Skip to main content Skip to secondary navigation
Main content start

Stanford journalists use data to understand racial bias in policing

The interdisciplinary Stanford Open Policing Project has found Black and Hispanic drivers are stopped and searched at higher rates across the U.S.

When law enforcement in Nashville wanted to better understand their traffic stop practices, they turned to a Stanford team to help understand the city's practices and help propose a way forward. The Stanford journalists and researchers found law enforcement in Nashville were disproportionately making traffic stops that impacted Black drivers.

Nashville police had also been making nearly seven times more stops per capita relative to the national average, many times for equipment violations like a broken tail light. After the Stanford analysis was released, Nashville police eventually modified its practices to be more equitable and efficient, reducing stops by 75 percent the next year.

That real-world change is one of numerous impacts the Stanford Open Policing Project has had across the United States. For more than seven years now, the interdisciplinary project has brought together data journalists and social scientists in the Stanford Journalism Program, Big Local News and Stanford Computational Policy Lab to collect, standardize and analyze traffic and pedestrian stop data from law enforcement agencies across the country. There is currently no comprehensive, national repository for this information, except for efforts like this.


View of map with filtered view of city traffic stops
The distribution of Nashville’s residential driving-age population (left) and aggregated traffic stops between 2010 and 2016 (right). Stanford researchers and journalists found Nashville police had been making considerably more traffic stops per capita than the national average, with stops disproportionately involving Black drivers. (Image: Stanford Open Policing Project)

The traffic stop is the most common interaction police have with the American public, and in encounters gone amiss, has resulted in high-profile deaths including Patrick Lyoya, Philando Castile and Sandra Bland — all of whom were Black. The Stanford team harnessed the rigor of statistical analysis with the power of data journalism, consistently finding across datasets that Black and Hispanic drivers are stopped and searched at higher rates than any other group of Americans.

This initiative has gathered over 200 million records from state and local police departments, and it has standardized and made the state- and city-level data publicly available for journalists and researchers.

“The more transparency we have, the better, so people can feel like they’re informed,” said Cheryl Phillips, director of Big Local News and Lorry I. Lokey Visiting Professor in Professional Journalism. “Hopefully police agencies will see the impact of these micro-interactions with the public.”

This project was the first for Phillips’ Big Local News initiative at Stanford. It demonstrated the utility of collecting and cleaning local datasets for journalists, which in turn lowers the costs of discovering and telling accountability stories.

In early 2015, Phillips teamed up with Sharad Goel, then an assistant professor in the Stanford School of Engineering. The goal was to collaborate utilizing the strengths of their expertise: Phillips and her student journalists could file public records requests and report on the data, and Goel and his student researchers could standardize and analyze the data at scale.

“What got me much more involved in these policy issues is seeing the type of impact you can have using data and statistics to improve outcomes through journalism, and just through making this information more broadly accessible,” said Goel, who is now a professor of public policy at Harvard Kennedy School.

In 2016, the Open Policing Project received $310,000 in initial grant funding from the Knight Foundation. It allowed the team to boost efforts to request records and standardize the data, which sometimes was messy and varied in depth from jurisdiction to jurisdiction. The public records requests also varied in effort — some agencies quickly provided the data, while others took months of negotiations to provide.

By 2017, Phillips and Goel, along with students, started publishing the results both in journalistic articles and academic papers that showed racial disparities across the country. They had assembled more than 60 million police reports from 20 states, publishing the cleaned data for journalists on a public database

The team then trained over 200 journalists on how to use the data and localize stories for their own communities at the Investigative Reporters and Editors conference in Phoenix and at a Poynter seminar in Chicago. Local journalists have since used the data in their own news stories. In Washington state, for example, Investigate West reported on Washington State Patrol’s stops near Indigenous reservations. The reporting spurred the state to resume funding analysis of the data, which had been on hiatus since around the Great Recession.

Phillips and Goel later teamed up to teach a “Law, Order and Algorithms” course at Stanford that used the traffic stop data and helped students better understand the impacts of public policy.

“Any interdisciplinary collaboration has challenges of navigating the norms of different disciplines,” Goel said. “At the same time, you’re really learning … the approaches that different people are bringing to different problems.”

The traffic stops database has only grown over the years. By 2019, there were nearly 100 million records, generating media coverage from major outlets like The Economist, NBC News, HuffPost and CNN. In 2020, the team published a paper in Nature Human Behaviour that discussed the nationwide analysis of racial disparities in policing.

The project has also found that Black motorists are less likely to be stopped after sunset, when “a veil of darkness” masks their race — findings covered in the Los Angeles Times, among other publications.

“The more we can look at the patterns of what’s happening with police use of force, police conduct, it just helps us evaluate it at a systemic level, so that changes can happen [for] interactions with people,” said Phillips, who is now also faculty director of the Stanford Computational Policy Lab.

The project continues to live on via Phillips’ Big Local News work and Goel’s research on public policy at Harvard. At Stanford, the Open Policing Project data is also being utilized as part of the Community Law Enforcement Accountability Network (CLEAN), a partnership between journalists, lawyers, computer engineers and academics to open and share police data.

“All across Stanford, researchers are working to transform unstructured information into structured data that tell you how institutions are operating,” said Jay Hamilton, Hearst Professor of Communication, Communication Department Chair and Stanford Journalism Program Director. “The Stanford Open Policing Project demonstrates that when you combine the best of journalism and data science, you can generate stellar reporting, outstanding research and policy change.”

Editor’s note: The author of this article, Vignesh Ramachandran, is a former contributor to the Stanford Open Policing Project.