POKING HOLES IN THE THEORY OF 'BROKEN WINDOWS'

If there were a Hall of Fame for influential public-policy ideas, then the
"broken windows" thesis would probably have its own exhibit. In an Atlantic
Monthly article by that name published in 1982, James Q. Wilson and George
L. Kelling popularized the idea that neighborhoods that neglected minor
signs of decay and disorder were opening the door to serious crime.

"One unrepaired broken window is a signal that no one cares," they wrote,
"and so breaking more windows costs nothing."

Nowhere has "broken windows" thinking been more conspicuous than in New
York City, where the rates of serious crime have fallen spectacularly over
the last 10 years. William Bratton, the former police chief, translated the
thesis into "order maintenance" -- aggressive enforcement of laws against
nuisance offenses -- and shared the credit with the authors of the Atlantic
Monthly article.

Recently, however, the "broken windows" thesis has been challenged by
scholars in sociology, criminology, and political science. Some researchers
believe that empirical evidence for the connection between disorder and
crime is weak and overblown. Others argue that New York City's success has
been oversimplified and distorted. The city's amazing drop in crime, they
say, reflects a complicated array of factors that are difficult to tease apart.

That complexity bedevils criminology in general. Scholars regard the
nationwide decline in crime rates over the last decade as a grand puzzle,
of which "broken windows" is only a part. Police departments have become
remarkably creative laboratories for innovative tactics and policies, many
of which go by the vague label "community policing." But to policymakers
and citizens eager to know whether smart policing can prevail over the
"root causes" of crime, social scientists have been forced to say: "We may
never know."

The essence of "broken windows" is that neighborhood disorder -- physical
decay, such as graffiti, litter, and dilapidation; and minor misconduct,
such as public drinking and vagrancy -- will, if left unchecked, signal
potential miscreants that no one is watching.

In the spring of 1990, that idea put on a uniform and hit the streets -- or
more precisely, the subway platforms. Mr. Kelling, now a criminologist at
Rutgers University at Newark, was advising Mr. Bratton on the
crime-disorder connection during his stint as chief of New York City's
transit police. "We decided to apply this concept to crime in the subways,"
Mr. Bratton writes in his 1998 memoir, Turnaround: How America's Top Cop
Reversed the Crime Epidemic (Random House). "Fare evasion was the biggest
broken window in the transit system. We were going to fix that window and
see that it didn't get broken again."

He stationed roving squads of plainclothes cops to catch turnstile jumpers.
To his delight, he discovered that many were carrying illegal weapons or
flouting arrest warrants. As arrests for such misdemeanors surged, subway
crime of all kinds receded.

After Mayor Rudolph W. Giuliani hired him as the city's police commissioner
in 1993, Mr. Bratton began a "quality of life initiative" to crack down on
panhandling, disorderly behavior, public drinking, street prostitution, and
unsolicited windshield-washing by the city's notorious squeegee people.
Combined with computerized tracking of crime hot spots and other innovative
policies, "order maintenance" was credited with reducing felony crime by 27
percent in its first two years.

Yet little empirical evidence has ever emerged for the idea that disorder,
left unchecked, causes crime. Advocates of the disorder thesis could point
to little more than research from the 1980's by Wesley G. Skogan, a
political scientist at Northwestern University. As he reported in his book
Disorder and Decline: Crime and the Spiral of Decay in American
Neighborhoods (University of California Press, 1990), Mr. Skogan analyzed
earlier surveys of residents in 40 neighborhoods throughout a handful of
large cities. He found that measures of social and physical decay
correlated with certain kinds of serious crime.

In Fixing Broken Windows: Restoring Order and Reducing Crime in Our
Communities (Simon & Schuster, 1996), Mr. Kelling and a coauthor write that
Mr. Skogan "established the causal links between disorder and serious
crime." But Mr. Skogan had been more cautious. "With only 40 cases to
untangle this web," he had written, "the high correlation between measures
of victimization, ratings of crime problems, and disorder make it difficult
to tell whether they have either separate 'causes' or separate 'effects.' ..."

In the 1990's, New York City led the nationwide decline in serious crime,
and began touting "order maintenance." As big cities trumpeted their
success with various kinds of community policing, Bernard E. Harcourt
decided to give the "broken windows" thesis a second look.

A law professor at the University of Arizona with a Ph.D. in political
science, Mr. Harcourt could not find any skeptical examination of the
evidence. In a 1998 article in the Michigan Law Review, he concludes that
Mr. Skogan's data don't hold up. The link between neighborhood disorder and
purse-snatching, assault, rape, and burglary disappears when poverty,
neighborhood stability, and race are factored out. Only the link to robbery
remains statistically significant.

But even that result is suspect, writes Mr. Harcourt. Of the 40
neighborhoods in Mr. Skogan's sample, the strongest link occurs in five
contiguous neighborhoods in Newark. Without those, the link disappears
completely. Furthermore, he wrote, the various surveys didn't ask exactly
the same questions, so few of the calculations included data from all the
neighborhoods.

What really seems to bother Mr. Harcourt, however, is not the proof but the
principles of order maintenance. "'Broken Windows' is an essay about
creating order," he says. "But when you look at the text, you see that the
order is achieved in part through disorder." The 1982 article, he notes,
approvingly quotes an unnamed officer as saying "we kick ass" -- that is,
chase known gang members out of a housing project before the gang gets the
upper hand.

And New York's "zero tolerance" for quality-of-life infractions, Mr.
Harcourt says, has clogged up municipal courts and sapped judicial
resources. He is expanding his critique for Illusion of Order: The False
Promise of Broken Windows Policing, to be published by Harvard University
Press in June.

When Ralph Taylor, a criminologist at Temple University, sought to connect
neighborhood crime with what he calls "incivilities," his results were full
of complications. In 1994-95, he measured crime and disorder in 30
Baltimore neighborhoods and compared them with data from 1981-82. Some
kinds of incivilities could be linked with crime or neighborhood decay, it
turns out, while others could not. Increased numbers of assaults appeared
to be connected with physical disorder, and more rapes with social disorder.

That inconsistency, he writes in Breaking Away from Broken Windows (just
published by Westview Press), suggests that "incivilities may not reflect
an underlying disorder, but rather a constellation of only loosely
connected, somewhat separate problems that may each require somewhat unique
policy responses."

"We have focused too much attention on too small a set of strategies," he
says. "We've been saying, 'Let's reduce physical problems and let's roust
disorderly and homeless people.'"

An ongoing study of crime, delinquency, and urban neighborhoods, led by
Robert J. Sampson, a sociologist at the University of Chicago, has also
bolstered the critics of "broken windows." In the mid-1990's, Mr. Sampson's
research team surveyed thousands of Chicago families with children, rated
nearly 24,000 city blocks for physical and social disorder, and tried to
correlate that information with five kinds of predatory crime. They found,
Mr. Sampson wrote in the American Journal of Sociology in 1999, that
disorder is only a "moderate correlate of predatory crime."

Once they accounted for other neighborhood characteristics thought to be
associated with crime, such as poverty and instability, the connection with
disorder disappeared for every category except robbery. They conclude that
serious crime and disorder might both be symptoms of deeper social and
economic disadvantages.

Although he is "hesitant to make causal claims," Mr. Sampson says, his data
so far suggest that "concentration of disadvantage and high instability are
associated with high rates of violence" in a given neighborhood. But that
effect is moderated in neighborhoods in which residents trust one another
more and take collective action to protect and improve their surroundings.

In Fixing Broken Windows, Mr. Kelling regards New York City as Exhibit A
for the effectiveness of "order maintenance." He's not alone. Eli B.
Silverman, a criminologist at John Jay College of Criminal Justice,
observed the New York Police Department's innovative tactics up close in
the 1990's and praised them in NYPD Battles Crime: Innovative Strategies in
Policing (Northeastern University Press, 1999). Rates of serious crime
dropped earlier, faster, and farther in New York City than elsewhere, he
concludes, due in part to order-maintenance strategies.

Yes, he's heard the rebuttals. Crime began to decline several years before
Mr. Bratton's innovations -- but not, he says, at an annual rate in double
digits. Violent crime declined just as drastically in San Diego, which
never used Brattonesque policies, and almost as quickly in many other
cities. But that's because police departments in all those places have
gotten smarter.

"The real revolution has been in policing. Police departments are better at
learning from each other than ever before," he says. "They all have in
common one thing: more intelligent policing."

But Andrew Karmen, a sociologist at John Jay, thinks that's giving police
tactics too much credit. "N.Y.P.D. or not N.Y.P.D.? That is the question,"
he writes in New York Murder Mystery: The True Story Behind the Crime Crash
of the 1990's (New York University Press) just published in January.

In his view, the two-thirds drop in the number of homicides during the
decade was due to many factors, including the end of murderous turf wars
among crack dealers. And even the evidence for effective policing is
riddled with caveats and exceptions. Crime rates in New York City peaked
around 1990, before most significant innovations in policing. Crackdowns on
gun possession generally cut the murder rate -- except when they don't, as
New York discovered in the midst of the crack wars of the mid-1980's.

Mr. Wilson, the coauthor of the "Broken Windows" article and a political
scientist now retired from the University of California at Los Angeles,
does not claim that his thesis has been proved. He was unavailable for an
interview with The Chronicle, but at last summer's meeting of the American
Political Science Association, Mr. Wilson said, "It's only a theory." And
he told The New York Times last year, "God knows what the truth is."

Mr. Kelling, however, has not been so reticent. "'Broken windows' remains a
hypothesis which has a lot of support in the form of anecdotes and case
studies," he says. "My evidence for this continues to be the constant,
repeated contact I have with citizens and communities. I don't care where
you go, citizens believe that there's this link ... that every time you
start to restore order, crime goes down."

He is working on a historical study of policing strategies of all 76 New
York City precincts. "If, as order maintenance goes up, crime goes down,
and you get the same phenomenon in all 76 different precincts -- that's not
scientific evidence, but it's awfully coincidental," he says. "People who
don't think it has an impact will have to explain a whole lot of impacts."

The debate over "broken windows" is more than just a dispute over the
effectiveness of a particular policy. Skeptics like Mr. Harcourt see
"broken windows" as a harmful, conservative philosophy masquerading as
pragmatic and progressive public policy.

Because "order maintenance" seems less punitive than get-tough sentencing
and incarceration policies, says Mr. Harcourt, it appeals to more-moderate
and liberal voters. Furthermore, he adds, the "disorder" thesis implies
that fixing surface symptoms of decay can replace a deeper, liberal concern
about "root causes" of crime -- such as poverty, discrimination, and a lack
of economic opportunities in the inner cities.

Such ideological conflicts explain why critics like Mr. Sampson assess the
drawbacks of "zero tolerance," Mr. Kelling says, as if the "broken windows"
thesis were all about rigid and draconian enforcement. Mr. Kelling suggests
his detractors reread his 1982 article. There is nothing in it, he says,
about "zero tolerance" of minor infractions. Rather, it reads: "The essence
of the police role in maintaining order is to reinforce the informal
control mechanisms of the community itself." That means collaborating with
civilians and training cops in the exercise of discretion.

The debate over "broken windows" is just one part of a larger dispute over
the lessons of the 1990's. The dramatic decline in crime offers scholars a
unique opportunity to test different theories about why crime rates rise
and fall.

One way of interpreting the evidence of the 1990's, say scholars, is the
reverse of Murphy's Law: Virtually everything that could go right, did.
Turf wars in the crack trade died down. The number of young males between
the ages of 18 and 24 -- the crime-prone years -- shrank. Unbroken economic
growth provided disadvantaged young people with attractive alternatives to
crime. In addition, scholars credit public policies, such as those that led
to higher rates of incarceration.

The uncertainty that surrounds the New York story is typical of
criminology. In The Crime Drop in America, an edited volume published last
October by Cambridge University Press, John E. Eck and Edward R. Maguire
survey the evidence behind a plethora of policing reforms, including
"problem-oriented policing," "zero tolerance," and aggressive efforts to
get illegal handguns off the streets.

Based on available evidence, Mr. Eck and Mr. Maguire, associate professors
of criminology at the University of Cincinnati and George Mason University,
respectively, could draw firm conclusions about none of them.

We may never know for sure how policing affects crime. True experiments
require control groups and random samples for reforms introduced one at a
time. That, in turn, rests on finding officials and citizens willing to
tolerate differential treatment for the sake of discovery. That's so rare
that textbooks and scholars today still cite a 30-year-old experiment as
the standard for research on policing tactics.

In the early 1970's, as a young criminologist, Mr. Kelling persuaded the
police department of Kansas City to let him test the effect of
drive-through police patrols on neighborhood crime. The department
increased patrols in five randomly chosen neighborhoods, reduced them in
five others, and made no changes in a third group, as a control.

The results were both valuable and surprising: Increasing patrols made no
difference to crime rates.

Even criminologists' best efforts, however, cannot replicate the
laboratory. Scholars have challenged the Kansas City study, on the grounds
that the results from the control areas might have been tainted by patrol
cars passing through en route to other neighborhoods. And a sample of 15
neighborhoods, Mr. Kelling admits, isn't a lot on which to base conclusions.

Nowadays, say scholars, police officials are more open than ever to
cooperating with scholars. But innovations in policing are driven by the
needs and hopes of politicians and police officials, who adopt and modify
and discard tactics all the time. That is no way to run an experiment.

John Jay's Mr. Karmen believes that firmer conclusions about how much
policing matters are within reach. The answers, he says, lie in municipal
filing cabinets and on hard drives, where police departments keep data on
urban crime patterns that they have not been eager to make public.

Others are less optimistic. The precipitous drop in crime offered a prime
opportunity to set up experiments like Mr. Sampson's, conduct surveys, and
observe policing up close. Now that recent statistics suggest that trend is
slowing, scholars wonder whether they have missed the chance of a lifetime.
"We'll be having this debate for the next 15 years," says Mr. Kelling.