Bad Data

In 2014, the death of Michael Brown (an eighteen-year-old unarmed African American man killed by a police officer in Ferguson, Missouri) reignited widespread concerns about racial profiling and unnecessary use of force by police. The incident also drew public attention to the fact that at the time, no one knew the number of people killed by police across the United States each year. Motivated and concerned by these events, our research team began an exploratory project to determine the state of data on police homicides in Los Angeles County. We began by accessing what we could find through data.gov and other federal and local agency websites, issuing Freedom Of Information Act (FOIA) requests to local and regional law enforcement agencies (requests that went unanswered), and setting to work downloading and parsing through .csv (comma-separated value) files from federal and local agencies in the relatively few cases where these were available.

Unsurprisingly, we found ongoing, pervasive, and complex issues of power related to the U.S. government’s data collection and transparency efforts. The FBI’s Supplementary Homicide Report, for example, does not require every jurisdiction or even every state to report police homicides, and those that do comply do so on their own accord without any outside corroboration of their data. The deceased are criminalized in their very designations, as the data only includes homicides ruled as justified. As we were conducting research and observing these trends in the data, we remained attentive to the stance of many members of the communities affected by police violence, who often question why their own stories are not enough, and why, for example, independent organizations should need to verify the antagonism of their ongoing experiences with law enforcement. In response, we decided to open this data to the public to try to collectively make sense of it.

Over fifty community members—from grassroots organizers and activists, to interested professors and students from institutions across Los Angeles, to private citizens—answered our call to collaboratively investigate the data. During our preliminary discussions, we questioned whether the datasets were counting the same phenomenon. Exactly what constitutes a police officer–involved homicide (POIH) is not simple or clear-cut, and enforcement agencies cannot effectively turn to the law to determine what defines this complex and often politically fraught category of event.

As the group began downloading and parsing through dirty .csv files provided by federal and local organizations—ranging from the FBI and the Department of Justice to local activist groups like the Youth Justice Coalition—we discovered that rather than serving as conclusive evidence, data can be the starting point to ask new questions that come to light through reinterpretations and creative practice. Participant reflections on events pointed to overarching issues of biased and uncorroborated data collection, as well as the mysterious processes through which events are categorized in datasets. Participants developed projects to counter these problematic practices with the data at hand, through processes of collecting more data or visualizing it in new ways in order to illustrate the toll of a life taken or to reveal gaps in datasets. Concerns over anticipatory governance (see Kitchin 2014)—the idea that data collection in the name of fostering accountability can lead to data use with different or more nefarious ends—surfaced and were a theme throughout the project. This project and related findings demonstrate how collaborative, critical investigation of POIH data can lead to an array of outcomes in the areas of analytical, qualitative, and visualization work.

In October of 2016, FBI director James Comey spoke to the International Chiefs of Police Conference in San Diego, California, declaring that despite recent national discussions on police brutality, “Americans actually have no idea” how often police use unnecessary force. He went on to note “a narrative that has formed, in the absence of good information and in the absence of actual data, and it is this: Biased police are killing black men at epidemic rates.” Yet an unceasing stream of videos that document fatal encounters with police, from the case of Eric Garner in New York in 2014 to that of Philando Castile in 2016, proves otherwise.

Comey’s statement came three days after the Department of Justice announced it would move forward with plans to collect improved data regarding deadly and unnecessary use of force in policing. Relatedly, in May 2015 then-President Barack Obama issued the White House Open Police Data Initiative, which grew to encompass over eighty organizations nationwide working to promote local data collection and dissemination with the aim of holding law enforcement accountable.

As data-driven solutions (such as those proposed by Comey in October 2016 and the White House Open Police Initiative) quickly migrate from the practices of science and business to be applied to complex issues of everyday life, it has become increasingly clear that data are used to determine all types of implicit and explicit political, cultural, and intellectual forms of governance.

These data-driven governance practices are fueled by the widely held belief that data objectively encapsulate and portray complex events, identities, or phenomena. While data might help offer a broader vantage in many cases, data should be understood first and foremost as a tool of expression imbued with power and influence. As data exudes increasing influence over more aspects of life, we offer our intervention as an example of new ways to interrogate data within communities of stakeholders to bring about meaningful change.

Reference 

Kitchin, Rob. 2014. The Data Revolution: Big Data, Open Data, Data Infrastructures and Their Consequences. Thousand Oaks, Calif.: SAGE.