Broken windows policing reduces serious crime

The claim

Broken windows theory, introduced by political scientist James Q. Wilson and criminologist George Kelling in a 1982 Atlantic Monthly essay, holds that visible signs of disorder — broken windows, graffiti, public intoxication, fare evasion, loitering — send a signal that no one is in control of a space, which in turn invites increasingly serious criminal behavior. The practical prescription is that police should aggressively enforce low-level disorder violations to maintain the social norm of order before it unravels. Proponents argue that this approach explains the dramatic decline in New York City crime beginning in the early 1990s, when Mayor Rudolph Giuliani and Police Commissioner William Bratton made order-maintenance policing central to the NYPD’s strategy. The theory has since been adopted in transit systems, business improvement districts, and cities worldwide, and it continues to inform debates about subway fare enforcement, encampment clearances, and quality-of-life enforcement.

The mechanism

Wilson and Kelling’s proposed causal chain has two links. First, physical and social disorder signals to would-be offenders that social controls are weak, reducing the perceived risk of escalation to serious crime. Second, disorder signals to law-abiding residents that the neighborhood is unsafe, causing them to withdraw from public space, which removes the informal surveillance that deters crime. Policing disorder is intended to break both links: it signals control, retains residents in public space, and — through stops, summonses, and arrests — incidentally surfaces individuals carrying weapons or wanted on warrants.

The mechanism is plausible in its first link. Robert Sampson and Stephen Raudenbush’s influential 1999 Science paper found that collective efficacy — the willingness of neighbors to intervene in disorder — predicted crime rates better than the disorder itself, suggesting that disorder is partly a proxy for social cohesion rather than a direct cause of crime. This reframing shifts the policy implication substantially: if collective efficacy is the operative variable, aggressive enforcement may undermine it by damaging community trust in police, reducing crime reporting, and concentrating enforcement costs on residents who are simultaneously the most likely victims of crime and the most likely targets of misdemeanor enforcement.

The second link — that police-enforced order prevents crime escalation — has received the most empirical scrutiny, and the evidence is mixed. Hot-spots policing, which concentrates patrol in specific high-crime microlocations, has stronger experimental support than disorder policing broadly defined. The conflation of these two strategies has made it difficult to isolate what, if anything, is attributable specifically to the broken windows mechanism versus increased police presence or deterrence-through-certainty.

The evidence

The NYC case and the timing problem

New York City violent crime fell roughly 75% between 1990 and 2010. Bratton and Giuliani attributed much of this to broken windows policing and COMPSTAT-driven accountability. The problem with this attribution is one of timing and concurrence. NYC crime began declining in 1990, before broken windows policing was systematically implemented. Criminologist John MacDonald and colleagues have argued that the crack cocaine epidemic — which drove a surge in violence beginning in the mid-1980s — was already abating by the early 1990s as communities developed norms against crack use and competing drug markets stabilized. National violent crime declined by roughly 40% between 1993 and 2001 across cities that used radically different policing approaches. San Diego implemented community-oriented policing without broken windows tactics and achieved comparable crime declines. Boston’s Operation Ceasefire, which targeted gang violence with focused deterrence rather than disorder enforcement, produced a homicide decline of 63% in two years.

Weisburd et al. systematic review

A 2016 systematic review by David Weisburd, Joshua Hinkle, Anthony Braga, and Alese Wooditch, published in the Journal of Research in Crime and Delinquency, examined 25 studies of disorder policing. The review found that while some studies showed crime reductions in treatment areas, the evidence was inconsistent across sites and methodologies, and effect sizes were generally small. The authors concluded that the disorder-to-crime causal mechanism proposed by Wilson and Kelling received limited support, and that the benefits observed in some studies were better explained by general deterrence effects of increased police presence than by the specific broken windows mechanism. Hot-spots policing — placing officers in defined crime microlocations regardless of disorder level — had substantially stronger evidentiary support in the same literature.

The stop-and-frisk quasi-experiment

New York’s stop-and-frisk program operationalized broken windows theory at scale: from roughly 97,000 stops in 2002 to 685,724 stops in 2011. In Floyd v. City of New York (2013), Federal Judge Scheindlin found that the NYPD conducted stops in violation of the Fourth Amendment, with 88% of stops of Black and Latino men finding no evidence of criminality. Following the Floyd ruling and a mayoral transition, stops declined by over 95% by 2015. If broken windows enforcement was causally responsible for suppressing crime, a 95% reduction in its primary instrument should have produced measurable crime increases. It did not: NYC crime continued to decline through 2015–2019. This near-natural experiment provides the strongest available evidence against the claim that aggressive disorder enforcement is necessary to sustain crime reductions, though proponents argue that the behavioral equilibrium established during the high-enforcement period persisted.

Racial disparity in enforcement

The enforcement costs of broken windows policing are not distributed evenly. ACLU and academic analyses of NYC stop-and-frisk data consistently found that Black and Latino men were stopped at rates disproportionate to their share of the population, and disproportionate to local crime rates even within precincts. Fare evasion enforcement on the NYC subway system showed similar patterns: a 2019 Legal Aid Society analysis found Black riders received approximately 70% of civil summonses despite comprising 35% of ridership. These disparities are not incidental — they follow from the discretionary nature of disorder enforcement, in which officer judgment determines what constitutes suspicious behavior or actionable disorder. Bernard Harcourt’s 2001 analysis in Illusion of Order argued that if minority communities bear a higher enforcement burden, the deterrent logic of broken windows fails: the signal sent is not “disorder will not be tolerated” but rather “certain populations will be policed regardless of behavior.”

Cross-national evidence

Japan achieves a homicide rate of approximately 0.2 per 100,000 — among the world’s lowest — through a policing model built on community officers (koban) embedded in neighborhoods, high clear-up rates, and intensive social support structures, not disorder enforcement. The Netherlands, with an incarceration rate of 62 per 100,000 (versus the US rate of 531), uses a tolerance-based enforcement philosophy for many low-level offenses and achieves crime rates broadly comparable to or lower than the US in most categories. Germany similarly emphasizes proportionality and social integration. The cross-national variation is most consistently explained by social trust, inequality, and social safety net strength rather than disorder policing intensity.

Community trust and crime reporting

A mechanism frequently omitted from broken windows analysis is the role of community trust in crime reporting. Naomi Murakawa and others have documented that in communities subjected to intensive disorder enforcement, residents — particularly undocumented immigrants, young men with prior arrests, and people in informal housing — become reluctant to call police as crime victims or witnesses. If aggressive enforcement suppresses crime reporting, official crime statistics will decline without a corresponding decline in underlying victimization. Kirk and Papachristos (2011) found in Chicago that legal cynicism — the belief that police are not legitimate or effective — mediated between neighborhood disadvantage and crime rates, with high legal cynicism neighborhoods experiencing more violence even after controlling for poverty. Broken windows enforcement, by generating large numbers of low-fruitfulness stops and summonses, may increase legal cynicism in precisely the neighborhoods it is intended to protect.

Who benefits

Mayors and police commissioners can claim credit for crime declines that have multiple causes, and broken windows provides a legible, attributable narrative. Bratton parlayed the NYC story into a career template replicated in Los Angeles and elsewhere. Transit authorities benefit from fare evasion enforcement both as revenue recovery and as a legally defensible basis for removing individuals from transit facilities — including homeless people who do not constitute criminal threats. Private security contractors and business improvement districts have deployed broken windows rhetoric to justify aggressive removal of unhoused individuals from commercial corridors. District attorneys offices have used the misdemeanor case volume generated by disorder enforcement to maintain staffing levels and prosecutorial leverage in plea negotiations. Politicians running on public-order platforms — across partisan lines — benefit from a theory that frames visible enforcement action as causally connected to public safety, regardless of whether the causal connection holds.

The counter

The broken windows framework should not be dismissed entirely. The core intuition that visible disorder affects residents’ perception of safety, and that perception of safety affects neighborhood investment, social cohesion, and willingness to use public space, has empirical support. Sampson and Raudenbush (1999), even while complicating the simple disorder-causes-crime story, confirmed that disorder and crime are correlated and that both respond to neighborhood collective efficacy. Physical remediation of disorder — cleaning vacant lots, repairing abandoned buildings — has shown positive effects in several randomized experiments, including Branas et al. (2018) in Philadelphia, which found that greening vacant lots reduced gun violence in adjacent blocks. This suggests the physical environment matters, even if aggressive misdemeanor policing is not the right instrument for addressing it.

The hot-spots policing literature, which is methodologically stronger than the broader disorder policing literature, does support the claim that concentrated police presence in defined microlocations reduces crime. If broken windows proponents interpret the theory as justifying targeted, place-based enforcement in documented crime concentrations, the evidence is more favorable than a blanket endorsement of citywide disorder enforcement would be. The dispute is partly about the specificity of the mechanism: presence and certainty of response, not the content of the offenses targeted, appear to be the active ingredients.

What the evidence does not support is the aggressive, racially disparate application of misdemeanor enforcement across large populations as a primary crime-control strategy, the claim that disorder policing rather than concurrent social and economic changes drove the 1990s crime decline, or the inference that reducing disorder enforcement necessarily increases serious crime.

References

Wilson, J. Q., & Kelling, G. L. (1982, March). Broken windows: The police and neighborhood safety. The Atlantic Monthly, 249(3), 29–38.

Sampson, R. J., & Raudenbush, S. W. (1999). Systematic social observation of public spaces: A new look at disorder in urban neighborhoods. American Journal of Sociology, 105(3), 603–651. https://doi.org/10.1086/210356

Weisburd, D., Hinkle, J. C., Braga, A. A., & Wooditch, A. (2016). Understanding the mechanisms underlying broken windows theory: The need for evaluation evidence. Journal of Research in Crime and Delinquency, 53(2), 113–160. https://doi.org/10.1177/0022427815617510

Harcourt, B. E. (2001). Illusion of order: The false promise of broken windows policing. Harvard University Press.

MacDonald, J. (2002). The effectiveness of community policing in reducing urban violence. Crime & Delinquency, 48(4), 592–618. https://doi.org/10.1177/001112802237169

Kirk, D. S., & Papachristos, A. V. (2011). Cultural mechanisms and the persistence of neighborhood violence. American Journal of Sociology, 116(4), 1190–1233. https://doi.org/10.1086/655754

Branas, C. C., South, E., Kondo, M. C., Hohl, B. C., Bourgois, P., Wiebe, D. J., & MacDonald, J. M. (2018). Citywide cluster randomized trial to restore blighted vacant land and its effects on violence, crime, and fear. Proceedings of the National Academy of Sciences, 115(12), 2946–2951. https://doi.org/10.1073/pnas.1718503115

Chalfin, A., & McCrary, J. (2018). Are U.S. cities underpoliced? Theory and evidence. Review of Economics and Statistics, 100(1), 167–186. https://doi.org/10.1162/REST_a_00694

Eck, J. E., & Maguire, E. R. (2000). Have changes in policing reduced violent crime? In A. Blumstein & J. Wallman (Eds.), The crime drop in America (pp. 207–265). Cambridge University Press.

New York Civil Liberties Union. (2012). Stop and frisk: The human impact. NYCLU. https://www.nyclu.org/report/stop-and-frisk-human-impact