In the late 1990s, Cynthia Lum was a beat cop in Baltimore’s eastern district, and her tenure was marked by skyrocketing rates of homicide, rape, robbery, assault, arson and theft. Like many young police officers trying to combat violent crime, Lum assiduously stuck with the tactics she knew best: respond to calls, remember your training, and follow procedure. And at the time, training and procedure were largely predicated on the presumed benefits of targeting minor infractions.
“We were doing a lot of what was called ‘aggressive policing’ and that meant making lots of arrests for misdemeanor crimes and doing a lot of stop and frisk,” Lum recalled. “Broken windows had just come out.”
Introduced in the March 1982 issue of The Atlantic magazine, “Broken Windows” was the brainchild of George L. Kelling, a criminologist, and James Q. Wilson, a political scientist. At its heart was the idea that physical and social disorder – a broken window, a littered sidewalk, public drunkenness – are inextricably linked to criminal behavior. By focusing on repairing the windows, cleaning up the streets, and dissuading crude behavior, Kelling and Wilson suggested, police departments can help to forestall more serious crimes from ever taking shape.
The idea, which seemed to make some intuitive sense, had immediate purchase on the popular mind, and it’s little surprise that the “broken windows theory,” as it became known, went on to inform policing strategies from New York City to Los Angeles and everywhere in between. It has also given rise to a whole genre of spinoff and ancillary theories – all of them imbued with the idea that focusing on low-level disorder will have some concomitant impact on broader categories of crime.
But even as Lum and other patrol officers dutifully followed what they believed to be sound policing policy, critics were picking apart broken windows and its offshoots, arguing that for all of their appeal, pat theories present far too simplistic a view of the push-and-pull between crime and police work, and often with little data to back them up. That’s starting to change, and researchers are now developing much more nuanced and statistically rigorous views of which tactics work, when they work (or don’t), and why.
The question now is how quickly such insights will permeate police culture overall, and Lum — now off the streets of Baltimore and working as an associate professor of criminology at George Mason University, where she directs the school’s Center for Evidence-Based Crime Policy — suggests that decades of blind faith in the orthodoxies of broken windows remains a powerful impediment to change.
“The correlation between disorder and crime is very obvious, you can see it all the time,” Lum said. “But whether or not the processes that Kelling argued about — that disorder begets violent crime, that there’s a causal connection between the two — that is something that is much more debated.”
That debate over broken windows reached a crescendo in June, when the New York City Department of Investigation’s Office of the Inspector General released an assessment of the police department’s longstanding emphasis on so-called quality-of-life policing — an outgrowth of broken windows. The analysis examined five years of arrest and crime data in a hunt for some statistical relationship between quality-of-life arrests — those made for such offenses as public urination, disorderly conduct, and drinking alcohol in public — and a reduction in felony crimes. The results were undeniable: “OIG-NYPD’s analysis has found no empirical evidence demonstrating a clear and direct link between an increase in summons and misdemeanor arrest activity and a related drop in felony crime,” the report stated.
The NYPD promptly refuted the OIG’s analysis, calling its methodology “deeply flawed.” Just this week, outgoing NYPD commissioner Bill Bratton — perhaps the best-known and earliest advocate of broken windows — countered the city’s study by issuing an NYPD-branded analysis supporting the theory.
On Wednesday, the New York Department of Investigation dismissed Bratton’s study as lacking in statistical muscle — and it reiterated the advice it gave the NYPD back in June: “Rely on a more data-driven approach” to determining what works and what doesn’t.
That kind of advice should have come much sooner, experts say. In the 30 years since the Atlantic article first appeared, there have been thousands of studies carried out in U.S. cities analyzing a variety of disorder-focused law enforcement strategies and whether they reduce criminality. But the quality of the studies, along with their results, are very much a mixed bag. Where one study might suggest that clearing out homeless encampments in Los Angeles reduces violent crime, it has proved all too easy to find another analysis of, say, aggressive traffic enforcement in Ohio, that finds no impact on more serious crimes at all.
“That’s part of the problem with ‘broken windows’ literature,” said Dan O’Brien, an assistant professor at Northeastern University’s School of Criminology and Criminal Justice. “There’s just so many studies, that people will point to whatever study supports their argument.”
Adding to the problem is that most of the research in this area hasn’t been very scientific, with conclusions often hinging on sporadic case reports or policy analyses rather than hard data, according to systematic reviews of the literature. This presents a challenge for researchers and police departments looking for empirical data on disorder policing. “Without any actual scientific theoretical development, you end up with something that people keep reinterpreting in different ways,” said O’Brien.
In one effort to get beyond all this, Lum and her colleagues have developed the Evidence-Based Policing Matrix, a visualization of the rigorous scientific research that she hopes will convince police departments to adopt disorder reduction tactics – whether they reference broken windows or not – that have been empirically shown to have an effect on reducing crime. Her matrix ranks more than 140 studies of intervention strategies according to how general or focused they were, how proactive or reactive they were, and who or what was being targeted.
“The matrix is like a starting point,” said Lum. “It’s like a visual systematic review. It takes a whole bunch of information and it distills it into something very quick that an officer can look at or a commander can use for strategic planning.” By organizing the disorder policing research in such a way, Lum is able to offer tailored, scientifically-proven strategies to the police departments with which she consults.
After discussions with Lum’s colleagues at George Mason in 2012, the Seattle Police Department, for example, refocused police patrols on high crime street blocks – a data-driven intervention known as “hot spots” policing. According to an editorial in The Seattle Times, “in the first week of the patrols, the number of 911 calls from the four ‘hot spots’ in the West Precinct — which includes downtown — dropped 60 percent during the 11 a.m. to 8 p.m. shift, traditionally when the most calls come in.” Similar hot spots experiments resulted in crime reductions in Minneapolis in the 1980s and Sacramento in 2011 and continue to be piloted in other cities. The research shows that crime can be reduced, at least moderately, by ramping up police presence in places with statistically higher rates of it. And a brief, 10 to 15 minute “dosage” of police presence in an area migh maximize deterrence, one study found.
So-called “problem-oriented” strategies, aimed at solving specific issues facing a community, have also been backed up with data. A randomized control trial of 83 crime hot spots in Jacksonville, Florida, for example, found a 33 percent reduction in street violence during a three-month period in 2009 when police enacted targeted solutions. These included erecting fences around apartment complexes to prevent car thefts, and blocking a highway off-ramp to prevent out-of-towners from driving into an area rife with violence and prostitution.
Importantly, experts say, both of those interventions managed to reduce crime, not just displace it.
Bruce Taylor, a co-author of the Jacksonville study and a senior fellow in public health at the University of Chicago’s National Opinion Research Center, said that problem-oriented policing, or POP, works because it relies on an evidence-based approach that involves interacting with the community and identifying specific reasons for violence. “It’s not really rocket science,” he said. “You look at the data, scan for the problems and form a response based on the nature of the data.” Taylor is careful to point out that POP is distinct from other strategies that excessively ticket individuals for misdemeanor crimes. “With POP, you’re not necessarily trying to arrest your way out of the problem,” he said.
The work of Taylor and Lum joins that of numerous other researchers who are now compiling a much more vivid and nuanced picture of policing that eschews catch-all theories in favor of more granular and situational assertions — ones that tacitly acknowledge the myriad variables that might make one approach effective in one place, but less so in another. “It’s not just what interventions might work,” Lum said, “but what dosages of interventions.”
A 2014 study in Philadelphia, for example, randomly addressed crime hot spots with one of several tactics: foot patrols in some, problem-oriented policing in others, or a focus on repeat, violent offenders in still others. It found that offender-focused policing reduced violent crime by more than 40 percent. Foot patrols and POP, at least for the duration of the study, failed to reduce violent crime. And a 2012 study in Glendale, Arizona was able to reduce crime at Circle K convenience stores by employing problem-oriented policing strategies that included surveillance, collaborating with management, and working with the community to prevent crime.
Taken together, Taylor and others suggested, these findings could be the seeds of a revolution in policing tactics — one that might finally move beyond broken windows and toward a more thoughtful complement of strategies that not only work, but are rooted in real evidence. “We’ve learned quite a bit about the importance of using data-driven approaches,” Taylor said — though he added that more research is needed. “We haven’t replicated these experiments in very many places.”
That’s going to take both time and dedication, Lum said.
“I see agencies genuinely wanting to try something new, but that is a sea change in the way they train their officers and what they do on a daily basis,” she said. “But if they don’t have the tools to think differently about crime or do something different about crime, then they’re going to revert back to what they know.”
With that in mind, Lum has begun taking her matrix on the road and talking to police departments around the country about incorporating the most statistically effective interventions into their own work. In addition to anecdotes from her time in Baltimore, Lum brings with her field training activities, performance review checklists, and even playbooks for on-duty patrol officers on how to be proactive, all filled with real-world interventions shown to reduce crime.
“When we work with police agencies, we start off by telling them this is not going to be easy,” Lum said. “This is like telling you that you have to put on your pants a different way every morning.”