What Works in American Policing Part 2—Evidence-Based Policing: Letting Data Lead

This is the second in a seven-part series examining the policing strategies that shape public safety in America, including what the evidence supports, what has fallen short, and what policymakers and practitioners should prioritize going forward.

Every strategy examined in this series rests on a single premise: that policing decisions should be guided by evidence rather than by tradition, politics, or anecdote. That premise is at the core of evidence-based policing—the idea that law enforcement can and should subject its practices to the same kind of rigorous scientific evaluation that has transformed medicine, aviation safety, and public health. It is not a tactic or a program; instead, it is a decision-making framework that asks agencies to target resources based on data, test interventions through controlled evaluation, and track outcomes to determine whether those interventions are delivering results.

The concept is deceptively simple. In practice, it represents a fundamental shift in how police agencies operate, allocate resources, and define success. And while it has gained significant momentum over the past two decades, the gap between the philosophy and its adoption across American law enforcement remains wide.

Where It Came From

The intellectual roots of evidence-based policing trace back to a series of landmark experiments that challenged assumptions the profession had held for over a century. Conducted between 1972 and 1973, the Kansas City Preventive Patrol Experiment was the first large-scale controlled study of a core police function. The experiment, which divided 15 patrol beats into three groups (no routine patrol, normal levels, and doubled or tripled presence), found no statistically significant differences in crime rates, citizen fear of crime, or satisfaction with police. Rather than proving police presence irrelevant, the study demonstrated that simply driving around in marked cars—the largest use of police resources at the time—was not the crime deterrent everyone assumed it to be. (A recent reanalysis suggests that methodological limitations of the era may have obscured the activity’s modest preventive benefits.)

A decade later, the Minneapolis Domestic Violence Experiment broke new ground in a different way. Conducted between 1981 and 1982 by Lawrence Sherman and Richard Berk, the study randomly assigned three police responses to misdemeanor domestic assault cases: arrest, separation of the parties, or mediation. Results showed that arrest significantly reduced recidivism compared to the alternatives. The experiment had an extraordinary policy impact, as over a third of surveyed police departments reported changing their domestic violence policies in response; however, it also illustrated the complexity of translating research into practice. Subsequent replication studies in cities including Charlotte, Colorado Springs, Miami, Milwaukee, and Omaha produced mixed results, with some finding that the deterrent effect of arrest varied by offender employment status and community context. The lesson was not that the original findings were wrong—it was that evidence-based policing demands ongoing testing, not one-time conclusions adopted as permanent doctrine.

That progression from initial finding to replication to refinement is not a flaw in the evidence-based model; rather, it is the model working as designed. A profession that relied on tradition would have adopted the original Minneapolis result as permanent policy and never looked back. Evidence-based policing instead subjected its own landmark finding to further scrutiny, discovered important boundary conditions, and produced a more nuanced understanding of when and why arrest deters domestic violence. The Kansas City experiment followed a similar arc. Its original conclusion that routine patrol had no measurable effect on crime stood for decades; however, beginning with the first hot spots experiments in the mid-1990s, a growing body of research demonstrated that police presence in targeted locations does reduce crime. A 2023 reanalysis using stronger statistical methods revealed modest crime prevention effects that the original methodology was unable to detect. In both cases, the framework corrected itself—exactly as it should.

Sherman formalized this iterative logic in a 1998 lecture, drawing an explicit parallel to the revolution that evidence-based medicine had brought to healthcare. His core argument was that policing’s reliance on tradition, intuition, and the prevailing “three Rs” model—random patrol, rapid response, and reactive investigation—could be replaced by practices grounded in systematic research. By 2013, Sherman had refined this into the “Triple-T” framework, which targets resources where harm concentrates, testing practices through rigorous evaluation, and tracking their delivery and effects in real time.

How It Works in Practice

Evidence-based policing is not a single program for agencies to adopt wholesale. It is a philosophy that shapes how decisions get made at every level, from the chief’s office to the patrol car. In an evidence-based agency, deployment decisions are informed by crime analysis rather than beat tradition. New initiatives are piloted and evaluated before being scaled department-wide. Performance is measured by outcomes like crime reductions, case clearances, and community satisfaction rather than activity metrics like arrests and traffic stops. And when an initiative fails to produce results, the agency adjusts rather than doubling down.

The institutional infrastructure supporting this approach has grown considerably. The Center for Evidence-Based Crime Policy at George Mason University developed the Evidence-Based Policing Matrix, a research-to-practice translation tool that organizes rigorous evaluations of police interventions along three dimensions: the nature of the target, the level of focus, and the degree of proactivity. One of the matrix’s most consequential findings is that police strategies are most effective when they are place-based, proactive, and focused. Interventions directed at specific micro-locations or particular street segments, intersections, or address clusters consistently outperform strategies targeting broader areas or general populations.

But the matrix also reveals that many common police practices—including generalized patrol, individual-focused interventions, and purely reactive strategies—show mixed results at best. Research shows that measurable gains in public safety happen when the right activities are directed toward the right targets at the right times.

These results are not merely theoretical. When the police department in Lowell, Massachusetts, used crime analysis to identify its highest-crime micro-locations and deployed tailored, problem-oriented interventions at those sites, a randomized controlled trial found a nearly 20 percent reduction in calls for service at treated locations—with no evidence that crime simply moved elsewhere. That study is one of more than 65 rigorous hot spots policing evaluations to conclude that focused police interventions at locations with high crime activity produce statistically significant reductions in crime. These are not isolated findings—they are the cumulative product of a research infrastructure that tests, replicates, and refines while continuing to grow.

Professional societies for evidence-based policing now operate in the United States, the United Kingdom, and Australia-New Zealand, providing practitioners with forums to share research, collaborate with academics, and develop their analytical skills. The Campbell Collaboration’s Crime and Justice Group produces systematic reviews that synthesize findings across multiple studies, and the National Institute of Justice’s CrimeSolutions clearinghouse rates over 700 programs and practices as effective, promising, or ineffective, giving policymakers accessible evidence of what the research actually supports.

Why Adoption Remains Uneven

Despite its growth, evidence-based policing faces persistent barriers to adoption that are as much cultural and structural as they are intellectual. Policing has a strong craft tradition in which experience, instinct, and “what has always worked” carry enormous weight. Officers who have spent their careers developing street-level judgment can be skeptical of research produced by academics who have never made a traffic stop, let alone a felony arrest. That skepticism is not always unfounded, as academic research does sometimes fail to account for the operational constraints officers face daily; however, it creates a default resistance that leadership must actively work to overcome.

Resource constraints compound the problem. While producing new, rigorous evaluations requires analytical capacity that most departments do not have, agencies need not generate their own research to be evidence-based—they need only to access, interpret, and apply existing evidence. Most of America’s roughly 18,000 law enforcement agencies employ fewer than 25 sworn officers with no dedicated crime analysts, no research partnerships, and no budget for evaluation. Without accessible pathways to existing evidence, the vast majority of American policing operates on tradition by default—not by choice.

The 911 system itself reinforces reactive habits. American policing is structured around call-and-response, leaving little institutional space for proactive work even though research estimates that anywhere from a quarter to the vast majority of patrol time is uncommitted to active calls. Redirecting that uncommitted time toward targeted activities informed by crime analysis—directed patrol at identified hot spots, follow-up contacts with repeat victims, problem-solving tasks assigned by supervisors—is one of the most significant untapped opportunities in American policing.

Political cycles pose another challenge. Evidence-based policing requires sustained commitment to strategies that may take months or years to show measurable results. Elected officials and police leaders under intense public scrutiny face pressure to show immediate action (e.g., a new task force, visible crackdown, or press conference). While these may be politically necessary, they are not rigorously evaluated interventions. The danger is that agencies can adopt the language of evidence-based policing without its substance, claiming data-driven approaches while continuing to deploy resources based on tradition, political pressure, or crisis response.

The Data Problem

Evidence-based policing depends on data—and the state of criminal justice data in the United States remains a serious obstacle. Many agencies still operate with fragmented records systems that cannot communicate across departments let alone across jurisdictions, siloing critical information about case outcomes, repeat offenders, and crime patterns in ways that prevent meaningful analysis.

Closing those gaps is not just an internal operational priority—it is a matter of public accountability. Data transparency enables the targeting, testing, and tracking that the framework demands internally, while externally, it is one of the strongest tools law enforcement has to build the community trust that every effective policing strategy depends on. Agencies that can show their communities what they are doing, why they are doing it, and what results they are achieving operate from a position of strength. Agencies that cannot leave a posture that speculation, distrust, and political opportunism will fill.

Assessment

Evidence-based policing is not a strategy that competes with other approaches; instead, it is the lens through which they should all be evaluated. Hot spots policing, community-oriented policing, predictive policing, intelligence-led policing, and problem-oriented policing all make claims about what works, while evidence-based policing asks each of them to prove it.

The strategy’s primary limitation is operational rather than conceptual, as implementing it requires investments in civilian analytical capacity, training infrastructure, data systems, and academic partnerships that many agencies currently lack. It also requires a cultural shift within departments that have operated for decades on intuition and experience only.

That shift is happening, albeit unevenly. Professional societies, researcher-practitioner partnerships, and implementation science tailored to policing are moving the profession in the right direction, though not at the pace the problems demand. Agencies facing historic staffing shortages, collapsing clearance rates, and eroding public trust cannot afford to deploy their limited resources toward strategies that have not been tested—or worse, on strategies that the evidence shows do not work.

The case for evidence-based policing is ultimately a fiscal and operational one. Taxpayers fund police departments to keep communities safe. Every dollar spent on an untested intervention is a dollar that might be wasted—and in an era of constrained budgets and staffing emergencies, waste is not an abstraction; rather, it is a patrol shift that goes unfilled or a detective position that goes unfunded. As a core government function, policing is among the most consequential public safety investments any jurisdiction can make. Given the immense demands placed on law enforcement today, the least we can do in return is to ensure that investment is guided by evidence we know works—for officers, for departments, and for the communities they serve.

bool(true)

bool(true)

string(3) "100"

Next in this series: Part 3—Hot Spots and Precision Policing: Putting Officers Where the Data Points