Q&A With Dr. Desmond Patton

Dr. Desmond Patton tells us about how he uses social media to study inner city violence and what sparked his interest in the work.

By Adam Piore, Columbia University News

At the University of Chicago, where he earned his Ph.D., Patton did research at an inner city school to examine how young African American men coped with the transition from eighth grade to high school. That is a time when many students tend to have trouble, but Patton noticed something else.

“What I saw in the two years I followed them was violence,” Patton said. “Every single one of them had some experience with violence, and not just one experience. They would be in class one day, best friend beside them, the next day the best friend was dead.”

Several years later, when it came time for him to pick a dissertation topic, he decided to look at what influences education beyond grades and test scores. He followed 20 high-achieving young men at a Chicago charter school to see how they navigated neighborhood violence and succeeded in school. “One of the chief things that I learned was that they were able to cognitively geocode their neighborhood and school,” he said.

Geocode? “They knew who to talk to, how to talk to them, what clothes to wear, which streets were violent. It was as if they were walking around with a GPS in their head, and they were able to use those same strategies in school to identify mentors and helpful individuals to hang out with,” he said.

Later, a few of the young men he followed tipped him off to a Twitter war between two Chicago rappers that ended in death.

“That was the jumping off point for me,” he said. Now, he studies social media: “What we’re trying to do is take a deep look at the conflicts, culture and language on social media, and then develop algorithms that can then detect those instances of aggression or threats or grief online.”

What can social media posts tell you about inner city violence?

We all know, from research and anecdotal evidence, how many hours and hours most youths spend on the internet, sharing personal information on social media platforms about who they talk to, where they live, what they’re doing. In neighborhoods like the West Side of Chicago that I’ve studied, young people are living in violent contexts, where trauma and stress are everyday parts of their lives, so that’s what they talk about. They’re not talking online about violence only, but what they do discuss is real world and it has profound consequences for their offline interactions

So what do you look for in these communications?

I am interested in understanding the context around language and online postings that trigger offline violence. Not just the 140 characters indicating aggression or grief, but sifting through many posts, images and videos to understand why are they making these postings in the first place? Does kill mean the same thing if you put it in a different context?

What is most important in these posts?

It is hard to determine online threats, because there are so many factors to be considered. Context and who is sending these social media messages are important. Is a certain tweet from a gang member, or someone known to be violent? There’s a big difference between someone who’s a big talker and someone who is actually willing to kill, but that’s sometimes hard to distinguish. I also look for emoji use, such as a gun or frowns compounded with threats, which lead to different interpretations. Speed of response, and how someone responded to a social media post are all something to be considered.

And what can you do with that information?

Our ultimate goal is to be able to provide these data points to violence-prevention organizations across cities that struggle with gun violence. We could give them social media data in real time so they assess and potentially intervene.

Does this work?

This past summer, we used a case-study method to predict with a high level of accuracy aggression and grief in social media data from known Chicago gang members in Chicago. This hasn’t really been done before. Most people who do big data work give you raw numbers. Our data analysis approach looks at tone, language, context, the author of the tweet, any events that precipitated it and then any aggressive posts that then lead to an issue offline.

What are some examples?

For example, someone may put up a post and say, “Oh, I miss Johnny. He was my best friend. RIP.” Then someone from an opposing group, or what young people call the opposition or the ops, may come and say, “Oh, well f--k Johnny, he’s nobody.” That simple kind of jive or poke then becomes the initial trigger for a host of communications from the opposing group that leads to a potentially violent situation offline. This is very different from the old days, when people would have arguments and battles in person; now the beef has already occurred and it can go straight to violence. Can we predict, correlating a cluster of aggressive communication online and cross-referencing Chicago crime records, what might happen? If we had millions of posts, which we are planning to do in a relationship with existing violence-prevention groups, perhaps we could, and there could be an intervention.

What are you working on now?

This is my second year here, and as soon as I arrived I started collaborating with data scientists. I’m currently working with Kathy McKeown, director of the Data Science Institute and Owen Rambow, a computational scientist there. We are applying for a grant from the National Institute of Child Health and Human Development, which would allow us to do research on a far-larger scale, and use a qualitative approach to fill in what’s been missing in a lot of the research. In that way, we could leverage the skills of young people from the community to interpret what’s happening on social media. For example, there was a situation that, according to social media, had occurred on Lamron Street in Chicago. There is no such street, but when I asked a young person he said, “Oh, that’s Normal Street,” a block where a certain gang faction hangs out.

What else are you doing?

Earlier this year, I published a paper with several co-authors that was a case study of a South Side gang member’s Twitter communications. It involved something called “internet banging,” which is when urban youth use social media to brag, insult and threaten their rivals, disseminating violence online that results in serious injuries and homicides. I also received a grant for work I’m doing with Jeffrey Lane, an urban ethnographer at Rutgers University, on social media policing. We’re looking at what kind of comments online might get someone arrested. What might be a song or rap lyric to a kid might be considered a threat to a cop. These have all sorts of implications for law, social media and other factors.

How did you get started in this research?

It seems so long ago now, but for my undergraduate honors thesis I studied a support group at UNC-Greensboro for high-achieving black males. As freshmen, they were recruited to join this group to help them kind of navigate their four years, and I was the anthropologist for that group. So I went to their meetings and interviewed them to understand what mechanisms helped them navigate and be successful.

What did you find that worked?

I think what always works is support from people who have had similar experiences, people who paved the way who can come back and give you some pointers and tips. And being able to have kind of a dual sense of self, being able to talk the talk and walk the walk in multiple settings and contexts. I think young people who don’t have that ability to kind of turn it on and off in these spaces probably have more challenges than those who would easily find themselves in different contexts.

How did you happen to concentrate on this group?

I was particularly interested in high-achieving young men because there is so much focus on young men who don’t do well. I knew there was a group of young folks who were doing well but living in the same violent contexts.

How do you think they managed that?

I think some of it is by chance. They had strong parents. Some came from two-parent families, most came from single-parent homes. But they all went to charter schools, which automatically said they had a champion somewhere, someone who decided these young men should not be in the neighborhood school—a mother, a father, a cousin, an aunt, someone. So they had a lot of support, and support is a big deal. They’re also just really smart guys. We focus a lot on what people look like in terms of grades and test scores, but there is a brilliance that allows people to navigate the space they find themselves in. That is an important and salient tool for young people who are growing up in tough contexts.