Legal Affairs
space
space
Debate Club
DEBATE CLUB 10/17/05

Is Broken Windows Policing Broken?

Bernard E. Harcourt and David E. Thacher

This Week's Entries: Monday | Tuesday | Wednesday | Thursday | Friday

Today, major cities from New York to Los Angeles are vigorously enforcing laws against vagrancy, panhandling, and other quality-of-life crimes. This practice is based on the "broken windows" theory of policing, which suggests that a reduction in minor crimes will lead to a decrease in violent ones.

While "broken windows" policing has grown steadily in popularity in the last 25 years, recent analyses suggest that the results don't support its claims. Critics have begun to argue that, by chasing jaywalkers, cities are wasting resources. Is it time to forget about broken windows?


Bernard E. Harcourt is Professor of Law at University of Chicago and author of the forthcoming book, Language of the Gun: Youth, Crime, and Public Policy. David E. Thacher is Associate Professor of Public Policy and Urban Planning at University of Michigan.

Harcourt: 10/17/05, 09:17 AM
It is definitely time for law enforcement to stop focusing on minor disorder and to target, instead, serious crimes involving guns and physical injury. Given the increased demands on police resources from terrorism (particularly following the bombings in London), and the decreased supply of police resources across the country (due to the scaling back of the federal COPS program and local and state budget troubles), it is even more important today to allocate properly the smaller and smaller crime-fighting dollars and officers.

As the National Research Council's report on Fairness and Effectiveness in Policing notes, although many policymakers continue to believe that broken-windows policing reduces serious crime, the empirical support for the idea now rests entirely on a 2001 Manhattan Institute study of New York City data by George Kelling (co-author of the broken windows essay with James Q. Wilson) and William Sousa (a graduate student of Kelling's at the time).

In a forthcoming article titled Broken Windows: New Evidence from New York City and a Five-City Social Experiment, Jens Ludwig and I conduct the most recent work-up of the New York City crime data, and explore new findings from an important five-city social experiment called Moving to Opportunity (MTO). We find no evidence for the proposition that disorder causes crime or that broken-windows policing reduces serious crime.

One argument we make in our paper is that Kelling and Sousa's interpretation of what happened across New York City precincts seems—how to put this diplomatically?—flat wrong. The pattern of crime reduction across New York precincts during the 1990s is entirely consistent with what statisticians call mean reversion: those precincts that experienced the largest drop in crime in the 1990s were the ones that experienced the largest increases in crime during the city's crack epidemic of the mid- to late-1980s. Jens and I call this Newton's Law of Crime: what goes up, must come down, and what goes up the most, tends to come down the most.

Jens and I also explore the empirical results from MTO, a social experiment underway in five cities, including New York, Chicago, and Los Angeles—three of the largest cities that implemented broken-windows style policing—as well as Baltimore and Boston. Under the MTO program, approximately 4,800 low-income families living in high-crime public housing communities characterized by high rates of social disorder were randomly assigned housing vouchers to move to less disadvantaged and disorderly communities. Using official arrests and self-report surveys, Jens and I compare the crime rates among those who moved and those who didn't and the results are clear, though disappointing: moving people to communities with less social or physical disorder on balance does not lead to reductions in their criminal behavior. Neighborhood order and disorder do not seem to have a noticeable effect on criminal behavior.

You suggest, David, in your excellent essay Order Maintenance Reconsidered, that "social science has not been kind to the broken windows theory." I think that's right. Jens and my most recent findings are consistent with the other research on broken windows that I summarize in Illusion of Order: The False Promise of Broken Windows Policing including Robert Sampson and Stephen Raudenbush's work on Chicago neighborhoods, Jeffrey Fagan's research on New York City, and Ralph Taylor's book on Baltimore neighborhoods. The bottom line is that there's no reliable empirical support for the broken windows theory or for broken-windows policing.

Thacher: 10/17/05, 12:29 PM
It's true that I said social science has not been kind to Broken Windows, but I think that's largely because it has asked the wrong questions. Somehow the question of whether police should take order maintenance more seriously got equated with the question of whether doing that would reduce crime. I think that's an interesting and a little dispiriting comment on our culture. It's consequentialism gone awry: nothing is worthwhile in itself; you have to show that it has some other good consequence before doing it. As if there were no reason for a cop walking by to do something about a guy urinating in the middle of the street in a commercial district; or to tell a man who propositions and harasses every woman who walks past him to cut it out; or to lay down some rules with a panhandler who gets in people's faces and won't take no for an answer; or to tell a guy lying down on the subway steps, blocking people who are trying to get to work, that he's got to get up—unless by doing all these things the cop would prevent a statistically-significant number of burglaries next month.

Police should do something about the things I mentioned because they are wrong. They make unfair use of public spaces. They frighten and offend people gratuitously. They destroy the sense of livability in a place, and they demoralize the people who live there. All these problems are particularly concentrated in our most disadvantaged neighborhoods, and we owe it to the people who live there to give them the same quality of life in their public spaces that the middle-class takes for granted. Of course it's true that police need to have priorities (if my cop is running past this parade of horrors on his way to a hot call about an armed robbery in progress, then yes, I agree, he shouldn't stop!), but it's important to remember that disorder is much more prevalent than crime. That's probably why it can affect our sense of safety more than crime itself does. Disorder is what you can see.

So that's my basic position, but I'd be derelict in my debater's duties if I didn't say something about Broken Windows. Does order maintenance reduce crime? I think the bottom line is that nobody knows. In my view, both the critics and the proponents of Broken Windows underestimate how hard it is to study an issue like this in a convincing way. Very little of the existing literature even comes close. Maybe I hang out with too many finicky econometricians, but I have a hard time seeing how we can conclude anything at all about Broken Windows from most of it.

The only exception I can think of is the study that Anthony Braga and his colleagues did in Jersey City a few years ago, and that study is much more encouraging than your summary of the evidence suggests. Braga's group picked out 24 areas in the city and matched them up into 12 pairs. Then for each pair, they flipped a coin to decide which area would get special police attention. What kind of attention, exactly? On paper it was what the policing world calls "problem-oriented policing," but on the ground it's clear that by far the most important element of what the Jersey City cops did was order maintenance. They dispersed loiterers, issued summonses for public drinking, and posted signs laying down neighborhood rules. The results? Crime and disorder fell dramatically in the 12 areas that got the treatment compared with the 12 that didn't. It's true that a few things other than order maintenance happened in some of the experimental areas, but if there are a few confounding variables in the Jersey City study, the rest of the literature is drowning in them. If we're going to talk about the best evidence we have on Broken Windows (beyond the whole body of research in social psychology that supports its assumptions), I think this is it, and it's very positive.

This Week's Entries: Monday | Tuesday | Wednesday | Thursday | Friday

Harcourt: 10/18/05, 09:11 AM
I'm glad you mention the excellent study Problem-Oriented Policing in Violent Crime Places by Anthony Braga, David Weisburd, and their colleagues. That study does find that increased police presence in violent crime "hot spots" has a good effect on crime. The trouble is, the study does not—nor does it attempt to—distinguish between a traditional deterrence explanation (i.e. increased police presence serves as a deterrent to crime by increasing the probability of detection) and the broken windows theory. The authors acknowledge as much. The bottom line from that study is that the police do matter—and that more is better, at least for crime reduction.

But we know that the police matter. A reliable body of empirical literature demonstrates, convincingly I believe, that increased police presence, spending, and resources do indeed reduce crime. The most recent evidence is my colleague Steve Levitt's research on the 1990s crime drop, in which Levitt demonstrates that increased police investment may account for up to one-tenth of the overall reduction in crime between 1991 and 2001.

The question we are addressing is not whether to put more officers on the beat, but rather what the same number of officers should do with the same amount of time. Should they be arresting, booking, processing, and clogging our courts with minor disorder offenders or instead investigating patterns of violent crime and gun shootings? I think the answer is clear: targeting police resources on violent and gun crime in the highest-crime "hot spots" is more effective. (Larry Sherman's research on Fair and Effective Policing is the places to go here).

Now I agree, David, that it would be nice not to have to jump over some guy lying down on the subway steps on the way to work. It would be even nicer if there were no homeless people and no homelessness. In fact, it would be even better if everyone lived in a palace and drove a Mercedes-Benz. But that's not the question. In the real world, there are limited resources. There's a trade off. If a police officer is booking a turnstile jumper at the precinct, she's not investigating that string of burglaries you mention. And the trade off has become even more important today given that police resources are stretched thin with the war against terrorism and budget cuts.

Why is there no reliable evidence supporting broken-windows policing? The answer is, there's no good reason to believe that disorder is causally related to crime. There are only two social scientific studies that explore the relationship between disorder and crime on large data sets—the Sampson and Raudenbush study of Chicago neighborhoods that I mentioned yesterday, and Jens Ludwig and my newest study on MTO—and both of these studies undermine the idea that disorder causes crime.

This raises probably our sharpest disagreement. You write that the social sciences have "asked the wrong questions" and suggest that they should instead weigh in, somehow, on the fact that graffiti and panhandling "are wrong." But that's a moral, political, legal, and aesthetic judgment, not a social scientific question. The social sciences have been asking the only appropriate questions for social scientific inquiry in light of the broken windows hypothesis. I'd be happy to debate whether disorder is "wrong," but realize, that doesn't call for scientific reasoning. It doesn't call for regression analysis or ethnography.

And ultimately, I don't think that "right" or "wrong" are the correct categories. Those terms don't begin to capture the complexity of order and disorder. Once we move away from the narrow social scientific inquiry, then a whole new set of questions arise, and they're different than the ones you mention. The proper questions are: How do we define disorder? Who defines disorder? How do we come to believe that any conduct is disorderly? In other words, what do we see as disorderly and at what price? You suggest that disorder, in contrast to crime, "is what you can see." That, I believe, is what the debate is really all about.

Thacher: 10/18/05, 01:44 PM
I certainly agree with you that the Jersey City study doesn't provide an unequivocal test of Broken Windows. But if it's confounding variables like overall level of police presence you're worried about, the rest of the literature is in much worse shape than Jersey City, no matter how large their datasets. Forgive me for picking on your MTO study as an example (it's the nefarious influence of those finicky econometricians!), but while it's true that people who moved to more orderly neighborhoods in MTO didn't have lower arrest rates compared with people who stayed in public housing, there were zillions of differences between the movers and the stayers other than the level of disorder immediately surrounding their apartments. For starters, the movers had their lives, friendships, expectations, and daily routines disrupted and challenged in a way that the stayers didn't. Worse, if your study casts doubt on Broken Windows, it also casts doubt on the idea that you can reduce crime by improving schools, job opportunities, police responsiveness, and neighborhood social capital, or by deconcentrating poverty and disbanding gangs, since the full interim MTO report suggests that the movers improved on all those dimensions compared with the stayers too. (I'll leave aside the serious unit of analysis problem involved in using arrest rates for individuals to investigate a theory about the factors that affect crime at places.) Better, I think, to conclude that too many things are going on in MTO to disentangle the real effect of any single factor like disorder. By comparison, in Jersey City the only systematic differences between the experimental areas and the controls were a coin toss and a group of officers who viewed the heart of their jobs as order maintenance. And crime fell.

The questions you raise about police priorities are clearly important, though I'm not convinced that the tradeoffs are as stark as you suggest. (The notion that expecting police to do something when they see a guy urinating in the middle of the street is comparable to wishing that everyone owned Mercedes-Benz's has me puzzled.) Are you really saying that a police officer walking past my parade of horrors ought to just keep walking? That police should stop maintaining order entirely and devote 100% of their time to burglary investigations and vehicle stops searching for guns (forgetting the fact that order maintenance itself helps police uncover contraband—a point I thought you acknowledged in Illusion of Order)? Mayors have lost their jobs for that, because the public expects police to maintain order.

And the public is right to expect it. Without some basic modicum of order, society is not just unpleasant but impossible. If, for example, we are going to have shared public spaces, everyone has to observe some basic norms of cooperation and civility in order for those spaces to remain usable at all. People simply abandon the subways, parks, and buses when disorder becomes severe. That's not a problem for the corporate executives who can drive straight to work from their gated communities to their suburban office parks, but the rest of us expect and need our public spaces to be livable. Forgive me for being a little polemical, but when did the left become the hardnosed realists with lowered expectations about what our public assets can aspire to?

You seem to suggest that order maintenance is about getting rid of the homeless and treating every form of panhandling as "wrong," but that's simply not true. As George Kelling has explained in detail, and as I've discussed too, serious order maintenance (as opposed to the absurd and ultimately impossible "zero tolerance" idea it's too often confused with) is not about excluding "undesirables" from public spaces but about ensuring that everyone abides by the standards of order that make public life possible, and treating them with the dignity involved in believing that everyone is a moral agent who can tell appropriate from inappropriate behavior. When those tasks go untended, as they have in places like the New York City subway during the 1980s (read Malcolm Gladwell's characteristically gripping account of how those conditions led to the bizarre and ultimately frightening celebration of Bernhard Goetz as a kind of "hero" after he shot four young black men on the subway during its darkest days), that's when the public demands that the homeless simply be gotten rid of—shunted into prisons or out to isolated skid rows where no one will have to see them. The exclusionary and punitive political attacks on the homeless in San Francisco over the past decade or so (such as the cuts and restrictions on general assistance benefits) followed widespread public sentiment that behavior by too many of them had become intolerable and the refusal of the mayor to take action against the chaos in city parks. Only by maintaining and, where necessary, enforcing some reasonable expectations about everyone's public behavior is it possible for a diverse society to share its public spaces peaceably. (Thomas Nagel defends a version of this idea in a brilliant essay.) The only alternative, as far as I can tell, is the segregation and exclusion that you rightfully seem to be worried about. Maybe order maintenance is like democracy: The worst option, except for all the others.

This Week's Entries: Monday | Tuesday | Wednesday | Thursday | Friday

Harcourt: 10/19/05, 09:12 AM
What I'm hearing you say, David, is that some basic modicum of order is necessary for urban living. At that level of generality, though, I don't think anyone could possibly disagree. You've watered down the broken windows theory so much, it no longer does any work. If it is true, as you suggest, that "nobody knows" whether order maintenance reduces crime, and if it is indeed true, as you suggest, that there is no real trade-off between broken-windows policing and "hot spots" policing, then there's not much left to the broken windows theory.

To be sure, ever since the publication of Wilson and Kelling's essay, there's been a lingering question about what it really takes to implement a broken-windows approach. Does it require zero-tolerance or discretionary policing or, for that matter, any policing at all? But George Kelling resolved that question in his 2001 Manhattan Institute report—the report that Jens Ludwig and I analyze in our forthcoming article. Kelling explicitly measures broken windows policing by the number of misdemeanor arrests in a precinct. In his own words, he uses "arrests for misdemeanors as our measure of 'broken windows' enforcement." (Incidentally, this is clear from the New York City experience: as I demonstrate in my Boston Review essay, as soon as Bill Bratton began implementing broken windows in 1994, adult misdemeanor arrests began climbing steadily).

Those arrests consume massive law enforcement and judicial resources. In New York City in 2004 alone, there were about 190,000 adult misdemeanor arrests. When I calculate the cost of each arrest, and factor in officer time and resources for transporting the arrestee to the precinct, booking, searching, charging, jailing, reviewing the charge, presenting, and releasing—not to mention the costs associated with prosecuting the charges in court—this translates into millions of dollars. Deborah Rhode of Stanford Law School estimated, back in 1995, that a typical prosecution for prostitution costs upwards of $2,000 on average. In New York City, quality-of-life prosecutions clogged the courts to such an extent that they had to invent new courts, called "community courts," to deal with the congestion—and it turns out, according to a recent NIJ Report on community courts, that those new courts are even more expensive!

At $2,000 a pop for judicial costs and, very conservatively, another $200 for officer time and precinct costs, New York City misdemeanor arrests in 2004 cost us over $400,000,000. I see a real trade-off here. Take that money, hire more police officers, and direct them to do "hot spots" policing of violent and gun crime: that will more effectively reduce violent and gun crime and improve our quality of life.

As for Jens and my analysis of the MTO data, you are certainly right that there are a lot of variables at play. Social scientists have developed a long list of theories for why dangerous, disadvantaged, disorderly neighborhoods should affect criminal behavior, and MTO changed neighborhoods in almost every imaginable way predicted by standard theories. There were substantial reductions in neighborhood poverty, perceptions of physical and social disorder, measures of collective efficacy, litter, trash, graffiti, and abandoned buildings, as well as, for the New York data, police precinct misdemeanor arrest rates. (Jens and I document those changes in Table 5.)

All of those indicators would predict a reduction in crime. Yet the actual impacts on criminal offending, particularly among males, are not what these standard theories would predict. If the broken windows theory were correct, then there must be some sort of change induced by MTO that generates equally large offsetting effects on crime—for men, but not for women. In the absence of any good hypotheses about a gender-specific offsetting factors, it's simply hard to look at the MTO data and conclude that there's much there to support the broken windows theory.

Now, if the argument boils down to the need for some basic modicum of order, then the difficult question is: How do we come to see certain conduct as disorderly? This is where things get complicated because disorder, it turns out, is in the eye of the beholder. In a fascinating new study called Seeing Disorder, Robert Sampson and Stephen Raudenbush demonstrate that the racial composition of a neighborhood affects people's perceptions of neighborhood disorder. Using Chicago neighborhood data, the authors compare an objective measure of disorder—systematic social observation from videotapes of Chicago neighborhoods—against subjective survey-response perceptions of disorder. What they find is that, as the proportion of African-American residents in a neighborhood increases, subjective perceptions of disorderliness increase as well—even for neighborhoods that have been measured as having the same objective amount of disorder. The same was true for neighborhoods with more Latinos, as well as neighborhoods with more concentrated poverty. Their study highlights one of the most troubling aspects of the broken windows theory—namely the idea that "disorder" has a well defined meaning and unambiguous signification.

Another difficult question is: Why do we think the police should be the agency dealing with minor disorder? I've been spending time recently on L.A.'s Skid Row, a truly dehumanizing neighborhood where thousands of destitute, mentally ill, and drug dependent people sleep on the sidewalks, pitch tents, make shelters and encampments out of discarded cardboard boxes, urinate and defecate in the street, and wander about trolling shopping carts overflowing with all their earthly possessions. (I describe the situation there in a forthcoming article, Policing L.A.'s Skid Row). The questions I keep asking myself are: How is it that we have turned these social problems over to the police? Why isn't it the job of social services? Should we arrest these street people for urinating and defecating in public, or should we build secure public toilets?

Thacher: 10/19/05, 12:28 PM
On the link between order maintenance and crime, I think we've gotten the key issues out on the table. My main point is just that while studies like MTO and Sampson and Raudenbush are really excellent for many purposes, they are just too hard to interpret as tests of Broken Windows (understood as the theory that disorder causes crime), so they can't "undermine" that theory. I don't claim that the Jersey City study and the large body of situationist social psychology that underlies Broken Windows prove that the theory is correct. But I do want to question what seems to be a gathering consensus in some corners of academia that Broken Windows has been discredited, or at least that it's on the run. If we look closely it turns out that its pursuers aren't really all that frightening, and that the most relevant evidence out there is actually consistent with Broken Windows. Regardless, as I indicated in my first post, any indirect effects of order maintenance on crime are just icing on a cake that we ought to be baking anyhow.

Your data on the cost of misdemeanor arrests are interesting, though I suspect you're not suggesting that police should ignore all misdemeanors (including many assaults, thefts, vandalism, etc). All the same, you are clearly right to raise the question of priorities. In response to that question, I gave reasons yesterday why police shouldn't abandon order maintenance to devote all their time to gun policing and burglary investigation. Maintaining order in our subways, parks, and sidewalks is no more dispensable than the physical maintenance of these public assets is. And more than physical maintenance, order maintenance has a moral impetus because disorder (properly defined) violates the fair terms of cooperation and civility that are an essential part of public life. I don't think this is a "watered down" claim—at minimum, it implies that police ought to take action if they come across any of the examples in my parade of horrors, and although I haven't heard your direct response on that issue, it sounds like you disagree.

You raise excellent questions about how the police and others come to define certain conduct as disorderly. How do we distinguish legitimate order maintenance from mere fussiness about the bustle of city life or, infinitely worse, from racial harassment? I agree completely with you that these are the most serious questions facing order maintenance policing, and they are especially pressing in light of the troubling results in the new Sampson and Raudenbush study. The potential for abuse doesn't mean order maintenance should be abandoned, any more than the potential for abuse in police interrogations and "hot spots" gun policing means that police should stop interrogating criminal suspects or searching for guns. But it does place an important responsibility on the police, the courts, and their overseers to develop standards that can guard against this potential.

New Haven, Connecticut is one example of a city that made an excellent start of that in the 1990s by developing detailed guidelines for order maintenance (described in Kelling's monograph). The guidelines carefully define what counts as "disorder" that warrants police intervention and they lay out a graduated stepladder of interventions that instructs officers to use "the least forceful means possible to achieve [their] purposes" ("while we will not hesitate to cite or arrest offenders, our approach, at all levels of the organization, will be to attempt to get citizens to obey laws and ordinances as unintrusively as possible").

Every police agency that plans to take order maintenance seriously ought to go through this kind of process. The guidelines that result ought to be made public for the sake of accountability and public input. And regardless of what the public asks for, the guidelines need to be consistent with clear moral principles that maximize the scope for personal liberty and, in particular, that recognize the special needs of the homeless, whose condition is defined by the fact that they have no private space where they can take care of their basic human needs (an issue Jeremy Waldron has discussed lucidly). I've tried to make a start at developing such principles in my JCLC article, partly by drawing from the remarkable work of your former colleague Joel Feinberg. Finally, guidelines should include clear statements about what order maintenance can not do—for example, that it must not be used to harass racial and sexual minorities, or to try to eject the homeless from public spaces altogether. I think it's by going through this kind of process that police can avoid the problems you are rightfully worried about while still taking order maintenance seriously.

You also raise a good question about why the police should be the agency dealing with disorder, but I'll have to get to that next time.

This Week's Entries: Monday | Tuesday | Wednesday | Thursday | Friday

Harcourt: 10/20/05, 09:13 AM
Let's turn to that question, then. Why should the police be the agency dealing with "social and physical disorder"? Take the case of L.A.'s Skid Row, which I describe in this essay. The 50-block area has all the markers of what qualifies as "disorder" under the broken-windows theory: graffiti, public urination and defecation, peddling, aggressive panhandling, prostitution, public drinking—you name it. One of the local organizations posted day-time photographs of Skid Row on the web that you can view here. "A woman in ragged clothes, pus and blood running from a burn on her leg, was lying on the sidewalk, mumbling incoherently," reports the L.A. Times. A destitute man in wheelchair, his head locked back, mouth gaping open, drooling, wheezing, with bloated ankles and fleshy, pussing wounds on the side of his legs. L.A.'s Skid Row is truly a shocking sight, and, as the sun sets, it gets even more disorderly. If you walk the streets at night, you literally have to wind around fifty tents and box encampments on a single side of a single street. George Kelling describes L.A.'s Skid Row as "a Third World country." (The analogy, tragically, carries a double-entendre. The homeless population on the Row is overwhelmingly of color. There is, after all, some order in that disorder).

Now, why would we think that the human and social problems that make up L.A.'s Skid Row are the province of the police? What is it about our generation that would make us turn to law enforcement as the way to deal with these issues? Other generations, after all, embraced very different models. Some took a social services approach and relied on social workers and public facilities. Others adopted a more medicalized model and set out to cure the disease and contagion. Others institutionalized street people in asylums. Perhaps there was a time when family or perhaps even clan were the mechanisms to deal with these issues. My point is not to advocate any of these other approaches. I'm just asking: Where did we get the idea that the police officer was the solution? When and why did we turn to law enforcement, to the jail, the penitentiary, in sum, to the carceral sphere?

I don't have an answer. But what I do know is that, even if the goal is simply to reduce surface "disorder," there are lots of creative ways to achieve that goal without resort to law enforcement and misdemeanor arrests. There are, in practically all of these cases of "disorder," practical alternatives. So, for instance, in the case of public urination and defecation, one solution would be to build public toilet facilities. The city of Los Angeles, magnanimously, placed a number of port-o-johns on street corners on Skid Row. But we all know how disgusting those things are—with their open containers of raw sewage and human excrement. Why not build proper public facilities—clean, secure, safe, and hygienic? On the South Side of Chicago, we have a number of such public facilities along the lake front. There is no mystery here.

If we're concerned about fare-beating, then let's design and put in place turnstiles that you can't jump—full height turnstiles, for instance. Here's what they look like. It's not rocket science. They're in use in many cities, including New York and Paris. If panhandling is the problem, then let's develop and implement more ready-and-willing-to-work programs. And if graffiti is the problem, then let's clean it off. Chicago, it turns out, has very little graffiti, but not because of misdemeanor arrests. The reason is "Graffiti Blasters": city sanitation crews that roam the city and paint over or spray off fresh graffiti. They're actually pretty remarkable. I've seen them on the South Side, early in the morning, roller painting and sand blasting the neighborhood. Actually, that's also how New York City dealt with graffiti in the subways: by cleaning them.

Why use "Graffiti Blasters" rather than the police? Because of the trade-off: using police officer time and resources, as well as judicial resources, to conduct misdemeanor arrests costs a lot more, doesn't work, and takes the officers away from more effective "hot spots" policing of violent and gun crime.

I think that even Bill Bratton recognizes it. Bratton was appointed police chief in Los Angeles on a platform that promised more broken-windows policing and greater attention to quality-of-life issues. During the selection process in the summer and fall of 2002, Bratton made a big deal of disorder. "I was amazed to find that of 9,000 persons in the Police Department, not a single one is focused on graffiti," Bratton told the L.A. Times. He told the press that he "will make graffiti a top priority for all officers."

Well, in April 2003, about six months after taking office, Bratton disbanded the eleven-member undercover LAPD transit police anti-graffiti unit—the Graffiti Habitual Offenders Suppression Team ("GHOST"). And this, despite their enforcement success: the unit apparently had made over 500 graffiti-related arrests the previous year.

Why did he disband the unit? Because those arrests—i.e. the officer salaries, benefits, overhead, precinct costs, and judicial resources—were simply a waste of money and resources as compared to what Bratton could do with eleven police officers to fight serious crime. Using those police officers to deal with the gang problems in L.A.—and the associated violence and gun crime—is a far better use of resources than the GHOST unit. It is likely to have a far greater impact on serious crime—and that really improves our quality of life.

Naturally, Bratton would never admit this. Just the other day, he was on L.A. public radio saying that Jens and I, in his words, "don't know what the hell they're talking about." But his actions speak louder than his words: Bratton knows there is a trade off. If you have a whole unit of eleven police officers making 500 graffiti-related arrests, you're not only sinking massive financial resources into something that can be done cheaper and more effectively by the sanitation department, you are losing the opportunity to engage in more effective "hot spots" policing of violent and gun crime.

Thacher: 10/20/05, 01:40 PM
You may be more sympathetic to order maintenance than you realize. Yesterday you seemed to suggest that the contemporary concern about "disorder" reflects fussiness or even racial intolerance. I thought you might agree with Norman Mailer's view that graffiti is a form of folk art (in some cases I actually agree with that myself), or that you'd want the money spent on the Graffiti Blasters to be reallocated to gun policing because the harms of graffiti pale next to the harms of gun crime. Today you share Kelling and Bratton's concerns about graffiti and the deplorable conditions on L.A.'s skid row that you describe so vividly in your paper. And you share their view that we ought to devote resources to them despite the never ending possibility that those resources could be shifted to crime control.

You also share their preferred solution. In Fixing Broken Windows, Kelling and Coles begin the story of how the New York City subway tackled the appalling disorder I mentioned Tuesday by describing Transit president David Gunn's "Clean Car Program," which rolled out new clean subway cars one-by-one and then immediately removed them from service for rapid repainting as soon as graffiti struck them. By 1989 every single car had been cleaned. Kelling and Coles conclude, approvingly: "The NYCTA succeeded because the Gunn administration abandoned the use of a law-enforcement strategy in dealing with graffiti."

Then in 1990 Alan Kieper became NYCTA president and hired Bratton as head of the transit police. They built on Gunn's success against physical disorder by tackling social disorder—the public drunkenness, sex, aggressive panhandling, defecation, and intimidation that generated immense fear in the subway for the people who were either hardy enough to keep using it or just had no other way to get to work. How should these behaviors be handled? Should police ignore them because they need to search for guns?

The appalling conditions in the subway were not simply a problem of "homelessness"—it is offensive to the majority of homeless who exercise self-restraint even under such demoralizing conditions to say so—but the transit authority and its police collaborated with the city Human Resources Administration and funded a nonprofit agency to reach out to the homeless and offer help. Disorder continued, but as Kelling and Coles describe, Kieper still saw scope for other non-police solutions. He established a new "station manager" program and charged these civilian employees with trying to maintain order by making physical changes to their stations (like improving lighting or blocking side-areas of the stations that were used for drug use and sex) and intervening informally when they saw disorder.

No one opposes realistic, effective, non-coercive solutions to the disorder that you, I, Kelling, and Bratton all agree needs attention. The question is what happens when, despite our best efforts of that sort, unacceptable disorder remains. Should the police ignore it because of the never ending possibility that they might target one more serious crime? Until 1990 that's exactly what the transit police said: They insisted that order maintenance was beneath them and that they needed to focus on robberies. Bratton disagreed. With Kelling's help, his transit police developed guidelines like those I described in New Haven, and then they made order maintenance a central part of their mission.

No one relishes the use of coercion to deal with problems like disorder, whether the coercion is exercised by police or by anyone else. Institutionalizing the mentally ill and many of the other alternatives to order maintenance policing that you describe in your second paragraph obviously raise their own civil liberties concerns, as you clearly acknowledge. (Michel Foucault, who you often draw from, recognized these dangers as clearly as anyone, as have those he inspired.) One advantage of police order maintenance is that no one can fail to recognize that it is coercive. Therapeutic solutions like institutionalization make it dangerously easy to jump too quickly to the claim that "it's for their own good," but it's harder to kid ourselves that way about police order maintenance. It puts us on our guard. It forces us, hopefully, to recognize the need to face the tough questions you rightfully asked yesterday: Do we really have good reason to intervene against this disorder? Is this really wrongful conduct? Or are we just being finicky, intolerant—even racist?

An important benefit of the recent attention to police order maintenance is that it has brought this aspect of policing out into the open and thereby forced us to ask what rules ought to govern it. It has stimulated departments like New Haven's and the NYC Transit Police to develop clear standards governing how this essential and ultimately unavoidable police function should be carried out. Idealistically averting our eyes from the need for police to maintain order when everything else has failed is the really serious danger that we need to be worried about.

This Week's Entries: Monday | Tuesday | Wednesday | Thursday | Friday

Harcourt: 10/21/05, 11:03 AM
The fact that the broken windows theory is probably wrong does not mean that we should encourage, promote, or value homeless encampments and public defecation. Being a skeptic of the disorder-crime nexus says nothing about whether you favor or disfavor graffiti or abandoned buildings or prostitution. There are plenty of independent reasons—compassion, hygiene, public health—why we should do something to alleviate the wretched conditions of homeless persons living in encampments on the streets of L.A.'s Skid Row. The reasons to address the human suffering on Skid Row are not related to the broken windows hypothesis. They have nothing to do with crime. Instead, they derive from the fact that those homeless men, women, and children, are human—all too human. They spring from our shared sense of humanity and our concern for the well-being of others.

The key trade-off that I have discussed in this debate is a trade-off about our crime-fighting dollars. We, as a society, allocate some money for education, some for public health, some for roads and tunnels and bridges, for national security and national emergencies, for charitable work—and some money to fight crime. It's those crime-fighting dollars that I am talking about. Those dollars are scarce, as they are for education and housing. My argument is that we need to be very careful in allocating those crime-fighting dollars, and that it is more effective to direct those scarce law enforcement dollars to targeted policing of violent and gun crime than it is to spend them on aggressive misdemeanor arrests: it is better to disband the Graffiti Habitual Offenders Suppression Team and to redirect those eleven police officers to deal with gang homicides. We get more bang for the buck.

Naturally, we as a society will have to make other trade-offs between, for instance, the money we invest in crime fighting and the money we invest in education and public health and in public art projects, in poverty relief, and in humanitarian aid. But just because targeted law enforcement is more effective than broken windows policing does not mean that we will invest all our resources in hot spots approaches. It does not mean that we should not invest money in housing and public toilets to deal with the plight of the homeless persons on Skid Row. What it does mean is that those dollars allocated to crime-fighting should not be squandered on ineffective measures.

Let me, then, summarize what I see as the terms of our disagreement:

Proposition 1: The broken windows theory—which holds that disorder causes crime, or, in the original formulation by Wilson and Kelling, that "disorder and crime are usually inextricably linked, in a kind of developmental sequence"—is not right.

I have argued, based on the most reliable social science evidence—including Sampson and Raudenbush's study of Chicago neighborhoods, Ralph Taylor's study of Baltimore neighborhoods, and Jens Ludwig and my study of the MTO findings—that there is no good reason to believe that disorder causes crime.

You have taken the position, essentially, that we don't really know whether disorder causes crime.

Proposition 2: Broken windows policing involves the aggressive use of misdemeanor arrests.

I have argued that the broken windows approach boils down to the use of aggressive misdemeanor arrests. I've relied on George Kelling, who measures the extent of broken windows policing in his 2001 Manhattan Institute report by the number of misdemeanor arrests in a precinct—in his own words, "arrests for misdemeanors [are] our measure of 'broken windows' enforcement." I've also demonstrated that broken windows enforcement has historically led to significant increases in misdemeanor arrests.

You have suggested that this is an overly simplistic picture of broken windows policing.

Proposition 3: There is no good evidence that broken windows policing reduces serious crime.

I have argued that the most reliable social science evidence—including Steve Levitt's study of the crime drop, Jeffrey Fagan's study of New York City, and Jens Ludwig and my study of New York City—demonstrates that there is no good reason to believe that broken windows policing is effective.

You have argued that "nobody knows," but that there are some indications in Anthony Braga, David Weisburd, et al.'s study Problem-Oriented Policing in Violent Crime Places that order-maintenance may affect crime rates.

The conclusion I draw is that there is a sharp trade-off: other reliable evidence—including the research of Larry Sherman on Fair and Effective Policing and the research collected in the review essay by Anthony Braga—suggests that targeted policing of violent and gun crime is a more effective measure for combating serious crime. As a result, we should not be directing our scarce law enforcement dollars toward broken windows policing.

I also believe that focusing on misdemeanor arrests has a negative impact on race relations in this country. As I show in Boston Review, in NYC and the U.S. more generally, adults arrested for misdemeanors are disproportionately African-American in relation to their representation in the community. In 2000, slightly over 50% of all adults arrested for misdemeanors in NYC were African-American. Slightly over 50% of adults arrested for disorderly conduct and 45.6% of adults arrested for loitering were African-American. Yet African-Americans in 2000 represented only 24.6% of the NYC population. Whites, who represent 48.8% of the population in NYC, account for only 15.5% of adults arrested on misdemeanor charges in 2000. These disparities hold true for large cities across the country. The decision to arrest misdemeanants—adopting that policy in preference to other policing strategies—is a choice with significant negative distributional consequences for African-Americans.

The central claim of the broken windows theory—the idea that disorder causes crime by signaling community breakdown—is flawed. The heart of the problem, I would argue, lies in the category of "disorder." We have come to identify certain acts—graffiti spraying, litter, panhandling, turnstile jumping, and prostitution—and not others—police brutality, accounting scams, and tax evasion—as disorderly and connected to crime. But the category of disorder does not have such well-defined boundaries, nor does it have such settled meaning. In a forthcoming book, Language of the Gun: Youth, Crime, and Public Policy, I try to demonstrate how we should go about trying to understand the rich symbolic dimensions of categories like "disorder." But given that we have reached the end of this debate, I will leave that to another time and, here, thank you, David, for your excellent research and writings, for your outstanding paper on Order Maintenance Reconsidered, and for this wonderful conversation.

Thacher: 10/21/05, 06:43 PM
I've enjoyed and learned a lot from our debate this week, Bernard, and let me start out by saying that I agree with much of what you've said in your last three posts. As I implied in my first post, I agree that any possible effect of order maintenance on crime shouldn't be a major reason for embracing it. I agree that the concept of "disorder" is complex and has to be approached with caution, and on Wednesday I laid out a detailed process that police should follow to meet that important demand. I agree that police intervention, and particularly arrests, ought to be treated as a last resort. I have endorsed that claim wholeheartedly, and I have quoted both Kelling and actual order maintenance policy guidelines that do the same. (I am sorry to put the point bluntly, but I think your repeated references to Kelling and Sousa's use of misdemeanor arrests as a proxy measure—I'd argue a flawed one—in a piece of quantitative research is disingenuous. Your continued claim that "broken windows boils down to the aggressive use of misdemeanor arrests" seriously muddies public understanding of what order maintenance is, and it encourages the dangerous confusion of "order maintenance" with "zero tolerance.")

Where we disagree is when we reach the question of what happens when, despite our best efforts outside of policing, serious and debilitating disorder remains in our public spaces. You assert that police dollars are "crime-fighting dollars," and therefore police shouldn't worry about disorder unless it affects crime. But that argument isn't helpful because its premise is exactly what's in dispute.

You, like many police themselves, believe that the one and only mission of policing is crime control—forgetting that police perform a huge variety of miscellaneous tasks including crowd management, traffic patrol, and responding to calls about neighbor disputes. Police are involved in all those things because they are the institution we charge with dealing with any situation "which ought not to be happening and about which something ought to be done now" (I'm quoting Egon Bittner)—a job that can only be performed successfully by someone who can resort to arrest if voluntary compliance fails. Disorder often falls into this category. To return to the examples I started with: If a man is flagrantly urinating in public (without at least trying find a discrete location if he has no other option), or aggressively demanding donations from frightened pedestrians (not politely asking whether they can spare some change), or harassing every woman who passes by (and refusing to stop when civilians ask)—then who other than the police are we going to turn to? Of course it's essential to my argument that disorder (properly defined) is in fact something that ought not to be happening, but over the week I have given several reasons why that's true. Does anyone really claim that the examples I just gave really aren't wrongful behavior? Bittner himself clearly recognized how central order maintenance was to the police role.

If the idea that all police dollars ought to go to crime control still concerns you, think about the money spent on crime control as part of our government's budget for parks, subways, and sidewalks that happens to flow through the police department (since it's the only institution with the necessary authority). As I said Wednesday, order maintenance is no more dispensable than the physical maintenance of these public assets. As the experience of the New York City subway during the 1980s illustrates, refusing to spend money on order maintenance because we believe the entire police budget ought to focus on crime control—precisely the position of the Transit Police during that period—can lead to an utterly appalling situation. Such problems, like the problem of public safety more generally, weigh far more heavily on the disadvantaged than on the wealthy, who don't have to ride the subways anyhow.

On the surface, it may always seem hard to understand why police should take disorder seriously, since each individual incident of disorder looks trivial compared with gun crimes and burglary. That, of course, was exactly the concern that drove Wilson and Kelling to write "Broken Windows." They tried to address it by articulating a causal theory about the serious consequences of disorder: That unchecked disorder sends out a message that a place is out-of-control, thereby inviting more disorder and even serious crime. It's the nature of such theories that the evidence is always likely to be ambiguous. My own guess is that there is probably some truth to it—more because it's consistent with much of what social psychology has taught us about human nature than because of anything in the inconclusive criminological literature. (At minimum, it defies reason to believe that the truly appalling disorder you describe so vividly on LA's skid row doesn't affect some people's propensity to commit crime there.) But as I said in my first posting, I think it was probably a mistake to connect order maintenance policing so closely to Broken Windows, for reasons I have explained elsewhere.

There is, however, a closely-connected normative theory that I think captures the spirit of the Broken Windows article in a less problematic way. Disorder is like littering or pollution: If I drop a candy-wrapper on the sidewalk it won't spoil the streetscape, and if I drive my car with a dirty engine it won't destroy the environment. Each individual case of littering or pollution is a minor annoyance; instead of worrying about these things, shouldn't the government be looking for guns? The trouble is that if everybody dropped candy-wrappers and drove cars with dirty engines we'd have much bigger problems, so we agree to fair terms of cooperation under which nobody is allowed to drop candy-wrappers and everybody is held to strict emissions standards. Then we back those requirements up with legal sanctions. Whether or not disorder causes more crime and disorder, disorder is wrong because these individually-trivial acts would destroy the livability of our public spaces if everyone engaged in them. As a matter of fairness, everyone ought to share the burdens of self-restraint.

The tragedy is that those burdens will fall especially heavily on the homeless, who have no private space of their own where they can take care of their basic human needs. I have argued that it is possible to develop guidelines for order maintenance policing that are sensitive to this dilemma, but it undoubtedly remains the central concern that police must always keep in mind as they perform this essential role.

This Week's Entries: Monday | Tuesday | Wednesday | Thursday | Friday

space