There’s A Racial Bias on Police Facebook Pages
There’s A Racial Bias on Police Facebook Pages
There was nothing overtly biased about the way the Wilkes-Barre Township Police Department described a mugging on its Facebook page in February 2019. The first post simply described a Black suspect who was alleged to have threatened a victim with a gun and demanded cash in this small community in northeastern Pennsylvania. Two later Facebook posts about the case congratulated the police on catching the suspect.
But two years before, when a white man had robbed a gas station at gunpoint and fled the scene, the police department’s social media response was completely different. There was no mention of the case on social media at all, according to John Rappaport, a professor of law at the University of Chicago who is part of a team studying racial bias in law enforcement social media accounts. Not before the suspect was arrested, to warn the public and seek their help in an arrest. And not after, to reassure the community that the suspect had been caught. “The crimes are quite similar,” Rappaport said. “[It undermines] any notion that crime severity is straightforwardly driving the department’s posting decisions.”
This is just one example of a larger pattern of bias that Rappaport’s team found when they analyzed nearly 14,000 Facebook pages maintained by law enforcement agencies across the United States. They found that police Facebook pages consistently overreport crimes by Black suspects relative to local arrests rates: Between 2010 and 2019, Black suspects were described in 32 percent of posts but represented just 20 percent of arrestees. It mirrors statistics that show white Americans overestimate the percentage of crimes committed by Black Americans by as much 20 to 30 percent compared to the actual figures (numbers that, themselves, already reflect a bias in who gets arrested versus who actually commits crimes).
And scientists say it’s reasonable to suspect those two sets of statistics are connected to one another. “We really framed the paper as being less about ‘are police departments behaving well or badly,’ and more about the perspective of the reader,” Rappaport said. That’s because these biased accounts are likely part of feedback loops, reflecting bigger issues in society as police both respond to — and perpetuate — the myths white Americans already believe.
Wilkes-Barre isn’t uniquely problematic, Rappaport said. And not all of the law enforcement agencies his team looked at engaged in biased posting. But the totality of the data showed clear patterns that extended nationwide. Only a few areas didn’t overrepresent Black suspects, relative to actual arrests, including part of the Black Belt region of the South, where Black people make up the majority of the overall population.
The racial disparity in posts compared to arrests differed by type of crime, but was present across a variety of serious offenses. Car theft, for example, had the smallest disparity: There was less than a percentage point of difference between the percent of local auto thefts involving Black suspects and the percent of Facebook posts about auto thefts involving Black suspects. But the differences were much larger with other crimes. While Black suspects made up 22 percent of all theft arrests, 32 percent of Facebook posts about thefts involved Black suspects.<a class="espn-footnote-link" data-footnote-id="1" href="https://fivethirtyeight.com/features/police-facebook-posts-racial-bias/#fn-1" data-footnote-content="
These findings are based on a sample of over 70,000 Facebook posts from police agencies that mentioned a selection of crimes and the race of the suspect(s); arrest data from the FBI’s Uniform Crime Report; and offense-level and offender demographic data from the National Incident-Based Reporting System.
Overall, Black people’s involvement in violent crimes was being overreported by law enforcement Facebook pages by 11 percentage points and involvement in property crime was being overreported by 8 percentage points.
These differences may seem small, but Rappaport and outside researchers said the impacts of being exposed to these disparities can be wide reaching. I spoke with three other scientists, unaffiliated with Rappaport’s research, who also study American beliefs about race and crime. They all told me this paper is representative of larger, systemic issues with how race, crime and punishment are viewed in this country.
At a time when relationships between traditional media, like newspapers, and police have become strained, social media allows law enforcement to regain more control over narratives of crime, said Sarah Britto, a professor of criminal justice administration at California State University, Dominguez Hills. In decades past, researchers found evidence of traditional media overrepresenting Black people as perpetrators of crime and under-representing crimes committed by white suspects. That’s changed — newer research suggests Black Americans are now underrepresented as both suspects and victims of crime in cable and network news.
But back when researchers were finding clear bias in traditional media, research also suggested that portraying Black people as criminals had an impact on how viewers thought about criminals and criminality. For example, a study in the late 1990s exposed Los Angelenos to a news report in which an alleged perpetrator was identified as Black, white or without identifying information. It found that, when the perpetrator was Black, white viewers’ support for punitive laws increased by 6 percent — while that support only increased by 1 percent when the perpetrator in the news story was white.<a class="espn-footnote-link" data-footnote-id="2" href="https://fivethirtyeight.com/features/police-facebook-posts-racial-bias/#fn-2" data-footnote-content="
White and Black respondents were limited to those who identified as white and Black alone, respectively; Hispanic respondents were a separate group.
But what the public already believes about race and crime could also be shaping what police post. Rappaport’s study opens up a whole new direction in research, said Tony Cheng, a professor of criminology at the University of California, Irvine. One of the things Cheng said he’d like to see studied in the future is the motivations and practices within police departments that create racial disparities in social media posts. He suspects that the nature of social media incentivizes police to seek traffic and “likes” as much as any other group or individual who is trying to build an audience. If a post produces a lot of engagement, the posters are likely to try to repeat the success with similar content. But that becomes a problem if the most popular posts are all about Black people committing crimes, Cheng said.
The irony here is that public communication through social media channels is often lauded as part of the best practices that improve transparency in policing, he told me. “[This shows us that] those very practices can be exacerbating racial biases in public information and crime information in ways that we wouldn’t even really think about,” Cheng said.
And those biases are powerful, said Nicholas Valentino, a professor of political science at the University of Michigan. There’s lots of research showing that, the more Americans perceive poor people to be overwhelmingly Black, the less support they have for social welfare policies aimed at helping the poor, he said. It makes sense that portraying Black people as more likely to be arrested for a crime than they actually are would have a similar impact on how Americans view crime policy.
“That’s not a controversial thing to say,” Valentino said. “What’s interesting here is that this is neither new, nor even unique to this domain of political communication. It’s widespread, and we’ve known about it for maybe 30 or 40 years.”