It's not just black people, but entire, predominantly black, neighborhoods that are disproportionately impacted by the Philadelphia Police Department's use of stop-and-frisk. That's a key finding of a new analysis of police data from 2014 to 2015 by Lance Hannon, a Villanova University professor of sociology and criminology who began analyzing publicly available police data after the presidential candidates clashed over the effectiveness of stop-and-frisk in debates last year.
He found that mostly black neighborhoods drew 70 percent more frisks than nonblack areas, yet yielded less contraband. And, he discovered, the elevated rate of frisking was consistent whether the predominantly black neighborhood was a high-crime area or a very low-crime area. Although many African American neighborhoods in the city have low crime rates, he said, "People, police officers, and nonpolice officers tend to judge the dangerousness of a place based on racial predominance. When they think of a black area as being dangerous, they are thinking of the outliers — and all the other neighborhoods that are relatively safe get painted with the same brush."
Mike Dunn, a spokesman for Mayor Kenney, said that the city is addressing the issue with implicit bias training and progressive discipline for officers who make bad stops. He said the overall number of stops has declined since last spring, while the number of stops identified by the city as problematic has declined by 72 percent.
Hannon's research makes the case that stop-and-frisk can be problematic not just at the individual level– but also at the neighborhood level. He says if that can't be fixed, perhaps it's time to end it.
What does your research reveal about stop and frisk that we didn't already know?
My study provided support for two findings in previous research. First, African Americans are more likely to experience a frisk when stopped. Second, African Americans are more likely to experience unproductive frisks – in which the frisk does not uncover any contraband or lead to an arrest.
My findings also go beyond what has been found in previous research. For example, I find that it's not just black individuals; anyone, regardless of race, in a black area is more likely to be frisked when stopped and more likely to be frisked unproductively.
I also looked at how the crime rate of an area differently matters in predicting the likelihood of a frisk based on the racial composition of the area. I found that the association between violent crime and the likelihood that an officer will frisk someone is contingent on the degree to which African Americans are present in the neighborhood. So in predominantly black neighborhoods, the association between the violent crime rate and the frisk rate is very weak. In areas that are not predominantly African American, the association is moderately strong and in the direction you would expect: higher violent crime rates mean higher frisk rates.
There is a puzzle here: Why is the violent crime rate significantly less predictive of frisking activity in predominately black neighborhoods?
Why do you think that is?
Actually, I think the answer to that question is very closely related to the answer to why law-abiding African Americans experience so many more unproductive frisks as individuals. The reason that has been talked about before by others is it appears that African Americans are held to a lower level of reasonable suspicion – so people who should not be perceived as dangerous or are objectively not dangerous are more likely to get swept up in a stop-and-frisk initiative and be unproductively frisked due to implicit racial bias. Not only is that happening at the individual level, but it's also happening at the neighborhood level.
Neighborhoods with low levels of criminal activity are erroneously considered dangerous due to the prevalence of African American residents. There seems to be good evidence that it's not just the stigma associated with an individual's race that potentially drives bias; the stigma associated with a neighborhood's race also affects bias.
In other words, there is racial profiling of individuals and there is racial profiling of neighborhoods.
Outside of the issue of racial bias, what do your results say about the usefulness of stop-and-frisk as a policing tool?
The data suggest that, regardless of racial composition, as the number of frisks increases in a neighborhood, the proportion of frisks that produce contraband decreases. Some might interpret that correlation as proof that frequent frisks deter people from carrying contraband.
But what I believe the data are actually showing is that, in places where you are doing more stopping and frisking, you are being tougher and harder — but you are not being smarter if your goal is to uncover weapons. That negative correlation represents that when you're overzealous, you make more bad judgment calls. Maximizing the quantity of stop-and-frisk seems to hurt its quality as a policing tool.
Why do you think this is a big deal?
If there's more stopping and frisking in minority neighborhoods, it's going to decrease the legitimacy of the police in the eyes of residents there and increase mistrust of the police. Police are going to perceive that mistrust and it's going to lead them to potentially stop and frisk more. It becomes almost a spiraling system that hurts everyone.
What do you think should be done?
We need to pay attention to how police officers can be racially biased, based not just on the race of the suspect but also based on the race of the neighborhood they're in. We need to start compiling statistics and monitoring outcomes based on that. ….
In addition, there are a lot of different ways to do proactive policing besides stop-and-frisk. So if stop-and-frisk isn't working, or if it's having these effects which are racially disparate and ultimately potentially undermining police legitimacy, then change. Stop using stop-and-frisk, and use something else.
Certainly, we see that, in New York, the crime rate has not actually gone out of control after stop-and-frisk was greatly curtailed. One clear policy implication is stop trying to do proactive policing via stop-and-frisk. Find another way.