In 2014, President Obama launched the Taskforce on 21st Century Policing in an effort to build trust between communities and police departments in the wake of the police murders of Michael Brown, Eric Garner, and Tamir Rice and the ensuing uprisings against police violence. On the recommendation of this task force, the Police Data Initiative was created to gather data from police departments around the country and make it publicly available as a means of increasing transparency and therefore trust. The Police Data Initiative's online repository includes datasets from 120 departments around the country.
Open/Opaque focuses on the use of force datasets available through the Police Data Initiative. A separate set of datasets covers officer involved shootings. I decided to focus on use of force rather than shootings, since the latter has been the focus of other very thorough digital humanities projects, such as the Washington Post's police shooting database. Other projects such as Mapping Police Violence and Fatal Encounters visualize police killings. There are not similarly high-profile projects that look at the much more mundane, but still serious and potentially fatal, force that officers exercise on a daily basis. And while resources on police killings can look to other sources to supplement incomplete or unavailable police data, non-fatal police use of force usually doesn't attract media coverage, making data collected and shared by the police the only compendium of this information.
Of the 17 datasets on police use of force available on the Police Data Initiative site, I selected those from larger police departments and with more complete data. For example, I decided not to work with the Seattle Police Department's dataset because it included so few data points for each incident. After creating visualizations for each of the selected datasets (see my section on methods below), I eliminated another dataset because the structure of the data was too different from the other datasets and I couldn't create comparable visualizations with it. The remaining use of force datasets that are represented by this project are: Cincinnati Police Department, Dallas Police Department, Indianapolis Metropolitan Police Department, Portland Police Bureau.
I used OpenRefine to clean my data, Tableau to visualize it, and Mobirise to construct this website.
Some datasets required more cleaning than others. In the case of the Dallas Police Department data, for example, the data was broken up by year into separate datasets, so I first combined the data into one document following the instructions here (thanks to Professor Posner for the help!) Otherwise, I tried to keep my cleaning of the data to a minimum, just enough to clean up any errors or inconsistencies.
Visualizing the data in Tableau helped me understand what insights the data offered. One of the biggest obstacles in doing this project was the inconsistency among the datasets (discussed more on the Findings page). What appeared to be an obstacle ended up being one of the project's most significant findings, however. I ended up leaning into the inconsistencies and trying to demonstrate them through the visualizations. I did try to create consistency within the project by using similar labeling for all visualizations, but for the most part I did not edit the language used in the actual data. In some cases, I was not even able to translate the police-jargon for myself. My hope is that, where viewers encounter opacity in the data, it will show the underlying flaws in the data itself and not in my visualizations of it!
I am a PhD student in Information Studies at UCLA. My research focuses on if, and how, communities can use archives and records to get accountability for state violence. This is my first digital humanities project, thank you for checking it out!
The web page was created with Mobirise