Election administration and individual political behavior were not immune to the massive global disruption of COVID-19. U.S. institutions and citizens faced early tests when primary elections were to occur when medical and social science offered limited guidance.
Unlike many states, Wisconsin’s attempts to postpone their elections during a state stay-at-home order were thwarted by the legislature. Additionally, the U.S. Supreme Court ruled ballots had to be postmarked by Election Day, limiting absentee options.Footnote 1 The election was subsequently described as a debacle,Footnote 2 fiasco,Footnote 3 and disaster.Footnote 4 Long lines and crowded polling places prompted health officials to express concerns with viral spread.
Using a survey experiment on a diverse national sample, we examine the effect of information from Wisconsin’s Health Services about possible COVID-19 infections due to in-person voting on comfort with various voting methods. Results indicate information about coronavirus exposures following Wisconsin’s election decreases comfort voting in-person early and on Election Day yet does not significantly increase comfort with mail voting.
Expectations
Psychologists agree that individuals are motivated, at least in part, by self-preservation (Muraven and Baumeister Reference Muraven and Baumeister1997). It is reasonable that individuals will feel anxiety at the prospect of contracting a deadly disease and that their self-preservation instinct will guide behavior under such circumstances. Indeed, a national survey weeks before Wisconsin’s election found large majorities worried about contracting coronavirus and took steps to avoid illness including by maintaining physical distance from others and staying home as much as possible.Footnote 5 In this context, we expect individuals to view news of infections linked to in-person voting through the lens of self-preservation and for this to impact comfort with voting methods. Individuals may also be guided by concern over infecting others, and thus altruism may impact comfort with voting in-person.
Political science has long attended to how structural aspects of the electoral process shape attitudes and behavior. For example, Alvarez, Hall, and Llewellyn (Reference Alvarez, Hall and Llewellyn2008) show confidence one’s vote was accurately counted varies by method. During the pandemic, self-interest or concern about infecting others may reasonably play a larger role, particularly for in-person voting which brings interactions with others. Given the increased health risks we propose:
H1: Comfort with in-person voting will decrease when individuals learn about potential COVID-19 spread during in-person voting.
Given attempts to postpone elections and increase absentee voting, it is reasonable that most believed mail voting did not pose health risks. But in-person voting worries might not translate into greater comfort with a method less familiar to most. Outside health, mail voting might raise concerns with fraud, undue influence, privacy, and potential for missing late-breaking news (Gronke et al. Reference Gronke, Galanes-Rosenbaum, Miller and Toffey2008). Additionally, while support for mail voting is high in states that have it (Southwell Reference Southwell2004), Alvarez et al. (Reference Alvarez, Hall, Levin and Stewart2011) show low support for entirely mail elections. Absentee voting has expanded greatly since Alvarez et al.’s study, suggesting overall support has grown as well. Hassell (Reference Hassell2017) finds that voters can be persuaded to choose mail over in-person voting, suggesting preferences on voting methods are malleable. Overall, whether comfort with mail voting is influenced by information about health risks is less clear, but we test the following:
H2: Comfort with mail voting will increase when individuals learn about potential COVID-19 spread during in-person voting.
Design
Between April 28 and 30, 2020, Qualtrics recruited a diverse sample of 1,313 adult citizens for an online survey and randomly assigned them to either a control or treatment condition.Footnote 6 All participants were first told “As you may know, Wisconsin recently held primary elections with in-person voting.” Treatment participants then read an excerpt from a recent Wisconsin Department of Health Services’ post:
“Today the Department of Health Services (DHS) announced new tracing mechanisms for local health departments to better track Wisconsin residents who may have been exposed to COVID-19 during Tuesday’s election.”
Participants then rated their comfort voting in-person on Election Day, during early voting, and voting by mail in the November election (in that order). Because comfort is linked to self-reported likelihood of voting (see online Appendix), comfort with particular vote methods are meaningful outcomes of interest.Footnote 7
Results
The probability of expressing comfort with each voting method by condition is displayed in Figure 1. For each method, we code very/somewhat comfortable as 1, and 0, otherwise, and use Ordinary Least Squares (OLS) regression. Since the dependent and independent variables are binary, this is a more efficient difference-of-means test (Hanmer and Kalkan Reference Hanmer and Kalkan2013).
Figure 1 Treatment Effect on Comfort with Vote Methods.
NOTES: Estimates calculated using OLS regression. Error bars are 95% confidence intervals. Respondents asked: “In November, how comfortable or uncomfortable would you be [voting in person at a polling place on Election Day/voting in person at an early voting location before Election Day/voting with a ballot you receive and return by mail]?” Dependent variables coded 1 for very/somewhat comfortable, 0 for neither comfortable nor uncomfortable and somewhat/very uncomfortable.
Subjects shown information about possible coronavirus exposure while voting expressed significantly less comfort voting in-person. The treatment reduced the probability of expressing comfort with in-person, Election Day voting by about 10% points (p < 0.001, two-tailed), and reduced comfort with early voting in-person by about 8% points (p = 0.005, two-tailed). Given the treatment did not provide evidence of actual infections, but just suggested the possibility, these are fairly large effects. The treatment did not significantly increase comfort with mail voting (p = 0.27, two-tailed). While the treatment did not directly target a shift to mail voting, structural differences between in-person and mail voting suggest future research might explore focused messaging to increase comfort with mail voting.
Conclusion
Running elections during COVID-19 requires balancing concerns over providing access and ensuring safety. An original experiment demonstrates that voting-linked coronavirus infections news reduces comfort voting in-person. Interestingly, comfort with mail voting, the main alternative, does not increase in tandem.
Other COVID-19 research suggests the need for tailored messaging to alter some attitudes and behavior (Kuipers, Mujani, and Pepinsky Reference Kuipers, Mujani and Pepinsky2020). That literature along with our results has implications for election administration during health crises. Given continuing media coverage of coronavirus infections, election officials should expect to tailor messages to specific features of voting methods to increase citizens’ comfort. This will likely involve convincing voters opting to vote in-person that they will be safe and alleviating concerns with mail voting.
Supplementary material
To view supplementary material for this article, please visit https://doi.org/10.1017/XPS.2020.38
Open access
