The phrase “I can’t breathe” — used by both Eric Garner and George Floyd in their fatal encounters with police — has become a rallying cry for a nationwide movement demanding an end to excessive use of force by police.
In the wake of Floyd’s death, there has been a renewed call from reform advocates to restrict police use of force. But measuring the impact of different restrictions can be complicated, making it hard to get a clear picture of whether they are effective. Some officers have also expressed concern that significant limitations may jeopardize their safety or prevent them from doing their jobs effectively.
Some cities are proposing additional regulations to their use of force policies. Others are implementing trainings on bias or de-escalation. President Donald Trump issued an executive order on policing last month that calls for a federal database to track incidents of excessive use of force. Two pieces of legislation introduced in Congress, also in June — one by House Democrats and another by Senate Republicans — sought to limit chokeholds and encourage different training and alternatives to force.
Researchers said that what the country knows about how police use force, as well as the success of proposals to reduce it, is limited. What is considered an unnecessary use of force can be different from department to department. Federal and state data tracking use of force is lacking, and the quality of policies and training — what skills or techniques they emphasize — also varies.
Research and data on use of force is lacking
The U.S. has more than 12,000 local law enforcement agencies, and none are required to report use of force incidents to the Justice Department.
In recent years, the federal government has made efforts to collect more data on police use of force, but participation is voluntary. The FBI’s National Use-of-Force Data Collection project, launched in 2019, received submissions from 40 percent of police agencies. The findings of the information gathered so far have not yet been published.
Most states do not have a standardized system for police departments to report use of force, said Seth Stoughton, an associate professor of law at the University of South Carolina who worked as a police officer in Florida for five years.
“We really don’t have any comprehensive data federally or at a state level,” Stoughton said, adding that the available federal data is “wildly inaccurate.” The most comprehensive tracking of police use of force or fatal encounters do not come from the federal government, he said, but rather from journalists.
In 2015, the Washington Post began tracking fatal shootings by on-duty police officers across the country. Between 2015 and 2020, the Post found more than 5,000 fatal officer-involved shootings. But not every fatal encounter involves a shooting, as exhibited in the cases of Garner and Floyd.
A report from the Guardian found that police killed 1,093 people in 2016, of whom 1,011 died of gunshot wounds. Another comprehensive database, by NJ Advance Media, tracked five years worth of use-of-force reports — 72,677 in total — from every local police department in New Jersey. Among the project’s findings: Ten percent of police officers accounted for 38 percent of all instances of use of force.
These three investigations do not assess whether force was justified in any of the cases.
Research indicates that of the roughly 60 million police-civilian encounters in the U.S. each year, about 1.8 percent may involve use of force, Stoughton said, but states and departments do not have uniform definitions of “force” and what interactions officers are required to report. Lower level uses of force like a shove or tackle to the ground are more likely to go unreported, Stoughton said.
The variation between police jurisdictions underscores the need for a national database to provide a centralized way to identify and compare trends, said Kami Chavis, a professor and director of the criminal justice program at Wake Forest University School of Law.
Police are more likely to use force against people of color
Researchers who spoke with the NewsHour said disparities exist in how police use force against people from different racial groups, but measuring these differences is complicated. Many studies analyze police killings against nationwide Census data, while some try to account for crime rates in a particular area.
The Washington Post report found that fatal police shooting rates were twice as high among Black Americans as they were among white Americans, and Hispanic Americans had the second highest rate of fatal police encounters. An analysis from The Guardian found that in 2016 police killed Native Americans at the highest rates (10.13 per million people), followed by Black people (6.66 per million). The rate for Latino people was 3.23 per million, and 2.9 per million for white people. The NJ Advance Media investigation concluded that statewide in New Jersey, a Black person was more than three times more likely to face police use of force. And a 2019 national study from university researchers found that Black people are 2.5 times more likely than white people to be killed by police.
One Bureau of Justice Statistics survey determined that in 2015, reported rates of nonfatal threats or uses of force were also higher for Black and Hispanic people — 3.3 percent and 3.0 percent, respectively — compared to 1.3 percent for white people.
But it’s hard to measure racial disparities, in part because every case has a unique set of circumstances, said Robin Engel, director of the International Association of Chiefs of Police/University of Cincinnati Center for Police Research and Policy. Simply taking the total number of Black people killed by police and comparing it to their overall population size is not a good comparison because it removes the situational factors that may play a role in the use of force, Engel said.
“I’m not claiming that [racial disparities] are not real. They are. It’s very clear there are racial and ethnic disparities,” Engel said. “But the reasons for those disparities is what we really need to better understand as social scientists so that we can better inform the solutions.”
One of the strongest predictors of whether police use force is civilian resistance, she said. Use of force is also more likely when officers are engaged in enforcement activities like making an arrest, Stoughton said. However, these predictors do not speak to different reasons people of color might be more likely to encounter an officer in the first place.
These reasons can include disparate decision making in how police are assigned to different neighborhoods, as well as whether and why people choose to call the police, Stoughton said. Incidents of Black people having the police called on them while sleeping on their university campuses, barbecuing and birdwatching, among other activities, have gained national attention. Numerous high-profile cases of police killings also involve Black people who were unarmed and engaged in nonviolent, low-level offenses.
Some of this behavior can be explained by racial bias, said Jennifer Eberhardt, a Stanford University psychologist and leading researcher on the science of bias. In a series of studies, she found that a group of police officers and a group of graduate students were each more likely to associate Black faces rather than white faces with images and words related to crime, such as knives or guns.
One California police chief told the NewsHour he believes racial disparities can be a problem in policing, but added that officers face a difficult task when making split second decisions about whether to investigate a potential crime or to walk away.
“Ultimately it’s about communicating the role the police have and the job of trying to maintain public safety,” said John Perez, chief of the Pasadena Police Department and research fellow with the National Police Foundation.
To assist with this goal, Perez’s department began working with a nonprofit in 2018 called Why’d You Stop Me, which seeks to foster positive civilian-police interactions by dispelling misconceptions about police to the community and training officers to understand community issues.
Efforts to regulate and reduce use of force in policing
Currently, 36 states have laws regulating lethal and non-lethal force, Stoughton and other researchers wrote in a piece for The Atlantic. More than three-quarters of those statutes were adopted in the 1970s, and most have not been amended recently, according to their findings.
For states without statutes, courts have the discretion to interpret use of force cases. Courts often evaluate use of force by referring to the Fourth Amendment, which is meant to regulate seizures, Stoughton said.
In recent years, some police departments and states have moved to limit officers’ use of force. One notable example is Cincinnati, which entered into an agreement with the Justice Department in 2002 that mandated sweeping changes to the city’s police department, including restrictions on use of force. Engel’s research found that between 1999 and 2014, Cincinnati saw a 69 percent decline in police use-of-force incidents, a 56 percent reduction in citizen injuries during police encounters and a 42 percent decrease in citizen complaints
The city of Camden, New Jersey, dissolved and rebuilt its police force in 2014. In 2019, Camden police adopted an 18-page policy that emphasized de-escalation and authorized deadly force only as a last resort.
“It started with two real principles that were laid as the cornerstones for how we would use force. One is that the sanctity of human life underpins everything that we do,” Scott Thomson, the city’s police chief until 2019, told the PBS NewsHour Weekend’s Hari Sreenivasan. “We review every incident of force that’s used with multiple layers of review from first line supervisor to the commander, to an internal affair review to the training unit review.”
Camden found that civilian excessive force complaints declined by 95 percent from a peak of 65 complaints in 2014 to three complaints in 2019.
After the 2018 police shooting death of Stephon Clark, a 22-year-old Black man, California enacted one of the country’s strictest use-of-force laws, which only allows police to use deadly force in the necessary defense of the officer’s or another person’s life. That same year, Pasadena police chief Perez implemented a 30-day review requirement for all use of force incidents in his department, which he said allows them to more quickly identify and discuss potentially unnecessary use of force incidents. Between 2018 and 2020, Perez said the number of use of force incidents decreased by 50 percent.
Tennessee, Delaware and Iowa also have laws that require officers to exhaust other “reasonable” means before using deadly force. And in the weeks since George Floyd’s death, several states and cities have moved to make changes such as banning chokeholds or no-knock warrants, Chavis of Wake Forest said.
“Now is the time for a meaningful change so that no one, especially black men and women, has to ever again think ‘that could have been me,” Isaiah McKinnon, a retired chief of the Detroit Police Department, wrote in a USA Today piece that recounted his experience being stopped by one of his own officers.
But a policy change is one of many things that may affect officer behavior, and may not change the rate of fatal encounters, Stoughton said. For example, he pointed out that the available data between 2015 and 2018 suggests officers killed more people per capita in Tennessee (about 3.6 per million people) than in Florida (2.9 per million), a state that gives broad authorization for officers to use deadly force.
Evidence indicates that in order for administrative policies to change officer behavior, the policy must clearly dictate what officers can and cannot do, must be widely communicated, and must be enforced, said Michael White, a professor at Arizona State University’s School of Criminology and Criminal Justice.
A 2016 report by the Use of Force Project looked at 91 police departments and assessed eight different force-related policies various departments had in place. Among these, the report found the policies most effective at reducing police-involved killings were those that require comprehensive reporting of when officers use force (25 percent reduction), those that require officers to exhaust all other reasonable means before using a firearm (25 percent reduction), and those that ban chokeholds and strangleholds (22 percent reduction).
The report found that police departments that had implemented four or more of the eight policies had 37 percent fewer police-involved killings than those with zero or one policy in place, and that departments with all eight policies in place would kill 72 percent fewer people, on average, between 2015 and 2016.
When it comes to bias or de-escalation training, research on their efficacy is virtually non-existent. Over the last decade police departments have shown a growing interest in both styles of training to mitigate use of force or address racial disparities. As of 2017, 16 states required officers to have de-escalation training, according to American Public Media, eight of which enacted the policies after the 2014 shooting of Michael Brown in Ferguson, Missouri. A CBS News report found that 57 percent of 155 police departments it contacted had implemented racial bias training in the five years since Brown’s death.
In January of this year, Engel and other researchers published a systematic review of studies that look at the effectiveness of de-escalation training. “And do you know how many we found for policing? None,” Engel said. “Not one study had been conducted to examine the impact of de-escalation training on officer attitudes or behavior.”
Of the 64 studies examining de-escalation training for other industries like nursing and psychiatry, Engel said the quality of the methodology was not strong, but the findings showed “slight-to-moderate individual and organizational improvements” as a result of the training. Based on this and anecdotal evidence, “there’s reason to be optimistic about de-escalation training,” Engel said.
On bias, there’s hardly any study on the effectiveness of training, Eberhardt said. Furthermore, she said, evidence indicates that simply becoming aware of a bias does not change behavior.
The composition and quality of the training varies, said Lorie Fridell, CEO of Fair and Impartial Policing, a company that has provided bias training to officers in New York City, Baton Rouge, Louisiana, Arlington, Texas and others. For Eberhardt, the most effective programs look to disrupt circumstances that may trigger bias. Fridell said her focus isn’t about eliminating bias but instead “managing bias.” Both strategies may involve prompting police to ask certain questions of themselves to figure out why they are engaging a particular civilian.
“Rather than simply informing people about the conditions under which bias is most likely to occur, we should be actually working to change those conditions,” Eberhardt said.
Fridell said the attitudes of officers in her company’s bias training range “somewhere between defensive and outright hostile.” But when she approaches bias as a human and societal problem rather than just a police problem, they become much more receptive, she said.
Bias and de-escalation training has received more broad support from law enforcement and lawmakers, but officers have voiced frustration over use-of-force restrictions and disciplinary action taken against them for using force. In 2019, the Crime and Justice Institute released a report on focus groups conducted with police officers in Baltimore. The summary said “officers fear and believe that too many documented uses of force will be used as evidence against them and result in disciplinary action, a criminal investigation, or restrict reassignment and advancement within the department.” It added that the officers said they felt less safe on the job and apprehensive about when to use force.
Perez of the Pasadena Police Department said he respects the Black Lives Matter movement’s calls for reform but he also understands some of the anxieties officers may have regarding strict limitations. Rigid policies that don’t take into account the challenges of policing could make officers hesitate to take actions to protect themselves during a confrontation, Perez said. “It could go too far,” Perez said. “It requires so much more discussion to make the changes that we need. We have to get the empirical evidence and look at it to protect the young officers who are working in the streets.”
Advocates, however, continue to push forward with proposals for sweeping systemic changes. As conversations about policing continue, Eberhardt said, it will take more than training or restrictions on excessive use of force, but they are an important start, she said. “It’s about addressing the entire context under which these police-civilian interactions occur.”