Contagious comments? How the pandemic has affected populist user comments in seven Western-European countries.

Covid-19 and the government measures taken to combat the pandemic have fueled populist protests throughout Europe. Social media played a key role in the emergence of these protests. This study argues that the topic of Covid-19 has triggered populist user comments on Facebook pages of mass media in seven Western European countries (AT, DE, FR, IT, NL, UK, SE). It asks how reporting about the Covid-19 crisis, and real-world aspects of managing the crisis have affected the scope of populist commenting on Facebook. Drawing on media psychology, this paper theorizes populist comments as an expression of ‘reactance’, sparked by repeated ‘fear appeals’ in posts about Covid-19. Since reactance results from restrictions of personal freedoms, this paper expects an interaction effect of reporting about the government measures and the actual stringency of the measures, which varied considerably in the selected countries. The paper derives several hypotheses from these claims and test them on a dataset of N=65,343 Facebook posts, posted between January 2020 and June 2021 on 21 Facebook pages of mass media from in seven Western European countries (AT, DE, FR, IT, NL, UK, SE), and about 3.5 million corresponding user comments. Automated content analysis methods were used to measure populist communication in user comments and message characteristics of the posts. For measuring populism, the comments were first machine-translated into German using the Google Translate API to allow comparability of the measurements. Secondly, a fasttext word vector model was trained on the full corpus of German-speaking text used in this study. Finally, this study implemented a method called Distributed Dictionary Representations (Garten et al. 2018) which augments short conceptual dictionaries with word vector information to obtain a gradual measurement of populist communication. To classify the posts along several categories, such as mentioning the topic Covid-19, government measures, or public health experts, this study developed and applied language-specific dictionaries. All measurements were carefully validated against a human-coded gold standard of 300 comments and 300 posts per language, coded by thoroughly trained native speakers. The paper uses real world data provided by the Oxford government response tracker. Hypotheses are tested using carefully modeled multilevel regression models. The paper discusses implications of the findings for crisis communication against the worrisome background of growing anti-vaxxer and covid-sceptic movements.