“Us” and “Them” across 60 Years of New York Times Terrorism Coverage

• In Event: Ingroups and Outgroups in the News

Abstract

Terrorist organizations create dramatic events to attract news coverage that instills strong feelings of threat and panic in a target population. Despite strong academic interest in the strategies and effects of terrorist activity, how these terror events are communicated to target populations remains less well understood. It is especially unclear whether news coverage of terrorist events tends to be presented in ways that might stimulate fear responses, or whether news coverage might instead cue ingroup solidarity as a response to violations of social norms by outgroups.

This paper draws on the lived history of hundreds of thousands of terrorist attacks around the world over 60 years to assess how ingroup and outgroup discourses have evolved in news coverage of terrorist events published in the New York Times. We also leverage known features of terrorist attacks as a natural experiment to examine how ingroups and outgroups are constructed in Times-produced news discourse describing terrorist activities. This allows us not only to analyze the effectiveness of various tactics in generating strategically valuable news coverage for terrorist groups, but more importantly to identify styles of reporting likely to promote pro-social ingroup responses to descriptions of terrorist activities. The ultimate goal of this analysis is to recommend strategies that journalists in democratic societies can use to report on terrorist attacks in ways that meet the informational needs of democratic citizens without advancing the strategic goals of terror groups.

Three unusual data sources provide the empirical foundation for this analysis. First, we deploy a variety of text analytics methods including sentiment analysis, topic modeling, and unsupervised machine learning over the entire population of news articles published in the New York Times from 1945 to 2005, a 60-year span containing approximately 15 million news stories. The Cline Center for Advanced Social Research at the University of Illinois obtained these data in 2007 from ProQuest Historical Newspapers for use in Cline Center research projects. Second, we identify theoretically important features of terrorist attacks using a master list of over one hundred thousand “ground truth” terror events compiled from the Global Terrorism Database (GTD) covering the period 1970 to the present. This allows us to identify news reports about terrorism that are associated with particular attacks so that we can explore how ingroup and outgroup discourses are associated with the severity and lethality of terrorist events. Third, we developed a master list of violent non-state actors that builds upon the open-access Terrorist Organization (TORG) Cross-Walk (Asal et al. 2015) with an expanded set of actors derived from the Cline Center’s SPEED project (Nardulli et al. 2015). We use this master list of violent non-state actors to analyze features of terrorist groups that might be related to news construction of outgroup identities.

This paper is the second of several planned analyses undertaken by the Cline Center for Advanced Social Research as part of an NEH-funded “Digging into Data Trans-Atlantic Partnership” grant project with collaborators from Mannheim University and the Free University of Amsterdam (see https://diggingintodata.org/awards/2016/project/responsible-terrorism-coverage-resteco-global-comparative-analysis-news-coverage). The paper’s contribution is novel in two important respects. First, while the extensive literature on media and terrorism contains a wide array of empirical research focusing on single types of events or single organizations, we find that very little of that research has been explicitly comparative across historical periods, organizations, and types of events, in part because this literature has rarely used computational methods for content analysis that would be required to produce historical comparisons over long spans of time. Second, we develop an automated method of analyzing ingroup and outgroup discourses associated with terrorist acts. This approach allows us to compare evolving discourses around the term “terrorism” as well as evolving coverage of terrorism attacks, and how group identities cued within that coverage vary as a function of the type of terror threat (e.g., domestic or non-domestic), lethality of terror events, and identity of perpetrators.

Authors

• Scott L. Althaus, University of Illinois at Urbana-Champaign

• Dan A. Shalmon, University of Illinois at Urbana-Champaign