Evidence-based policy — the systematic application of scientific research findings to the design, implementation, and evaluation of government programs and interventions — has been one of the most consequential intellectual developments in criminology over the past three decades, transforming how the field understands its relationship to the institutions it studies and the policymakers it aims to inform. The aspiration is straightforward: decisions about which policing strategies to deploy, which correctional programs to fund, which prevention initiatives to invest in, and which reforms to pursue should be grounded in the best available evidence about what works rather than in intuition, ideological preference, political calculation, or the historical inertia of existing practice. The reality is considerably more complex, as any honest assessment of the gap between what criminological research has established and what criminal justice policy actually does must acknowledge.
Criminology and Public Policy as a domain of inquiry places evidence-based policy at its analytical center — examining not only what the evidence supports but how evidence enters or fails to enter policymaking, what institutional and political conditions make evidence-based reform possible or impossible, and what obligations criminologists carry when their research speaks directly to contested questions about crime, punishment, and social investment. This article addresses those questions in depth, examining the intellectual foundations of evidence-based criminological policy, the methodological infrastructure that has been built to support it, the institutional channels through which research reaches or fails to reach decision-makers, and the genuine limitations of the evidence-based framework that honest engagement with its practice requires.
Introduction
The evidence-based policy movement in criminology drew its intellectual energy from the broader evidence-based medicine movement that transformed clinical practice beginning in the 1970s and 1980s, proposing by analogy that criminal justice practice should be governed by systematic research evidence rather than clinical intuition or received practitioner wisdom. The analogy was powerful but imperfect: medical interventions can be tested in randomized trials that are ethically, practically, and legally feasible in ways that many criminal justice interventions are not; the outcomes of medical treatment are more readily measured and less politically contested than the outcomes of criminal justice programs; and the institutional pathways from medical research to clinical practice — through regulatory bodies, professional licensing standards, and continuing education requirements — have no direct equivalent in criminal justice, where thousands of largely autonomous agencies make independent program and strategy decisions with minimal external accountability.
Lawrence Sherman’s 1998 report to the United States Congress, Preventing Crime: What Works, What Doesn’t, What’s Promising, commissioned by the National Institute of Justice and produced with colleagues at the University of Maryland, is widely regarded as the founding document of the modern evidence-based criminology movement. The report applied a systematic research quality rating scale — what became known as the Maryland Scientific Methods Scale — to evaluate the strength of evidence supporting criminal justice programs across policing, prevention, corrections, and courts. Its blunt findings — that many widely adopted programs lacked credible evidence of effectiveness, that some popular interventions actively increased recidivism, and that a smaller number of programs had genuine experimental support — were both scientifically rigorous and politically disruptive (Sherman et al., 1998). The report created the intellectual framework for a generation of systematic evidence synthesis in criminology.
The institutionalization of evidence-based criminological policy has proceeded through multiple channels since Sherman’s report: the establishment of the Campbell Collaboration’s Crime and Justice Review Group, which has produced systematic reviews of criminal justice interventions using the rigorous methodological standards of the Cochrane tradition; the development of the CrimeSolutions clearinghouse by the Office of Justice Programs, which rates criminal justice programs as Effective, Promising, or No Effects based on systematic review criteria; the growth of embedded researcher programs that place academic criminologists within police departments, corrections agencies, and courts to generate evidence in direct response to operational questions; and the expansion of the What Works Centre for Crime Reduction at the College of Policing in the United Kingdom, which disseminates systematic review evidence to police practitioners in accessible formats. Together, these developments constitute a substantial institutional investment in making criminological evidence available, accessible, and usable for criminal justice decision-makers.
The Methodological Foundation: What Counts as Evidence
The Hierarchy of Evidence and Its Application to Criminal Justice
The evidence-based policy framework rests on a hierarchy of research designs whose position in the hierarchy reflects the strength of causal inference each design supports. At the apex sits the randomized controlled trial, in which individuals, places, or agencies are randomly assigned to intervention and control conditions, and in which the equivalence of treatment and control groups before the intervention allows post-intervention differences to be attributed to the intervention itself rather than to pre-existing differences between groups. The logic is simple and powerful: if those who received a policing strategy, correctional program, or prevention intervention were assigned to that condition by chance rather than by self-selection or administrative decision, observed differences in outcomes can be attributed to the intervention.
The Maryland Scientific Methods Scale operationalized this hierarchy for criminal justice evaluation research, grading study quality from Level 1 (cross-sectional correlation) through Level 5 (randomized controlled trial). Studies graded at Level 3 or above — those with at least a comparison group, even if not randomly assigned — are generally considered sufficient for drawing tentative policy conclusions, while studies at Levels 1 and 2 are treated as preliminary or exploratory evidence (Sherman et al., 1998). The Campbell Collaboration’s systematic reviews apply similar quality criteria, and the CrimeSolutions clearinghouse requires at least one Level 3 study with positive findings for a program to receive a “Promising” rating, and at least one Level 4 or two Level 3 studies for an “Effective” rating.
Several design innovations have substantially expanded the range of criminal justice questions that can be addressed with rigorous causal inference beyond simple randomized trials. Regression discontinuity designs exploit arbitrary cutoff rules — the age threshold distinguishing juvenile from adult court jurisdiction, the score threshold triggering mandatory sentencing enhancements, the census population threshold determining federal drug prosecution eligibility — to compare individuals or cases just above and just below the threshold, who are assumed to be otherwise equivalent. Difference-in-differences designs compare outcome changes in jurisdictions that adopted a policy to contemporaneous changes in jurisdictions that did not, using the untreated jurisdictions to control for secular trends that would have affected outcomes regardless of the policy. Instrumental variable designs exploit sources of as-if random variation in treatment assignment — the random assignment of defendants to judges with different sentencing tendencies, the quasi-random timing of prison releases due to capacity constraints — to identify the causal effect of criminal justice processing on subsequent outcomes.
What the Evidence Has Established
Across three decades of systematic evidence accumulation, criminological research has established several propositions with sufficient rigor and replication to constitute genuine policy-relevant knowledge rather than merely promising preliminary findings. Hot spots policing — the concentration of patrol resources at the small number of addresses and street segments where crime is most densely and most stably concentrated — is supported by the strongest body of randomized experimental evidence in the policing literature, with multiple trials across diverse cities consistently demonstrating crime reductions at targeted locations with minimal displacement to adjacent areas (Weisburd et al., 2016). Cognitive-behavioral programs in correctional settings — structured interventions that teach offenders to identify and modify the distorted thinking patterns associated with criminal behavior — show consistent meta-analytic evidence of recidivism reduction across diverse program models, populations, and implementation contexts (Lipsey et al., 2007). Drug courts — specialty courts providing intensive supervision and treatment as alternatives to conventional prosecution for substance-involved offenders — are supported by multiple randomized trials and extensive quasi-experimental evidence demonstrating reduced recidivism and incarceration at comparable or lower cost than conventional prosecution (Mitchell et al., 2012).
Evidence has also established what does not work, with implications equally important for policy. Scared Straight programs — confrontational interventions in which at-risk youth are exposed to incarcerated adults describing prison life — show consistent evidence of net harm across multiple randomized trials; participants in these programs subsequently offend at higher rates than comparable youth who receive no intervention (Petrosino et al., 2013). Boot camps modeled on military basic training — a popular punitive intervention in the 1990s — show no consistent evidence of recidivism reduction and in some evaluations show negative effects. Mass incarceration itself — the expansion of imprisonment far beyond what crime control rationale would support — is associated in the research literature with diminishing marginal returns at current American incarceration levels, such that the additional crime control achieved by marginal increases in incarceration is outweighed by the criminogenic consequences of incarceration on individual prisoners, their families, and their communities (Travis et al., 2014).
The evidence base for focused deterrence — the strategy of communicating credible and specific consequences to identified high-risk individuals through group notification meetings that combine deterrent messaging with offers of social service support — has grown substantially since David Kennedy’s work in Boston in the 1990s. Systematic reviews and meta-analyses covering more than a dozen American cities find consistent violence reductions following focused deterrence implementation, with the largest effects concentrated in cities where the strategy was implemented with the highest fidelity to the original model (Braga & Weisburd, 2012). The mechanism is deterrence at the individual level — specifically targeted at the small number of gang members and high-risk individuals responsible for a disproportionate share of gun violence — rather than the general deterrence of broad enforcement that raises sanction risks for entire populations. This targeting logic aligns with the concentration principle of hot spots policing and collectively reflects a broader shift in the evidence-based criminology movement toward precision and targeting rather than broad-based enforcement escalation.
The evidence on community violence intervention programs — which deploy credible messengers, often with histories of violence or incarceration, to interrupt conflicts and connect high-risk individuals to social services — has grown more robust in recent years, though it remains at an earlier stage of methodological development than hot spots policing or drug courts. The Health Alliance for Violence Intervention (HAVI) network’s implementation data and several rigorous quasi-experimental evaluations of programs including Cure Violence and hospital-based violence intervention programs have documented reductions in violent incidents in targeted communities, with the Abdul Latif Jameel Poverty Action Lab (J-PAL) policy evidence summary identifying promising effects from multiple evaluations. The methodological quality of this evidence base is improving rapidly, driven partly by federal investment in community violence intervention through the American Rescue Plan and partly by the research infrastructure that HAVI and partner universities have developed to support multi-site evaluation.
Table 1. Evidence-Based Criminological Policy: Program Effectiveness Ratings
| Intervention | Evidence Standard Met | Primary Evidence Base | Effect Direction | Key Qualification |
|---|---|---|---|---|
| Hot spots policing | Strong — multiple RCTs across cities | Sherman & Weisburd (1995); Braga et al. systematic review | Positive — crime reduction at targeted locations | Modest displacement effects possible; legitimacy costs require management |
| Cognitive-behavioral therapy (corrections) | Strong — meta-analytic support across programs | Lipsey et al. (2007); Andrews & Bonta RNR research | Positive — recidivism reduction | Fidelity to program model critical; strongest for moderate-to-high risk |
| Drug courts | Strong — multiple RCTs; systematic review | Mitchell et al. (2012); Marlowe meta-analysis | Positive — recidivism and incarceration reduction | Effect varies with eligibility criteria and program quality |
| Focused deterrence / Group Violence Intervention | Moderate-strong — multiple quasi-experiments | Kennedy (2011); Braga & Weisburd (2012) | Positive — violence reduction | Effectiveness tied to credibility of enforcement and social service delivery |
| Mentoring programs (youth) | Moderate — systematic review with mixed results | DuBois et al. (2011) | Modestly positive when implemented well | Quality of mentoring relationship matters greatly; dosage effects |
| Scared Straight | Strong negative — multiple RCTs | Petrosino et al. (2013) | Negative — increased recidivism | Should not be implemented; consistent iatrogenic effect |
| Boot camps | Moderate-strong negative — systematic review | MacKenzie et al. systematic review | No effect or negative | Emphasis on punishment without rehabilitative content is ineffective |
| Risk-Need-Responsivity (RNR) correctional programming | Strong — meta-analytic support | Andrews & Bonta (2010); Bonta & Andrews (2016) | Positive — recidivism reduction | Requires validated risk assessment; adherence to responsivity principle |
The Evidence-to-Policy Translation Problem
Structural Barriers to Evidence Use
The existence of strong criminological evidence does not automatically produce evidence-based policy. The gap between what criminological research has established and what criminal justice policy actually does is substantial, persistent, and structural — reflecting not primarily the ignorance or bad faith of policymakers but the institutional conditions of democratic governance that make evidence-based policy difficult to achieve even when research findings are clear, well-communicated, and widely known.
The political economy of criminal justice policy creates systematic incentives that compete directly with evidence-based decision-making. For elected officials, crime provides an arena for demonstrating toughness — a politically valuable signal to constituents who experience crime fear as a primary safety concern — and the toughness signal is most effectively conveyed through punitive responses (longer sentences, more police, stricter supervision) that criminological research consistently finds less effective than their political prominence suggests. Conversely, the responses that research most strongly supports — upstream social investment, treatment alternatives, diversion from prosecution — are politically vulnerable to accusations of being “soft on crime” that evidence of effectiveness does not, in the current political environment, adequately counter. The political economy of crime thus systematically rewards evidence-resistant policy choices and penalizes evidence-consistent ones.
The institutional fragmentation of American criminal justice compounds this problem. With approximately 18,000 separate law enforcement agencies, 50 state correctional systems, 3,000-plus county jails, and thousands of local courts all making independent program and strategy decisions, there is no central mechanism through which research evidence is systematically incorporated into criminal justice practice. Unlike the Food and Drug Administration, which must approve new pharmaceutical interventions based on clinical trial evidence before they can be marketed, no equivalent regulatory body exists for criminal justice programs. Agencies can adopt programs with no evidence of effectiveness, continue programs with evidence of ineffectiveness, and ignore programs with strong evidence of effectiveness without any external accountability. The default position is institutional inertia — doing what has been done before — broken only by high-profile crime events that generate political pressure for visible response.
The problem is further compounded by the asymmetric political costs of evidence-based reform. When an evidence-supported program reduces recidivism but a participant subsequently commits a serious crime, the political accountability falls on the official who adopted the program — not on the officials who maintained the status quo with a worse overall record. This asymmetry means that evidence-based innovation carries political risk disproportionate to its expected benefit, creating rational incentives for officials to prefer conventional approaches even when evidence favors alternatives. The political visibility of program failures relative to the statistical invisibility of program successes — recidivism rates expressed as population statistics are far less politically salient than a single high-profile crime by a program participant — systematically biases criminal justice policymaking against evidence-supported innovations whose aggregate benefits are real but whose individual failures are vivid and politically exploitable.
The media environment in which criminal justice policy is made reinforces these dynamics. Crime coverage in American media is consistently high relative to actual crime rates, skewed toward violent and sensational incidents, and episodic rather than analytical — producing public environments of crime fear that are disconnected from empirical crime trends and that create political pressure for punitive responses regardless of evidence about their effectiveness. The evidence-based policy movement has invested substantially in research communication and in building relationships with journalists who cover crime and criminal justice, but the structural economics of crime coverage — that dramatic individual incidents generate more audience engagement than statistical analyses of program effectiveness — have proven resistant to the kind of systemic reorientation that would genuinely support evidence-based policymaking.
The Role of Evidence Clearinghouses and Translation Infrastructure
The institutional response to the evidence-to-policy translation problem has been the development of evidence clearinghouses, practitioner-accessible synthesis platforms, and embedded research programs that reduce the cognitive and logistical barriers to evidence use. The CrimeSolutions clearinghouse, maintained by the Office of Justice Programs and continuously updated as new evaluation evidence accumulates, provides practitioners with standardized effectiveness ratings for criminal justice programs across policing, prevention, courts, corrections, and community supervision. The Blueprints for Healthy Youth Development registry at the University of Colorado Boulder performs a similar function for violence prevention programs, identifying programs that meet rigorous evidence standards and providing implementation guidance that supports quality replication.
The What Works Centre for Crime Reduction at the College of Policing represents perhaps the most ambitious evidence translation initiative in international policing, systematically reviewing evidence on crime reduction interventions and producing practitioner-accessible summaries calibrated to the needs of operational decision-makers rather than academic audiences. Its crime reduction toolkit rates interventions on both the strength of evidence supporting them and the size of the effects demonstrated, providing a two-dimensional assessment that addresses the limitation of evidence hierarchies that focus on study quality without weighting effect magnitude. Similar initiatives have been developed in juvenile justice, through the Model Programs Guide of the Office of Juvenile Justice and Delinquency Prevention, and in community violence intervention, through the Health Alliance for Violence Intervention (HAVI) toolkit.
Embedded researcher programs — placing academic criminologists within criminal justice agencies to generate evidence in direct response to operational questions — represent a structural solution to the communication gap between academic and practitioner knowledge. The National Institute of Justice‘s Research and Evaluation on Violence Against Women program and its Sentinel Events Initiative have pioneered embedded approaches that develop research agendas in direct collaboration with practitioners, ensuring that the questions being studied are those that practitioners most need answered and that the findings are communicated in formats that practitioners can actually use. Lawrence Sherman’s concept of translational criminology — applying the medical model’s distinction between basic research and translational research to criminal justice, with explicit institutional investment in the translation function — has influenced how the National Institute of Justice structures its research portfolio.
Implementation Fidelity and the Scaling Problem
Why Evidence-Based Programs Fail When Implemented
One of the most important and most practically challenging findings in applied criminological research is that programs that demonstrate effectiveness in controlled evaluations frequently fail to produce equivalent effects when scaled and disseminated to new implementation contexts. This implementation science problem — the observation that the conditions of successful program evaluation often cannot be replicated at scale — is as important for evidence-based policy as the question of what works in the first place.
Cognitive-behavioral programming in corrections provides the clearest illustration. The meta-analytic evidence for CBT is robust and consistent — well-designed programs reliably reduce recidivism by 10–30% relative to comparison groups — but the real-world implementation of CBT-based programs in American correctional settings frequently departs substantially from the model that produced the evidence. Program dosage is often reduced to accommodate resource constraints; facilitator training and supervision are inadequate to maintain the quality of delivery that effective CBT requires; programs are applied to low-risk offenders who the Risk-Need-Responsivity model predicts will not benefit; and the organizational culture of correctional settings is often incompatible with the therapeutic relationship that effective CBT requires. The result is that CBT programs in routine implementation often produce effects substantially smaller than — and in some cases indistinguishable from — zero, despite the strength of the evidence from controlled evaluations.
The implementation gap reflects a broader truth about evidence-based policy: the evidence establishes that a program can work under favorable conditions, but implementation determines whether it does work in specific organizational contexts. Evidence-based policy that ignores implementation is evidence-based in its program selection but not in its program delivery — and the latter determines the actual outcomes that citizens and communities experience. The emerging field of implementation science, drawing on research from public health, education, and criminology, has developed frameworks for understanding what implementation fidelity requires, what organizational conditions support it, and how fidelity monitoring can be used to maintain program quality over time.
The implementation gap also reflects a deeper tension between the standardization that evidence-based practice requires and the discretion that professional criminal justice practice has historically exercised. Effective cognitive-behavioral programs require consistent delivery of structured curriculum by trained facilitators with adequate supervision — conditions that require practitioners to follow protocols rather than exercise professional judgment in the delivery of each session. Police officers implementing problem-oriented policing strategies are expected to follow analytical protocols and evidence-based response options rather than defaulting to the patrol habits and enforcement strategies they developed through experience. Correctional officers managing offenders in treatment programs are expected to maintain the supportive, pro-social modeling climate that RNR-based programming requires rather than the punitive command-and-control stance that correctional culture often promotes.
These requirements create real organizational tensions. Professional practitioners who have developed expertise through years of experience may experience evidence-based protocols as deskilling — as the substitution of external prescription for professional judgment — even when the protocols produce better average outcomes than individualized discretion. The organizational change management challenges of implementing evidence-based programs are as real as the methodological challenges of evaluating them, and implementation science has documented that organizational leadership commitment, staff buy-in, adequate training and supervision, and ongoing fidelity monitoring are all essential for maintaining the quality of evidence-based program delivery over time. Evidence-based policy that ignores these organizational dynamics risks producing technically correct program choices that are organizationally undermined in implementation.
Normative Dimensions of Evidence-Based Policy
What Evidence Can and Cannot Settle
Evidence-based policy is most productive when it is clear about what research evidence can and cannot establish. Research can establish what the consequences of different policy choices are, with varying degrees of certainty depending on the quality and quantity of available evidence. It can document that mandatory minimum sentences do not produce deterrent effects commensurate with their incarceration costs; that cognitive-behavioral programs reduce recidivism more effectively than punitive interventions; that hot spots policing reduces crime at targeted locations. These are empirical claims about consequences that research methodology is designed to evaluate.
Research cannot, however, settle the normative questions that determine which consequences count as most important and how trade-offs among competing values should be made. Whether a 10% reduction in recidivism is worth the financial cost, the liberty costs imposed on participants, and the potential displacement of resources from alternative investments is not an empirical question — it is a normative one that involves value judgments about liberty, safety, and social investment priorities that democratic processes rather than scientific evidence are the appropriate mechanism for resolving. Whether drug use should be treated as a criminal justice problem or a public health problem, whether violent offenders should be incapacitated longer than evidence of deterrent benefit supports because retributive justice demands it, and whether racial disparities in criminal justice are acceptable if the processes that produce them are individually procedurally fair are all normative questions that evidence bears on without settling.
The most intellectually honest form of evidence-based policy acknowledges these normative dimensions explicitly, providing policymakers with clear evidence about consequences while being transparent about the value judgments embedded in the selection of outcomes to measure and the weighting of competing values. This requires criminologists to distinguish clearly between empirical claims, where their scientific expertise is relevant and authoritative, and normative claims, where their views deserve no special deference over those of any other thoughtful citizen. The conflation of these two kinds of claims — presenting normatively contested policy positions as if they were straightforward empirical conclusions from research — is among the most common and most damaging failures of evidence-based policy advocacy.
Conclusion
Evidence-based criminological policy represents both a genuine intellectual achievement and a work in permanent progress. The methodological infrastructure built over the past three decades — the Campbell Collaboration’s systematic reviews, the CrimeSolutions clearinghouse, the embedded researcher programs, the What Works Centre for Crime Reduction — has produced a substantially more rigorous and more accessible evidence base than criminology possessed in any prior period. The core findings that have emerged from this infrastructure — about hot spots policing, cognitive-behavioral programming, drug courts, focused deterrence, and the iatrogenic effects of programs like Scared Straight — are as well-established as any criminological knowledge and deserve substantially more weight in criminal justice decision-making than they currently receive.
The translation challenge — moving from evidence to policy — requires addressing structural features of democratic governance that no amount of evidence synthesis can by itself overcome. The political economy that rewards punitive responses regardless of evidence, the institutional fragmentation that prevents systematic evidence incorporation, and the implementation conditions that undermine the effectiveness of programs whose evidence base is strong — these are organizational and political problems that require organizational and political solutions alongside the continued production and synthesis of evidence. Evidence-based policy advocates who focus exclusively on generating and communicating evidence while ignoring these institutional barriers are solving the easier part of a harder problem.
The most important contribution that criminology can make to evidence-based policy is perhaps the most fundamental: insisting on clarity about what is and is not known, what the evidence actually supports, and where the crucial choices are normative rather than empirical. A criminology that overstates the certainty of its findings to achieve policy influence, or that presents normative positions as if they were empirical conclusions, ultimately undermines the credibility that makes criminological evidence worth attending to. The discipline’s most durable policy contribution is not any specific program endorsement but the cultivation of the critical empirical intelligence that makes democratic deliberation about crime and punishment more honest and more productive than it would otherwise be.
References
- Andrews, D. A., & Bonta, J. (2010). The psychology of criminal conduct (5th ed.). LexisNexis/Anderson.
- Braga, A. A., & Weisburd, D. (2012). The effects of focused deterrence strategies on crime: A systematic review and meta-analysis. Journal of Research in Crime and Delinquency, 49(3), 323–358. https://doi.org/10.1177/0022427811419368
- DuBois, D. L., Portillo, N., Rhodes, J. E., Silverthorn, N., & Valentine, J. C. (2011). How effective are mentoring programs for youth? A systematic assessment of the evidence. Psychological Science in the Public Interest, 12(2), 57–91. https://doi.org/10.1177/1529100611414806
- Kennedy, D. M. (2011). Don’t shoot: One man, a street fellowship, and the end of violence in inner-city America. Bloomsbury.
- Lipsey, M. W., Landenberger, N. A., & Wilson, S. J. (2007). Effects of cognitive-behavioral programs for criminal offenders. Campbell Systematic Reviews, 3(1), 1–27. https://doi.org/10.4073/csr.2007.6
- Mitchell, O., Wilson, D. B., Eggers, A., & MacKenzie, D. L. (2012). Drug courts’ effects on criminal offending for juveniles and adults. Campbell Systematic Reviews, 8(1), 1–87. https://doi.org/10.4073/csr.2012.4
- Petrosino, A., Turpin-Petrosino, C., & Guckenburg, S. (2013). Formal system processing of juveniles: Effects on delinquency. Campbell Systematic Reviews, 9(1), 1–88. https://doi.org/10.4073/csr.2010.1
- Sherman, L. W., Gottfredson, D. C., MacKenzie, D. L., Eck, J., Reuter, P., & Bushway, S. D. (1998). Preventing crime: What works, what doesn’t, what’s promising. National Institute of Justice. https://www.ojp.gov/pdffiles/171676.pdf
- Travis, J., Western, B., & Redburn, F. S. (Eds.). (2014). The growth of incarceration in the United States: Exploring causes and consequences. National Academies Press. https://doi.org/10.17226/18613
- Weisburd, D., Telep, C. W., Hinkle, J. C., & Eck, J. E. (2016). Is problem-oriented policing effective in reducing crime and disorder? Criminology & Public Policy, 9(1), 139–172. https://doi.org/10.1111/j.1745-9133.2010.00617.x