Racial disparities in interactions with law enforcement are of critical concern. Yet data limitations hinder nuanced understanding of the role race plays in these encounters. This research note addresses the conceptual and methodological challenges of measuring race in the context of law enforcement interaction, with a special focus on use of lethal force, emphasizing the importance of ascribed race—how individuals are perceived racially by others—in shaping outcomes. We evaluate multiple approaches to ascribing race to civilians involved in lethal force incidents, including human coding by student analysts and police cadets, and algorithmic classification using name- and image-based models. We find that combining human judgments with algorithmic tools yields high concordance with administrative race data for a five-category classification, though lower concordance was found for Native Americans. A three-category approach improved reliability but sacrificed granularity. Based on these findings, we propose a hybrid classification system incorporating coder agreement and algorithmic scores. This approach offers a transparent and substantively relevant method for researchers examining racial disparities in policing and beyond.
Jennings et al. (Wed,) studied this question.