Black Americans are more than twice as likely as white Americans to be killed by police officers.1 Researchers agree that racism almost certainly plays a role in that disparity. But “racism” is too broad an explanation to reveal much about the more immediate causes or to point to a way to reduce police killings of black people like the recent ones in Baton Rouge, Louisiana, and a suburb of St. Paul, Minnesota.
Researchers who have studied the issue say that racism manifests itself in different ways, requiring a range of solutions. If the disparity arises because bias among police officers makes them more likely to fire guns at black people than at white people who pose equal threats, for example, then the answer could lie in hiring, training and firing: test recruits for bias, train officers to not exercise bias and fire officers who demonstrate bias.
But if the disparity is due more to systemic police practices than the prejudices of individual officers, then the answer could be to change those practices — for instance, ensure that departments don’t concentrate car checks that are unlikely to turn up anything illegal and could turn violent in predominantly black neighborhoods.
And if the disparity is because there are relatively more police interactions with black people, because black people commit a disproportionately large share of reported crimes, then the answer could be to address the systemic causes of the crime disparity, including urban poverty. (No one said the solutions would be easy.)
Researchers say that these and many other factors underlie the disparity in killings but that identifying how much each factor contributes to the burden of police violence borne by black Americans isn’t possible based on the data available.
“Each of these factors all nudge reactions in the same direction: Greater expectations of crime and greater police confrontation among minority than majority members of the community,” said Keith Payne, a professor of psychology and neuroscience at the University of North Carolina who has studied police bias. “Even if the effect of each factor is small, their cumulative effect could be enormous. I suspect that is what we see when we look at the overall climate right now: the cumulative effect of dozens of factors all pushing lightly in the same direction.”
Charles Epp, a political scientist at the University of Kansas, thinks most scholars in the field would say the convergence of black people and police officers in places of concentrated disadvantage plays a major role, although he added that the decisions of departments and officers also are significant and interconnected. “A more aggressive style of policing” in those areas “almost certainly contributes to more rapid escalations toward use of deadly force,” said Epp, co-author of the book “Pulled Over: How Police Stops Define Race and Citizenship.”
Michelle Phelps, a sociologist at the University of Minnesota, said it may not be possible to determine exactly how much each factor plays a role in the racial disparity in police killings. “But I think we can say more generally which of these are driving more or less of the trends we see and, critically, which we can ameliorate through policy choices,” said Phelps, who studies criminal justice.
Saying racism likely plays a role in black Americans’ being killed at a higher rate than the national average doesn’t mean it plays a role every time an officer kills a black person. Among the black people killed by police officers this month are the suspected shooters in attacks on officers in Dallas and Baton Rouge that killed five officers and three officers, respectively. Police are trained to use lethal force when necessary to protect themselves and the public.
As incomplete as national data is for people killed by police officers — their number and the circumstances of their deaths — the data on people who interact with police officers and aren’t killed is even more limited. The FBI plans to start collecting data on use of force from police departments in January, according to spokesman Stephen Fischer. For now, though, it’s hard to answer basic questions about risk. How many people who are armed are encountered by police and perceived to be dangerous, what percentage of them are killed by police officers, and how does that differ by race of the people who are armed? That’s all impossible to answer.
Take, for instance, a high-profile study released this month, though not yet peer-reviewed, by Harvard University economist Roland G. Fryer Jr. and covered in The New York Times and many other media outlets. It found that police are no more likely to shoot a black suspect than a white suspect in encounters in which using deadly force could be deemed justified — such as when suspects attack officers or resist arrest. The implication: Black people are shot at a higher rate relative to their population simply because they have more high-risk encounters with officers.
Other researchers took issue with some of the research methods, including Fryer’s reliance on police reports, which don’t always depict uses of force accurately. One of their biggest criticisms was of the study’s scope: It could only make a reasonably strong claim about shootings in Houston. Shootings are rare enough that a single city’s data isn’t likely to provide much insight. And even if it does, something idiosyncratic about that city could have driven the finding. Maybe Houston has a police force that is unusually free of bias. Maybe police officials in the city knew that the results would reflect well on their officers and so they were more inclined to share the data.2
The larger point is that without a nationwide effort to collect data on every police interaction, it’s up to individual departments to decide whether to collect and publish this kind of information — and then it’s up to researchers like Fryer and his team, who took 30 to 45 minutes per arrest record to manually enter data on each of around 1,000 arrest records for Houston. And that doesn’t only mean we won’t know about police interactions with civilians in cities whose departments don’t report data. It means we have no way of knowing whether the departments that do report data are representative of the ones that don’t.
“Those departments willing to provide information voluntarily may differ systematically from those that do not,” said Rajiv Sethi, an economist at Barnard College who critiqued Fryer’s study and has called for more data on police encounters with civilians.
If departments that have less bias and use less force are more likely to share data with researchers, that would tend to make academic findings of bias conservative, said Phillip Atiba Goff, a professor in policing equity at John Jay College of Criminal Justice. In other words, they would be likely to understate bias. But Goff has found that there are plenty of reasons other than fear of looking bad for departments not to share data. In his role as the co-founder and president of the Center for Policing Equity, a research group that studies policing, Goff is working to build a database on police stops and use of force. He said that he has received commitments to share data from police departments that collectively serve one-third of Americans but that some departments are held back by software that won’t allow them to extract data. “That’s the level of technological retrograde we’re dealing with in law enforcement,” Goff said.
The many factors that might contribute to the racial disparity in police killings are hard to disentangle. “It is very difficult to empirically separate the effects of poverty, geography, race, race bias and policing tactics,” Payne said.
And factors can create or reinforce one another. For instance, a greater proportion of black Americans than white Americans live in poverty. Poverty is positively correlated with certain kinds of crime. This might lead officers who weren’t biased before joining a police force to become biased because of whom they encounter and arrest, Payne said.
Payne wants to see more controlled experiments, ones in which researchers can change just one variable in a lab setting and see what happens, to try to weigh the different effects of racism on the disparity in police killings. These can be valuable for measuring what’s known as implicit bias, for example. Tests have shown that officers who have been shown pictures of people and of objects are quicker to identify objects as weapons after they’ve been shown a photo of a black person than one of a white person. However, in simulations of interactions with the public, officers showed no bias against black suspects when deciding whether to fire their weapons.
These findings are from laboratory experiments. The real world doesn’t have controlled conditions, but it provides ample evidence that black people face a greater risk of being killed by police officers. For instance, while the vast majority of police officers are white, the vast majority of officers shot by other officers are black. And law-abiding black Americans interact with the police at a far higher rate than other Americans. Black people are disproportionately the subject of random stops by police officers, called “low hit rate” encounters by some researchers because of the small probability that officers will find a weapon. Data from Kansas City, San Francisco and New York City shows that black people are far more likely than white people to be stopped. And this isn’t just a function of the overlap between areas of high crime rates and ones with a high black population: Black people are stopped more often even after controlling for an area’s crime rates.
Beyond the data, there is the testimony of black Americans, including some in uniform or in political office, about experiencing a disproportionately large share of encounters with police. All but one of 25 current or former black New York Police Department officers interviewed by Reuters in 2014 said they’d been racially profiled. Republican Sen. Tim Scott of South Carolina said in a floor speech this month that he’d been stopped and pulled over all his life by officers.
These tense encounters can spiral into violence. A study co-authored by Goff this month found that even after adjusting for crime rates, police officers were more likely to use force with black people than with white people. (Fryer found the same thing for uses of force that didn’t involve firearms.) One reason for that is that officers’ racial bias — conscious or not — “may shape subtle aspects of the whole interaction in ways that are more consequential than the last stage of pulling the trigger,” Phelps said. There are many individual examples of police officers expressing explicit racism in emails. These add up to less than comprehensive statistics about the extent of racism among American police officers but are perhaps more than just a series of anecdotes, especially considering that other officers who are racist may never have put it in writing.