Have heads cooled? Changes in radical partisanship from 2020–2022

Abstract

ABSTRACTRecent polling shows significant levels of radical partisanship among the US public in the form of violent partisan schadenfreude and violent partisan attitudes. Previous research shows that the partisan intensity underlying these sentiments increases around presidential elections and when political actors use violent rhetoric. Has lethal mass partisanship declined in the years after the 2020 election, especially now that one of the main purveyors of violent rhetoric has left office? This paper uses panel data from the Understanding America Study (UAS) to show how levels of radical partisanship have changed from the post-election period of 2020 to the same time period in 2021 and 2022. The results demonstrate that rates of violent partisan attitudes have declined in 2021 and 2022 compared to 2020, especially amongst those who were most supportive of former President Donald Trump. However, changes in partisan schadenfreude have no discernable pattern. The results demonstrate both that radical partisanship is not on a uniformly upward trajectory but likely decreases and increases with changes in political context, and that elements of radical partisanship do not vary in parallel. Disclosure statementNo potential conflict of interest was reported by the author(s).Notes1 Kalmoe and Mason (Citation2019) originally termed the attitudes that make up the focus of this paper as “lethal mass partisanship.” Kalmoe and Mason (Citation2022) have since referred to these attitudes as “radical partisanship.” For the sake of clarity, I use their more recent terminology.2 The UAS received all necessary ethics approvals (approval number: BRANY IRB # 22-065-1044).3 Exact question wording and coding details can be found in the appendix.4 All schadenfreude and violence question items were only asked of those who had a Republican or Democrat partisan affiliation.5 See appendix for demographics.6 Statistical significance was ascertained using a weighted difference in means test.7 These results do not change substantially when including all participants.8 While variables such as race and gender may be seen as time invariant, data for these variables is collected on a quarterly basis for UAS respondents with the opportunity to update past answers.9 The question on approval for Trump (post-November 3rd) comes from a UAS post-election survey (UAS 318). The survey had 7,279 respondents with a response rate of 80.91%(USC Dornsife Center for Economic and Social Research Citation2020) .10 These models are not estimated with fixed effects since the question about Trump support was only asked in 2020 and is therefore time-invariant.11 See appendix for this and other robustness checks.