Can We Predict Urban Gun Homicides?
According to CDC data, more than 13,000 people die from gun homicides every year—and most of them are people of color who live in urban areas. Many of them are children.
Public health researchers often compare gun deaths to other public health problems like HIV, cigarette smoking, and opioid overdoses, and they study firearms with similar approaches. But as scientists seek to understand the causes and solutions for gun deaths, can we also learn to predict them…and even intervene before they happen?
[What we do—and mostly don’t—know about guns.]
Desmond Patton, a social work researcher at Columbia University, is studying how social media conversations among Chicago youth who are grieving losses in their communities seem to predict retaliatory gun violence. And Andrew Papachristos, a sociologist at Northwestern University, says knowing gunshot victims can reliably increase your risk of becoming one. They explain the patterns in the data, and what implications their work may have for potential interventions.
Desmond Patton is an assistant professor of Social Work and the Director of the SAFE Lab at Columbia University in New York, New York.
Andrew Papachristos is a professor of Sociology at Northwestern University in Evanston, Illinois.
IRA FLATOW: This is Science Friday. I’m Ira Flatow. Last week, we explored some of the gaps in gun research as teenagers around the nation gravitated to Washington for the March For Our Lives, but as one of our guests pointed out last week, while more than 30,000 people die per year from gun injuries, that story plays out differently depending on whom you are and where you live. More than 80% of white people who die from gun injuries are suicide victims. More than 80% of black people who die from gun injuries are the victims of homicides.
This week, research that focuses on the homicides– and that’s what we’re going to be talking about. Not just mass shootings like the one in Parkland, Florida that led to this recent wave of youth activism, but homicide in urban centers like Chicago and Baltimore and Philadelphia and beyond, where more than 80% of gun homicides occur. If gun violence is a public health issue, is there a public health intervention, like a vaccine, for homicide? Or, at the very least, can we predict it using social media. When and where the next homicide will be.
One of my next guests has research out this week in the Journal of Digital Medicine that says that, yes, social media could point the way. When young people in Chicago who are affiliated with gangs lose a loved one to gun violence, their first tweets are about grief, but that grief turns to aggression and further violence within two days. Study author Desmond Patton uses machine learning algorithms to detect and categorize these conversations, and other research is demonstrating the degree to which homicide is contagious. If you know a gunshot wound victim, you have a higher chance of becoming one yourself. Gun violence is contagious like other public health problems. That’s how this theory goes. Here to explain their work and what may come of it are my guests Desmond Patton, Assistant Professor in the School of Social Work at Columbia University, author of the social media study. Welcome to Science Friday.
DESMOND PATTON: Thank you so much for having me.
IRA FLATOW: Andrew Papachristos is Professor of Sociology at Northwestern University who studies how networks can transmit gun violence. Welcome to Science Friday.
ANDREW PAPACHRISTOS: Thanks, glad to be here.
IRA FLATOW: You’re welcome. Desmond, as I just said, gun violence in the urban areas kills more than 13,000 people per year, about a third of the total number of killed by guns in a year. What drew each of you to try to bring the data to solving this problem? Desmond, let’s ask you first.
DESMOND PATTON: Sure, so I did my PhD at the University of Chicago, and I spent about a year with young black men who were high-achieving. They were doing really well in school but I wanted to understand how they were staying connected to school while also navigating violence in their community, and what became very clear doing that ethnographic study is that they were cognitively geocoding their neighborhood, trying to make sense of all of the people and things and spaces and places that they had to navigate and understand very well in order to stay safe, and this also becomes apparatus, if you will, that they use when they’re engaging in their digital life as well. So they need to remain connected to the community, and so they also are mapping individuals and mapping conversations to stay safe as well.
IRA FLATOW: So you think that the social media is a fertile ground to be studied about gun violence?
DESMOND PATTON: So I look at social media as an environmental context so that we can understand the pathways to violence, and in my research I’ve learned that young people are initially using social media to communicate with one another, to share the news, and to talk about the throes of their everyday life, and when that experience is wrapped into an ecology of violence and hyper-exposure to violence, the conversation that they talk about also include their exposure to violence.
IRA FLATOW: And it sends out flags that you might follow?
DESMOND PATTON: Absolutely, but I want to be clear that I think it’s also important to not just focus on threats and aggressive communication on social media because that may lead you down a trope, as well. I think what I’ve learned is that there is a pathway, and that oftentimes expressions of grief and trauma and pain and experiences of loved ones are the things that people initially come to social media to discuss.
IRA FLATOW: Yeah because sometimes, as you point out about– I’ve seen your Ted Talk on this– is that we do, sometimes we focus too much on just the violent aspect of it.
DESMOND PATTON: Absolutely.
IRA FLATOW: Andrew Papchristos, what brought you to this line of work?
ANDREW PAPACHRISTOS: So I think, much like Desmond, I was born and raised and educated in Chicago in the ’80s and ’90s, and you couldn’t escape the dialogue around gun violence and gun homicide in particular, and as my sort of work developed, I kept coming back to, what is it that’s driving violence? What’s keeping it at high levels in some neighborhoods rather than other neighborhoods? And more importantly, what can be done by it? And, as you suggest at the beginning, one of the things that gun homicide does in this country is it drives massive disparities in health, mental health, and just mortality outcomes, especially among black and white young men, and so you could be in one part of Chicago, and if you are white your chance that– you’ll live seven or eight or 10 years longer, and we know this sort of at birth, and so the idea, as I started getting into the research in this area, was really trying to unpack ways to understand this epidemic beyond the sort of political framing of gun violence epidemic because it is an epidemic. And if it’s an epidemic, it should follow certain rules. It should be transmitted in certain ways. It should spread in certain ways. It should be concentrated in certain ways, and so the work was really trying to sort of leverage that sort of information to figure out and think about ways we can stop these sort of cascades of gun violence.
IRA FLATOW: So you can– it is an epidemic, as in diseases– you can put a number about how it’s spreading. Can you do that also for gun violence?
ANDREW PAPACHRISTOS: I think that that’s what my research really is focusing on. So one of the things that the sort of network-based approach uses– and I’m talking about sort of human social networks. Not just social media networks, but who you’re related to, who you engage in certain behaviors with. One of the things that shows is that gun violence victimization is even more socially-concentrated than just spatially-concentrated, meaning that most gun violence acts are happening in very small networks, even within, say, high-crime communities, and the science is suggesting that, in fact, it is transmitted, which is if I get shot, if you’re one of my associates, your likelihood of getting shot just skyrockets within the time period around that shooting, and then subsequently your friend’s friend and your friend’s friend friend, and it literally builds out like a cascade over time.
IRA FLATOW: But is there an intervention someplace? You know, with a disease you try to quarantine people or you try to stop it with a vaccine or treatment. Is there an intervention period in your disease that you’re studying?
ANDREW PAPACHRISTOS: So I think there are lots of points of intervention. Of course, the best one is the biggest one, which is growing up in a community that is safe and healthy you know, full-stop. That’s one of the biggest, sort of, vaccinations to use your phrase, but there are sort of emergency responses, which is, can you stop these cascades? And, while lots of interventions are not using these sorts of analytics formally, good preachers, teachers, cops, social workers, educators– they live in these networks, and what we see is interventions, whether they’re public health or otherwise, that can focus efforts on sort of small populations, small geographic areas, or sort of insert themselves in these networks with, say, trauma care or violence interruption. You can see fairly significant short-term reductions in gun violence, which is they can slow things down, save a life, but what they don’t do is they don’t do the big things. They don’t fix communities or schools, and you have to kind of do both at the same time. It’s not one or the other.
IRA FLATOW: Desmond, your work looks at how, as you say, youth in Chicago talk to each other on Twitter and teaching machine learning to follow– how hard is it to teach a machine?
DESMOND PATTON: It is extremely hard, and so I’ve had the good pleasure of working with the Data Science Institute at Columbia University and I partnered with Kathy McEwen, who is one of the leading data scientists in our country, and we have thrown the gold standard data science techniques at this problem, and the issue is it doesn’t understand the complexity, the nuance and culture that’s embedded in the [? tone ?] communication that I look at. And so to circumvent that, we had to develop our own ontologies and, more importantly, involve community in the translation and interpretation of social media communication.
So that means we hire young people from Chicago to translate and interpret the data and to also validate our own judgments or tagging of the data, if you will.
IRA FLATOW: How good is Twitter at actually understanding the personalities involved?
DESMOND PATTON: I think what’s important to know about Twitter is that people bring the understanding of their world and their experiences to Twitter, and it’s really coming upon us to try to understand it the best way we can, and so what we do is we go through a set of procedures to really extract context in our analysis of Twitter data, and then we talk to community members to validate our interpretations so as to not misinterpret a rap lyric for a threat, if you will.
IRA FLATOW: What kind of would be a typical tweet that you would be looking at?
DESMOND PATTON: So we look at the gamut of tweets, and so we look at communication about love and happiness and joy, and we look at a lot of pain and trauma as well, and what we come across a lot are rap lyrics and threats and taunts, but also pain and responses to trauma as well
IRA FLATOW: And Andrew, what happens once we can predict an individual person’s risk? What do we actually, what can we do about it?
ANDREW PAPACHRISTOS: Well I think one of the things that’s important as you try to understand people’s exposure to risk and their level of risk is, in fact, we have lots of things that work already around trauma reduction, around violence reduction, around promoting mobility. So part of, I think, one of the benefits of these sorts of approaches is getting the right resources to the right people in a timely fashion, which also requires a lot of political maneuvering, which is getting people to share information, share responsibility, and being able to mobilize them and also, to Desmond’s point, to be able to have some human intelligence and human thought going behind something and not just looking at a printout of numbers or scores, but be able to say, hey, that’s a rap lyric. That’s not, you know, that’s not something Andrew said, or that was something that happened 20 years ago, not something that happened 20 minutes ago. So it’s really important to just move beyond sort of individual numbers and figures, but have it in that sort of iterative human context as well.
IRA FLATOW: And Desmond, is there a role for law enforcement in any interventions to prevent gun violence or does the science suggest otherwise? Well, I think that law enforcement has already been involved, particularly in the use of social media as evidence for understanding future criminal behavior. I think part of the challenge is, how do we develop more contextually-driven techniques to make sure that when we’re looking at text and we’re looking at images, that we are embedded in an understanding that is local and that is as accurate as possible. So I think that it’s really important to engage in interdisciplinary collaborations to push each other to make sure that we’re not over-policing these communities.
IRA FLATOW: Right, that brings me to this question. Last week, we focused on some of the difficulties public health researchers have in getting data to study gun violence. To that end, what’s on your data “wishlist”? What would you study? If I give you a blank check and you could study, what do you need to know?
DESMOND PATTON: I mean, I would love to be able to look at all social media platforms. Some of the– when I talk to youth about the platforms that they are engaged in, one of them is Snapchat, but it’s really hard to get data from Snapchat because–
IRA FLATOW: Because?
DESMOND PATTON: –because, well, A, the videos have been very quick and so it’s very hard to capture those videos, and we found a real rigorous way, and so I think trying to tap into the variety of platforms will be very helpful in triangulating our approach to analysis.
IRA FLATOW: Andrew, what would you like to have?
ANDREW PAPACHRISTOS: Well on the data question I could just send you a bunch of unfunded proposals–
–we could talk about, but–
IRA FLATOW: Don’t send it to me.
ANDREW PAPACHRISTOS: So, I mean, I think the first barrier, the first thing I’d love to do would be to link data sets across different systems. So, by default, I think criminal justice data is often the best kept data, but to be able to link individuals across, say, educational, public health, and health care and criminal justice systems– that would be the first step, and some cities like Chicago have great data infrastructures and others just don’t. The second thing I would add, which is part of the current public dialogue, is of course I would love to have information on the guns used in crimes, which most often are illegally-funneled guns. They are not, sort of, the guns that are used legally in Chicago, which is always on this national stage here. You know, one of the second sources of guns is of course Indiana, where you can literally walk across the street and be in a different state and under different laws. So the ability to look and see how illegal firearms move within networks and markets, that would really go a long way for intervention and prevention processes.
IRA FLATOW: That’s what we were talking about last week. This is Science Friday from PRI, Public Radio International. Talking about gun violence with Desmond Patton, Andrew Papachristos. Despite the students from Parkland, Florida, the ones organizing the Never Again movement making a specific point about urban homicides, in some of their interviews it feels like we’re not hearing as much about this in the conversation. We’re not hearing a lot about urban gun violence in this conversation coming from teenagers. I mean, I think what is so powerful about these kids from Parkland is that they are speaking out and they’re being listened to. We don’t– certainly, as much as we, you know, care about these mass shootings, is an urban homicide from these kids is not getting much attention. Do we, is there a way for these kids to see the power that the Parkland kids are getting and them speaking out more?
DESMOND PATTON: Well I think what’s interesting about this whole scenario is that the Parkland students have been engaged in conversations with youth from the South Side of Chicago, and a part of the challenge is we recognize that mass shootings are a societal crisis but we’re not– when we talk about violence that happens in poor communities of our own youth of color, we’re not equating the same conversations or resources behind it, and I think people get really–
IRA FLATOW: They’re not getting the same stage.
DESMOND PATTON: Absolutely not.
IRA FLATOW: They’re not getting the same megaphone.
DESMOND PATTON: Right. It’s a very important issue
IRA FLATOW: Yeah would you agree, Andrew? What to do about that?
ANDREW PAPACHRISTOS: So I would give a whole-hearted amen to that, and I actually think, for those of us that have been doing this research for a long time, you know, we had the same reaction after Sandy Hook, which is you have literally dead white suburban school kids that started a movement and yet we’ve made very little progress since then. And so I think making sure that both of these things are happening and on the national dialogue at the same time is essential, and like Desmond, I actually want to applaud the student organizers who have gone far to include those voices in many of the marches.
I know here in Connecticut, we went to the Hartford march and they shared the stage, which is great, and I think that everyone should make a conscious effort to keep all these sorts of gun violence, including suicide, sort of on the national stage. And to go to your research point, by the way, there are different types of guns. There are different types of populations, and they’re all tragic and many of them are actually, in fact, preventable.
IRA FLATOW: Do you think that people are listening any more now than before? Desmond, you’re sort of looking at me like I’m crazy.
DESMOND PATTON: You know, I attended the march here in New York City and I was really pleased to see so many people and so many different types of people. I was also wondering, well, where have they been? And I also noticed a lot of people very happy and showing a lot of posters and being very excited, and I felt very solemn in the moment because I know too many people that have been shot. This isn’t something that’s very abstract for me.
IRA FLATOW: This is deadly serious.
DESMOND PATTON: This is deadly serious, and so I think people are listening. The impact of that listening is, I think we have some time to see what happens from that.
ANDREW PAPACHRISTOS: I would just add to it to build on that momentum. We know we’re going to fail if we don’t make any changes, right. For either of these sorts of types of violence we’re talking about, we need to really keep it going and lead to change, which actually– it happened in some states after Sandy Hook and some of the other shootings. We’re at a moment, but what’s going to happen? And I think it’s up to a lot of us to keep the dialogue going and make a lot of it at least data-informed.
IRA FLATOW: I’m glad we had some part in that dialogue. Desmond Patton, the Assistant Professor of the School of Social Work at Columbia. Andrew Papachristos, Professor of Sociology at Northwestern University. Thank you both for taking time to be with us today.
DESMOND PATTON: Thank you for having us.
ANDREW PAPACHRISTOS: It’s been a pleasure.
Christie Taylor was a producer for Science Friday. Her days involved diligent research, too many phone calls for an introvert, and asking scientists if they have any audio of that narwhal heartbeat.