On pins and needles: anxiety, politics and the 2020 U.S. Presidential election

ABSTRACT

Contemporary U.S. politics is characterized by a high degree of political polarization and conflict. Consequently, scholars have become increasingly interested in understanding how political factors and events impact different dimensions of health, such as anxiety. Using data from a nationally-representative, two-wave panel survey conducted before and after the 2020 U.S. presidential election, we develop a measure of political anxiety and examine how levels of political anxiety changed following the election. In general, we find that levels of political anxiety decreased following the presidential election. We then examine individual-level factors that influence post-election levels of political anxiety. Those who are highly politically engaged, interested in politics, and who score highly on negative emotionality felt more political anxiety than their counterparts after the election. Those who voted for Donald Trump, conservatives, and African Americans reported feeling less political anxiety than their counterparts following the election. Our findings regarding vote choice and ideology are somewhat surprising in light of previous research on the impact of electoral loss. We conclude with a discussion of what might be driving some of our counterintuitive results and provide ideas for future research.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Notes

1 Interestingly, during and after the 2016 and 2020 presidential elections, many media outlets published stories on the development of political or election anxiety. See, for example: https://www.nbcnews.com/better/lifestyle/election-stress-disorder-how-cope-anxiety-political-tensions-intensify-ncna1146951, or https://www.vox.com/science-and-health/2016/10/13/13259938/election-stress-aniexty-ahhh-poll, or https://time.com/4299527/election-mental-health/, or https://www.gq-magazine.co.uk/lifestyle/article/political-anxiety-disorder

2 https://www.psychiatry.org/newsroom/apa-public-opinion-poll-2020

3 https://www.apa.org/news/press/releases/2020/10/election-stress

4 The respondents were matched to a sampling frame on gender, age, race, and education. The frame was constructed by stratified sampling from the full 2017 American Community Survey (ACS) 1-year sample with selection within strata by weighted sampling with replacements (using the person weights on the public use file).

5 YouGov provides a weight to correct for any imbalances in the sample. Our analyses make use of the sampling weight.

6 We recognize that other scholars may construct similar measures using alternative items, and we encourage refinements along these lines in extensions of this work.

7 In the Supplementary Materials (Tables 2a and 3a), we include correlations matrices showing how the eight items are correlated with each other in each survey wave. We also note that within each survey wave, we conducted factor analyses of the items. In both waves, there is evidence that the items load on one factor. In wave 1, the eigenvalue for the first factor was 4.21 (.26 for the second). In wave 2, the eigenvalue for the first factor was 4.56 (.32 for the second factor).

8 In the Supplementary Materials (Figures 1a, 2a, 3a, 4a), we include histograms showing the distribution of each anxiety item for each wave and histograms showing the distribution of the overall political anxiety scales for each wave.

9 Descriptive statistics for all variables included in this study are provided in the Supplementary Materials (Table 1a).

10 In the Supplementary Materials (Tables 4a, 5a, and 6a), we include for interested readers tables showing the influence of our independent variables on each of the eight items that make up our political anxiety scale.

11 An alternative possibility is that the decrease in anxiety among Republicans could be due to increases in anger, but it is beyond the scope of the current study to unpack these mechanisms completely. Still, we hope subsequent research will adjudicate these possibilities more directly (e.g., by measuring different types of emotion in the context of the same study).

12 We code vote choice as 1 = Trump voter and 0 = Biden voter (with those voting for some other candidate coded as missing; only 3% of our sample reported voting for some other candidate). To check the robustness of our results, we also generated a measure of vote choice where 1 = Trump voter, 0 = Biden voter, and 0 = voted for some other candidate. Comfortingly, when we use this alternative coding, we get nearly the same results as are shown in Model 1, Table 3.

Additional information

Funding

This work was supported by University of Nebraska-Lincoln.