Supplementary material: Document similarity analysis

These files contain supplementary data generated in response to peer review. Peer reviewers requested that we compare online and other kinds of journalistic content, to see if there were any differences in the extent to which digital media related to the emergency imaginary. K-means clustering indicated that there seemed to be a closer relationship between digital media and the emergency imaginary, so we sought to explore why this might be. One hypothesis, drawn from previous research, was that "emergency" reporting was informed by a group of major 'Western' wire agencies, known for their "on the spot" reporting - Agence France Presse, Associated Press, and Thomson Reuters. However, since that research was conducted, other non-Western agencies, such as Xinhua and Interfax, have grown considerably. So, we divided our original corpus of news texts into digital and non-digital content per country, and conducted document similarity tests between these corpora and the copy produced by a selection of wire agencies. The file “similarity_analysis.xlsx” shows the results of that analysis. The variable “to.Vprop” is the measure of document similarity, as indicated by the RNewsflow package. It is the percentage of media items in our corpus from a given country/type of media that were very similar or identical to previously published copy from one of our selected wire agencies. In the written article, we mapped out which wire agencies were dominant in the media content disseminated by outlets in specific countries. In file “all_similarities.xlsx” we present the results of the same analysis for all news sources in the sub-corpus. It includes 245 sources from 20 countries. For each news source, the variable “to.Vprop” measures the percentage of articles in the “from” column that are identical or highly similar to those in the column “to.” Finally, the file “sources_corpus.xlsx” contains details of the sources we used for this analysis, which country they were from, how we classified them (digital/not digital) and how many media items were included in this analysis.