Shootings And The Problem With Profiling

Originally published on December 16, 2012 1:43 pm
Copyright 2016 NPR. To see more, visit


And at moments like these, people often wonder if they should have seen something, in the behavior of the gunman, that might have predicted he could commit such an act of violence. We're joined now, in the studio, by NPR science correspondent Shankar Vedantam. Thanks very much for being with us today.

SHANKAR VEDANTAM, BYLINE: Glad to be here, Scott.

SIMON: And we want to ask you: Every time there is an incident like this - and as we have noted, there has certainly been more than one - people begin to look for what are called profiles. Now, we want to be careful about this because we know somebody can hit all the marks on a profile, and wind up being a very worthy citizen. What are some of the signatures, though; common traits that people have noticed?

VEDANTAM: Well, if you look at the series of incidents that have happened in recent years, there are several things that stand out, in terms of patterns. These have always been conducted - the shooters have always been men. They have usually been young men; usually, young white men. Many have shown longstanding interest in guns. Many of them have been mentally disturbed. Many of them have been described as loners. Some of them show an interest in video games. Some of them have a history of being bullied. But all these chactacteristics describe events that have taken place in the past, which is why I call them patterns.

What they are not is actually, not very good at predicting what's going to happen in the future. What a profile - a really effective profile does, it tells you who the next shooter is going to be; what the next target is going to be. And we're simply not very good at doing that

SIMON: If patterns can't give us that kind of information, is there nevertheless something that can be deducted from these patterns that people have begun to notice, that could be useful; that could enable law enforcement, mental health officials, family members and friends?

VEDANTAM: Well, it's potential that - there's a lot of stuff that's potentially valuable in patterns, in - you know, surfacing things that get passed along to authorities. But to basically turn the pattern into a profile - where you are using it as a predictive mechanism - is very, very difficult to do because the characteristics I just mentioned to you don't fit dozens of people. They probably fit hundreds or thousands - or even tens of thousands of people.

So the problem, from a scientific point of view, is that these are highly unusual events. And unusual events - especially highly unusual events - are precisely the ones for which you cannot predict very precisely exactly what's going to happen next.

SIMON: Shankar, we have to ask this morning about that phrase. Are these - can we really say these are unusual events? There have been - well, this was the seventh incident - school shooting, just this year.

VEDANTAM: You know, I agree. As a parent, I have to say it doesn't feel like these are unusual events. These feel like they are happening with incredible regularity. And it's hard, I think, at points like this, to step back and say, let's look at the evidence, and let's look at the data. But if you do do that, what you do find is that - you know - most children who die, aren't being murdered. They die in traffic accidents and unintentional injuries, and drowning. Most of the homicides that occur among children don't occur in schools. They occur outside the school. And most of the homicides that occur in schools, don't take place in mass shootings.

So if you were to look at the data, the odds that any one school woud potentially be affected by a mass shooting - you know, you would have to wait thousands of years for that school to potentially be affected. That's not to downplay it; it's not to be blase about it. But it is to put the risk in to some perspective.

SIMON: Given that these are statistically rare events, and that we certainly don't have a scientific way of identifying anybody who might be at that point of bursting, ahead of time, is there still something else that authorities - I'll repeat it - medical experts, family and friends should be doing?

VEDANTAM: Well, the growing consensus among authorities - and law enforcement authorities, in particular - is to move from this idea of profiling, to the idea of what they call threat assessment. Because if you look at all these recent incidents, one of the things that stands out is that most typically, these events have taken time to plan. It has taken time to set up. And very often, there is another person who knows what the plan is; or maybe there are people who know some element of the plan that the shooter has in place.

Now, very often, this information might be hazy, or it might be disjointed. And one of the strategies of threat assessment is to encourage people to report what it is that they are seeing. This is the basis of the - you know, the "See Something, Say Something" campaign that we hear all the time in subway stations and metro stations and airports. And the idea is that once you get specific information that's behavioral; it's not targeted to the indivi - to sort of a profile, but to the behavior of people; you then analyze that information. You say, is this a credible threat? Is this person just upset? Does he have a plan? Is this a serious threat? Is it a very serious threat? Should law enforcement be involved? And the idea is, you don't go from zero to 60 in one second; but you take in the information, you analyze the information, you escalate your response gradually.

SIMON: And increasingly, we've seen in the past few incidences, people tend to post things in social media; that's even if they're not talking to people.

VEDANTAM: Indeed. So that's certainly one of the places where we can look for the kinds of information that could give us clues that something might be happening.

SIMON: NPR science correspondent Shankar Vedantam.


SIMON: And you're listening to NPR News. Transcript provided by NPR, Copyright NPR.