In a study published this summer, researchers found that Oakland police officers were significantly less likely to use respectful language when talking to black motorists, and were more likely to speak respectfully to white drivers.
The researchers analyzed transcribed dialogue from footage taken by Oakland police using body cameras. Police officers wear these cameras, which record video of their interactions during traffic stops and other policing activities. The study “tells us about the everyday encounters people have [with] the police, and how those differ for Black and White citizens,” wrote Nick Camp, one of the researchers, in an email.
“To date we’ve been mostly looking at what’s been reported by the police in terms of their interactions with civilians,” said Jack Glaser, a professor of public policy at UC Berkeley who is familiar with the study, “and now we’ve got some more realistic data based on their actual conversations.”
The 981 encounters scrutinized for this study took place during April, 2014, in Oakland, and involved black and white motorists. The researchers, from Stanford University’s linguistics, psychology, and computer science departments, wanted to see if there was a discernible difference in the level of respect with which officers spoke to the people they pulled over, based on the motorists’ race.
To analyze the footage systematically, the researchers first took a small portion of the 36,738 “usable officer utterances” (gleaned from 183 hours of footage), and had “raters” participating in the study rank them on several “dimensions of respect.” The researchers then generated a scale of which words and phrases were considered most respectful and which least respectful. These ranged from reassuring the motorist, which the raters associated highly with respectful language, to the officer telling the driver to put their hands on the wheel, which registered as disrespectful with raters.
Based on these participants’ ratings, the researchers then built a model that they could use to rate the rest of the transcribed interactions. This “computational linguistic model” wrote Rob Voigt, one of the researchers, in an email, “is fundamentally a computer program we wrote that uses statistics about the human judgments of respect.” This model, he continued, could “automatically estimate the level of respect in any officer utterance.”
The model allowed the researchers to get “estimated ratings for every officer utterance in every interaction in our dataset, which would be prohibitively expensive and time-consuming to get by hand,” Voigt wrote.
The researchers then used their model to analyze the remainder of their data—all 36,738 “usable officer utterances.” They searched the police-motorist interactions for language associated with respect, such as “apologizing, gratitude, and expressions of concern for citizen safety,” they wrote.
In a paper published in June in the Proceedings of the National Academy of Sciences of the United States of America, the researchers posed their central question: “Controlling for contextual factors of the interaction, officers’ language more respectful when speaking to white as opposed to black community members?”
Yes, they concluded. Even when taking into account the motorist’s race, age, and gender; the officer’s race; whether a search was conducted; and whether the stop ended in warning, citation, or arrest, the researchers still found that the interactions of officers with white motorists were more respectful.
“Police officers speak significantly less respectfully to black than to white community members in everyday traffic stops,” they concluded.
Specifically, the researchers wrote, “White community members are 57 percent more likely to hear an officer say one of the most respectful utterances in our dataset, whereas black community members are 61 percent more likely to hear an officer say one of the least respectful utterances in our dataset.”
Interactions also rated as more respectful when they involved officers speaking to older motorists, and when citations were given. But during stops where a search was conducted, officers were rated as less respectful, they concluded.
Stops of black drivers were more likely to involve a search and/or arrest than stops of white drivers. That said, police officers’ interactions with black people “tend to be more fraught,” the researchers wrote, “even when no arrest is made and no use of force occurs.”
The researchers concluded that the officer’s race, the geographic location of the stop, the crime rate in the area of the stop, and the density of businesses in the area of the stop “did not contribute a significant effect.” They also found that the severity of the motorist’s offense couldn’t be used to predict the level of respect an officer would display, they wrote.
Civil rights advocates and people who study policing say this study is useful because having statistical evidence of bias may convince lawmakers and other powerful people that change is needed. “We have to look at policing patterns, where and what kinds of enforcement is happening in communities of color,” said Elisa Della-Piana, legal director for the Lawyers’ Committee for Civil Rights, a group of legal experts working to advance racial, economic, and immigrant justice for oppressed groups throughout the San Francisco Bay Area.
Della-Piana said there should be less policing of communities of color, and that the number of what she called “the low-level interactions,” such as detentions, stops-and-frisks, and traffic stops, especially needs to be reduced because the frequency of those interactions erodes trust between community members and law enforcement.
Before this study, everyday interactions between the police and community members they serve couldn’t be systematically measured, the researchers noted. But as Voight pointed out, “Thousands of hours of body camera footage are collected daily around the country, far more than could ever be analyzed by hand.”
As the researchers note at the beginning of their paper, previously body-camera footage had mostly been studied to help determine whether a police officer involved in a violent encounter— usually with a black person—was guilty of committing a crime.
The Stanford study used footage that’s much more commonplace—that taken during routine traffic stops. “By best estimates, more than one quarter of the public,” over the age of 16, the authors wrote, “comes into contact with the police during the course of a year, most frequently as the result of a police-initiated traffic stop.”
When reached for comment on the study, Oakland Police Department Public Information Officer Johnna Watson said that Oakland officers continue to work with the Stanford researchers, and that she was unable to “accommodate this request at this time.”
Dr. Prince White, policy director at Oakland’s Urban Peace Movement, an Oakland group working to change the community conditions that lead to experiences of violence and mass incarceration for people of color, said the study “rung true” for him. “This is something we already know. It’s something I know, as being a black man,” he said. “I’ve lived in California and been stopped by the police more times than I could ever count.”
The study shows “ that there’s a lot of discretion, and a lot of power, in the hands of the police department, in how they treat the community,” he said.
“Simple interactions, whether or not somebody’s saying ‘sir’ or ‘ma’am’ to you, versus ‘bro’ and ‘dude’—those are different forms of communication,” White said. “And this is what this really verifies. Something that I say quite often, as a communications person, is that, when you open up the space for communication, people put themselves in it.”
“The amount of disrespect that police show to people of color, and communities of color, as compared to white people—it’s long been true that this is a problem, that widespread disrespect and bias is a problem,” said Della-Piana.
Della-Piana said that it’s important to study racial differences in policing because, “it is often true that even when something had been true for a long time, that a study can bring the attention of lawmakers. So, it is one tool, one way to gather the kind of support that’s needed to change long-held practices.”
Della-Piana thinks the study “will likely be cited in potential court cases, for judges to have a better understanding.”
Glaser, the professor of public policy at UC Berkeley, stressed that he hopes the study doesn’t encourage police officers to outwardly be more respectful while still disproportionately stopping and arresting people of color. “We don’t want a situation where cops are just being really friendly and respectful to people, but they’re still discriminating, and causing these disparate outcomes,” he said. “You need both: You need better interactions, and fairer interactions.”
Black people “are subjected to more contact with police than they should be, and as a consequence, they are subjected to more of the negative effects of that,” said Glaser. “It’s very destructive to the social fabric, and it needs to be addressed.”