In an unprecedented, data-driven analysis of some 1,000 U.S. court decisions, Reuters revealed how a legal doctrine called qualified immunity insulates police from having to compensate victims of brutality. A 50-year-old creation of the Supreme Court, the legal defense is meant to protect government employees from frivolous litigation, but Reuters found it shields cops even when judges think they used force that could be considered excessive. The Reuters analysis of some 1,000 court decisions also found that the doctrine is applied unevenly, depending on where the victims live and that the obstacles to justice are especially pronounced for Black citizens.
The series had political and legal impact in the months after it published as the United States grappled with widespread protest over police tactics. U.S. Reps. Ayanna Pressley and Justin Amash introduced the “Ending Qualified Immunity Act” in June. In a press release announcing the new bill, Pressley cited our findings.
Our work was cited in the first congressional hearing that examined racial injustice and police brutality following George Floyd’s death. Sherillyn Ifill, president and director-counsel of the NAACP Legal Defense and Educational Fund highlighted Reuters’ investigative findings in written testimony.
Our findings have been cited in multiple petitions at the Supreme Court, a Supreme Court amicus brief, an appeal before the 10th circuit, three university law journals, and a legal textbook.
Our work also was cited by dozens of news organizations in the U.S. and elsewhere, including the Australian Associated Press, BBC News, and France’s Le Monde. The New York Times newsroom, opinion page and Editorial Board all separately highlighted our findings.
John Oliver referenced the series in a segment about police reform. SCOTUSBlog, a well-read site that probes the inner workings of the Supreme Court, profiled our reporters and their work. LINK And the Washington Post interviewed one of our reporters for a video explainer about qualified immunity.
We compiled three databases, to understand how qualified immunity plays out in all three levels of the U.S. judicial system – district courts, appellate courts and the Supreme Court — ultimately recording facts about more than 1,000 judicial decisions.
For each judicial decision in our database, reporters recorded dozens of facts about the case, such as whether the civilian was armed, the type of force officers used, and whether the civilian was injured or died in the encounter with police. We also recorded whether the court granted qualified immunity and some specifics about that decision.
Our reporters used R and SQL Server to analyze the data and tested all data findings against a binomial logistic regression model to confirm that our findings were statistically significant, robust, and could not be explained away by variations in other details about each case. For example, the increase in qualified immunity grants over time was significant and could not be attributed to other differences, such as which circuits heard the cases.
Our visual editors dived into the data, as well, sorting it in ways that enabled us to explain the courts’ rulings. To create a process diagram, we used R to categorize the information into tranches and a custom drawing algorithm to generate the Sankey lines. Svelte was used for the scrollytelling interface, D3 to make visualizations in canvas.
We also used police video as data, editing lengthy camera footage down to significant seconds in time. Those were integrated into the story to make fluid pauses, allowing readers a window into the moments when police brutality was captured on tape. We used color to link videos and data. We also used Datawrapper, Illustrator and AI2HTML.
What was the hardest part of this project?
The primary obstacle was the lack of structured data that could reveal insights into how qualified immunity plays out in the court system. We had to build our own by reviewing more than 1,000 lawsuits claiming excessive force by police.
While we were able to reasonably identify relevant lawsuits at the appellate level, there was no standard way to identify cases at the district level. We received raw docket listings from Westlaw, another Thomson Reuters company. That data, with more than 90 million entries, has no field or code that signifies a case involved qualified immunity. Because of that, our data reporters went through a tedious process of identifying potential cases and building queries that would flag likely suspects.
From there, the team read and reviewed thousands of cases to whittle our universe to hundreds of relevant opinions. Because court documents are unstructured, we had to create our own structured database, where we coded dozens of elements for each case to allow for quantitative analysis. This process was extremely time consuming and difficult, as individual lawyers and judges structure arguments and opinions differently. Relevant information was often buried deep on the docket, requiring reporters to read multiple documents and dig through exhibits for each case.
After analyzing the data we amassed, we then had to explain the workings of a complex legal doctrine to readers. By working closely with our visual editors, we were able to develop an engaging, easy to follow animation that accomplished that without the need for dense text.
What can others learn from this project?
Our project offers several important take-aways for other journalists:
1. If people are bemoaning a lack of data, that can be a great opportunity for data journalists. This project got started when Supreme Court Sonia Sotomayor, in a written dissent, faulted her colleagues for overwhelmingly favoring police petitions for immunity in excessive force lawsuits. Justice Samuel Alito shot back that Sotomayor had no data supporting her view. That’s where we stepped in.
2. Don’t be dissuaded by a lack of structured data. It’s a lot of hard work, but sometimes unstructured data – court opinions in this case – can be reviewed, categorized and turned into tables for analysis.
3. Think creatively about presenting complex concepts and findings. It would have been very hard for readers to grasp ours if we had not produced our animation that stepped through the process judges follow in deciding cases and showed how the Supreme Court’s rulings resulted in cops winning immunity more often.