What does "broken windows" policing mean? By Matthew Yglesias

www.slowboring.com

14 - 18 minutes

I like to write about cities. In recent years, this has meant writing a lot about crime and policing debates in which the concept of “broken windows policing” looms large.

On one side of these debates are people who are absolutely convinced that the problem of urban crime has been solved, conceptually, and that the solution is a set of policies called “broken windows policing.” Those on the other side are equally convinced that “broken windows policing” comprises a set of policies whose merits have been decisively refuted by empirical evidence.

I don’t think either side is correct, because I don’t think “broken windows policing” or “broken windows theory” clearly refers to a specific set of policies. There are strong feelings and affective polarization around the general idea of toughness on crime. But going back to the specific original claims of broken windows theory and the various empirical tests that have been proposed, the whole thing is a mess. I think totally ignoring this term and instead debating specific proposals would be a lot more useful.

That said, the people want a take! So it’s helpful, I think, to start at the beginning.

In March of 1982, George Kelling and James Q. Wilson published an article in The Atlantic titled “Broken Windows.” This is as close as one can come to an official description of broken windows policing and the theoretical and empirical basis for it. It’s useful, as an exercise, to try to purge your mind of everything you think you know about broken windows theory, about the 1990s crime drop, about George Floyd, etc., and just read what the article says.

In their own telling, the theory is based on an experiment conducted in 1969 that was literally about broken windows:

    Philip Zimbardo, a Stanford psychologist, reported in 1969 on some experiments testing the broken-window theory. He arranged to have an automobile without license plates parked with its hood up on a street in the Bronx and a comparable automobile on a street in Palo Alto, California. The car in the Bronx was attacked by “vandals” within ten minutes of its “abandonment.” The first to arrive were a family—father, mother, and young son—who removed the radiator and battery. Within twenty-four hours, virtually everything of value had been removed. Then random destruction began—windows were smashed, parts torn off, upholstery ripped. Children began to use the car as a playground. Most of the adult “vandals” were well-dressed, apparently clean-cut whites. The car in Palo Alto sat untouched for more than a week. Then Zimbardo smashed part of it with a sledgehammer. Soon, passersby were joining in. Within a few hours, the car had been turned upside down and utterly destroyed. Again, the “vandals” appeared to be primarily respectable whites.

Here, again, is another statement of the thesis regarding literal windows:

    Social psychologists and police officers tend to agree that if a window in a building is broken and is left unrepaired, all the rest of the windows will soon be broken. This is as true in nice neighborhoods as in rundown ones. Window-breaking does not necessarily occur on a large scale because some areas are inhabited by determined window-breakers whereas others are populated by window-lovers; rather, one unrepaired broken window is a signal that no one cares, and so breaking more windows costs nothing.

The claim here is that the presence of broken windows signals the lack of consequences for breaking additional windows, leading rapidly to further window breakage. That’s an interesting hypothesis, though as someone more familiar with contemporary policing debates than those of the late 1960s, it seems to have surprisingly little direct connection to anything — there was never a movement afoot to decriminalize window-breaking, at least not that I’m aware of. And even though a bunch of people got shot in my neighborhood this summer, I’m pleased to report that almost none of the buildings have broken windows.

Of course the article itself goes on to make some broader claims.

But it’s very challenging to reconstruct from the piece exactly what they are trying to argue. The most specific debate they are engaged with seems to be a somewhat obsolete one about whether police officers should primarily engage in foot patrols or be dispatched via squad cars. But they also remark on everything from decriminalizing prostitution (they’re against it) to handling the mentally ill (de-institutionalization was bad) to smoking on mass transit (kick the smoker out but don’t arrest him). They even ask the very contemporary question “how do we ensure, in short, that the police do not become the agents of neighborhood bigotry?”

This makes for a very interesting piece of long-form journalism. But also, I think, for an incredibly confused debate over whether broken windows policing is a good idea — because it’s really unclear what the idea even is.

I usually tell people that it’s better to familiarize yourself with the broad contours of a whole literature than to focus on just one study, and the problem with the literature on broken windows theory is that few of the papers seem to be asking the same question.

For example, Hope Corman and Naci Mocan published a paper in 2002 that found, in New York City, that “the broken windows hypothesis has validity in case of robbery and motor vehicle theft.” But what hypothesis? I wish they’d researched whether the prevalence of broken windows has a causal impact on robbery and motor vehicle theft, but they didn’t. In fact, as far as I can tell, nobody in the literature on broken windows policing has done any studies that relate to broken windows, even though the hypothesis as stated in the article is not a metaphor. What Corman and Mocan find is that areas with a larger increase in misdemeanor arrests saw a larger decline in robberies and motor vehicle thefts. I don’t blame them for characterizing this as “broken windows” policing because lots of people use the phrase that way. But Kelling and Wilson don’t make any specific claims about misdemeanor arrests. And the papers that are more skeptical of so-called broken windows policing don't re-rerun this same test in other cities to see if there’s external validity — they look at different things.

Bernard Harcourt and Jens Ludwig, for example, have a study in which they look at the results of the Moving to Opportunity program, which randomly assigned certain low-income families to new neighborhoods. They find that being sent to live in a more orderly place had no impact on a person’s odds of committing crimes down the road, which they take as disconfirmation of broken windows.

In a later paper, Harcourt and Ludwig look specifically at misdemeanor marijuana arrests in New York City and find no causal relationship to declines in more serious crimes. They recommend that marijuana possession be reclassified as an administrative violation rather than a misdemeanor. This, again, they characterize as a knock against broken windows.

On the flip side, a paper from Jacinta Gau, Nicholas Corsaro, and Rod Brunson offers “confirmation of a portion of broken windows theory, in that disorder may inspire fear partially as a result of its detrimental impact on neighborhood cohesion and shared expectations for social control.” Ren, Zhao, and He have a similar finding: “Study results indicate that disorder incidents had a significant impact on perceptions of disorder, while perceptions of disorder tend to increase fear for personal safety, which in turn reduces sense of collective efficacy.” But note that these papers don’t specifically contest what Harcourt and Ludwig are saying. By the same token, David Weisburd, Joshua Hinkle, Christine Famega, and Justin Ready characterize their results as going against broken windows policing — they found that surging police officers into crime hotspots “had no significant impacts on fear of crime, police legitimacy, collective efficacy, or perceptions of crime or social disorder.”

But, again, my point here is that these papers aren’t so much finding different results as testing different things. Weisburd et al. don’t test whether misdemeanor arrests lead to reductions in robberies, and Ren, Zhao, and He don’t test whether perceptions of disorder lead to increases in violent crime.

I of course don’t blame any of the individual researchers for this disorderly research landscape, but I do wish someone would come in and impose some order on the research agenda. Because clearly one of the things that happened here is that broken windows leaped from the pages of The Atlantic into the political rhetoric of Rudy Giuliani’s administration (probably via the Manhattan Institute) and the debate thus became murky and ideologically loaded, in part because Rudy’s team obviously did more than one thing.

As Bench Ansfield wrote in the Washington Post back in 2019, Zimbardo was pretty clearly trying to make a left-wing point with his observation.

A lot of people’s folk understanding of crime is that some people are just bad criminals by nature and what we need to do is get tough and lock them up. This idea is bound up with racial stereotyping and with a kind of implicit assumption that the “toughness” won’t be visited on your son or his friends if they slip up; it’s those people that we need to get tough on. Zimbardo says this is wrong; that he can turn Palo Alto passers-by into maniacs just by creating a criminogenic environment. Kelling and Wilson say Zimbardo is right, and that’s why we need to lock up drunks, addicts, and other sources of urban disorder. It’s through this mechanism that it becomes a right-wing take. Harcourt and Ludwig say in their MTO paper that Kelling and Wilson are wrong, which on the surface is a left-wing take, but in many ways takes us back to the pre-Zimbardo view that some folks are just bad news.

If you wiped your mind of any knowledge of the “broken windows” debate and heard about the Moving to Opportunity program and also heard that some suburbanites were worried that introducing MTO families into their orderly neighborhood would cause crime, you’d assume that was the right-wing position. Indeed, when Raj Chetty and the Opportunity Insights team looked again at MTO and found larger benefits for kids who moved at younger ages, that was universally understood as a progressive take. But crucially, that team doesn’t invoke the magic words “broken windows” or suggest that their finding is in any way relevant to policing debates.

By the same token, there’s a line of research that argues that cleaning up vacant lots has crime control benefits.

Every paper that I have seen with this conclusion frames it as demonstrating that there are non-policing, non-carceral approaches to crime reduction that we should be trying. I think that reflects, in part, the fact that the overwhelming majority of academics are on the left, so given an interesting empirical finding, they tend to emphasize its alignment with progressive concerns. But if I were Jon Bel Edwards trying to pitch the Louisiana state legislature on funding a vacant lot remediation program for my state’s very violent cities, I would tout this as an example of broken windows theory. It is, after all, far and away the closest thing to a literal test of the hypothesis.

At any rate, I don’t want to claim to be able to read people’s minds, but I think it’s probably best to understand the country as having an ideological debate about the merits of aggressive policing in which the so-called “broken windows” theory is just a bystander. The idea that neighborhood-level disorder is criminogenic is ideologically neutral or even left-wing. What’s right-wing is the idea that the best way to address neighborhood-level disorder is by having the police arrest people. But if you say “Harcourt and Ludwig showed neighborhood-level effects aren’t causal, we just need to round up the usual suspects and throw away the key,” nobody is going to congratulate you on your progressive bona fides as an opponent of broken windows theory.

This whole thorny situation is why I’ve been very reluctant to engage with the rhetoric around “broken windows.” To me, it’s a kind of empty signifier that doesn’t lend itself to discussing the merits of specific policies.

There are two things associated with tough policing that I think are correct, neither of which particularly hinges on the Zimbardo hypothesis:

    Policing is a superior mechanism of upholding neighborhood standards compared to land use regulation.

    You need to find ways to deter people from illegally carrying concealed handguns on their person or in their car. 

If you go to a shopping mall, you’ll find that private-sector shopping mall landlords find it worth their while to invest in private security personnel who generally don’t tolerate much bullshit from the clientele. You can’t do graffiti, you can’t shoot heroin, you can’t deal drugs, and you can’t run around screaming at people. I don’t think malls conduct themselves that way because of esoteric academic theories — they think that customers prefer to visit a clean and orderly mall and that it is therefore worth their while to maintain certain standards. I think that in practice, when progressives demand de-policing of public spaces, they are pushing middle-class people to spend all their time in neighborhoods that zone out the poor while driving back and forth to privately secured malls and office parks. A thriving city needs to be a place where people want to spend time, and that is a fully sufficient rationale for investing in basic order maintenance.

You also need to worry about people getting shot. You can certainly hope that a general atmosphere of orderliness helps prevent that, but the reality is that a lot of people getting shot will contribute to a sense of disorder.

As liberals like to say, the sheer quantity of people walking around with guns is a big factor here, but most people doing shootings are not legal handgun owners. The problem is that enforcing the law is difficult. Superman could fly over to Gotham, use his X-ray vision to sweep the city, and just start scooping up every concealed weapon in the city. But civil libertarians would raise significant questions about the privacy and due process implications of that. It seems to me that being fairly strict about things like traffic violations, fare-beating, open containers of alcohol, shoplifting, and other “low-level” crimes is a reasonable middle ground between “don’t enforce the gun rules” and “build an all-pervasive surveillance regime.” That’s compatible with the idea that the punishment for breaking minor laws doesn’t have to be especially severe, because the point of making the arrest is to check for gun possession crimes. But you do need to punish those or there’s no deterrence.

I think this is a completely different causal mechanism than what’s proposed in “broken windows,” but it is basically convergent with their conclusion.