Brentin Mock is a staff writer at CityLab. He was previously the justice editor at Grist.
Two new studies have revived the long-running debate over how police respond to white criminal suspects versus African Americans.
In the U.S., African Americans are 2.5 times more likely to be killed by police than white people. For black women, the rate is 1.4 times more likely.
That’s according to a new study conducted by Frank Edwards, of Rutgers University’s School of Criminal Justice, Hedwig Lee, of Washington University in St. Louis’s Department of Sociology, and Michael Esposito, of the University of Michigan’s Institute for Social Research. The researchers used verified data on police killings from 2013 to 2018 compiled by the website Fatal Encounters, created by Nevada-based journalist D. Brian Burghart. Under their models, they found that roughly 1-in-1,000 black boys and men will be killed by police in their lifetime. For white boys and men, the rate is 39 out of 100,000.
In fact, people of color in general were found more likely to be killed by police than their white counterparts.
The study was published on Monday in the Proceedings of the National Academy of Sciences, or PNAS, a journal that recently drew controversy for publishing another study on police killing disparities. That study, led by Michigan State University psychology professor Joseph Cesario, published on July 22, found that violent crime rates and the racial demographics of a given location are better indicators for determining a police killing victim’s race.
As Cesario explained in a press release:
Many people ask whether black or white citizens are more likely to be shot and why. If you live in a county that has a lot of white people committing crimes, white people are more likely to be shot. If you live in a county that has a lot of black people committing crimes, black people are more likely to be shot.
The two studies are just the latest salvos in a long-running debate over whether police violence towards African Americans is better explained because of racial prejudice or because black people are really violent enough to justify extra police force. The Cesario study, with its focus on crime rates, seems to fall in the latter camp. Both rely on media-generated police shootings data—Cesario’s uses databases produced by The Washington Post and The Guardian.
Several academics have challenged Cesario’s methodology, namely his decision to “sidestep the benchmark” of using population to calculate racial disparity. It has been questioned whether using population is an appropriate benchmark in these kinds of analyses: Critics of this technique believe that population-benchmarking is flawed because it assumes black and white people have an equal likelihood of encountering police. (An example of population-benchmarking is, as Cesario’s study explains, stating: “26% of civilians killed by police shootings in 2015 were Black even though Black civilians comprise only 12% of the U.S. population. According to this 12% benchmark, more Black civilians are fatally shot than we would expect, indicating disparity.”)
Instead of using population, Cesario analyzed variables such as the race of the police officers, crime rates, and the racial demographics of locations where police shootings happened in 2015. From that, he derived that black and Latino victims of police killings were more likely to have been shot by black and Latino cops, and that ”might not be due to bias on the part of Black or Hispanic officers, but instead to simple overlap between officer and county demographics.”
The problem with this, as Princeton professor Jonathan Mummolo, explained on Twitter, is that it still rests on the assumption that black and white officers encounter black civilians in equal numbers, or in even temperaments—which they don’t.
The study claims its approach “sidesteps the benchmark debate”—the problem of picking a baseline to use to evaluate shooting rates across racial groups. We show this is not true. The study implicitly and wrong assumes black/white civilians encounter police in equal numbers.2/N— Jonathan Mummolo (@jonmummolo) August 1, 2019
What do the recent mass shootings tell us, if anything, about this?
There’s also something to be said for what the victims were doing when the cops shot them. Cesario points out that, “The vast majority—between 90 percent and 95 percent—of the civilians shot by officers were actively attacking police or other citizens when they were shot”—and that there were more white civilians who were committing such attacks when police killed them than were African Americans. In fact, white people were more likely to be armed when police killed them, as Cesario’s study acknowledges—“if anything, [we] found anti-White disparities when controlling for race-specific crime,” reads the study.
The three most recent mass shootings—in Gilroy, California; Dayton, Ohio; and El Paso, Texas—represent extreme examples of armed white men on the attack, but looking at them through the lens of Cesario’s findings is still revealing. Police shot two of the perpetrators, killing one of them. Connor Betts, the shooter who killed nine people and injured 27 more on August 4 in Dayton, was stopped by police bullets less than a minute after his attack began. Police fired at the Gilroy shooter, Santino Legan, but he ultimately succumbed to self-inflicted wounds. Patrick Crusius was arrested “without incident” after killing 22 people and injuring dozens more in El Paso.
In only one of these cases did police actually shoot and kill an armed white suspect who was on the attack: Betts in Dayton. Even that case is murky, though. Betts wore a mask, hearing protection, and body armor—his race was likely not apparent from a distance, and the entire melee happened very quickly. But even if one interprets the fact that police shot at two of those three shooters as evidence of the “anti-White disparity” Cesario mentions, one could also argue that it takes whites committing large-scale acts of terror with automatic weapons for police to respond in the same way that police have responded to, say, a teenager walking away-from-police-while-black.
The limitations of the data
Another way to determine whether racial bias is a factor is by examining police behavior when their target is unarmed and not on the attack. This is what University of Nebraska at Omaha criminology professor Justin Nix examined in his 2017 study on police killings. Nix’s research, which Cesario cites often in his own study, also focuses on police shooting-killings in 2015, when police killed nearly twice as many white people that year (495) than they did black people (258). But 15 percent of the black people police killed that year were unarmed, compared with just 6 percent of white people who were unarmed when killed by police. The study also found that 24 percent of African Americans and 32 percent of other non-white racial groups were not attacking police officers when they were killed, compared to 17 percent of white people. This was interpreted as “preliminary evidence of an implicit bias effect,” against African Americans and people of color.
Nix, however, is cautious about deriving any firm conclusions from his own findings or Cesario’s because the data on police shootings in general is too limited. The FBI finally launched its database on police-involved shootings just this year, which is why researchers rely on databases created by journalists. And even the subset of data that academics have been working with—police shooting fatalities—have their own range of limitations.
Cesario declares in his press release that “violent crime rates are the driving force behind fatal [police] shootings,” but Nix says that is “pretty strong language in light of the limitations,” especially if looking at when police deploy lethal force at the local level.
“I don’t think the conclusions are warranted based on their analysis,” said Nix. “You can’t restrict the data to just fatal shootings. Another problem is that when doing these bird’s-eye views, you lose nuance from city to city. Policing is a local thing and there’s no reason to believe that everything is the same across the board.”
For example, Nix would want numbers not only on how many times a police officer shoots their weapon, but every time they draw their gun. “You need a benchmark that says how often they were in certain circumstances where they could have shot but did not. That gets us closer to the likelihood of racial bias.”
Nix recently updated his analysis on police shootings using fatal and non-fatal shootings from the 47 largest metros from 2010 to 2016, using a dataset produced by VICE. That analysis found wide variation between the cities—in St. Louis, 16.8 percent of police shootings were fatal; in Phoenix, 51.9 percent were; and, in Tampa, all three of its police shootings were fatal.
What the studies don’t tell us
Cesario’s study centers the characteristics of the police officer over the victim, concluding essentially that since black and Hispanic police are as likely or more likely to kill people of color as white officers, that the race of the police officer doesn’t matter. But it’s not clear whether that matters in determining whether police bias exists at all. As Philip Atiba Goff, president and cofounder of the Center for Policing Equity, told NPR, “Racism is not a thing that white people can have and black people can’t. And nobody’s research would suggest that it does.”
Looking at individual police characteristics doesn’t tell the public anything about the links between structural racism—both within a police department and throughout society—and police violence. Boston University School of Public Health scholar Michael Siegel found that connection in his study last year, which analyzed data on police killings between 2013 and 2017. States that have higher rates of racial segregation, incarceration, educational attainment, economic disparity, and unemployment also tend to have higher levels of police violence against African Americans, Siegel found.
Nor does looking at the racial characteristics of individual police shooters tell the public anything about why American law enforcement as a system finds unarmed nonwhite civilians threatening enough to shoot and kill more often than unarmed whites. They don’t explain why police choked Eric Garner, who had no weapon and posed no threat, or why police shot and killed Philando Castile while he was restrained by a seatbelt in a parked car. Meanwhile, Patrick Crusius committed one of the largest terrorist attacks on Latino Americans in U.S history and was apprehended “without incident” while still at the scene of the crime. The point is not that police should have also killed Crusius, but that Garner and Castille should still be alive.