ABSTRACT
Terrorism, poverty, cancer: Citizens fear many things. Some of these fears are fueled by fear-related news as a form of sensational media coverage. As research on the variety of fears depicted in the news is scarce, this study analyzes the construction of fear-related news in the US and the UK from 1990 to 2017. By combining unsupervised machine learning in the form of topic modeling (N = 15,487) and manual content analysis (N = 1013), it explores the prevalence of themes and topics. It also analyzes the fear-inducing presentation of news through the use of fear appeals, specifically which (severe) threats are emphasized by the media. Results indicate that the media do not only concentrate on fears of violence and crime, but also on fears of economic downturn, political unrest, or social fears concerning unemployment. The most prominent threat emphasized across topics is death, followed by political and economic threats. Topic prevalence is highly volatile and coverage has not become more fear-inducing over time. Overall, this study contributes to a better understanding of how news media may foster fear: through mirroring a variety of economic, political and social fears and emphasizing specific threats across them.
Acknowledgements
The authors would like to thank three anonymous reviewers for their very helpful comments as well as Mike S. Schäfer and Korinna Lindemann for their valuable input to this paper.
Disclosure Statement
No potential conflict of interest was reported by the author(s).
Notes
1 Our codebook can be found in the Supplementary Material (Document A2). We originally differentiated between minor and major severity of a threat but later combined both into one measure of severity. To understand threats better, we also coded the presence of the source of the threat (entity acting out the threat) and the victim of the threat (entity being victimized) using closed and open categories. Both had satisfying reliability values (λ = .89 and λ = .90, respectively) but are not reported as the papers focuses on topics, themes and threats.
2 In the results section, we report whether we find a monotonic increase in percent of articles mainly dealing with, for example, “Immigration” over time. Since topic modeling allows for multiple topics to occur in articles, we verified the robustness of our results by comparing them to the same test based on a modified θ-matrix describing the conditional probability with which a topic K is likely to occur in a given year across all articles and report robust results.