Connections Among People: Tracking and Preventing Violence through Social Network Analysis

Sociologist Andrew Papachristos focuses his studies on urban neighborhoods, social networks, street gangs, violent
crime, and gun violence. A Robert Wood Johnson Health and Society Scholar at Harvard University, Papachristos discusses
how social network analysis can aid crime prevention.

Hi, I’m Sarah Schweig at the Center for Court Innovation, and today I’m speaking with Andrew Papachristos. We
are currently in an executive session in Washington, D.C. sponsored by the California Endowment, Community Oriented
Policing Services, and the Center for Court Innovation that brought together public health experts and law enforcement
representatives for a conversation about public health approaches to public safety. Andrew Papachristos is a sociologist
who focuses his studies on urban neighborhoods, social networks, street gangs, violent crime, and gun violence. As
a Robert Wood Johnson Health and Society Scholar, Andrew will expand his use of network analysis to the study of
crime epidemics in U.S. cities, paying particular attention to the way violence diffuses among populations of youth.
Thank you for speaking with me today.

It’s my pleasure.

SCHWEIG:  So let’s start out by talking
about crime epidemics. How does treating violence like a contagious disease affect approaches to crime in law enforcement,
and how does it help us understand, in general, how crime occurs?

So I think when the phrase ‘crime epidemics’ is typically brought up, it’s brought up as sort of like a moral panic
to sort of motivate people into action, and usually the action is political or law enforcement. I think we mentioned
in our talk today one of the actions when you have a crime epidemic is to call out the National Guard, quite literally.
I think what both I’m trying to do, as well as people in this room are trying to do, is understand what the idea
of an epidemic means in public health—so we don’t mobilize the National Guard when we have an outbreak of influenza
or an STD—and in fact, taking some of the science that’s used in contagious disease studies and apply them to violence,
as in who is most susceptible to getting shot. And to use more precise models, especially around social networks
to really understand if crime is a disease, then how does it spread? What is its form? What is the way that it is
transmitted? So to actually take this sort of analogy, which is often political, and actually take it at its sort
of real value, and understand what these mechanisms of contagion might be. 

 The public probably thinks of social networks as Facebook and stuff like that. Can you talk a bit about
what you mean when you say, ‘using social network analysis,’ and also the kinds of relationships are you looking
at—and why are you looking at those relationships?

So social network analysis is a science that’s been around for 60+ years. It really started in anthropology and sociology
where people were interested to understand the connections among groups and individuals, and how that affects action,
political action, social action, health and behaviors, and the difference between that and sort of regular science
is—rather than just understand individuals in isolation—network analysis really tries to understand how the connections
among people affect their behavior. So how your friend’s choice of political candidates affects your choice, or whether
it’s eating habits or social influence, or how you get a job.  And so the idea of understanding connections
and how that affects your behavior.  When you talk about the connections that you have, like Facebook, who
are your friends, it is actually—the science behind it is much like social network analysis, especially the mathematics. 
What you are interested in, who are your friends, who are your family, who did you go to school with, who were you
in clubs with? As well as sort of—did you buy the same book as other people? So Amazon uses that type of analysis
to understand people who bought this, they also like this, based on what other people like you have done.

In the case of today’s discussion about crime, we’re trying to understand how certain types of risky relationships—so
sort of offending relationships or gang relationships affect some of these outcomes like mortality: who gets shot
and killed. And like other types of behavior like intravenous drug use or sex behavior, you want to understand the
relationships that matter. So who you share needles with, who you’re sleeping with, in much the same way we’re trying
to understand who you hang out with and how that affects all these sorts of bad outcomes.

Right. And you mentioned an offending population and during the talk today you talked about co-offending populations—how
does mass incarceration, that our country is dealing with right now, tend to affect the offending networks?

PAPACHRISTOS:  So it’s a great question. I haven’t looked at it sort
of in this analysis yet. But what it does do, we know—from the evidence as well as what I would hypothesize—is it
disrupts these positive networks—family, school, romantic relationships, jobs—and then essentially filters you into
a different set of networks when you come out. So the capacity to get a job being an incarcerated felon, the capacity
to maintain relationships with your children, and that then getting passed down to them, intergenerational transition
of things like poverty, really incarceration is having a long effect on these networks both now and in the future.
And hopefully it’s something we can get at empirically. 

That’s really interesting. And sort of in the vein of where you want this to go in the future, where do you see this
research moving and how might it affect law enforcement approaches on the ground? And also maybe talk a little bit
about what jurisdictions would need to best implement these ideas in the most useful ways.

Yeah, I mean my guess is that in 5 to 10 years—probably 10—sort of network mapping for police will be where sort
of hotspot mapping is now. I think that intuitively people understand more now—because of things like Facebook and
Amazon and the Internet—how connections matter. I think we’ve always known, you know, it’s not what you know but
who you know, the good old boys club, like all of those sort of images and sort of analogies now have a basis in
scientific analysis. So I think that it’s only gonna go up. I think there’s a critical mass of people beginning to
sort of galvanize around it. I think the difference is it’s not quite as intuitively simple as geographic map. 
So you have to actually get people past a certain learning curve. I think that there are all kinds of tools for both
enforcement that can be developed technology-wise, as well as for other public health and violence prevention initiatives.
Again, so this has been done on smaller scales with HIV and needle-exchange programs, sex work, and hopefully the
same models can be applied, but it can’t be cut and paste. It has to be specific to the epidemic you’re examining,
right?  So crime is like, but not the same as, a needle exchange, right? So understanding the context is
essential. And that’s sort of the caution to moving forward.

And you mentioned also moving some of these initiatives during your talk toward Milwaukee and San Francisco, some
of that analysis. Do you want to talk a little bit about that, just sort of as a last point?

Yeah, so I think that what I’m seeing these last couple years is more and more jurisdictions that are eager to try
something that’s much more directed, especially given the limited resources. But not just directed, but is directed
and can have a huge sort of effect for a small sort of intervention. I just received the NSF award to expand the
work in Chicago and Boston, to Cincinnati and potentially San Francisco, but even already cities like Atlanta, Milwaukee,
Minneapolis, Brooklyn Park, there are a bunch of cities that are already eager to do it and the capacity is just
not there. The big barrier is that it’s not a point and click approach just yet, right? So with crime mapping, you
put in a bunch of addresses you get your picture. There’s a—there’s quite a bit more backend to this type of technology
and we’re not at that stage yet, where people with basic computer skills can do it. So it requires an analyst. And
so as we train more analysts, it will become more pervasive, I think.

Great. Thank you so much for speaking with me today.

My pleasure.

SCHWEIG:  I’m Sarah Schweig and I’ve been speaking
with Andrew Papachristos. To learn more about the Center for Court Innovation, please visit 
Thank you for listening.

January 2012