Can math stop murder?

In besieged Chicago, how police are tapping Big Data to try to curb gang violence.

Cover Story

It was an unusual encounter from the start. On a cold evening in February, Chicago police officer Maria Peña knocked on the door of a house in one of the city’s most crime-ridden neighborhoods. She wanted to talk to a leader of the Latin Kings, a gang on the city’s South Side, as well as his mother. Ms. Peña was not there to question or arrest the man. She wanted to save his life.

The gang leader opened the door. “He was a little wary to see me,” says Peña, commander of the Chicago Police Department’s 10th District. Nevertheless, his mother welcomed Peña and several other local officials. She invited them to sit around the kitchen table with her son, his girlfriend, and a 6-year-old niece. The kitchen was tidy, the atmosphere cordial. But the talk turned blunt quickly.

“I looked at the mom and said, ‘Honestly, do you want to bury your son?’ ” And, she told her, it might not be just him. Peña recounted a recent retaliation shooting in the neighborhood in which an innocent bystander – a 10-month-old girl – fell victim.

“Do you want your little granddaughter being the next victim?”

She urged the son to jettison his gang-related activities and offered a variety of social services to help him do it. 

Since then, Peña is not sure if the gang leader changed his ways or whom he associates with. What she does know is that he’s still alive – and shootings related to the Latin Kings have stopped, at least temporarily, in her district.

Peña was basing her warnings in part on intelligence the police had gleaned from computer algorithms. Her visit on that raw night was part of a pioneering attempt by the Chicago Police Department (CPD) to harness the power of Big Data to stop crime before it happens. 

Armed with a plethora of statistics on everything from gun violations to individual parole and arrest histories, police here are trying to create a national model that will help them predict where shootings might occur and who might be involved – both victims and offenders. Then they use the information to reach out to people in the neighborhoods in the hope of preventing the guns from ever being brandished.

“I don’t think the families always know these guys are out there doing what they are doing,” says Peña. “When you have these kind of conversations with their families, and the [gang members] see the hurt in their mother’s eyes, I think that hurts them, because the family means everything to them.”

Chicago’s experiment is part of an emerging new movement of preventive policing that is sweeping through precinct houses across the country and even around the world. Cities from Los Angeles to Atlanta, Seattle to Santa Cruz, Calif., are trying to harness the power of math to curb various forms of crime, whether it’s robbery, gun violence, or drug dealing. Authorities in Kent, England, are using sophisticated computer models to guide police, too. 

But the CPD is pushing the science in new directions. Beat cops and veteran precinct captains in the nation’s second largest police force are teaming up with number crunchers and leading university crime researchers in their quest to better understand gun violence and engage with neighborhood residents.

Chicago’s initiative, though still nascent, has shown some success so far. Since last July, police have carried out 66 house calls similar to the one Peña made. In only two cases were gang members they visited later involved in shootings – which is progress for a city reeling from a national reputation for bloodshed. 

It’s also one reason the initiative here is now being looked at by a number of police departments in other cities, such as New Orleans and Toronto. While no one sees Big Data as the answer to crime – and the program isn’t without controversy – people here see it as a powerful new tool to combat urban violence.

“Police in major American cities have endured generations in which good and serious people came to work every day and felt they were not doing any good in solving the violence issue...,” says David Kennedy, director of the
 Center for Crime Prevention and Control at the John Jay College of Criminal Justice in New York, who is involved in the Chicago project.  “Now, the fact that you can do good and help people and possibly save lives is transformational.”

• • • 

Preventive policing moves cop work one more step from the era of batons to the era of bits and bytes. In the 1970s and ’80s, many cities dealt with a surging homicide problem by trying to sweep it off the streets. Police flooded besieged neighborhoods. They made large numbers of arrests. Courts sent the criminals to jail – often for substantial periods of time.

Then came the era of community policing, in which police were more rooted in neighborhoods, both protecting and interacting with residents. This approach was less reactive, adding an element of preventiveness to law enforcement. Now comes preventive policing, which combines enforcement, community engagement, and analytics. It’s what Garry McCarthy, Chicago’s police superintendent, calls “community policing almost on steroids.”

In Chicago, the shift to a more numbers-based approach began with the arrival of Mr. McCarthy, whom Mayor Rahm Emanuel appointed superintendent three years ago. A beefy “cop’s cop” as comfortable around numbers as he is doing roll call, McCarthy came at an inauspicious time: The city was dealing with the aftereffects of a police torture scandal that was continuing to siphon millions in settlement fees. The police union was upset at budget cuts and declining manpower, and Mr. Emanuel was intent on shedding the label “murder mayor,” the moniker critics gave him when the number of homicides spiked above 500 his first year in office. 

At the outset, McCarthy represented something different for Chicago: a superintendent from outside the CPD ranks with a background in crime statistics. He had overseen the highly touted CompStat program in New York City, which created a new way of managing crime numbers and holding precinct commanders accountable. Later, while police superintendent in Newark, N.J., he saw homicides fall 28 percent on his watch. Early in his new role in Chicago, McCarthy did something unheard of in this city where top officials operate with a tribal allegiance to each other: He apologized for decades of racial targeting by the police.

“What he said was, ‘I get it and I’m sorry and we’re going to try to do better.’ No one has said that,” says Andrew Papachristos, a Yale University sociology professor who grew up in Chicago. “McCarthy has moved the culture of the police into a place that keeps Chicago in the front, and not in the middle, of the pack.”

At the core of Chicago’s new approach is both collaboration between cops and academics and the numbers they gather. Just as the social network model is effective in helping create a portrait of Facebook users – their tastes, habits, and the people they interact with – the data analysis in Chicago is giving beat officers a more comprehensive understanding of the neighborhoods they police. 

McCarthy says incorporating the statistics into preventive policing will move Chicago away from targeting an entire community to focusing on individuals most likely to either cause, or fall victim to, gun violence. In an interview, the superintendent says that accountability is no longer just the domain of district commanders: Beat officers are now being trained to use data to figure out where violence might erupt  and how to stop it.

“I see what we are doing in Chicago as creating a model that is really going to a whole other level of intelligence-based policing,” he says. “It requires a new way of thinking.”

The marquee program is the Chicago Violence Reduction Strategy, a partnership funded by the MacArthur Foundation and led by John Jay College, with Yale University, the University of Chicago, and other academic institutions playing major support roles. Although currently used in only four police districts on the city’s South Side, the strategy is expected to spread.

Fashioning the program required understanding Chicago’s gang culture. Decades ago, gangs here were large criminal organizations that dominated expansive areas of turf and fought for control of the drug trade. Today, they are neighborhood factions that control only a few blocks. According to the Chicago Crime Commission, 70 to 100 gangs now inhabit the city, encompassing about 100,000 members. Many have different and changing reasons for existence, which makes it hard for the beat cop to understand, much less control, them. 

Enter Mr. Kennedy and the digiteers. His team helped the police create a database that monitored which gangs were active, which were quiet, which were fighting – and with whom. “You can’t even understand what is going on until you have that level of specificity,” he says.

To get that information, John Jay researchers interviewed key personnel across the city – mainly beat cops – about what they saw on the streets and built that into the current CompStat system. 

Many seasoned officers were reluctant to cooperate. They didn’t understand why they needed to be that granular about gang activity, and many were more apt to trust their instincts than a computer algorithm. “Getting [the value of this] across can be really hard, especially in a city like Chicago, where you have four generations serving in the same police district,” says Professor Papachristos of Yale. 

But he says views began to change when the John Jay team returned with the results: a systematic snapshot of gang factions that could yield predictions of where violence might break out next. 

• • • 

While some officers still don’t embrace the new approach, the results have been reasonably encouraging so far: Police data show shooting incidents – both fatal and nonfatal – dropped 24 percent in 2013 compared with the year before. Murders fell 18 percent, from 503 to 414. 

Shooting incidents through June 30 of this year ticked up 5.5 percent, though the number of murders dropped 5 percent compared with last year, according to the CPD. Still, crime remains stubborn: The city experienced 82 shootings – and 14 deaths – over the July 4th weekend alone.

The numbers overall, while not perfect, reflect a tactical shift in how police respond to shootings. In 2013, Papachristos and Yale colleague Christopher Wilderman calculated that, between 2006 and 2011, the homicide rate of “a high-crime African American community in Chicago” was 55.2 per 10,000 people. That number was about four times the citywide rate during the same period.

The old mind-set would suggest that this neighborhood was a dangerous place and that many people there were at risk of being either a victim or perpetrator of a crime. The traditional response might have been to dispatch a large number of police to the area, turning the neighborhood into something of a battle zone.

But Papachristos and Mr. Wilderman’s data showed that a very small percentage of the population was probably causing all the problems and a lighter, more targeted police footprint would be more effective. According to their research, arrest records show that 85 percent of all gunshot homicide victims had at least one previous arrest. Victims and perpetrators of violence also tend to know each other and operate within the same social networks. Using both arrest and homicide records, the researchers found that the population of the community that had been arrested during the five-year period consisted of 24,110 people, about 30 percent of the total population.

Drilling down further, they calculated that 41 percent of all gun homicide victims were located within the social network of arrestees. The percentage that those victims represented within the community as a whole was 3,718 people, or only 4 percent of the population.

“Crime is more concentrated within networks of people than actual places,” says Papachristos. “So all of a sudden you’re talking about 70 to 80 percent of shootings taking place in networks that represent just 3 to 5 percent of the population. Those at risk are a couple hundred people. You now have a sense of who they are.”

• • • 

Knowing the numbers is one thing. Getting people to not pull out their guns is another. Chicago police are taking several steps to try to head off violence before it happens, which may be the most difficult part of preventive policing.

They have expanded “call-ins,” group meetings with repeat offenders on probation or parole, during which they remind them of the risks of committing another crime. The city has stepped up foot and bicycle patrols in 20 high-risk zones – ones that represent just 3 percent of city land but account for 20 percent of gun violence. Authorities have also put into place a system to solicit feedback from crime victims about their experience with the police. 

“When you talk to the public, yes, they want crime reduced, but they are more concerned about fairness and respectfulness with their interaction with police,” says Dennis Rosenbaum, a criminologist at the University of Illinois at Chicago who has helped oversee the feedback program.

“When the public feels ... the police are being fair in their authority and aren’t judging them on the basis of race, gender, and ethnicity, they are much more likely to cooperate, much more likely to provide them with intelligence to solve crimes,” he adds.

The most recent and perhaps revolutionary development in the CPD’s outreach effort is the door-to-door campaign, such as the one officer Peña was a part of. 

Years ago, the scene would have been unimaginable: The doorbell rings at the home of a gang member who has been arrested multiple times. An officer says a district commander wants to chat with the man and his family. Can they come in?

Drawing on research by Papachristos and his team that focuses on victims and arrest records, the police have created a “hot list” of people to visit who may be most at risk of becoming either the next offender or victim. It is based on an analysis of individuals’ criminal histories, prison records, open court cases, and victims’ social networks. [Editor's note: The original story incorrectly identified who drew up the "hot list."]

Police present the visit as an information session and an opportunity for change. Accompanying the district commander are representatives from social agencies who offer to connect the family with health-care services, classes to earn a high school equivalency certificate, or job-training and placement programs.

The officer also conveys something more sober: what the legal consequences would be if a person with that kind of record – and the police detail what the individual’s record is – were to commit another crime.

“We just lay it out like that with them, and close by saying the people of your neighborhood do not want violence – put down the gun,” says Cmdr. John Kenney, executive director of the CPD’s Bureau of Organizational Development, which runs the custom notification program.

The visitors make sure the others around the kitchen table hear about the individual’s rap sheet, too, and what the consequences of more criminal activity would be for him and the family. 

“It’s terrifying to the families,” says Chris Mallette, executive director of the Chicago Violence Reduction Strategy program at John Jay College. “We can show them, with almost certainty, the probability of their son or brother or grandson being the next gunshot victim based on their social network.” 

• • • 

Not everyone is enamored of preventive policing, however. Some doubt it will ever have a meaningful effect on lowering gun violence. Others worry that it is just a distraction from what is ultimately needed to improve some inner-city neighborhoods.

“Much like any other program in Chicago, they are not dealing with the crux of the problem, which is poverty,” says Tracy Siska, executive director of the Chicago Justice Project, a nonprofit group. Police may offer counseling and social services during interventions, but Mr. Siska says those are hollow choices.

“You are not getting a guy out of a gang until you bring him a job,” he says. “I think [house calls] are a decent idea, and in the short term, they may have an impact. But it’ll only become long term if you can get those people out of those lifestyles.”

Johnny Outlaw believes they need jobs, too, but thinks the new police initiative is a definite step in the right direction. He leads an effort to find work for repeat offenders and provide them with legal services. In private, many gang members tell him that they want to put down their guns but either are afraid of retaliation or don’t know how to go straight. Mr. Outlaw says that he’s had “top-flight gangbangers” break down sobbing in his office – “physical tears out of these cats” – because they feel they’re out of options.

“They say, ‘I’m tired of shooting people, I’m tired of robbing people. I want to do something with my life.’ I say, ‘If you work with me, we’ll find you a job. But you need to put down that gun,’ ” he says. 

They show up at Outlaw’s matchbook-sized office at Teamwork Englewood, a nonprofit that partners with the CPD, on the campus of Kennedy-King College. Outlaw has gone out with Chicago police on some of their house visits, wearing a bulletproof vest over one of his natty suits. He has transitioned gang members into jobs ranging from manual labor at warehouses to driving trucks cross-country.

Others fault Big Data policing for putting a new face on an old problem: profiling. Critics argue that a person’s past shouldn’t define what he will do in the future – that a person shouldn’t be stigmatized or singled out for his or her rap sheet. 

But police here refute those arguments. “Profiling is stopping someone because of their race or color or creed,” says McCarthy. “We’re doing an empirical analysis using a scientific formula, which is quite the opposite of profiling.” 

In fact, he says, the Big Data approach shows that the vast majority of people in troubled neighborhoods are not doing anything nefarious, prompting police to view the communities as less of a threat. 

“What it should be teaching us is everybody in Chicago is not a criminal,” says McCarthy. “If 5 percent of the community is responsible for violent crime, then 95 percent of the individuals are good people, and that provides a larger understanding of the community, which often goes unsaid.”

In the end, the effectiveness of the new wave of policing will likely hinge on one thing: whether it reduces crime. While some of the initial results are promising, many say it’s too early to determine how much – or even whether – it will reduce violence. But what is certain is that Chicago will provide one of the nation’s premier tests of how well the new science of policing works.

Already, it’s got the department thinking more creatively about preventive law enforcement. It’s changing police perceptions of neighborhoods. McCarthy, for one, has no doubts about what the denouement will be. 

“I think this is what every police department in the country needs,” he says. “This can be used anywhere and can really move the scale.” ρ