What Is Broken Windows Theory? Origins, Policing Impact, and Why the Evidence Falls Apart

Where It Came From

In March 1982, social scientists James Q. Wilson and George L. Kelling published an essay called “Broken Windows” in The Atlantic Monthly. (Wilson & Kelling, 1982) The title was a metaphor: if a broken window in a building goes unrepaired, it signals that nobody is watching, nobody cares, and more windows will follow. Left unchecked, the theory went, visible disorder in a neighborhood would invite serious crime.

The intellectual scaffolding for the essay came partly from a 1969 experiment by Stanford psychologist Philip Zimbardo. Zimbardo parked a car with its hood up and no license plates in the Bronx, and an identical car in Palo Alto, California. (Zimbardo, 1969, cited in Wilson & Kelling, 1982) In the Bronx, a family stripped the battery and radiator within ten minutes. In Palo Alto, the car sat untouched for days — until Zimbardo himself smashed it with a sledgehammer, at which point bystanders joined in. Wilson and Kelling used this to argue that disorder signals permissiveness, and permissiveness breeds more disorder.

The argument had intuitive appeal. It also had no original empirical data behind it. Wilson and Kelling were not reporting the results of a study. They were advancing a hypothesis — one that would nonetheless become the theoretical foundation for urban policing strategies implemented across the United States within a decade.

How It Shaped American Policing

The theory sat in academic circles through most of the 1980s. Then crime rates in New York City hit historic highs in the late 1980s and early 1990s, and broken windows found its moment.

In 1994, Mayor Rudy Giuliani took office and appointed William Bratton as NYPD Commissioner. Both credited Wilson and Kelling’s framework directly. (Britannica, 2024) The resulting strategy targeted minor infractions — turnstile jumping, public drinking, loitering, panhandling, graffiti, prostitution — as the entry point to preventing serious crime. Misdemeanor arrests in New York climbed from approximately 133,000 in 1993 to over 205,000 by 1996. (Simply Psychology, 2025)

Crime did fall in New York during this period — dramatically. Giuliani and Bratton pointed to broken windows policing as the reason. The story spread. Cities across the country adopted variations of the same approach, applying intensive misdemeanor enforcement as a crime-reduction strategy. The theory had become policy, and the policy had become, in certain circles, doctrine.

Beyond New York, the policy spread through a network of political appetite and institutional momentum. Stop-and-frisk — the practice of stopping and searching pedestrians based on “reasonable suspicion” — became its most visible and controversial instrument. In the decade leading up to 2013, NYPD officers conducted millions of such stops. The constitutionality of the program was ultimately challenged and ruled unconstitutional by a federal court in Floyd v. City of New York (2013). (Wikipedia, Broken Windows Theory)

The Impact on Communities

The communities that absorbed broken windows policing were not random. The theory’s application concentrated enforcement in low-income neighborhoods, and those neighborhoods were disproportionately Black and Latino. This was not an accident or an anomaly. It was measurable and documented.

Research by Columbia Law School professors Jeffrey Fagan and Garth Davies found that the strongest predictor of where stop-and-frisk activity concentrated was not the presence of disorder — it was racial composition and poverty levels. (Fagan & Davies, 2000) Policing, their data showed, was not tracking broken windows. It was tracking Black and brown people in poor neighborhoods. “Our empirical evidence suggests that policing is not about disorderly places,” they wrote, “but about policing poor people in poor places.”

The harms were not only constitutional. Research has linked aggressive order-maintenance policing to poor academic outcomes for minority youth, psychological trauma from repeated stops, eroded trust in institutions, and the criminalization of poverty and homelessness. (Legewie & Fagan, 2019; Geller et al., 2014)

There is also the matter of what a misdemeanor arrest actually costs someone. A ticket or summons for a quality-of-life offense can escalate into a bench warrant for a missed court date, a warrant into an arrest, an arrest into a fine that grows with fees, a fine into a license suspension, a suspension into job loss. That sequence — well-documented in low-income communities — has nothing to do with disorder and everything to do with how the system compounds disadvantage. (Vera Institute, 2022)

Applying broken windows policing also produced what researchers at Texas Southern University called “unintended consequences” — specifically, the diversion of limited public resources away from social programs and toward enforcement and incarceration, undermining the community trust that effective policing actually requires. (Texas Southern University, 2015)

Why the Theory Doesn’t Hold

The premise was never verified. That is not a polemical statement — it is the conclusion of multiple independent researchers who went back and looked at the data.

Columbia Law professor Bernard Harcourt has been the most systematic critic. In his 2001 book Illusion of Order: The False Promise of Broken Windows Policing and subsequent research co-authored with economist Jens Ludwig, Harcourt reanalyzed the data that was most often cited as supporting broken windows — including the work of political scientist Wesley Skogan — and found that the disorder-crime link largely disappeared when researchers controlled for poverty, residential instability, and race. (Harcourt, 2001; Harcourt & Ludwig, 2006) The connection between fixing broken windows and reducing serious crime was not there.

Harcourt and Ludwig also examined a natural experiment: the federal Moving to Opportunity program, which moved thousands of low-income families from high-disorder public housing into lower-disorder neighborhoods. Broken windows theory predicts those families should commit less crime once exposed to orderly environments. They did not. Crime rates remained the same. (Harcourt & Ludwig, 2006)

Critics also raised a structural problem with the theory from the beginning: it gave enormous discretion to individual officers to determine what counted as “disorder” and who counted as “disorderly.” Legal scholar Gary Stewart wrote that arguments for low-level police intervention, including broken windows, often function “as cover for racist behavior” because vaguely written ordinances allow officers to determine who engages in disorderly acts — and that determination has consistently produced racially skewed outcomes. (Stewart, cited in Wikipedia)

University of Michigan criminologist David Thacher put it plainly: broken windows theory “frames trivial misbehavior as the beginning of something much more serious.” That framing changes how officers perceive the people in front of them — not as individuals engaged in a minor infraction, but as potential escalators to serious crime. The consequences of that framing are not abstract. (PBS Frontline, 2016)

What Actually Works

The question after broken windows is not “more enforcement or less” — it’s a more specific question about what enforcement is for and what problem it’s actually solving. The research points in a clear direction.

Community Violence Intervention

Community Violence Intervention (CVI) programs employ credible messengers — people with direct experience in the criminal justice system — to interrupt cycles of retaliatory violence before they escalate. These programs engage people at the highest risk of gun violence and operate outside the police response chain. The Vera Institute of Justice has documented that CVI investment improves safety and saves cities millions compared to enforcement-heavy approaches. (Vera Institute, 2022)

Behavioral Health Response Teams

A significant portion of 911 calls involve mental health crises, substance use, homelessness, and other situations that do not require an armed law enforcement response. Cities including Denver, Eugene (Oregon), and Olympia (Washington) have implemented civilian crisis response teams that handle these calls. Early data from these programs show reduced costs, reduced arrests for low-level offenses, and higher rates of people being connected to actual services. (Vera Institute, 2022)

Problem-Oriented Policing

Unlike zero-tolerance enforcement, problem-oriented policing focuses on identifying the specific, local conditions that produce repeat incidents at specific locations — and addressing those conditions with community input. It is targeted, not sweeping. The National Academies found evidence that this approach, unlike aggressive misdemeanor enforcement, can be effective. The key distinction is specificity and community partnership versus generalized crackdown. (Braga et al., 2024)

Investment in Root Causes

The research on what actually reduces crime over time consistently points toward the same factors broken windows ignores: stable employment, adequate housing, access to behavioral health treatment, and educational opportunity. Communities with stronger social and economic infrastructure have lower rates of serious crime — not because of policing strategy, but because the conditions that produce desperation and disorder are less severe. (Vera Institute, 2022; Simply Psychology, 2025)

Pre-Arrest Diversion

Where low-level offenses are involved, evidence-based pre-arrest diversion and citation programs — rather than arrest — maximize public safety outcomes without the cascading consequences of a criminal record. They also free officer capacity for serious crime response. (R Street Institute, 2023)

Broken windows theory was a magazine hypothesis that got handed a police department. The communities that bore the cost of its implementation had no say in that handoff, and few mechanisms to push back until the harm was well underway. Understanding where it came from — and what the evidence actually says — is the starting point for any serious conversation about what policing is for and who it serves.