The paper supplies conceptual scaffolding for the field to develop cumulative, mechanism-focused evidence that may optimize Cure Violence implementation at scale.

Ziminski et al review how the term community violence intervention is used in practice, examining various web pages and reports to map the field’s working vocabulary. ¹⁶ Using scoping-style methods, the authors distil 9 commonly referenced CVI approaches and emphasize that various approaches operate at different levels and draw on distinct theoretical traditions, which helps to explain why CVI programs can look and behave differently. The authors review existing language and provide a roadmap for future research to help policymakers grasp these differences.

McCarthy et al address a central weakness in the hospital-based violence intervention (HVIP) evidence base: the field’s heavy reliance on reinjury rates and cost savings as proxies for success. ¹⁷ By foregrounding the lived expertise of Violence Prevention Professionals (VPPs) and the voices of clients and healthcare collaborators, the authors reframe “effectiveness” to include the relational, psychosocial, and systems-level work that can enable healing and violence prevention. The study provides practical infrastructure for future evaluations. Using a VPP-led, participatory qualitative design, the authors propose a logic model and conceptual framework to link core HVIP activities (eg, hospital integration, foundational client engagement, basic case management) to short- and long-term outcomes (eg, sustained partnership with hospitals, ongoing commitment, and client-level change). The authors suggest a pathway to comparable, fidelity-focused evaluation and stronger, more nuanced evidence for HVIPs within the broader CVI ecosystem.

Research Agenda

The field of community violence reduction continues to face an urgent need for credible, policy-relevant research. In a politicized research environment where studies are conducted by investigators competing for funding, with funding organizations competing for attention and influence, the existing body of evidence is not determined by policy relevance, theoretical salience, or methodological rigor. Programs providing individual-level interventions enjoy a head start. Achieving measurable outcomes and demonstrating program effects is typically easier, quicker, and cheaper when a program is designed to change individual behavior rather than to adjust social norms or improve community conditions.

Evaluations of individual services tend to be stronger because data for those investigations are more readily available, standardized, and collected directly from participants through surveys, assessments, or administrative records. Greater data access enables control over sample definition, more frequent measurement, and clearer attribution of observed change to an intervention. Program leaders in law enforcement and social services derive benefits in a competitive research environment when their efforts focus on the individuals “most at risk” of violence.

Program models designed to intervene at the neighborhood or community level are more challenging to evaluate. Sample sizes are inherently smaller. Data collection methods are more complex and less exact. Measuring meaningful change typically requires aggregating data from multiple sources—usually administrative or ecological—where consistency and coverage can vary. Privacy protections and legal regulations (eg, HIPAA) complicate the use of individual records in studies largely focused on community-level measures. Establishing appropriate comparison areas and attributing outcomes to a program rather than to concurrent social or policy shifts poses additional methodological challenges.

Articles in this Special Collection address some of the challenges facing community-led CVI models, and the authors consider interim steps and strategies to build a future evidence base that includes programs leveraging community-level mechanisms alongside individual services and supports. What are the conceptual origins and theoretical foundations of community-led violence interventions? What strategies should or should not be considered community-led or community-level? What evaluation evidence exists for such models? What can researchers do to strengthen the evidentiary nexus of research, policy, and practice in ways that support the political viability and growth of community-led CVI approaches?

Not all CVI models should be judged by their success in delivering resources to individuals and achieving behavioral change in those individuals, but all CVI strategies involve some work at the individual level. Models designed to create change at the community level must form relationships and partnerships with neighborhood residents. They help change community norms by supporting the communication and conflict-resolution skills of people most likely to be involved in interpersonal conflicts. They rely on case management to connect individual residents with food, housing, and employment. The effect of these intervention strategies, however, can be evaluated at the neighborhood level and not only at the individual level. Determining the best evaluation strategy must be guided by an intervention’s theory of change and its expected relationships between efforts and outcomes. These challenges will continue to affect the existence and viability of community-resourced and community-led CVI models. Researchers need to address them for the CVI field to reach maximum effectiveness and impact.