Hard-Right Social Media and Civil Unrest

Abstract

Does the use of right-wing social media translate into right-wing protest and violence? Recent research linking social media activity to political attitudes, participation in large, isolated demonstrations, and the occurrence of hate crimes suggests a relationship between right-wing social media activity and subsequent contentious events and violence. However, analyzing this relationship at countrywide scales over many months has not been possible in the absence of reliable georeferenced social media and corresponding contentious politics data. We create a spatial panel dataset measuring 57,505 right-wing social media posts and 1,765 incidents of right-wing unrest in all United States counties from January 2020 through January 2021. Using spatial regression analysis, we find that right-wing social media activity in a given month increased the frequency of right-wing unrest events in the following month. A content analysis of 1.5 million Parler comments reveals that discussions of political events were relatively rare, suggesting that while right-wing social media can be used to coordinate political unrest, this may not fully explain the positive relationship between online and offline activity. Instead, a subsequent analysis of politically moderate and conservative counties indicates that the main effect we observe is driven by shifting perceptions of norms and not changes in individuals’ beliefs or preferences. Our results help explain how social media affect political unrest beyond the conventional coordination mechanism and inform an emerging literature on social media, political attitudes and action, and violence.

How Symbols Influence Social Media Discourse: An Embedding Regression Analysis of Trump’s Return to Twitter

“…This difference may be due to the analysis of different kinds of symbols; unlike Trump’s reinstatement, the symbols of eliteness explicitly convey a notion of relative social status. Moreover, Karell et al (2023) argued that the symbols’ effect is due to how it interacts with the content’s lexical dimension. This suggests a related opportunity for future work: after distinguishing content’s lexical and symbolic properties, and perhaps variants of these properties, researchers can theorize and analyze their interactions.…”

“…For example, in their study on religious figures’ endorsement of tweets promoting intercommunal tolerance in Lebanon, Siegel and Badaan (2020) argued that although endorsements may be written, their influence depends on invoking, or symbolizing, their authors’ religious standing. Karell et al (2023) provide an example of a nonwritten symbol by highlighting how accounts’ markers of status—in this case, a gold badge conveying elite status on Parler—can influence other users’ interpretation of these accounts’ posts.…”

“…Building on three earlier discussed literatures—the work indicating that online content’s symbolic properties play some role in shaping users’ discourse (e.g., Anspach 2021; Karell et al 2023; Siegel and Badaan 2020), the research describing Trump’s significance as a politicized symbol (e.g., Brooks and Harmon 2022; Graham et al 2021), and the research emphasizing politicized symbols’ ability to cue social and political boundaries (e.g., Scoville et al 2022; Smirnova 2018)—we construct our first hypothesis.…”

Offline events and online hate

“…Second, the effects of an offline event on online hate speech may be difficult to predict, and may quickly evolve and broaden, which further suggests a broad and rapid mitigation effort in response. Because online hate speech levels are predictive of offline hate crimes [4][5][6][7][8][9], the potential exists for an offline trigger event to be followed by a broad cascade of online hate speech that, in turn, precedes hate crimes of a nature unrelated to the initial event. Third, because the effects of an offline event on different types of platforms are often similar, content moderators on mainstream platform could advance their efforts by systematically observing ongoing developments on other, fringe platforms.…”

“…In turn, highly charged offline trigger events tend to be followed by sharp increases in online hate speech [2,3]. Such spikes in online hate speech have, further, been shown to predict similar spikes in offline violent hate crimes [4][5][6][7][8][9].…”