Punishment in these late modern times is marked by two striking developments. The first is a stunning increase in the number of persons incarcerated. Federal and state prison populations nationwide have increased from less than 200,000 in 1970 to more than 1,300,000 in 2000, with another 600,000 persons held in local jails.1 Today, approximately 2 million men and women are incarcerated in prisons and jails in this country.The intellectual rationale for this increase is provided by “incapacitation theory”—the idea that a hardcore 6 percent of youths and young adults are responsible for the majority of crime and that locking up those persistent offenders will significantly impact crime rates.

A second dramatic development is the popularity of “order-maintenance policing,” an approach to policing that emphasizes creating and maintaining orderly public spaces. This policy is driven by the “broken windows theory”—the idea that tolerating such minor infractions as graffiti spraying, aggressive panhandling, prostitution, public urination, and turnstile jumping encourages serious violent crime by sending a signal that the community is not in control. The broken windows theory has ignited a virtual “revolution in American policing,” also known as the “Blue Revolution.”2 And its popularity in this country is apparently matched by its appeal abroad. In 1998 alone, representatives of over 150 police departments from foreign countries visited the New York Police Department for briefings and instruction in order-maintenance policing. For the first ten months of 2000, another 235 police departments—85 percent of them from abroad—sent delegations to One Police Plaza.3 In New York City, where broken windows policing was introduced under the administration of Mayor Rudolph Giuliani and his first police commissioner, William Bratton, the order-maintenance strategy produced an immediate surge in arrests for misdemeanor offenses.

These two developments have common intellectual roots in the second half of the twentieth century, primarily in the writings of James Q. Wilson, and before him, Edward C. Banfield.4 The popularization of incapacitation theory can be traced in large part to Wilson’s 1975 collection of essays, Thinking About Crime, where he famously wrote: “Wicked people exist. Nothing avails except to set them apart from innocent people….We have trifled with the wicked, made sport of the innocent, and encouraged the calculators. Justice suffers, and so do we all.”5 Like Banfield before him—who, in The Unheavenly City, similarly advocated “abridg[ing] to an appropriate degree the freedom of those who in the opinion of a court are extremely likely to commit violent crimes”—Wilson argued strenuously for increased prison sentences for hardcore offenders.6

The broken windows theory can also be traced to Wilson, who, with his co-author George L. Kelling, wrote an influential Atlantic Monthly article in 1982 called “Broken Windows,” a nine-page anecdotal essay that revolutionized policing. Wilson there spelled out and popularized the idea of cracking down on minor disorder as a way to combat serious crime.

Conservative policymakers tend to argue for both incapacitation and broken windows policing. William Bennett, John DiIulio, Jr., and John Walters, in their provocative 1996 work, Body Count: Moral Poverty…and How to Win America’s War Against Crime and Drugs—the book that coined the expression “super-predators”—advocate for both lengthened sentences for hardcore offenders and order-maintenance policing. The first step in the war on crime and drugs, they argue, is to stop “revolving-door justice.” “The truth,” they declare, “is that virtually all of those in prison in this country are just what most average Americans suppose them to be—…duly tried and convicted violent and repeat criminals…[T]he problem is that they too often escape justice entirely, get released early, commit more crimes, and manufacture more misery for victims, their families, and society.”7 At the same time, Bennett, DiIulio, and Walters praise and recommend to others New York City’s model of policing, what they call the “Bratton miracle.”8

What is striking, though, is that among progressives, liberals, and the full spectrum of Democrats, order-maintenance policing has been hailed as the only viable and feasible alternative to the three-strikes and mandatory minimum laws that have resulted in massive incarceration. The power and appeal of broken windows policing for liberals derives precisely from its opposition to incapacitation theory. Order maintenance is billed as a milder public-order measure. It represents, according to some progressives, one of the only “politically feasible and morally attractive alternatives to the severe punishments that now dominate America’s inner-city crime-fighting prescriptions.”9 This is precisely what makes broken windows policing “progressive.”

As a result, order maintenance is popular today in a way that massive incarceration is not. Whereas the prison boom has received a lot of criticism in the media, among public officials, and in the academy, broken windows policing has received far less attention and scrutiny. With few notable exceptions, order maintenance continues to receive extremely favorable reviews in policy circles, academia, and the press.

This popularity was powerfully demonstrated during the 2001 mayoral campaign in New York City, especially before 9/11. In the early Democratic primaries, all the candidates were jockeying to be the next Giuliani on crime. The first question at the August 28, 2001 Democratic mayoral debate—what if the squeegee men come back?—had a domino effect on the candidates: they fell over each other pledging to maintain the crackdown on quality-of-life offenses. Mark Green, who ultimately won the Democratic primary, sought, obtained, and heavily publicized the endorsement of William Bratton, who appeared slated to return as police commissioner in a possible Democratic administration. As former mayor Edward Koch remarked, “Giuliani has so impacted New Yorkers by the reduction in crime that any candidate who doesn’t agree with that has no chance.”10

After the election, the Manhattan Institute—a conservative New York think tank—began waging a renewed campaign in support of broken windows policing. A few days before Michael Bloomberg’s inauguration as mayor, the Institute issued a new report on New York-style policing. Written by lead author George Kelling—co-author (with James Q. Wilson) of the “Broken Windows” article and of the book Fixing Broken Windows—the report is titled Do Police Matter? An Analysis of the Impact of New York City’s Police Reforms. The report contends that economic improvements and the end of the crack epidemic had no significant effect on overall drops in violent crime in New York City. In contrast, order maintenance policing exerted “the most significant influence” on violent crime trends.11

The Manhattan Institute report was issued with a simulcast editorial comment by the authors in the New York Post. “So what does all this mean?” they ask. “First, it means that New Yorkers should stop listening to critics who contend that police tactics matter little, if at all, in determining crime rates.” More poignantly, the authors note, “these critics have been parroting what is virtual dogma in criminal-justice circles, that crime is caused by ‘root causes’ such as racism, poverty and social injustice. This study places the ‘root cause’ theory of crime in serious jeopardy.”12 The New York Post carried its own editorial the same day, capturing in its pithy title the thrust of the Manhattan Institute report: “It’s the Cops, Stupid.”13

Two days after Bloomberg’s inauguration, the Manhattan Institute caught the ear of The New York Times, arguing on the Op-Ed pages that broken windows policing “is not just smart policing; it’s smart politics.” In an editorial entitled “A Policing Strategy New Yorkers Like,” Kelling emphasizes the popularity of enforcing quality-of-life laws, especially among minorities. “African-Americans, Hispanics and Asian-Americans all favor quality-of-life enforcement even more strongly than whites,” Kelling claims. And for those who thought that Giuliani and Bratton were the chief architects of broken windows policing, Kelling offers a new narrative: it was Raymond Kelly, Bloomberg’s new police commissioner, who, as police commissioner for former Mayor David Dinkins, started broken windows policing in the fall of 1993.14

Questionable social science

The result is that today, broken windows theory seems more popular than ever. And the breadth of its popularity reflects the fact that it is understood as an alternative to the incarceration explosion we have witnessed in the latter part of the twentieth century. It has, in effect, captured the liberal imagination.

The difficulty is that there is no good evidence for the theory that disorder causes crime. To the contrary, the most reliable social scientific evidence suggests that the theory is wrong. The popularity of the broken windows theory, it turns out, is inversely related to the quality of the supporting evidence. The most recent study by the Manhattan Institute is a good illustration.

The goal of the Manhattan Institute study is to separate out “the relative contributions of police actions, the economy, demographics, and changing drug use patterns on crime” in New York City. The study’s innovation is to achieve this separation by treating the city as 75 separate and comparable entities. “Rather than one city,” they explain, “we view New York as 75 separate entities, corresponding to the 75 police precincts.”15 Because economic conditions, for example, vary considerably across the 75 precincts, Kelling and his co-authors could look across the 75 cases and see if favorable economic conditions are associated with reduced rates of violent crime.

The aim of the research, then, is to determine the relative impact of four factors—broken windows policing, economic indicators, young-male population shifts, and the decline in crack cocaine consumption—on violent crime in the 75 precincts of New York City. Is the decline in violent crime fully explained by an improved economy, a drop in the number of young men in the population, and reduced crack cocaine use? Or does the increase in arrests for minor offenses also play a significant role?

Naturally, the complexities in a study of this kind begin with the need to provide a precise measure of all the relevant factors. To measure broken windows policing, the authors use precinct-level reports of total misdemeanor arrests. To measure the effect of the crack epidemic, they use borough-level reports of hospital discharges for cocaine-related episodes. For the number of young males, they use precinct-level school enrollment data. And for unemployment, they use borough-level gross unemployment data. Finally, to measure violent crime, they use precinct-level reports of four violent offenses (murder, rape, felonious assault, and robbery). In all cases, they use data from 1989 to 1998.16

Their findings are striking. They conclude that the “measure of ‘broken windows’ policing is the strongest predictor of precinct violent crime in the model.” Looking at the change over time, they find that neither demographics nor drug use patterns are significantly related to the drops in precinct violent crime over the ten-year study period. Unemploymentis related to changes in crime rates, but in the opposite direction to what is generally assumed. As it turns out, “an increase in borough unemployment is related to a decrease in precinct violent crime,” Kelling concludes.

The bottom line: “The average NYPD precinct during the ten-year period studied could expect to suffer one less violent crime for approximately every 28 additional misdemeanor arrests made.” Over the period from 1989 to 1998, increased rates of misdemeanor arrests therefore prevented, by their calculation, over 60,000 violent crimes or 5 percent of the overall decline.17 The study offers, in the authors’ words, “the most-definitive possible answer to the question of whether police mattered in New York City during its intense crime-drop.”18 It accomplishes this remarkable task with essentially one regression analysis and one page of statistical discussion. And the answer is: Yes.

But their case is remarkably weak and inconclusive. First, there are numerous alternative explanations that may account for the associations between the number of misdemeanor arrests and violent crime, including, for instance, the deterrent effect of police presence or even, possibly, some effect from incapacitating potential offenders. The study tells us nothing about the causal mechanism, if there is one, and does not begin to address the symbolic effect of orderliness. Moreover, it is not at all clear that individual precincts are the proper comparison groups. The Manhattan Institute study focuses on individual precincts and asks, in effect, whether high unemployment in a particular precinct is associated with a high rate of violent crime in that same precinct. But some of the precincts are geographically limited—the sixth precinct in Greenwich Village, for example, covers only .79 square miles. So the effects of employment, drug use, and policing are likely to spill over from one precinct to another. But if a high rate of unemployment in one precinct leads criminals from that precinct to focus their activities on other precincts where the economy is doing better and people are working, then the Manhattan Institute study would conclude that economic performance does not generate serious crime.

Even if we were to accept the research design, though, the authors do not have the proper data, and so could not carry through on their apparently ingenious idea. Unemployment data do not exist at the precinct level, so they instead use borough-level data: in effect, they assume that every Manhattan precinct faces the same economic conditions. Similarly, hospital discharges for cocaine-related episodes (even if that were a reliable index of crack as opposed to powdered cocaine consumption) are also borough-level data: in effect, the authors assume that every Brooklyn precinct has the same drug problem. So the report is essentially comparing apples and oranges: precinct-level data for crime, broken windows policing, and demographics, and borough-level data for gross unemployment and cocaine consumption. As Alfred Blumstein of Carnegie Mellon University recently commented, the data used to measure unemployment are too blunt and unreliable to do any damage to the idea that the economic boom and drop-off in crack cocaine use were important contributors to the national and municipal drop in violent crime.19

The Manhattan Institute report is just the most recent unsuccessful attempt to establish the broken windows claim. The previous research that George Kelling credited with “empirically verifying the ‘Broken Windows’ hypotheses”—Wesley Skogan’s 1990 study Disorder and Decline: Crime and the Spiral of Decay in American Neighborhoods—has also been discredited on the specific issue of establishing a connection between disorder and crime.20 With data from thirty neighborhoods, Skogan found a strong connection between residents’ perceptions of neighborhood disorder and their robbery victimization, holding constant the level of poverty and stability in the neighborhood and the race of the residents. But Skogan failed to reveal the absence of any similar connection between disorder and burglary, rape, physical assault, or purse-snatching victimization—four other crime variables in his data set.21

The other set of studies that has traditionally been offered in support of the broken windows theory—studies that focus on the effect of aggressive police arrest practices on crime rates—offers little if any support. The debate here was in part triggered, again, by James Q. Wilson in his 1968 book on the Varieties of Police Behavior, and in his study with Barbara Boland on the effects of police arrests on crime.22 Wilson and Boland hypothesized that aggressive police patrols, involving increased stops and arrests, have a deterrent effect on crime. A number of contributions ensued, both supporting and criticizing these findings, but, as Robert Sampson and Jacqueline Cohen suggest, the results thus far have been “mixed.”23 There has been some excellent new work; but it is unable to distinguish between the broken windows theory and more traditional explanations such as increased police presence, contact, and surveillance.24 To a certain extent, the problem is endemic to the design of these studies. As Sampson and Cohen acknowledge with regard to their own work, “[i]t is true that our analysis was not able to choose definitely between the two alternative scenarios.”25

The most comprehensive and thorough study of the broken windows theory to date is Robert Sampson and Stephen Raudenbush’s 1999 study entitled Systematic Social Observation of Public Spaces: A New Look at Disorder in Urban Neighborhoods. Their study is based on extremely careful data collection. Using trained observers who drove a sports utility vehicle at five miles per hour down every street in 196 Chicago census tracts, randomly selecting 15,141 streets for analysis, they were able to collect precise data on neighborhood disorder. Sampson and Raudenbush found that disorder and predatory crime are moderately correlated, but that, when antecedent neighborhood characteristics (such as neighborhood trust and poverty) are taken into account, the connection between disorder and crime “vanished in 4 out of 5 tests—including homicide, arguably our best measure of violence.” They acknowledge that disorder may have indirect effects on neighborhood crime by influencing “migration patterns, investment by businesses, and overall neighborhood viability.” But, on the basis of their extensive research, Sampson and Raudenbush conclude that “[a]ttacking public order through tough police tactics may thus be a politically popular but perhaps analytically weak strategy to reduce crime.”26

The New York story

If we look at the criminological evidence, the results are no more helpful to broken windows proponents. The basic fact is that a number of large U.S. cities—Boston, Houston, Los Angeles, San Diego, and San Francisco, among others—have experienced significant drops in crime since the early 1990s, in some cases proportionally larger than the drop in New York City’s crime. But many of these cities have not implemented the type of aggressive order-maintenance policing that New York City did. One recent study found that New York City’s drop in homicides, though impressive, is neither unparalleled nor unprecedented. Houston’s drop in homicides of 59 percent between 1991 and 1996 outpaced New York City’s 51 percent decline over the same period, and both were surpassed by Pittsburgh’s 61 percent drop in homicides between 1984 and 1988.27 Another study looked at the rates of decline in homicides in the seventeen largest U.S. cities from 1976 to 1998 and found that New York City’s recent decline, though above average, was the fifth largest, behind San Diego, Washington, D.C., St. Louis, and Houston.28

A straight comparison of homicide and robbery rates between 1991 and 1998 reveals that although New York City is again in the top group, with declines in homicide and robbery rates of 70.6 percent and 60.1 percent respectively, San Diego experienced larger declines in homicide and robbery rates (76.4 percent and 62.6 percent respectively), Boston experienced a comparable decline in its homicide rate (69.3 percent), Los Angeles experienced a greater decline in its robbery rate (60.9 percent), and San Antonio experienced a comparable decline in its robbery rate (59.1 percent). Other major cities also experienced impressive declines in their homicide and robbery rates, including Houston (61.3 percent and 48.5 percent respectively) and Dallas (52.4 percent and 50.7 percent respectively).29

Many of these cities, however, did not implement New York-style order-maintenance policing. The San Diego police department, for example, implemented a radically different model of policing focused on community-police relations. The police began experimenting with problem-oriented policing in the late 1980s and retrained the police force to better respond to community concerns. They implemented a strategy of sharing responsibility with citizens for identifying and solving crimes. But while recording remarkable drops in crime, San Diego also posted a 15 percent drop in total arrests between 1993 and 1996, and an 8 percent decline in total complaints of police misconduct filed with the police department between 1993 and 1996.30

San Francisco also focused on community involvement and experienced decreased arrest and incarceration rates between 1993 and 1998. San Francisco’s felony commitments to the California Department of Corrections dropped from 2,136 in 1993 to 703 in 1998, whereas other California counties either maintained or slightly increased their incarcerations. San Francisco also abandoned a youth curfew in the early 1990s and sharply reduced its commitments to the California Youth Authority from 1994 to 1998. Despite this, San Francisco experienced greater drops in its crime rate for rape, robbery, and aggravated assault than did New York City for the period 1995 through 1998. In addition, San Francisco experienced the sharpest decline in total violent crime—sharper than New York City or Boston—between 1992 and 1998.31

Other cities, including Los Angeles, Houston, Dallas, and San Antonio, also experienced significant drops in crime without adopting as coherent a policing strategy as New York or San Diego. The fact is, there was a remarkable decline in crime in many major cities in the United States during the 1990s. New York City was certainly a very high performer. But numerous major U.S. cities have achieved substantial declines in crime using a variety of different policing strategies. It would be simplistic to attribute the rate of the decline in New York City solely to the quality-of-life initiative.

A number of other factors seem to have contributed to declining crime rates in New York City, including a shift in drug use patterns from crack cocaine to heroin, favorable economic conditions in the 1990s, new computerized tracking systems that speed up police responses to crime, a dip in the number of eighteen to twenty-four-year-old males, as well as shifts in adolescent behavior. There have also been important changes at the NYPD, including a significant increase in the sheer number of police officers. Mayor Dinkins hired over two thousand new police officers under the Safe Streets, Safe City program in 1992, and Giuliani hired another four thousand officers and merged about six thousand Transit and Housing Authority officers into the ranks of the NYPD. As a result, from 1991 to 2000, the NYPD force increased almost by half, up by 12,923 police officers (including those transferred from Transit) from a force of 26, 856 police officers in 1991 to 39,779 in 2000. Excluding the Transit merger, the police force grew by almost a quarter.32 As a result, the NYPD now has the largest police force in the country and the highest ratio of police officers to civilians of any major metropolitan area.

Moreover, beginning in January 1994, Giuliani and Bratton instituted major changes in police management. They started relying more heavily on new computer technology to compile crime statistics, to convert the data into maps and charts that inform the police about crime patterns in different precincts, and to monitor and review police performance at the district level through regular meetings. The data and meetings allow the police to target their enforcement to changing crime trends. Bratton also shook up his management team, cut out layers of bureaucracy, aggressively promoted younger and more ambitious managers, and delegated more authority to precinct commanders.33 In addition, Giuliani and Bratton implemented a number of different police strategies—including gun-oriented policing and enhanced drug enforcement initiatives—that significantly overhauled the way the NYPD approached crime fighting.

So even if we were prepared to attribute a significant role in the drop in crime in New York City to changes at the NYPD—despite comparably sharp drops in crime in cities that did not institute similar policing strategies—we would still need to ask if it was the broken windows strategy that did the trick, or if instead the drop in crime owes to the increased surveillance afforded by a more aggressive policy of arrests and stop-and-frisks. The answer to this question has to be the increased surveillance.

What the quality-of-life initiative gives the NYPD is a legal reason to seize, search, and run checks on persons committing or suspected of committing minor offenses. This has important consequences for the detection and prevention of crime, which was powerfully demonstrated in the now-famous case of John Royster. Royster was accused of fatally beating a flower shop owner on Park Avenue—as well as several brutal attacks on women, including an infamous assault on a piano teacher in Central Park that left her severely impaired. Royster was identified after he was arrested for turnstile jumping in the subway. Upon arrest, a computer matched his prints with fingerprints left at the scene of the Park Avenue murder. This is not a case in which the restoration of public order sent a message to criminals that they can’t commit serious crime. It is not the broken windows theory. It is the old-fashioned idea that more police contact, more background checks, and more fingerprinting will produce better crime detection.

Soon after the original strategy was first tested in New York subways in 1990, William Bratton realized the full potential of order-maintenance crack-downs and sweeps. “For the cops,” Bratton exclaimed, they were “a bonanza. Every arrest was like opening a box of Cracker Jack. What kind of toy am I going to get? Got a gun? Got a knife? Got a warrant? Do we have a murderer here? Each cop wanted to be the one who came up with the big collar. It was exhilarating for the cops and demoralizing for the crooks.”34 The first quality-of-life experiment in the New York City subways demonstrated the benefits early on. With misdemeanor arrests up more than 50 percent in New York City, and with routine fingerprinting and record checking, the quality-of-life initiative produced a significant increase in arrests on outstanding warrants and a reduction in gun carrying.

The bottom line is that the broken windows theory—the idea that public disorder sends a message that encourages crime—is probably wrong. As Sampson and Raudenbush observe, “bearing in mind the example of some European and American cities (e.g., Amsterdam, San Francisco) where visible street level activity linked to prostitution, drug use, and panhandling does not necessarily translate into high rates of violence, public disorder may not be so ‘criminogenic’ after all in certain neighborhood and social contexts.”35

The title of the new Manhattan Institute report—Do Police Matter?—asks a good question, with an obvious answer: Yes. The police are uniquely suited to identifying crime trends and patterns and to implement innovative problem-solving techniques to deal with emerging crime situations. Few other governmental agencies or private companies have the immediate information, know-how, human resources, technology, and skills to perform these tasks. And the Manhattan Institute report offers a few good illustrations of problem-solving policing at its finest in New York City. When Brooklyn’s Flatbush precinct experienced a rash of automobile airbag thefts in February and March 2000, for example, the precinct executives implemented numerous strategies—including placing a decoy car in an at-risk area, equipping airbags with tracking devices, readjusting officer deployment, and publicizing these efforts—that resulted in success.

But that question is too easy and is not the right one to ask. The real question is whether an agency like the NYPD needs to implement a policy of making large numbers of arrests for minor misdemeanors and public disorder violations as a primary strategy for combating violent crime. The answer to that question is equally clear: No. The best social scientific and criminological evidence suggests that minor social and physical disorder is not causally related to violent crime.

Troubling consequences

But the Manhattan Institute report does, perhaps unwittingly, reveal the true face of broken windows policing. Since the publication of Wilson and Kelling’s Atlantic Monthly article, there has always been a lingering question concerning the implementation of a broken windows approach. After all, if the aim is improved public order, couldn’t that be achieved with urban renewal projects, homeless shelters, and social workers, as well as or instead of more police arrests? Now we know, from a reliable source. The only number used in the Manhattan Institute report to measure the extent of broken windows policing is the number of precinct-level misdemeanor arrests. The authors made a “decision to use arrests for misdemeanors as our measure of ‘broken windows’ enforcement.”36 The broken windows theory, it turns out, is not so much about public order, as it is about arresting people for misdemeanor and public disorder offenses.

And, of course, that is what we have seen in New York City. Adult misdemeanor arrests in the city have increased throughout the 1990s. According to the New York State Division of Criminal Justice Services, in 1993, the year before Giuliani and Bratton began implementing broken windows policing, total adult misdemeanor arrests stood at 129,404. By the year 2000, the number was up to 224,663—an increase of almost 75 percent. What is particularly interesting is that the vast majority of those arrests were for misdemeanor drug charges, which are up almost 275 percent from 27,447 in 1993 to 102,712 in 2000. Driving while intoxicated (DWI) arrests were down by almost 40 percent over the period (from 5,621 in 1993 to 3,432 in 2000), while other misdemeanor arrests increased by only a quarter (from 96,336 in 1993 to 118,520 in 2000).37 At the same time, the NYPD implemented an aggressive stop-and-frisk policy. Between 1997 and 1998, for instance, the Street Crime Unit—with approximately 435 officers at the time—stopped and frisked about 45,000 people.38

The trouble is, policing strategies that deliberately emphasize arresting misdemeanor and public order offenders—rather than issuing warnings or implementing alternative problem-solving techniques—have significant racial consequences. The fact is that in New York City, and the United States more generally, adults arrested for misdemeanors are disproportionately African-American in relation to their representation in the community. In 2000, for example, slightly over 50 percent of all adults arrested for misdemeanors in New York City were African-American (113,336 of the total 224,663 adults arrested). Slightly over 50 percent of adults arrested for disorderly conduct (19,563 of the total 38,780) and 45.6 percent of adults arrested for loitering were African-American. For prostitution and drug possession, the proportions are 40.7 percent and 51.7 percent respectively. (For DWI, interestingly, the proportion is only 22.3 percent).39 Yet African Americans (in 2000) represented only 24.6 percent of the New York City population.40 Persons of Hispanic descent represented 31.5 percent of all adult misdemeanor arrests, whereas they constituted only 25.1 percent of the city’s population.41 In contrast, European Americans represented 48.8 percent of the population, and accounted for only 15.5 percent of adults arrested on misdemeanor charges in 2000.

These disparities hold true for large cities across the United States. In 1999, for instance, 43.4 percent of adults arrested for vagrancy in large metropolitan areas were African-American; 34.2 percent, 39 percent, and slightly over 40 percent of those arrested for disorderly conduct, prostitution, and drug abuse charges, respectively, were African-American. (Again, curiously, only 10.7 percent of those arrested for driving under the influence were black). Yet African Americans represent less than 15 percent of the total population of these metropolitan areas.42

The point is not that the police are consciously targeting black misdemeanants, but simply that more blacks are arrested for misdemeanors given their proportion in the overall population. In other words, the decision to arrest misdemeanants—adopting that policy in preference to other policing strategies—is a choice with significant distributional consequences for African Americans.

Additionally, there is good evidence that New York City’s policy of aggressive stop-and-frisks was in fact implemented in a racially discriminatory manner. In 1999, New York State Attorney General Eliot Spitzer, with the assistance of Columbia University’s Center for Violence Research and Prevention, analyzed 174,919 stop-and-frisk “UF-250” forms—the forms that NYPD officers are required to fill out in a variety of stop encounters—from the period January 1, 1998 through March 31, 1999. Spitzer found that the raw number of stops was higher for minorities—African Americans and Hispanics—than whites relative to their respective proportion of the population. Spitzer then reanalyzed the raw numbers, this time taking account of the different crime rates and the population composition in different precincts, and found significant disparities across all precincts and crime categories: “in aggregate across all crime categories and precincts citywide, blacks were ‘stopped’ 23% more often (in comparison to the crime rate) than whites. Hispanics were ‘stopped’ 39% more often than whites.” Spitzer concluded from the data that “even when crime data are taken into account, minorities are still ‘stopped’ at a higher rate than would be predicted by both demographics and crime rates.”43

Broken-windows policing has come with other price tags as well. New York City, for instance, experienced illegal strip searches, mounting financial liability on police misconduct charges, clogged courts, wasted resources, and many traumatic encounters for ordinary citizens. The simple fact is that arrests and prosecutions are expensive: a typical prostitution prosecution—one of the offenses targeted by the quality-of-life initiative—costs upwards of $2,000. That’s a lot of money for a single transaction. Moreover, a policy of arrest may have unintended consequences. Someone arrested for turnstile jumping may be fired for missing work; and strained police-civilian relations can create friction between the community and the police force that may be detrimental to solving crimes.

What does this say about order-maintenance as a police technique? First, it is an illusion to think that broken windows policing operates through increased orderliness. Second, order-maintenance probably contributes to fighting crime through enhanced surveillance. Third, broken windows policing comes at a significant cost, with negative distributional consequences for African Americans and other minorities. Fourth, there are alternative strategies of problem-solving policing.

Propensities and human nature

The lack of empirical evidence is a signal that something is wrong with broken windows at the theoretical level. To understand what is wrong, it may be useful to return to the two trends in punishment in modern society, and consider the similarities between broken windows and the incapacitation theory. The surprising fact is that, despite the rhetoric of progressives and liberals, broken windows and incapacitation theory share the same assumptions and underlying logic about the disorderly or delinquent—namely, that these people are by nature disposed toward deviance and that they have a compelling propensity to commit crime. Both theories adopt a common theoretical framework: they both seek to build on the classical model of criminology by incorporating a thicker conception of human preferences based on notions of class culture, human nature, and social norms.

The classical or utilitarian model of criminal behavior was based on the idea that people are not simply impulsive; they act more or less rationally, with an eye to avoiding pain and achieving pleasure. So reducing crime meant, among other things, increasing both the certainty and severity of punishment. Both incapacitation and broken windows theory build on this fundamental assumption. “There is an element of calculation—indeed, a very considerable one—in practically all criminal behavior,” Edward Banfield emphasized. “To be sure, impulse characteristically enters into some types of crime more than into others, but an element of rationality is hardly ever absent.”44 James Q. Wilson and his co-author in Crime and Human Nature, Richard Herrnstein, were equally clear: “Our theory rests on the assumption that people, when faced with a choice, choose the preferred course of action,” they explained.45 Wilson and Herrnstein’s conceptual project was precisely to enrich the classical model by infusing sociological and political insights, while retaining the premise of rational action. Proponents of broken windows policing follow in these footsteps. They too assume rational action on the part of the disorderly, informed by the cues and norms of disorderliness.

More importantly, both theories deploy similar means for enriching the rational action model. First, they each assume a much richer and more deeply ingrained set of human preferences. They each imagine the propensity to commit crime to be a stable social and psychological characteristic. Banfield developed the idea of the “lower-class individual” who is present-oriented to a fault and unable fully to appreciate future benefits and costs. The attitudes, perceptions, and values of the “lower-class individual” are deeply embedded in his psyche and ways, and shape his behavior. Wilson focused on the notion of “human nature” and the “constitutional factors” that shape the way we act and evaluate our options. Proponents of the broken windows strategy develop a theory about the expressive nature of our preferences—the social meanings that people value—and accordingly divide the world into “committed law abiders” and “individuals who are otherwise inclined to engage in crime.”46

The shared outlook is that people in different social groups have different character traits, patterns of behavior, values, tastes, and perceptions. These deep-seated propensities to criminality, recalcitrant to change, make both theories tick. They explain and identify the 6 percent hardcore offenders, the super-predators that need to be incapacitated. But they also explain the distinction between the disorderly and the law-abiders that makes the broken windows theory work. Broken windows, after all, is grounded on these categorical distinctions. The central premise of the theory is that disorder operates on honest people and on the disorderly in different ways. Neighborhood disorder leads honest people to move out of the neighborhood or to lock themselves in their homes. But neighborhood disorder leads the disorderly to move into the neighborhood and commit crimes. Order and disorder operate differently on law-abiders and the disorderly.

Second, both theories argue for more varied interventions aimed at preventing people with deep, relatively fixed propensities to criminality from acting on those propensities. Because these propensities are resistant to change, it may no longer be cost-effective, for instance, to try to rehabilitate or otherwise reconfigure the present-oriented. In fact it may no longer be possible. “The fact is,” Banfield emphasized, “that no one knows how to change the culture of any part of the population—the lower class or the upper, whites or Negroes, pupils or teachers, policemen or criminals.”47 As a result, policy making must focus instead on changing the inducements to commit crime.

Banfield proposed, among other things, eliminating the minimum wage and reducing the school-leaving age in order to remove impediments to the employment of the unskilled and unschooled. He advocated a negative income tax for the more competent poor and intensive birth-control guidance for the rest. These policies were intended to change the situational inducements to crime, by giving youths work and relieving poor women of the burdens of child-rearing. More generally, he advocated incarceration for those “who in the opinion of a court are extremely likely to commit violent crimes.”48 In his essays in Thinking About Crime, James Q. Wilson picked up on this suggestion and strenuously argued for incarcerating a larger number of habitual offenders. Drawing on his earlier study of police departments, Wilson also argued, in the later “Broken Windows” essay, for greater attention to neighborhood cues of order and disorder.

In sum, the broken windows theory has the same intellectual roots and the same conceptual structure as incapacitation theory. The notion of ingrained propensities to commit crime lies behind both the idea of a hardcore habitual offender who must be incarcerated and of a disorderly person who must be arrested, controlled, and relocated.

The theories reflect, more than anything, our modern conception of the disorderly as an unattached, young, most often racialized other, with a powerful tendency to commit crime and cause disorder. It is the youth or young adult, threatening, defiant, suspicious, often black, wearing distinctive designer-label clothes. Or the down-and-out street person in a dirty oversized coat. Or the squeegee man, the panhandler, the homeless person, the turnstile jumper, the public drunk. The “disreputable or obstreperous or unpredictable people.”49 “Unattached males, the homeless, and the aimless [who] live in boarded up buildings, seedy residential hotels and flophouses.”50 “[I]ndividuals who are otherwise inclined to engage in crime.”51 Or, as James Q. Wilson wrote in 1968, the “teenager hanging out on a street corner late at night, especially one dressed in an eccentric manner, a Negro wearing a “conk rag” (a piece of cloth tied around the head to hold flat hair being “processed”—that is, straightened), girls in short skirts and boys in long hair parked in a flashy car talking loudly to friends on the curb, or interracial couples—all of these are seen by many police officers as persons displaying unconventional and improper behavior.”52

We know the disorderly when we see them. We can identify them easily. From their clothing, their habits, their glitter, their grit, their smell, their look. We just know who they are—entre nous.

A broken theory

Or do we really? After all, what exactly is the distinction between eccentricity, nonconformity, unconventionality, difference, disorder, and criminality? What makes distinctive clothes, youthful exuberance, or loitering disorderly? Why does order maintenance focus so heavily on certain types of street disorder and not others? Police brutality is a form of disorder, yet it appears nowhere as a target of broken windows policing. Everyday tax evasion—paying cash to avoid sales tax, paying nannies under the table, using an out-of-state address—is disorderly. So are public corruption, sham accounting practices, nepotism, insider trading, and fraud. Why does broken windows focus on the dollar-fifty turnstile jump rather than on the hundred-million dollar accounting scam? And what exactly is the meaning of neighborhood disorder? Sure it may signal that a community is not in control of crime. But it may also reflect an alternative subculture, political protest, or artistic creativity. An orderly neighborhood may signal commercial sex, wealthy neighbors with personal bodyguards, foreign diplomats, a strong mafia presence, or a large police force.

The central claim of the broken windows theory—that disorder causes crime by signaling community breakdown—is flawed. The categories of “disorder” and “the disorderly” lie at the heart of the problem. Those categories do not have well-defined boundaries or settled meanings. When we talk about “disorder,” we are really referring to certain minor acts that some of us come to view as disorderly mostly because of the punitive strategies that we inflict as a society. 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 broader patterns of serious crime. Hanging out on the front steps of a building or loitering with neighbors only signals that the community is not in control if hanging out or loitering is perceived as violating certain rules of conduct. But, of course, that depends on the neighborhood—and in some, in fact, it reflects strong community bonds and informal modes of social control. Graffiti only signals that the neighborhood is indifferent to crime if graffiti is viewed as violating the rules of the community. But graffiti is sometimes understood to be political or artistic expression or social commentary.

Order-maintenance policies, ironically, have the effect of reinforcing the idea that disorder causes crime. The “squeegee man,” for instance, has become the barometer of crime control in New York City primarily because of the quality-of-life campaign. And one might speculate that, as order maintenance has become more and more popular, the categories of the disorderly and law-abider may have become slightly more fixed in meaning. The broken windows theory has, in this sense, a self-reinforcing logic: it helps shape the perceptions, emotions, and judgments we form about people who are homeless, hustling, or panhandling. Still, the best social scientific evidence suggests that there are mixed signals associated with disorder—disorder does not correlate with crime in most tests. In sum, it is an illusion to believe that the order in order maintenance is necessary to combat crime.

When we peel away disorder from crime, it becomes clear that the model of orderliness offers to contemporary society, first, a classic mode of enhanced surveillance. The criminological evidence suggests that it is not the order, but the surveillance associated with broken windows policing that has probably contributed to the drop in crime in a city like New York. It is the increased arrests, background checks, fingerprint comparisons, stop-and-frisks, line-ups, and informants. It is, in effect, the increased police-civilian contacts. Are they necessary? The answer is no. Many cities in the United States have experienced remarkable drops in crime during the 1990s without implementing similar order maintenance strategies. The costs are simply too steep, especially to minority communities. And there are alternatives, such as the problem-solving policing discussed, in part, in the new Manhattan Institute report.

Second, order maintenance policing provides a way to enforce an aesthetic preference, under the guise of combating serious crime. Order maintenance is a way to get homeless people, panhandlers, and prostitutes off the street. It is a way to keep the avenues clean of graffiti and litter. Broken windows policing is a way to repossess red light districts and displace street vendors, panhandlers, and ordinary street life. It is, in effect, a type of “aesthetic policing” that fosters a sterile, Disneyland, consumerist, commercial aesthetic. It reflects a desire to transform New York City into Singapore, or worse, a shopping mall. The truth is, however, that when we lose the dirt, grit, and street life of major American cities, we may also threaten their vitality, creativity, and character.


1 Kathleen Maguire and Ann L. Pastore, eds., Sourcebook of Criminal Justice Statistics 2000 (Washington, D.C.: U.S. Government Printing Office, 2001), Table 6.1 and 6.27. Available at:<http://www.albany.edu/sourcebook> [12 March 2002].

2 Christina Nifong, “One Man’s Theory Is Cutting Crime in Urban Streets,” The Christian Science Monitor, 18 February 1997; Michael Massing, “The Blue Revolution,” The New York Review of Books, 19 November 1998, 32.

3 Elissa Gootman, “A Police Department’s Growing Allure: Crime Fighters From Around World Visit for Tips,” The New York Times, 24 October 2000.

4 Edward Banfield was James Q. Wilson’s professor and mentor at the University of Chicago, and, later, colleague and co-author in the Government Department at Harvard University.

5 James Q. Wilson, Thinking about Crime (New York: Basic Books, 1975), 209.

6 Edward C. Banfield, The Unheavenly City Revisited (Boston: Little, Brown, 1974), 270.

7 William J. Bennet et al., Body Count: Moral Poverty…and How to Win America’s War Against Crime and Drugs (New York: Simon & Schuster, 1996), 91.

8 Bennett, Body Count, 128–30.

9 Dan M. Kahan and Tracey L. Meares, “Law and (Norms of) Order in the Inner City,” Law & Society Review 32 (1998): 805, 806.

10 Adam Nagourney, “Democrats Run In the Shadow Cast by Mayor,” The New York Times, 2 September 2001.

11 George L. Kelling and William H. Sousa, Jr., Do Police Matter? An Analysis of the Impact of New York City’s Police Reforms, Civic Report No. 22 (New York: Manhattan Institute, December 2001), 9.

12 George L. Kelling and William H. Sousa, “Tough Cops Matter,” New York Post, 19 December 2001.

13 Editorial, “It’s the Cops, Stupid,” New York Post, 19 December 2001.

14 George L. Kelling, “A Policing Strategy New Yorkers Like,” The New York Times, 3 January 2002.

15 In 1994, a precinct was divided in two, resulting in seventy-six precincts existing today. To maintain consistency over the studied period, the authors use the original seventy-five precincts. See Kelling, Do Police Matter? 1, 4.

16 The method they use is “hierarchical linear modeling,” which is a variant of regression analysis.

17 Kelling, Do Police Matter? 9.

18 Ibid., 1.

19 See Kevin Flynn, “Study Says a Slumping Economy Doesn’t Mean Crime Will Rise,” The New York Times, 20 December 2001.

20 George L. Kelling and Catherine M. Coles, Fixing Broken Windows(New York: The Free Press, 1996), 24.

21 There are a number of other weaknesses in Skogan’s data and study. For example, the survey question that was posed to neighborhood residents about robbery victimization was not neighborhood specific. In addition, there was a set of five neighborhoods in Newark that exerted excessive influence on the robbery victimization findings. For criticism of Wesley Skogan’s study, see Bernard E. Harcourt, Illusion of Order: The False Promise of Broken Windows Policing, (Cambridge, Mass.: Harvard University Press, 2001), 59–78.

22 James Q. Wilson and Barbara Boland, “The Effect of the Police on Crime,” Law & Society Review 12 (1978): 367–90; Wilson, “The Effects of the Police on Crime: A Response to Jacob and Rich,” Law & Society Review 15 (1981): 163–9.

23 Robert J. Sampson and Jacqueline Cohen, “Deterrent Effect of the Police on Crime: A Replication and Theoretical Extension,” Law & Society Review 22 (1988): 166.

24 Anthony A. Braga et al., “Problem-Oriented Policing in Violent Crime Places: A Randomized Controlled Experiment,” Criminology 37 (1999): 541–80.

25 Sampson and Cohen, “Deterrent Effect of the Police on Crime,”185.

26 Robert J. Sampson and Stephen W. Raudenbush, “Systematic Social Observation of Public Spaces: A New Look at Disorder in Urban Neighborhoods,” American Journal of Sociology 105 (2000): 637, 638.

27 See Jeffrey Fagan, Franklin Zimring, and June Kim, “Declining Homicide in New York City: A Tale of Two Trends,” Journal of Criminal Law & Criminology 88 (1998): 1280–6.

28 See Ana Joanes, “Does the New York City Police Department Deserve Credit for the Decline in New York City’s Homicide Rates? A Cross-City Comparison of Policing Strategies and Homicide Rates,” Columbia Journal of Law and Social Problems 33 (1999): 303–4.

29 See Fox Butterfield, “Cities Reduce Crime and Conflict Without New York-Style Hardball,” The New York Times, 4 March 2000. Statistics compiled by Alfred Blumstein.

30 Judith A. Greene, “Zero-Tolerance: A Case Study of Police Policies and Practices in New York City,” Crime and Delinquency 45 (1999): 182–5.

31 Khaled Taqi-Eddin and Dan Macallair, Shattering ‘Broken Windows’: An Analysis of San Francisco’s Alternative Crime Policies (San Francisco: Justice Policy Institute, 1999), 4–5, 7, 8, 10, 11. Available at: <http://www.cjcj.org/jpi/windows.html> [12 March 2002].

32 See U.S. Department of Justice, Federal Bureau of Investigation,Uniform Crime Reports 1991 (Washington, D.C.: U.S. Government Printing Office 1992), Table 78; Uniform Crime Reports 2000(Washington, D.C.: U.S. Government Printing Office 2001), Table 78.

33 See William J. Bratton, Turnaround: How America’s Top Cop Reversed the Crime Epidemic (New York: Random House, 1998).

34 Bratton, Turnaround, 154.

35 Sampson and Raudenbush, “Public Spaces,” 638.

36 Kelling, Do Police Matter? 23.

37 New York State Division of Criminal Justice Services, Criminal Justice Indicators New York City: 1992–2000,<http://criminaljustice.state.ny.us/crimnet/ojsa/areastat/areast.htm> [12 March 2002].

38 Jeffrey Rosen, “Excessive Force: Why Patrick Dorismond didn’t have to die,” New Republic,10 April 2000, 26; Eliot Spitzer, The New York City Police Department’s ‘Stop & Frisk’ Practices: A Report to the People of the State of New York From The Office Of The Attorney General (New York: Office of the Attorney General of the State of New York, Civil Rights Bureau, 1999), 56–9. Available at: <http://www.oag.state.ny.us/press/reports/stop_frisk/stop_frisk.html> [15 March 2002].

39 Marge Cohen, New York State Division of Criminal Justice Services, fax to author, 4 January 2002, data from UCR system and Computerized Criminal History system.

40 U.S. Bureau of the Census, The Black Population: 2000: Census 2000 Brief (Washington, D.C., 2001).

41 Cohen, New York State Division of Criminal Justice Services, fax to author; U.S. Bureau of the Census, 2000 Census of Population and Housing: Profiles of General Demographic Characteristics (Washington, D.C.: U.S. Government Printing Office, 2001), Table DP-1, 345.

42 U.S. Department of Justice, Bureau of Justice Statistics, Sourcebook of Criminal Justice Statistics 2000 (Washington, D.C.: U.S. Government Printing Office, 2001), Table 4.12, 372; for demographic data, see U.S. Bureau of the Census, Profiles of General Demographic Characteristics, Table DP-1, 2. Note that 13.2 percent of the population inside metropolitan areas is African-American.

43 Spitzer, ‘Stop & Frisk’ Practices, 89, 123.

44 Banfield, The Unheavenly City Revisited, 181.

45 James Q. Wilson and Richard J. Herrnstein, Crime and Human Nature(New York: Simon and Schuster, 1985), 43.

46 Dan M. Kahan, “Social Influence, Social Meaning, and Deterrence,”Virginia Law Review 83 (1997): 349, 370–1.

47 Banfield, The Unheavenly City Revisited, 263.

48 Ibid., 269–70.

49 Wilson and Kelling, “Broken Windows,” 30.

50 Wesley G. Skogan, Disorder and Community Decline: Final Report to the National Institute of Justice (Evanston, Ill.: Northwestern University Center for Urban Affairs and Policy Research, 1987), 86.

51 Kahan, “Social Influence,” 370.

52 James Q. Wilson, Varieties of Police Behavior, (Cambridge, Mass.: Harvard University Press, 1968), 39–40.