In a new study currently under review for publication, our team evaluates the conditions under which news media describe a violent event as terrorism. We hypothesize that a perpetrator’s race, religion, or other racialized characteristics, which should have no impact on whether an event is considered terrorism, shape whether media characterize or associate an event as terrorism. We develop an original corpus of news coverage of U.S. mass shootings, with more than 4,000 news articles, to test our hypotheses against other factors associated with media’s terrorism designations.
We use statistical regression, qualitative analysis, and unsupervised natural language processing to evaluate relationships between mass shooters’ characteristics and terrorism designations. We find that racialized perpetrators, i.e., specifically those understood as Muslim or Middle Eastern, are more likely to be considered terrorists and covered unsympathetically, relative to “white American” perpetrators. With mass shootings on the rise and media treatment of terrorism shaping event data and policymaker perceptions, understanding racial biases in media coverage is crucial to urgently identify and immediately rectify.
In this work, we compared different existing definitions of terrorism and analyze terrorism’s social meaning over time and use those definitions to develop a theoretical framework for how racial heuristics might contaminate media, official, and academic designations of terrorism. In addition, we conducted extensive data collection and hand-coding of U.S. mass shooting events for use to develop an approach for applying unsupervised natural language processing tools to model “personas” among mass shooters. Finally, we advance knowledge about U.S. mass shooters’ diverse racial identities, motivations, and designations as terrorists. To our knowledge, we are the first to systematically evaluate the relationships between racial heuristics and media’s terrorism designations.
This research has an important limitation. We find a small universe of positive cases for our outcome variable; only 8.5% of mass shootings in our data qualify as terrorism. The same is true of our explanatory variable; “racialized foreigners” composed only 2-10% of shooters, where most mass shooters are white American men. Nevertheless, evidence from our statistical and qualitative analyses suggest that Muslim and/or Middle Eastern perpetrators are more likely than other perpetrators to receive designations of terrorism. Our unsupervised narrative analysis additionally suggests that media treatment of white perpetrators is more likely to contain sympathetic elements, while non-white perpetrators appear to receive more unambiguously negative frames.
Our research team includes:
Dallas Card, PhD student in the Machine Learning Department at Carnegie Mellon University
Sarah K. Dreier, NSF Interdisciplinary Postdoctoral Fellow at University of Washington’s Department of Political Science
Noah A. Smith, Senior Data Science Fellow at the eScience Institute, Professor of Computer Science & Engineering at the University of Washington, Adjunct in Linguistics, Affiliate of the Center for Statistics and the Social Sciences (website)