DAVID GREENE, HOST:
OK. Imagine for a moment that you're a beat cop driving around in your cruiser, looking for signs of trouble. Well, what exactly do you look for? Suspicious individuals? Abandoned homes? What about random violence? Well, maybe it's not so random after all. NPR's Shankar Vedantam joins us regularly to talk about interesting social science research. He spoke to Steve Inskeep about a study that tries to predict where murders take place.
STEVE INSKEEP, HOST:
So what do you look for to find where a murder is going to take place?
SHANKAR VEDANTAM, BYLINE: Well, I recently spoke with a researcher who thinks she has the answer to that question. Her name is April Zeoli, and she works for Michigan State University. And she told me that we often have an unspoken assumption when it comes to thinking about crime and homicide.
APRIL ZEOLI: We think of individuals who commit homicide as being unlike the rest of us. They are crazy or substance users or, you know, had a bad childhood. There is some reason specific to the individual that they are committing homicide.
VEDANTAM: It's sort of the "Law & Order" on TV approach to crime, I guess.
ZEOLI: Correct, yes.
VEDANTAM: So this is why the cop on the patrol beat is often on the lookout for individuals who might be trouble. The problem with the approach is that it doesn't look at larger patterns in crime. And Zeoli is an interesting figure because she's a public health researcher who is in the criminal justice department. And she says if you have an infectious disease on your hands - if you have AIDS or the flu - you just don't look at people one-by-one and say, why did this one individual become sick.
You look for much larger and bigger patterns. And so, what she asked herself is: What would homicide look like if we thought about it as a public health problem.
ZEOLI: We looked at homicide as an infectious disease. And to spread, an infectious disease needs three things: a source of the infection, a mode of transmission, and we need a susceptible population.
INSKEEP: OK, this sounds really interesting, thinking of it as a susceptible disease of murder as contagious. But, of course, it's not a virus. It's a thing that individuals do. How does this make sense?
VEDANTAM: Well, at some level it's an analogy but it's also, I think, more than an analogy. So what Zeoli and her colleagues did was they looked at all the homicide that's taken place in Newark, New Jersey, between January 1982 and September 2007. So that's a 25-year period. There were 2,366 murders in all. And Newark, in general, has had a homicide rate that's three times higher than the general U.S. population.
Now, if crime is just about individual behavior and individual motives, then the murders should have generally been fairly random. And it should be very difficult to predict when and where murder is going to take place. But, in fact, when Zeoli looked at these 2,000-plus murders, what she found is that homicide spread through Newark very much like an infectious disease. And in fact, it followed predictable rules almost as if it was an infectious disease.
INSKEEP: Predictable rules meaning what, did it happen in specific neighborhoods and specific places? When there's one murder there will be several more and spreads out from there, that's what that means?
VEDANTAM: Yes, and so there are mechanisms of transmission. So as one neighborhood becomes troubled the next neighborhood eventually, in five years time, could become troubled. And this, at some level, is predictable. And this is the power of the infectious disease model, which is it doesn't just tell you what the hotspots are right now - where the crime is happening right now. It promises to tell you where the crime might be taking place in the future.
One of the ways it does that, actually, is it looks at something that epidemiologists would call positive deviance. In other words, when you have an epidemic, one of the things you really want to study is areas that are not affected by the epidemic. You want to look at those areas and study everything you can about them.
ZEOLI: We actually had some areas within Newark that were resistant to homicide, despite being surrounded by areas with high homicide rates. So we need to investigate why those little islands exist.
VEDANTAM: So, Steve, basically what studying these areas of positive deviance can give us - and this is to use the language of infectious disease research. If you find areas that are resistant to an epidemic, it gives you clues about how to inoculate other areas and prevent the epidemic from spreading.
INSKEEP: Shankar, thanks very much.
VEDANTAM: Thank you.
GREENE: That's Shankar Vedantam talking to Steve Inskeep. And you can follow Shankar on Twitter @hiddenbrain. You can also follow this program @morningedition, @nprinskeep or nprgreene.
(SOUNDBITE OF MUSIC)
GREENE: You're listening to MORNING EDITION from NPR News.