Police radio transmissions contain personally identifiable information that could pose privacy risks for members of the public, especially Black males, according to a new study by researchers at Penn State and the University of Chicago.
"This study provides a window into police activity as events unfold," said Shomir Wilson, associate professor of information sciences and technology at Penn State and study co-author. "We found that because police radio transmissions disproportionately involve Black suspects, there's a proportionally higher privacy risk for Black people in these communications."
The researchers studied a total of 24 hours of human-transcribed and annotated broadcast police communications transmitted on a single day in three Chicago dispatch zones, or regions used to coordinate police activity. According to U.S. census data, one zone was majority non-Hispanic white, one majority Hispanic and one majority non-Hispanic Black. The team found that broadcast police communications mentioned males nine times more frequently than females and that Black males were most often mentioned of all groups, even in the majority white zone.
The researchers presented their findings at the 27th Association for Computing Machinery Conference on Computer-Supported Cooperative Work and Social Computing on Nov. 9-13 in Costa Rica. The team received a diversity, equity and inclusion recognition from the conference's awards committee.
"The typical police radio transmission is short and serves a coordinating purpose, something like 'Car 54, where are you?'" said Chris Graziul, research assistant professor at the University of Chicago, study co-author and one of two principal investigators leading the project. "These transmissions try to communicate what's happening and describe who's involved. In the process, sensitive information is often disclosed."
The researchers obtained 9,115 transmissions - what they called "utterances" - that occurred when police or dispatch communicated via radio broadcast. They manually transcribed the transmissions and then randomly chose 2,000 utterances from across the three zones to analyze further. They developed a qualitative annotation scheme to label the text. They divided the annotated data into six categories, ranging from event information, such as "residential alarm break in" or "traffic stop," and procedural transmissions, such as the "car 54" example, to casual transmissions like "Morning, squad."
