Read our COVID-19 research and news.
BOSTON--Greg Ridgeway doesn’t spend much of his time talking to the press, but the topics of his research are all over the news. A statistician at the University of Pennsylvania, Ridgeway has worked with police departments and oversight organizations on ways to measure racial bias in police stops, gauge police-community relations, and reduce gun violence. His presentation today in a session on statistics and criminology here at the annual meeting of AAAS, which publishes Science, focused on a study he recently published identifying which police officers are more likely shoot, based on data from the New York City Police Department (NYPD). Ridgeway sat down with Science to discuss that project and the experience of tackling controversial issues with data. This interview has been edited for clarity and length.
Q: What made you want to apply statistics to policing?
A: I started at RAND [Corporation] in Los Angeles [California] in 2000 … and I had this one project looking at gun violence in East L.A., trying to reduce gun violence in this very small neighborhood of Boyle Heights. Team members with me devised a new intervention strategy, and gun violence went down. … I think that got me really excited about the power of using data analysis to do something about crime and the justice system.
Q: One of your recent studies look at individual factors that make a police officer more likely to shoot. How did that come about?
A: The NYPD asked me to do analysis of their use of force … and I started reading about these situations where there are two officers on the scene, and one shoots and one doesn’t. Perhaps it’s just that one was a little closer, or one saw something the other didn’t. But what if it was consistently certain features about officers that predict which of the two was going to be the shooter? … Because they were matched on the same scene, we get away from excuses [like] "Well maybe it’s because someone has a higher rank or someone is working more difficult neighborhoods." All of those confounding explanations can’t work here, because they’re on the same scene of the same shooting.
Q: What did you find from the 106 shootings you analyzed?
A: These are not random events. There are features of officers that do increase or decrease the risk of shooting. A key one is … a lot of negative marks in a file. We keep finding the shooters are the ones that already were crashing cars, they were losing their badges, they weren’t showing up. … The other feature that had not come up before in other studies is the age at which officers are recruited to become police officers. … Years of experience were not predictive of being a shooter, but the age at which they joined the police department was. So that means those who join when they are 21 years old have an elevated risk of being a shooter, and that risk persists through their career.
Q: You also found that black officers were 3.3 times more likely to shoot than other officers.
A: I opted not to put out a press release on this one, simply because I thought that would end up being the dominating story. In fact, Jim Fife [a key researcher in this area], when he first noted 30 years ago that black NYPD officers were twice or three times as likely to shoot, faced a lot of criticism, even though he also said they’re working in different environments. I don’t have the luxury of sort of explaining away that finding.
Q: So what do you make of that finding?
A: I can’t explain it. No explanation for that. There’s also no policy recommendation I would ever make based on that. Because at the same time, I believe we do need to encourage and diversify our police departments for a variety of other reasons—transparency, and so that our departments resemble the communities they police.
Q: Do you worry the study could be used to dismiss concerns about white officers shooting black men?
A: I think [the study] focuses the problem. I don’t think we need to make this about white officers shooting young black men. It’s about police officers shooting young black men.
Q: Could the finding be specific to NYPD?
A: New York is the only place I’ve done this and the only place I’ve been able to do it. I’ve tried other cities, and they often don’t collect sufficient data of all the officers that are on the scene.
Q: Have any of your earlier findings been misinterpreted?
A: Yeah, I had this headline in The New York Times. It was something like, “Analyst says NYPD is stopping a lot of blacks—and they should.” … Nowhere in the report did it say they should stop more black people. It said in some aspects, the police seem race-neutral in areas where they were believed to be racially biased. In other places where they were believed to be racially biased, we indeed found, yep, they are. … And that kind of nuance is difficult to communicate succinctly.
Check out our full coverage of AAAS 2017.