Using Social Networks To Track And Predict Gun Violence

A new study shows that lethal gun violence in cities is socially contagious, but that you can also track it through public records.

Using Social Networks To Track And Predict Gun Violence
[Image: Gun via Shutterstock]

On a hot, August night in a community gym space on the South Side of Chicago, roughly 40 young men, or “group members” as they were called, gathered on bleachers in front of a screen. “Does anyone know why you’re here?” asked Chris Mallette, executive director of Chicago’s Group Violence Reduction Strategy. Parole agents, relatives, or maybe pastors had told the men that they needed to show up, but not much more. “You’re here because you’ve been identified as a member of a violent street organization,” Mallette continued.


“Then there’s a minor disturbance if you will,” Mallette tells me a few months later, recounting the strange pattern that he says always follows the initial allegation. Some in the group get upset, asking Mallete where he got his data. Others accuse him of profiling. But Mallete isn’t there to have police make arrests. He, community members, and some law enforcement partners are there to warn the men about the importance of changing their behavior, and offer various resources for support. But the attitude in the room changes when a slide pops up on the screen, a map showing how each of the individuals in the room is precisely connected to one another through a common social network of arrests and other public records.

“And at that point there’s a collective sigh and a collective slump,” Mallete explains. “And never again is there ‘conversation.'”

The scene Mallette describes is something called a call-in, one feature in a larger approach to reducing gun homicide that’s increasingly taking hold in American cities. Instead of broad-based profiling, Mallete and his colleagues’ methods rely on using community resources to influence tiny social networks, the kind demonstrated in the power point, which new research shows are responsible for the sweeping majority of gun deaths in urban areas.

David Kennedy, director of the Center for Crime Prevention and Control at John Jay College of Criminal Justice, first started pinpointing much of cities’ lethal gun violence to small networks in Boston nearly 20 years ago. But new research with some of Kennedy’s outside partners, along with the Chicago police department and a MacArthur Foundation grant, shows that gun violence in major cities may work the spread of disease. It shows that carefully treating individuals in social networks, not broad-based racial profiling, could reduce gun violence in an unprecedented way.

“When you take a broad categorical distinction like race or gender, or wearing baggy pants or a baseball cap, most people with those risk factors never shoot anybody,” explains Andrew Papachristos, a Yale University sociology professor and author of a new study published in the American Journal of Public Health that looks at Chicago’s gun violence epidemic. “In the case of gun homicide, the case is most young black men never shoot anybody, and most young black men never get shot.”

Papachristos’s study came to two major conclusions after analyzing gun homicide data in Chicago between 2006 and 2011. The first was that violence was actually hyper-concentrated in small networks of people connected by arrest records–and that any given homicide victim, on average, was 5.4 social ties away from an individual in that social network. Cut out one degree of separation, and risk of death decreased by 57%. By this finding, Papachristos’s work also suggested that gun violence was socially contagious, and functioned similarly to the spread of HIV among needle-sharing networks. “Your friend’s behavior and your friend’s friend’s behavior can have a real impact on your own risk. People that look like random victims could be in the wrong place at the wrong time because of a friend,” Papachristos says.


Take the extreme example of Jonylah Watkins, the six-month-old that was shot while her father was changing her diaper in a parked minivan in Chicago. When Papachristos pulled up Watkins’s father’s network, he found that two-thirds of Watkins’s father’s friends had been shooting victims. “Here’s a baby who was shot and killed, not because of anything she did, but because her father was part of her social network,” Papachristos says.

By targeting networks of people connected by robberies and certain drug trafficking arrests, Kennedy, in conjunction with local law enforcement, saw gang homicides in New Orleans decrease by 22% within the first six months of 2013 compared to the first six months of 2012. But the Violence Reduction Strategy just started work in New Orleans this year. Kennedy claims that the lowest level of impact they’ve seen using this strategy over a longer period of time is a decrease in lethal gun violence by a third.

“From the very beginning, we worked to identify those groups, and who was allied with whom,” Kennedy says. “Both the group level and individual analyses will produce very small groups relative to the rest of the population. It’s on the level of a fraction of a percent in the most violent neighborhoods.”

The Violence Reduction Strategy’s approach marks a stark contrast to the claims made by New York City Police Commissioner Raymond Kelly on the efficacy of the NYPD’s extensive stop and frisks. In August, a federal judge ruled that the NYPD’s practice amounted to unconstitutional racial discrimination. “If you only use race as the way to combat this, you’re casting your net so broadly, and it had a really harmful effect to how people view people police,” says Papachristos. “The network approach is the opposite.”

Papachristos is currently looking into applying the social network approach in New Haven, Hartford, and Bridgeport, Connecticut. More cities, like Palo Alto, Cincinatti, South Bend, Indiana, and Peoria, are also experimenting with this kind of analysis.

Still, Mallette, Papachristos, and Kennedy acknowledge that these strategies are no silver bullet, and keeping them effective gets more challenging over time. They’re also only looking at a small slice of a particular type of violence, and not necessarily the underlying causes–like poverty, lack of education, or other systemic injustices.


Interventions aren’t necessarily straightforward either. The tone and assembly of Mallete’s interventions play a critical role–he looks at everything from the placement of the food to the lineup of chairs and speakers in those settings as meticulously a forensic expert, and is careful to balance offerings of community support with warnings of real legal repercussions.

Papachristos also clarifies that he’s only looking at arrest records, not social media, or even school records. And he’s certainly not trying to build a predictive algorithm to try and target people, Minority Report-style.

“My dream would be to make this a system where a shooting happens, you can look at the networks, and flood it with resources,” Papachristos says. “That’s not just arresting your way out of a problem.”

About the author

Sydney Brownstone is a Seattle-based former staff writer at Co.Exist. She lives in a Brooklyn apartment with windows that don’t quite open, and covers environment, health, and data.