http://www.nytimes.com/2016/07/13/u...stions-about-his-police-force-study.html?_r=0
A new
study on police force found no bias against black civilians in police shootings in 10 cities and counties, including Houston. It did find bias against blacks in every other type of force, like the use of hands or batons. The study provoked debate after it was posted on Monday, mostly about the volume of police encounters and the scope of the data it used. Below, the author of the study, Roland G. Fryer Jr., a professor of economics at Harvard, answered questions from readers.
What about the chance of a police encounter?
Mr. Fryer’s study looked only at what happens once the police have stopped civilians, not at the chances of being stopped in the first place. Many readers questioned whether that was the right denominator. Other research has shown that blacks are more likely to be stopped by the police.
Lee Buttala from Ashley Falls, Mass.,
asked, “Is it possible that the statistics on shooting are misleading because the police are less likely to stop white people generally.”
Wendy Maland from Chicago
put it this way: “The question isn’t — once police identify you as a potential criminal, how are you treated? — the question is — who is being treated as a criminal?”
Mr. Fryer: I agree that blacks are more likely to be stopped, more likely to be harassed and more likely to be arrested.
Ideally we would be able to set up an experiment to understand potential differences before an encounter. Unfortunately, that would require us to randomly assign civilian race in encounters of police, which isn’t possible!
Given this limitation, we need to make the best out of available data. There are two important things I want to note:
1. The types of encounters that lead to police shootings in the videos that we have all seen are not the most common that actually occur in the data. In Houston, for instance, most of the officer-involved shootings come from calls for service resulting from burglaries or violent crimes, not from chasing down people with broken taillights.
2. I totally agree that deciding who to stop in a police stop is highly problematic and there certainly may be racial bias in that decision. So let’s think about the officer-involved shootings in which there’s a robbery in progress or a violent crime. Those are less likely to be plagued by selection bias in the decision of who to harass or stop. Analyzing only those cases yields similar results. Moreover, when we analyze only cases in which the officer-involved shooting began with a routine stop or a traffic stop, we do not find bias. But these results are susceptible to your point that there’s more traffic stops of blacks.
Why the focus on Houston?
Mr. Fryer’s evidence on shootings came from 10 cities and counties. In these places, he examined questions like whether officers were quicker to fire at black suspects and whether black civilians in officer-involved shootings were less likely to be armed. The data from these places supported his counterintuitive finding, that there wasn’t racial bias in the use of lethal force. But his most comprehensive data — including times when the police did not shoot — came from Houston.
Mr. Phil in Houston
wrote, “Houston and NYC are certainly the most racially diverse cities in the country, yet was Houston the best comparator?” (Houston and New York City are indeed among the most racially diverse cities in the country.)
Mr. Fryer The most comprehensive set of officer-involved shooting data is from the Houston Police Department. For this reason, we contacted HPD to help construct a data set of police-civilian interactions in which lethal force may have been justified. If we had the data from other cities, we would definitely use it.
My sincere hope is that the type of analysis being done in our paper will lead other police departments and community groups to understand the types of data we need to answer these important questions and work together to be more transparent and make that data public.
How is your work different from an earlier analysis?
Many readers asked about previous studies, in particular a paper published in PLOS ONE by Cody T. Ross. That paper,
“A Multi-Level Bayesian Analysis of Racial Bias in Police Shootings at the County-Level in the United States, 2011–2014,” found that the chance of being black, unarmed and shot by the police was about 3.5 times the chance of being white, unarmed and shot by the police. It was based on a crowdsourced data set, the U.S. Police-Shooting Database, that includes some nonfatal shootings. (About two-thirds of the shootings in Mr. Fryer’s data set were nonfatal.)
The USPSD covers the entire country, but it is not comprehensive. It has information from a variety of departments on 16 civilians shot by the police in Houston and other parts of Harris County, Tex., from 2011 to 2014. Mr. Fryer’s data shows 177 shootings by the Houston Police Department in those years.
The questions the papers asked were different, particularly in Houston. As Mr. Ross wrote, “The USPSD does not have information on encounter rates between police and subjects according to ethnicity. As such, the data cannot speak to the relative risk of being shot by a police officer conditional on being encountered by police.”
John H. in Chicago
wrote, “Please compare these results versus the ones published by Ross.”
Mr. Fryer: Our paper does not attempt to overturn previous analyses; its guiding novelty is the granularity of the micro-data.
For nonlethal force, why were the estimates from police data and civilian data so different?
Mr. Fryer looked at two sources of data on nonlethal force. The first was from
stop-and-frisk records in New York City. The second, from the perspective of civilians, came from a national survey, the
Police-Public Contact Survey, or PPCS.
Hydraulic Engineer in Seattle
wrote, “The last graph on citizen-reported encounters — ‘Use of Force in All Types of Police Encounters, According to Civilians’ — requires much more explanation. It shows much more bias toward blacks than the earlier graphs using police reports of encounters with ‘compliant’ citizens.”
Mr. Fryer: There are several potential explanations for the quantitative differences between our estimates using Stop and Frisk data and those using PPCS data. First, the baseline probability of force in each of the data sets is substantially different. Second, the PPCS is a nationally representative sample of a broad set of police-civilian interactions. Stop and Frisk data is from a particularly aggressive form of policing in a dense urban area. Third, the PPCS data is gleaned from the civilian perspective. Finally, granular controls for location are particularly important in the Stop and Frisk data and unavailable in PPCS.
In the end, the answer is likely somewhere in the middle, and — importantly — both bounds are statistically and economically important. In other words, even taking the Stop and Frisk data at face value, we should all care about these differences.
Do we have to trust police reports to believe your study?
“‘Police narratives’?!”
wrote Paula Robinson from Peoria, Ill. Larry Evans
asked,
“If a Narrative crashes in the forest, does it make any sound?”
Mr. Fryer: As we state in the paper, there are certainly high-profile cases in which the facts stated by officers differed substantially from the videos. So let’s take a minute and think about how this sort of misreporting might bias our findings.
In nonlethal uses of force, we find racial differences even when we use police accounts of the interactions, so perhaps the “true” estimates of the differences are even larger. In other words, one might think the differences are “at least” as big as we state.
Then we have to consider the lethal uses of force. Surprisingly — at least to me — the finding was true whether or not we relied on police accounts of the encounter. Let me pause there.
I expected to find that there were large racial differences in lethal uses of force but that once you accounted for what the police said happened that those differences would lessen or perhaps even go away. I didn’t have faith in police reports.
What we actually found was that there were no racial differences in the basic differences analysis. It didn’t matter whether we took context — as captured by officer reports — into account or not; there was no racial bias in either analysis.
Yes, police reports can be tricky to understand and decide how to use, but they don’t seem to affect our analysis meaningfully in this instance.
Mr. Fryer hopes his analysis — and reader examination of it — leads to better data on the police use of force. “It’s not the finish,” he said. “It’s the start.”
:dahell50mayne:
