Tweet no harm: Offer solutions when alerting the public to voter suppression efforts

Abstract

Voter suppression was a major issue during the 2018 U.S. Midterm elections. Civil rights organizations, advocacy groups, and celebrities used Twitter to urgently warn voters about potential problems at the polls. However, in raising awareness of these normatively problematic matters, these messages had the potential to negatively impact key measures of democratic health. We conduct a survey experiment exploring the effect of exposure to tweets about voter suppression on confidence in elections, democratic legitimacy, and vote intent. Tweets that highlight voter suppression without a solution depress confidence in elections across all party groups, suggesting that solely warning people about problems at the polls without concrete actions they can take in response may do more harm than good. In addition, tweets highlighting voter suppression increase democratic legitimacy among Republicans relative to Democrats, suggesting that the former see such efforts as beneficial. There is no impact of either treatment condition on vote intent. Normative implications for democracy are discussed.

Keywords

Confidence in elections democratic legitimacy political communication vote intent voter fraud voter suppression

Introduction

The months leading up to the 2018 U.S. midterm elections featured widespread efforts to suppress minority votes, including voter roll purges, strict photo ID requirements, and polling location changes, many of which had the potential to shift electoral outcomes. Hundreds of thousands of Georgians were purged from the voting rolls while the Georgia Secretary of State delayed the registration of tens of thousands of African Americans; Native Americans in South Dakota were prevented from voting due to a targeted address-based technicality; a majority-Hispanic town in Kansas had their only polling location closed (Lopez, 2018; Newkirk, 2018; Sullivan, 2018). According to a report from the Center for American Progress that cataloged voter suppression efforts, “eligible voters across the country were dissuaded or actively prevented from casting ballots that would have counted” (Root & Barclay, 2018).

In response, a range of political elites, including advocacy organizations, elected officials, journalists, and opinion leaders spoke out against these tactics on Twitter to alert people to the threat of voter suppression. By “voter suppression,” we specifically refer to “any attempt to prevent or discourage certain Americans from registering to vote or casting their ballot” (League of Women Voters, n.d.), which is a federal crime (Federal Bureau of Investigation, n.d.-b). Voter suppression is a real and urgent problem: over the past decade, states have passed nearly 100 laws intended to restrict access to voting by “[making] it harder for Americans to register, stay on the voter rolls, or vote as compared to existing state law” (Singh & Carter, 2023). One report documents 1688 polling place closures between 2012 and 2018, which are often geographically targeted to make it logistically impossible for older, rural, disabled, or minority voters to cast a ballot (civilrights.org, n.d.). Other voter suppression tactics include closing early voting locations, failing to repair broken voting machines, running out of paper ballots, and improperly training poll workers; these choices can result in hours-long waits to vote or the need to vote provisionally, all of which can deter individuals from successfully casting ballots (Gardner & Reinhard, 2018; Wilder & Baum, 2022).¹

It is normatively good to raise awareness of these suppression efforts in an attempt to both warn the public and potentially generate outcry against them. Indeed, such efforts to suppress legitimate voters are harmful because they violate expectations of procedural democracy (Strömbäck, 2005, p. 334):

The basic claim procedural democracy exacts upon citizens and politicians is that they respect the rules and procedures of democracy. The right to vote, the freedom of expression and of the press, and the other basic requirements, must always be protected and respected. . . . How people choose to spend their time and their mental energy is up to themselves, as long as they do not violate the basic democratic freedoms and rights.

However, good-faith efforts to raise awareness of voter suppression may make the public feel less confident in elections: after all, if people are being prevented from voting, how can the public trust the results? And if they cannot trust the results, what does that mean for the legitimacy of our democracy? If our democracy is not legitimate, why bother voting at all?

These questions motivate the following experiment that tests the impact of exposure to elite political communications about voter suppression on key measures of democratic health: confidence in elections, democratic legitimacy, and vote intent. Despite well-documented and widespread voter suppression efforts, there are woefully few studies that explore the impact of exposure to information about them (e.g. Barney & Rhodes, 2017; Biggers, 2021; Clayton, 2023; Hill, 2021; Kane, 2017). These studies tend to focus on psychological outcomes such as anger or reactance, or policy attitudes such as support for specific voter suppression policies such as strict photo ID rules. Conversely, research on elite communications about untrue and/or conspiratorial allegations of voter fraud, such as former president Donald J. Trump’s claims that “fraudulent and illegal votes” cost him the 2020 election (Olson, 2024), tends to focus on broader measures of democratic health, such as confidence in elections or democratic legitimacy (e.g. Albertson & Guiler, 2020; Berlinski et al., 2021; Clayton et al., 2021).

We seek to determine if elite communications about voter suppression efforts have the same type of consequences on democratic health outcomes as voter fraud messages. To be clear, voter suppression and voter fraud are vastly different, in terms of intent and incidence rate. Voter suppression was defined above; actual voter fraud is defined by the federal government as circumstances such as illegally voting in the name of a dead person or someone who has moved; intentionally falsifying information during voter registration; receiving money in return for voting or registering; voting twice in a federal election; or an election official allowing unqualified voters cast ballots (usa.gov, n.d.; Federal Bureau of Investigation, n.d.-a). The term “voter fraud” in academic literature usually refers to “noncitizen voting, double voting, and posing as someone else” in order to cast a ballot (Sheagley & Udani, 2021, p. 1). We emphasize here that actual, verifiable cases of illicit voting by non-citizens or those who are ineligible are exceedingly rare (see Beaulieu, 2014; Berlinski et al., 2021; Kane, 2017). Thus, evidence suggests that while voter suppression efforts are widespread, voter fraud is relatively rare.

This article takes up the challenge of informing the public about actual and ongoing efforts to block rightful voters from registering and casting a ballot. We believe messages alerting the public to these efforts are important and necessary. Nevertheless, we investigate the potential that raising awareness of voter suppression will negatively impact democratic health by depressing attitudes toward elections. We draw on prior work regarding exposure to both voter suppression and voter fraud information to develop hypotheses.

Below we report the results of a survey experiment conducted during the 2018 U.S. Midterm election that varies exposure to different types of actual tweets about voter suppression to determine their impact on three outcomes: confidence in elections, democratic legitimacy, and vote intent. Given the partisan nature of the issue, we consider overall effects as well as differences within and between partisan groups. Two sets of tweets are tested against a placebo control: tweets that only make the reader aware of suppression efforts, and those that also offer solutions, such as asking for a provisional ballot or calling a hotline. Tweets used in the study were sent in the month leading up to the 2018 election by “elite communicators” (prominent organizations and individuals who have large followings on Twitter); the tweets themselves also received high levels of user engagement, indicating widespread reach.

Exposure to messages that only highlight voter suppression depresses confidence in elections across all participants regardless of party. Conversely, messages that include individual-level solutions in addition to information about voter suppression were indistinguishable from a placebo. Furthermore, voter suppression threats increase democratic legitimacy among Republicans relative to Democrats, demonstrating the two parties’ divergent perceptions of how voter suppression helps or hurts them electorally. However, despite these normatively troubling results, there is no impact on voter intent. Implications for democracy are discussed.

Elite communications about voter suppression

The political communication literature generally agrees that elite communication can shape public opinion, particularly when it comes from ideologically congruent sources (e.g. Gilens & Murakawa, 2002; Zaller, 1992). “Elites” traditionally refer to “individuals and organizations, including politicians and political officials, policy experts, interest groups, religious leaders, and journalists,” (Gilens & Murakawa, 2002, p. 16). In our contemporary, social-media-saturated environment, this definition can be expanded to include people who “have greater political influence and power than other citizens do,” such as online influencers who play an opinion-leading role (Marland et al., 2018, p. 5), thus fitting within Gilens and Murakawa’s (2002, p. 16) conception of elite communicators as “those whose views are communicated through the media.”

In this article, we consider the specific case of elite communications pertaining to voter suppression. Ample prior work shows that elite communication can shape attitudes on procedural issues pertaining to voting and elections. Below, we review the survey experimental literature specific to voter suppression, and then to voter fraud.

Exposure to voter suppression information

Most prior survey experimental work assesses the impact of informing people about specific forms of voter suppression legislation—such as strict photo ID requirements—on emotions or support for those specific policies. Exposure to information about voter suppression generates anger and reactance, particularly among targeted populations such as minorities, young people, and Democrats (Biggers, 2021; Hill, 2021; Valentino & Neuner, 2017). When these specific subgroups learn about such laws, they feel angry (Biggers, 2021; Hill, 2021; Valentino & Neuner, 2017). This anger appears to be motivating, as it positively predicts individuals’ stated likelihood of voting, and may be one reason why actual voter ID laws have not generated substantial decreases in participation (Valentino & Neuner, 2017; see also Clayton, 2023). In short, when impacted voters learn about restrictive ID laws, they perceive it as a threat to their liberty; this makes them angry and reactant, and in turn those emotions make them more likely to say they will vote (Hill, 2021).

Other work looks at policy outcomes: exposure to information about specific voter suppression efforts has been found to increase support for them among White voters (Barney & Rhodes, 2017), and reduce support among Black Americans (Biggers, 2021). Republicans increase support for photo ID laws when told about voter fraud; Democrats increase support when laws are viewed to help them electorally (Kane, 2017; see also Clayton, 2023). Finally, the few experiments that look at voting attitudes find that exposure to voter ID laws raises perceived threats to the right to vote (Biggers, 2021) and negatively impacts external efficacy, or perceptions of government responsiveness, but has no impact on government trust or attitudes toward voting processes (Scacco et al., 2016).

However, none of this work asks how voter suppression information impacts broader attitudes toward the health of our democracy as a whole; these outcomes are usually relegated to studies about exposure to information about voter fraud.

Elite communication about voter fraud

Allegations of voter fraud have become worryingly commonplace over the last decade in American politics. While the candidacy and eventual presidency of Donald J. Trump brought this issue to the forefront due to his repeated allegations that elections are “rigged” (Dale, 2022), purported “fraud” has been used as a justification for voter identification laws for well over a decade (Kane, 2017). Notably, both Republicans and Democrats both believe that fraud takes place, though what they define as “fraud” varies greatly based on partisanship (Sheagley & Udani, 2021). Republicans tend to view “fraud” as illicit voting perpetuated by immigrants and non-citizens. Conversely, Democrats say “fraud” is practiced by governmental officials and pertains to elite manipulation; their definition corresponds to what is publicly (and in this article) referred to as “voter suppression.”² However, both Republicans and Democrats perceive that these illegitimating activities are perpetuated by the opposing party (Sheagley & Udani, 2021). Furthermore, partisans only view such tactics as problematic if they are perceived to harm their party; troublingly, they are not concerned when it is thought to impacts the other side (Beaulieu, 2014).

Much of the elite “voter fraud” rhetoric over the past several years in the United States has come from right-leaning entities (individuals, news organizations) that specifically promote conspiratorial and untrue allegations. Several survey experiments have assessed the impact of exposure to claims of voter fraud, tweets from Donald J. Trump, or statements that elections are “rigged” on measures of democratic health. Exposure to elite Republican claims of voter fraud reduces confidence in elections, though effects are concentrated in co-partisans (Berlinski et al., 2021). Another study found that exposure to former President Trump’s attacks on the legitimacy of the 2020 election “erode[d] trust and confidence in elections” and increased beliefs that elections are rigged among Trump supporters (Clayton et al., 2021, p. 1).

However, the effects are not limited to Republicans: both Democrats and Republicans exhibit a decrease in democratic norms such as a willingness to accept results when exposed to information that their party was victimized by election fraud (Albertson & Guiler, 2020). Other work has found that Trump’s tweets attacking democratic institutions actually generate a pushback among Democrats and have no impact on Republicans, possibly because their support for democratic values has already declined precipitously over the last decade (Bowler et al., 2023).

We review this work here because while the focus of this article is the impact of exposure to factual information about actual efforts to suppress qualified voters, we see the existence of attitudinal shifts as a result of exposure to voter fraud claims as evidence that elite communication about voting issues can potentially impact voters’ broader attitudes toward democratic health. Our question is whether these shifts occur when the content of the elite communication is true (voter suppression) rather than false (allegations of voter fraud and election rigging).

Elite communications about voter suppression during the 2018 U.S. midterm elections

As detailed above, voter suppression is an ongoing problem in the United States, where some states have passed laws disenfranchising voters through ID requirements, closed or changed polling places with minimal notice, and generally made it difficult for certain populations to vote (see Daniels, 2020; Hardy, 2019; Wilder, 2021). During the 2018 U.S. midterm election, voting rights organizations and allies attempted to raise awareness of these types of ongoing voter suppression efforts. A wide range of elite communicators used Twitter to spread messages about voter suppression, including national political organizations, lawmakers, activists, and even celebrities. Many of these accounts have large followings on Twitter, including members of the general public who may in turn spread these messages onward to their followers. All of these messages emphasized the threat of voter suppression (those that only raised the alarm are henceforth referred to as “threat tweets”). However, some of them also offered an individual-level solution if voters encountered problems at the polls: a hotline number to call, or the ability to request a provisional ballot, for example (henceforth “solution tweets”). Below, we review our outcomes of interest and whether we expect them to differ based on random assignment to view either one set of voter suppression tweets (“threat” or “solution”) or a placebo control.

Democratic health outcomes

We specifically concern ourselves with the impact of exposure to voter suppression information on three attitudinal measures of democratic health: confidence in elections, democratic legitimacy, and vote intent. Each of these is directly related to the normative impacts of voter suppression on elections: if people are being denied their right to cast a ballot, can we be confident that elections reflect the will of the people (confidence in elections)? Does raising this issue lead individuals to question the legitimacy of the results of the election, or potentially decrease their stated intent to vote under the assumption that it won’t be accurately counted? We take up each in turn.

Confidence in elections

Confidence in elections refers to the perception that votes will be counted and reported accurately. Traditionally, the United States has enjoyed high levels of voter confidence relative to other countries. However, the 2016 U.S. Election brought in a “perfect storm” of electoral threats, both real and imagined, and confidence in elections plummeted (Norris, 2019). Prior work finds that elite messages about voter fraud suppress confidence in elections (Albertson & Guiler, 2020; Berlinski et al., 2021; Clayton et al., 2021). Furthermore, specific information about different types of election fraud, such as ballot mishandling, is particularly likely to cause decreased confidence (Vonnahme & Miller, 2013).

We extend this work to find out if elite messages about voter suppression impact confidence in elections. We expect “threat” tweets to suppress confidence in elections relative to the placebo control group, by raising the issue of whether all voters will be able to cast a ballot.

Hypothesis 1 (H1). Threat tweets will suppress confidence in elections relative to the control group.

However, given that “solution” tweets contain two types of messages—“the threat exists, and here’s what to do about it”—we cannot predict the net effect of the messaging. Like threat tweets, we expect solution tweets to make the public more aware of voter suppression efforts relative to the control group. However, we cannot predict whether including individual-level solutions will mitigate the harms of this exposure, as these tweets still make individuals aware of voter suppression efforts while also offering a potential response.

Research Question 1 (RQ1). Will solution tweets impact confidence in elections relative to (a) the control group or (b) the threat condition?

Democratic legitimacy

Democratic legitimacy measures whether individuals will view the outcome of an election as legitimate if their preferred candidate loses. This follows conceptually from our previous outcome of interest: “How confident voters are that their ballots are counted correctly is a normative issue within a representative democracy as a lack of confidence threatens the perceived legitimacy of an elected government” (Alvarez et al., 2008, p. 764). This is a loaded question in the United States, which has enjoyed centuries of democratic elections and accumulated what Norris (2013, p. 576) terms “deep reservoirs of regime legitimacy.” However, in the wake of the 2016 U.S. presidential election, confidence in elections dropped and Americans’ satisfaction with democracy and faith in democratic institutions has wavered. Prior work shows that observed electoral fraud is negatively associated with satisfaction with democracy (Fortin-Rittberger et al., 2017) and that allegations of election interference can make people less likely to accept a loss as legitimate. We extend this work to awareness of voter suppression.

If subjects infer from the treatments that qualified individuals will be prevented from voting due to suppression efforts, they may then question the legitimacy of outcomes. We expect this relationship to hold for the threat condition versus the placebo control condition. However, again we cannot predict what happens when individuals are told about a threat and offered a solution.

Hypothesis 2 (H2). Threat tweets will suppress democratic legitimacy relative to the control group.

Research Question 2 (RQ2). Will solution tweets impact democratic legitimacy relative to (a) the control group or (b) the threat condition?

Vote intent

Vote intent represents an individual’s stated desire to do their civic duty; it is often higher than self-reported turnout, which in turn is higher than validated turnout estimated from public records (Achen & Blais, 2016). This discrepancy reflects the strong social norm that “everyone should vote,” thus respondents may choose to give a socially desirable response (Ansolabehere & Hersh, 2012). However, as this is a survey experiment conducted before the 2018 midterm election on an anonymized sample, we lack access to validated turnout; furthermore, self-reported vote intent, self-reported turnout, and validated turnout are highly correlated (Achen & Blais, 2016, though see Rogers & Aida, 2014).

Since we explore how exposure to voter suppression messages impacts subjects’ attitudes, we want to know if this extends to vote intent as well: do threat messages decrease people’s stated intent to cast a ballot? In the behavioral literature, the impact of voter suppression messaging on turnout is decidedly mixed. Survey and field experiments demonstrate that making Black voters aware of suppression efforts “raises perceived threat to their franchise,” though they have no impact on actual voter behavior (Biggers, 2021, p. 1161). Information about voter suppression increases intent to vote among people who get angry about it, but reduces voter intent among those who do not (Hill, 2021). One study does find behavioral changes as a result of online engagement with election misinformation: “Those promoting conspiracy theories questioning the legitimacy of the US electoral process were, at the same time, somewhat less likely than defenders to participate in it.” (Green et al., 2022, p. 1). Other work finds a link between lower confidence in elections and lower likelihood of voting (Alvarez et al., 2008).

Of our three outcomes, vote intent is perhaps the most alarming and normatively troubling: if telling people that voter suppression exists makes them less likely to vote, how can civic organizations and political elites raise awareness without harming participation? Given that prior research outcomes in this area are mixed, we pose the following:

Research Question 3 (RQ3). Does treatment condition impact subjects’ stated intent to vote?

Partisanship

Voter suppression efforts are a largely partisan issue: such laws and efforts are overwhelmingly driven by Republican lawmakers to disenfranchise likely Democratic voters, specifically Black and young voters. The elite cues literature leads us to expect that effects should differ between partisan groups (e.g. Zaller, 1992). Indeed, prior work finds that the impact of voter fraud messages is conditional on subjects’ partisanship: partisans view this issue through the lens of whom it helps (Albertson & Guiler, 2020; Berlinski et al., 2021; Clayton, 2023; Clayton et al., 2021). Republicans may view voter suppression as a worthwhile effort to prevent unauthorized individuals from voting; awareness of such practices may increase their confidence in elections and democratic legitimacy. Conversely, Democrats may view these efforts as aimed at targeting their co-partisans; awareness may negatively impact their confidence in elections and legitimacy—or may enrage Democrats enough that they demonstrate increased intent to vote. In short, we expect partisan differences. This tracks with prior work on the impact of voter ID laws: in states with ID requirements, Democrats report lower confidence in elections and Republicans higher (Bowler & Donovan, 2016).

The elite communication literature on voter suppression and voter fraud alike tends to find differences among partisans. Perceptions of electoral legitimacy, or whether votes are cast and counted correctly, are associated with partisan news use, again with Fox viewers reporting lower scores, likely because of the network’s focus on questioning elections (Grant et al., 2021). Support for voter ID laws—which Republican lawmakers tend to favor and Democrats oppose—is also stronger among Fox News viewers, who themselves overwhelmingly identify as Republican (Wilson & Brewer, 2013). Finally, while exposure to Trump’s rhetoric against liberal democracy does not increase anti-democratic attitudes among Republicans, it does cause Democrats to become more supportive of democratic institutions (Bowler et al., 2023).

Complicating this matter is the role of the “winner effect” in which partisans on the prevailing side of an election decide it was legitimate after all (Grant et al., 2021; Levy, 2021; Sances & Stewart, 2015; Sinclair et al., 2018). In the media, the 2018 U.S. midterms were widely predicted to be a landslide for Democrats as anti-Trump voters stormed the polls (e.g. Scott, 2018). Republicans may not have expected to win, and Democrats may have been even more concerned if they did not.

Generally, however, we expect partisan differences in attitudinal outcomes following prior work on elite communications about voter fraud specifically for confidence in elections and democratic legitimacy, with Democrats exhibiting more negative outcomes. We are not sure if there will be any measurable partisan impacts on intent to vote, however:

Hypothesis 3 (H3). Partisanship will moderate the impact of treatment relative to the control group on (a) confidence in elections and (b) democratic legitimacy.

Research Question 4 (RQ4). Will partisanship moderate the impact of treatment on intent to vote?

Methods and materials

A multiple-wave, placebo-controlled survey experiment was conducted during the fall of 2018 to address these hypotheses and research questions. In Wave 1 (11–15 October 2018), we measured subjects’ baseline outcome variables, as well as partisanship and demographics. All participants in Wave 1 were invited back for Wave 2 (31 October–5 November 2018), with a recontact rate of 48.20%. In Wave 2, subjects were randomly assigned to a messaging condition: placebo, threat, or solution. In all conditions, the stimulus consisted of screenshots of eight randomly ordered tweets. After the stimulus, subjects answered a series of questions about political efficacy, confidence in elections, and democratic legitimacy. Further details about the study procedure and pre-test are available in the Supplementary Materials (https://osf.io/wdbz5).

Pre-test and procedure

Experimental stimuli were created using screenshots of real tweets about potential voter suppression, such as voter roll purges and identification requirements. The use of screenshots is common given the logistical challenges of Twitter exposure and the need to minimize other confounding variables (e.g. Berlinski et al., 2021; Meeks, 2017).

First, a total of 25 tweets were collected dated between 3 and 22 October 2018 that were published by non-profit organizations, elected officials, and Twitter users with substantial followings. Tweets were selected by searching Twitter for keywords including “vote,” “voting,” “voter ID,” “suppression,” and “polls.” All pre-tested tweets received at least 1000 retweets, implying substantial reach. Tweets were categorized as presenting a “solution” if they provided information about how to address problems at the polls; tweets categorized as “threat” only mentioned potential sources of voter suppression. Screenshots of all 25 tweets are available in the Supplementary Material (Figure S1).

Next, tweets were pre-tested on MTurk, where workers rated them according to credibility and trust. The five “threat” and five “solution” tweets with the descriptively highest scores for credibility and trust were selected for the experimental stimuli. Scores for all pre-tested tweets are available in the Supplementary Material (Table S1).

The final stimuli consisted of a series of eight tweets displayed on a page in a manner akin to a Twitter feed. In the placebo condition, subjects saw a random draw of eight celebrity, entertainment, sports, and trivia tweets from a pool of 12 placebo tweets. None of the placebo tweets contained information relating to voting or politics. In the “threat” condition, subjects saw a random draw of four of the five “threat” tweets and a random draw of four of the 12 placebo tweets. In the “solution” condition, subjects saw a random draw of four of five “solution” tweets and a random draw of four of the 12 placebo tweets. The use of a random subset makes it less likely that effects are due to one particular tweet and mimics the experience of Twitter where not all users see all tweets.

The order of tweets in the experimental timeline was consistent across conditions and determined by random assignment; subjects saw the tweets in the following order: treatment, placebo, treatment, placebo, placebo, treatment, treatment, placebo (Figure 1). See Supplementary Materials for further details about the experimental manipulation and screenshots of all tweets.

Figure 1.

Examples of tweets used in experiment.

Measurements

Outcomes were measured during Waves 1 and 2, with subjects’ pre-test measures used as a statistical control. Survey items are available in the Supplementary Materials.

Dependent variables

Confidence in elections was measured with a four-point ordinal variable adapted from Gallup (McCarthy & Clifton, 2016), “How confident are you that, across the country, the votes will be accurately cast and counted?” Wave 1: Not at all confident (8%), 2 = Not too confident (16%), 3 = Somewhat confident (41%), 4 = Very confident (35%); (M = 3.03; SD = 0.91). Wave 2: 1 = Not at all confident (4%), 2 = Not too confident (17%), 3 = Somewhat confident (53%), 4 = Very confident (26%); (M = 3.01; SD = 0.76).

Democratic legitimacy was measured with a four-point ordinal scale, “If your preferred congressional candidate ends up losing, would you accept the election as legitimate?” Due to the low rate of response to “Definitely not,” answers were collapsed into “No” and “Yes” in both waves for purposes of analysis (Wave 1: 82% yes; Wave 2: 87% yes; see Supplementary Materials [Section D]).

Voter intent was measured using a three-level categorical variable, “Do you plan to vote in the upcoming election?” Wave 1: I plan to vote (79%), I already voted (3%), I don’t plan to vote (17%); Wave 2: I plan to vote (62%), I already voted (24%); I don’t plan to vote (14%).

Covariates

Partisanship was measured using the seven-point American National Election Studies (ANES) scale ranging from Strong Democratic to Strong Republican. Leaners were collapsed with partisans; for analysis, true Independents and those who selected “Other” are combined (Democrats: n = 490, 45.0%; Republicans: n = 427, 39.2%; Independents: n = 173; 15.9%). Other demographic controls (age, sex, education, income level) were measured in Wave 1 of the survey to avoid priming effects. The full text of the survey questions is available in the Supplementary Materials.

Participants

The subject pool was recruited by Survey Sampling International (SSI) based on age and sex. As this is a multiple-wave study, only participants who completed both waves are included in the analysis.³ The average age of the 1090 subjects who completed both waves was 50.0 years; 53.1% were female, and 42.8% reported holding a college degree. In terms of race, 80.0% were White, 8.1% African American, and 6.4% Hispanic. Balance tests found no significant associations between condition and age, sex, partisanship, or ethnicity (see Supplementary Materials).

Analysis Method

We conduct a series of linear regression analyses to determine the impact of assignment to treatment on outcomes.⁴ To exploit the advantages of our two-wave design, we control for Wave 1 measures when estimating outcomes in Wave 2, which allows us to better estimate changes due to treatment and dramatically increases statistical power (Bock, 1985; Clifford et al., 2021), as well as control for demographic variables. For vote intent, we exclude subjects in Wave 3 who reported already voting from the analysis.

Since our hypotheses and research questions engage with differences between treatment conditions, we evaluate results solely in terms of pairwise comparisons between treatment arms for main effects and then look at pairwise comparisons within treatment groups and party groups to explore moderating effects. To correct for the number of comparisons, we have performed Tukey adjustments on each model; unadjusted p values are available in the Supplementary Materials. We also report full regression model results in the Supplementary Materials (Tables S4, S7, S10).

Results

Exposure to voter suppression threats produces a decline in confidence in elections relative to the placebo and solution conditions; this effect is consistent across party groups (Table 1). In addition, exposure to threat messages increases legitimacy for Republicans relative to Democrats, generating a polarizing effect, especially in contrast to the placebo condition (Table S8). However, despite these normatively troubling findings, we find no impact on intent to vote.

Table 1.

Pairwise comparisons: Linear regression, effect of treatment on confidence in elections, main effects and within party (Tukey-adjusted).

Pairwise comparison	All	Democrats	Republicans	Independents
Threat vs Control	−0.14 (0.05) p = .010	−0.16 (0.07) p = .380	−0.10 (0.07) p = .900	−0.16 (0.12) p = .918
Solution vs Control	−0.02 (0.05) p = .889	−0.04 (0.07) p = .999	0.01 (0.08) p = .999	−0.06 (0.12) p = .999
Solution vs Threat	0.11 (0.04) p = .039	0.12 (0.07) p = .755	0.12 (0.08) p = .836	0.10 (0.11) p = .992

Difference in Coefficients (SE).

Confidence in elections

Messages alerting the public to voter suppression reduce confidence in elections, an effect that appears consistent across partisan groups (though partisan pairwise comparisons are not significant). Results are plotted in Figure 2 below and reported in Table 1 below and in the Supplementary Materials (Tables S4, S5, S6).

Figure 2.

Treatment effect on confidence in elections, overall and by party.

Following expectations, exposure to voter suppression threats reduces confidence in elections relative to the placebo control condition (H1 supported); threat messages also reduce confidence relative to solution messages (RQ1b). There was no difference between the placebo and solution messages (RQ1a). This effect is rather large: the coefficient of 0.14 is equivalent to a 4.4% decline in confidence between the control and threat treatment.

However, there is no moderating effect of partisanship (H3a unsupported). As Figure 2 shows, all parties follow a similar pattern in response to treatment: a decline in confidence in elections in the threat condition relative to the control and solution conditions.⁵

Democratic legitimacy

There are no main effects of treatment on democratic legitimacy (H2 not supported, RQ2a, RQ2b) (Tables 2 and S7, S8, S9). However, this finding conceals troubling evidence that exposure to threat tweets increases perceived legitimacy among Republicans (H3b supported). Threat messages increase legitimacy relative to the control for Republicans, though pairwise comparisons are non-significant after adjusting for multiple comparisons. However, among subjects in the threat condition, we see a significant difference for Republicans relative to Democrats (Table S8, Tukey-adjusted p = .045). For Republicans, the threat condition boosts their perceived legitimacy relative to the control group; for Democrats, it suppresses it relative to the control groups (see the difference between the triangle and the square for each group in Figure 3). This is a polarizing effect, in which the two parties move in opposite directions when exposed to the same stimulus. As a result, showing voter suppression alerts to Republicans results in them seeing the election as more legitimate than it does Democrats.

Figure 3.

Treatment effect on democratic legitimacy, overall and by party.

Table 2.

Pairwise comparisons: Linear regression, effect of treatment on democratic legitimacy, main effects and within party (Tukey-adjusted).

Difference in coefficients (SE)	All	Democrats	Republicans	Independents
Threat vs Control	0.02 (0.02) p = .553	−0.02 (0.03) p = .999	0.07 (0.04) p = .542	0.01 (0.06) p = .999
Solution vs Control	0.02 (0.02) p = .746	0.02 (0.04) p = .999	0.02 (0.04) p = .999	−0.02 (0.06) p = .999
Solution vs Threat	−0.01 (0.02) p = .949	0.04 (0.03) p = .943	−0.05 (0.04) p = .891	−0.03 (0.06) p = .999

Vote intent

The scant good news is that the treatments had no significant or sizable impact on vote intent (RQ3; Tables 3 and S10, S11, S12), so while people in the threat condition may express lower confidence in elections, the treatment does not decrease the likelihood that they state an intention to participate. There were no significant moderating effects of partisanship (Figure 4); effects were the same within treatment group and within-party group (RQ4).

Figure 4.

Treatment effect on vote intent, overall and by party.

Table 3.

Pairwise comparisons: Linear regression results, effect of treatment on vote intent, main effects, and within party (Tukey-adjusted).

Pairwise comparison	All	Democrats	Republicans	Independents
Threat vs Control	0.00 (0.02) p = .999	0.02 (0.03) p = .999	0.00 (0.03) p = .999	−0.04 (0.05) p = .997
Solution vs Control	−0.01 (0.02) p = .909	−0.00 (0.03) p = .999	−0.02 (0.04) p = .999	−0.01 (0.05) p = .999
Solution vs Threat	−0.01 (0.02) p = .895	−0.02 (0.03) p = .999	−0.02 (0.03) p = .999	0.05 (0.05) p = .999

Difference in Coefficients (SE).

We do note that for Independents in the threat condition, intent to vote is descriptively lower: a covariate-adjusted 76.5 predicted likelihood of voting, relative to 80.4 in the control condition. The impact of perceived partisan suppression efforts on Independents who do not align with a party merits future exploration.

Discussion

Tweets about voter suppression that only warn the public about problems they may encounter at the polls harm overall confidence in elections and have a polarizing and normatively troubling impact on partisans’ views of democratic legitimacy. However, these effects disappear if messengers include individual-level solutions in their tweets, and none of the voter suppression awareness messages have an impact on vote intent. The advice for groups looking to sound the alarm about threats to voting is clear: Include individual-level solutions such as calling a hotline or requesting a provisional ballot. Telling people what to do about the problem of voter suppression appears to ameliorate the harm that arises from alerting them to it.

We underscore here that we do not believe the tweets included in our study about voter suppression were sent with the intent to decrease confidence in elections or lower democratic legitimacy—they were most likely sent to inform the public and garner media attention on the issue. However, tweets that lacked individual-level remedies such as calling a hotline or requesting a provisional ballot had the perhaps-unintended consequence of reducing confidence that votes would be accurately cast and counted. This effect appears consistent across all partisan groups.

Questioning electoral procedures is a normatively valid response: learning that legitimate voters may be blocked from casting a ballot should make people concerned about whether votes are accurately cast and counted. However, if this information makes people so dispirited that they fail to vote, then the problem is only magnified. The lack of impact on vote intent is a sign that while this information is troubling, its consequences may not extend to deterring individuals from participating. Thus, civil rights groups, political elites, and public figures should continue to sound the alarm, while making sure to also offer individual-level remedies to those who find themselves impacted.

Confidence in elections

The finding on confidence in elections offers some optimism in this time of partisan rancor: Republicans were no more or less impacted than Democrats. Prior work on voter fraud finds a strong partisan effect on confidence in elections (Albertson & Guiler, 2020; Berlinski et al., 2021; Clayton et al., 2021) in which the nature of the threat or the partisanship of the candidate disadvantaged by the threat determined who felt less confident. Our study featured a variety of election threat tweets and were drawn from different Twitter accounts. The variety in our study might better approximate the way election threats come across social media feeds in the real world. Nevertheless, the decrease in electoral confidence across all partisan groups in reaction to threatening tweets is a cause for concern, particularly because citizens are not uniformly defenseless in the face of voter suppression. People can check their registration in advance of Election Day and request a provisional ballot if necessary. While individual action is insufficient given systemic efforts at voter suppression, people can overcome some barriers.

Democratic legitimacy

Our results pertaining to democratic legitimacy have normatively worrisome consequences, particularly for a diverse and pluralistic society. Republicans assigned to the “threat” condition who were shown tweets about voter purges, ID requirements, and votes not being counted exhibited an increase in perceived democratic legitimacy. What this article terms “suppression” appears to correspond to what Republicans categorize as “preventing voter fraud,” as articulated by Sheagley and Udani (2021). Partisans tend not to view such activities as problematic if they are not perceived to impact their party (Beaulieu, 2014). We note here that these tweets were not explicitly partisan in nature—they did not say that Democrats would be victims of suppression. However, Republicans appear to have perceived it as such. Three tweets in the stimuli came from prominent African Americans (Congresswoman Marcia Fudge, Democratic operative Donna Brazile, and Grammy- and Oscar-winning rapper Common). Over 80% of African Americans consistently identify as Democrats. Republicans may have seen the tweets and assumed that messages from elite African Americans were intended to warn other African American voters.

Republicans in the sample appear to have perceived that suppression efforts would be aimed at likely Democratic voters, and thus prevent Democratic candidates from receiving their intended votes. This caused Republicans in the study to perceive such elections as more legitimate, stating that they were more willing to accept the results, compared to Republicans in the control condition and Democrats in the threat condition. Voter suppression efforts are often targeted at minorities and young voters, who tend to vote Democratic (Tyson, 2018). Legislation that is perceived to result in voter suppression is often framed by Republicans as preventing widespread fraudulent voting by disallowing individuals who support Democrats to cast ballots (Wines, 2017). From a policy perspective, findings suggest that it may be difficult to marshal Republican support for legislation preventing voter suppression efforts since partisans associate it with confidence in elections and democratic legitimacy.

Vote intent

The lack of impact on voter intent is a positive sign, that these messages do not appear to dissuade individuals from expressing their likelihood of casting a ballot. The experimental conditions may have been too mild to shift voting intentions, though our study materials were taken from social media in the context of a real election. It is also possible that social desirability bias, or the internalized norm that voting is the “right” answer in the survey context, prevents people from shifting their voting intentions in the wake of threat, just as we suggest in overestimates the number of voters in our sample. Participants might have reacted to electoral threats by thinking their vote doesn’t matter. Finally, threats to voting rights might have angered some participants, causing them to respond with a heightened commitment to vote (Valentino & Neuner, 2017). While this could occur for a subset of the study, and mask depressed turnout in another group of respondents, we do not find support for countervailing effects through conditioning effects by partisanship. In the present study, the null effect here suggests that voting intentions are resilient to threatening tweets.

Voter fraud versus voter suppression messaging

As noted in our literature review, much of the work on voter fraud messaging looks at democratic health outcomes, whereas most prior work on voter suppression messaging considers emotional outcomes and policy support. One meta-question this article raises is whether it is possible to talk about voting procedures at all, whether real or fake, without harming democratic health; to begin to address this we compare our results to prior work on exposure to voter fraud information. In terms of confidence in elections, we find negative effects from exposure to threats of voter suppression, but no impact from exposure to solutions. This echoes prior work on voter fraud, exposure to which reduces confidence, particularly among targeted partisans (Albertson & Guiler, 2020; Berlinski et al., 2021; Clayton et al., 2021; Vonnahme & Miller, 2013). Turning to democratic legitimacy, we find negative effects among Republicans relative to Democrats in the “threat” condition, which is similar to partisan effects when individuals are told that elections are “rigged” by perceived political opponents (Albertson & Guiler, 2020). Finally, we find no impact on vote intent; prior work shows that increased voter intent is a by-product of anger generated by awareness of voter suppression (Hill, 2021; Valentino & Neuner, 2017). Together, this work suggests that only raising awareness of problematic voting procedures—whether true, in the case of suppression, or false, as in most allegations of voter fraud—has negative consequences on democratic health attitudes.

Don’t not talk about voter suppression

However, scholars and practitioners alike would be incorrect to read this article as an admonition to refrain from raising the alarm about voter suppression. Quite the opposite: we encourage activists and academics alike to continue speaking up about normative threats to democracy such as efforts to block qualified individuals from casting a ballot. We note again that including the individual-level solution mitigated the worst of the normative harms that arose from exposure to only the threat of suppression.

Prior work on exposure to voter fraud messaging shows an expected decline in democratic health measures and participatory behaviors, even though such claims are largely conspiratorial and untrue (e.g. Albertson & Guiler, 2020; Berlinski et al., 2021; Clayton et al., 2021; Green et al., 2022). In such cases, the problem is simple: don’t spread baseless allegations of voter fraud.

Here, however, our stimuli are raising awareness of actual ongoing efforts to suppress legitimate voters. We find negative consequences for confidence in elections arise from doing so. This is a valid normative response. In this manner, our results echo research by Scacco et al. (2016) who find a decline in external efficacy resulting from exposure to advocacy efforts to message against voter identification laws. When people heard about the laws, they perceived government as less responsive. Again, this is a normatively valid response, and the public is not “wrong” to express these negative sentiments when faced with real efforts to block legitimate voters from casting a ballot. Talking about threats to democracy remains a challenge: do it incorrectly, and one risks causing additional unintended problems for democratic health.

Limitations and next steps

Our study faced several limitations. While the survey sampling firm promised a representative population based on sex and age, subjects were older than the median age of American adults (50.0 vs 38.6) and had a higher share of non-Hispanic whites than the national population (80.0% vs 62.6%). We controlled for demographics in our models to address this. While we expect that the treatment effects in our research hold for other populations (Druckman & Kam, 2011; Mullinix et al., 2015), it is possible that the stimuli have an even greater effect on younger and minority voters, who are often the target of voter suppression efforts, and also more likely to use social media and thus be exposed to the stimuli content. Such comparisons were underpowered with our current sample. Subsequent studies of this nature may want to intentionally over-sample minority and young subjects to enable these subgroup analyses.

Due to multiple-testing adjustments, many of our pairwise comparisons within party or treatment groups are non-significant. Our sample of over 1080 American adults is large, though our party groups somewhat less so (Democrats: n = 490, 45.0%; Republicans: n = 427, 39.2%; Independents: n = 173; 15.9%), and the need to prevent Type I error meant that many significant p values in the initial regression results did not survive a Tukey adjustment. In practice, we suggest that scholars look at the size of the differences and consider sizable differences between- or within-party groups as avenues for future research with larger sizes and fewer planned comparisons to preserve statistical power.

In terms of design, we made a choice to look at actual messages that circulated in the real world, rather than concoct messages specifically for this study because we want to understand what happens when civic groups, opinion leaders, and high-profile individuals talk about threats to elections. There is a trade-off here between internal validity (constructing fake tweets that precisely control the language used) and reality (reflecting how this phenomenon occurs in the real world) and in this study, we chose the latter over the former and thus constrained our treatment tweets to those that were actually sent by political actors over whom we have no control. Other scholars might want to replicate this study using lab-created tweets to more precisely control for variations in language, sender, and so on, and determine under what conditions these effects persist. Along those lines, while we tested five tweets in each condition, we cannot be certain that our findings generalize to all tweets about voter suppression with or without solutions. Future research should use even more stimuli variation in terms of content, source, length, and other variables, and may want to test voter fraud messages directly against voter suppression messages or compare the effects of true and artificial warnings about voter suppression. Finally, our study focuses on the attitudinal consequences of reading about voter suppression; we leave behavior effects for future research.

Conclusion

Advocacy groups need to communicate to the public about efforts to suppress qualified voters from casting a ballot. It is equally as important for these messages to include individual-level solutions to prevent their efforts from unintentionally reducing confidence in elections. Tweets that only highlight suppression efforts depress confidence in elections overall; tweets that emphasized individual-level solutions perform no differently than those in the placebo condition. Advocacy groups should make sure to include mechanisms by which individuals can actively mitigate harm to avoid creating even greater harm in which more voters who oppose suppression efforts themselves are dissuaded from voting, thus exacerbating the problem.

Supplemental Material

sj-docx-1-ctp-10.1177_20570473241270602 – Supplemental material for Tweet no harm: Offer solutions when alerting the public to voter suppression efforts

Supplemental material, sj-docx-1-ctp-10.1177_20570473241270602 for Tweet no harm: Offer solutions when alerting the public to voter suppression efforts by Katherine Haenschen, Bethany Albertson and Sharon Jarvis in Communication and the Public

Footnotes

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This project was supported by The Democracy Fund, Grant #R-201805-02425.

ORCID iD

Katherine Haenschen

Supplemental material

Supplemental material for this article is available online at .

Notes

Author biographies

Katherine Haenschen is an assistant professor jointly appointed to the departments of communication studies and political science at Northeastern University. Her research focuses on the intersection of digital media and political behavior.

Bethany Albertson is an associate professor in the government department at the University of Texas at Austin. Her primary research areas are political psychology and American politics.

Sharon Jarvis is a professor in the communication studies department at the University of Texas at Austin. She researches the intersection of language use, politics, and persuasion.

References

Achen

C. H.

Blais

(2016). Intention to vote, reported vote and validated vote. In Elkink

J. A.

Farrell

D. M

. (Eds), The act of voting (pp. 195–209). Routledge.

Albertson

Guiler

(2020). Conspiracy theories, election rigging, and support for democratic norms. Research & Politics, 7(3), 2053168020959859.

Alvarez

R. M.

Hall

T. E.

Llewellyn

M. H.

(2008). Are Americans confident their ballots are counted? The Journal of Politics, 70(3), 754–766.

Ansolabehere

Hersh

(2012). Validation: What big data reveal about survey misreporting and the real electorate. Political Analysis, 20(4), 437–459.

Barney

D. J.

Rhodes

J. H.

(2017). Framing voting restrictions: How implicit racial, colorblindness, and government fairness frames influence White support for stringent election policies [Conference session]. Midwest Political Science Association Annual Meeting, Chicago, IL.

Beaulieu

(2014). From voter ID to party ID: How political parties affect perceptions of election fraud in the US. Electoral Studies, 35, 24–32.

Berlinski

Doyle

Guess

A. M.

Levy

Lyons

Montgomery

J. M.

Nyhan

Reifler

(2021). The effect of unsubstantiated claims of voter fraud on confidence in elections. Journal of Experimental Political Science. https://doi.org/10.1017/XPS.2021.18

Biggers

D. R.

(2021). Can the backlash against voter ID laws activate minority voters? Experimental evidence examining voter mobilization through psychological reactance. Political Behavior, 43(3), 1161–1179.

Bock

R. D.

(1985). Multivariate statistical methods in behavioral research. McGraw-Hill.

10.

Bowler

Carreras

Merolla

J. L.

(2023). Trump tweets and democratic attitudes: Evidence from a survey experiment. Political Research Quarterly, 76(3), 1325–1339.

11.

Bowler

Donovan

(2016). A partisan model of electoral reform: Voter identification laws and confidence in state elections. State Politics & Policy Quarterly, 16(3), 340–361.

12.

civilrights.org. (n.d.) Democracy diverted: Polling Place closures and the right to vote. The Leadership Conference on Civil and Human Rights. https://civilrights.org/democracy-diverted/

13.

Clayton

(2023). Understanding public attitudes toward restrictive voting laws in the United States. Research & Politics, 10(4), 20531680231206611.

14.

Clayton

Davis

N. T.

Nyhan

Porter

Ryan

T. J.

Wood

T. J.

(2021). Elite rhetoric can undermine democratic norms. Proceedings of the National Academy of Sciences of the United States, 118(23), Article e2024125118.

15.

Clifford

Sheagley

Piston

(2021). Increasing precision without altering treatment effects: Repeated measures designs in survey experiments. American Political Science Review, 115(3), 1048–1065.

16.

Dale

(2022, June 16). 10 Trump election lies his own officials called false. CNN. https://www.cnn.com/2022/06/16/politics/fact-check-trump-officials-testimony-debunking-election-lies/index.html

17.

Daniels

G. R.

(2020). Uncounted: The crisis of voter suppression in America. New York University Press.

18.

Druckman

J. N.

Kam

C. D.

(2011). Students as experimental participants: A defense of the “narrow data base.” In Druckman

J. N.

Greene

D. P.

Kuklinski

J. H.

Lupia

(Eds), Cambridge handbook of experimental political science (Vol. 1, pp. 41–57). Cambridge University Press.

19.

Federal Bureau of Investigation. (n.d.-a). Election crimes. https://www.fbi.gov/investigate/public-corruption/election-crimes

20.

Federal Bureau of Investigation. (n.d.-b). Election crimes and security. https://www.fbi.gov/how-we-can-help-you/scams-and-safety/common-scams-and-crimes/election-crimes-and-security

21.

Fortin-Rittberger

Harfst

Dingler

S. C.

(2017). The costs of electoral fraud: Establishing the link between electoral integrity, winning an election, and satisfaction with democracy. Journal of Elections, Public Opinion & Parties, 27(3), 350–368.

22.

Freedman

D. A.

(2008). Randomization does not justify logistic regression. Statistical Science, 23, 237–249.

23.

Gardner

Reinhard

(2018, November 6). Broken machines, rejected ballots and long lines: Voting problems emerge as Americans go to the polls. The Washington Post. https://www.washingtonpost.com/politics/broken-machines-rejected-ballots-and-long-lines-voting-problems-emerge-as-americans-go-to-the-polls/2018/11/06/ffd11e52-dfa8-11e8-b3f0-62607289efee_story.html

24.

Gerber

A. S.

Green

D. P.

(2012). Field experiments: Design, analysis, and interpretation. W.W. Norton.

25.

Gilens

Murakawa

(2002). Elite cues and political decision-making. Research in Micropolitics, 6(1), 15–49.

26.

Grant

M. D.

Flores

Pedersen

E. J.

Sherman

D. K.

Van Boven

(2021). When election expectations fail: Polarized perceptions of election legitimacy increase with accumulating evidence of election outcomes and with polarized media. PLOS ONE, 16(12), Article e0259473.

27.

Green

Hobbs

McCabe

Lazer

(2022). Online engagement with 2020 election misinformation and turnout in the 2021 Georgia runoff election. Proceedings of the National Academy of Sciences, 119(34), Article e2115900119.

28.

Hardy

(2019). Voter suppression post-Shelby: Impacts and issues of voter purge and voter ID laws. Mercer Law Review, 71, 857.

29.

Hill

(2021). Information about youth voter suppression fails to induce significant voter backlash effect [Working paper]. https://youngamericans.berkeley.edu/2021/09/backlash-to-voter-suppression/

30.

Kane

J. V.

(2017). Why can’t we agree on ID? Partisanship, perceptions of fraud, and public support for voter identification laws. Public Opinion Quarterly, 81(4), 943–955.

31.

League of Women Voters. (n.d.). Fighting voter suppression. https://www.lwv.org/voting-rights/fighting-voter-suppression

32.

Levy

(2021). Winning cures everything? Beliefs about voter fraud, voter confidence, and the 2016 election. Electoral Studies, 74, 102156.

33.

Lopez

(2018). Voter suppression really may have made the difference for Republicans in Georgia. Vox. https://www.vox.com/policy-and-politics/2018/11/7/18071438/midterm-election-results-voting-rights-georgia-florida

34.

Marland

Lawlor

Giasson

(2018). Political elites in the age of digital media. In Marland

Lawlor

Giasson

(Eds.), Political elites in Canada: Power and influence in instantaneous times (pp. 5–28). UBC Press.

35.

McCarthy

Clifton

(2016). Update: Americans’ confidence in voting, election. Gallup. https://news.gallup.com/poll/196976/update-americans-confidence-voting-election.aspx

36.

Meeks

(2017). Getting personal: Effects of Twitter personalization on candidate evaluations. Politics & Gender, 13(1), 1–25.

37.

Mullinix

K. J.

Leeper

T. J.

Druckman

J. N.

Freeze

(2015). The generalizability of survey experiments. Journal of Experimental Political Science, 2(2), 109–138.

38.

Newkirk

(2018, October 24). Voter suppression is the new old normal. The Atlantic. https://www.theatlantic.com/politics/archive/2018/10/2018-midterms-and-specter-voter-suppression/573826/

39.

Norris

(2013). Does the world agree about standards of electoral integrity? Evidence for the diffusion of global norms. Electoral Studies, 32(4), 576–588.

40.

Norris

(2019). Do perceptions of electoral malpractice undermine democratic satisfaction? The US in comparative perspective. International Political Science Review, 40(1), 5–22.

41.

Olson

(2024, February 2). Two scholars revisit Trump’s election fraud claims. Cato Institute. https://www.cato.org/blog/two-scholars-revisit-trumps-election-fraud-claims

42.

Rogers

Aida

(2014). Vote self-prediction hardly predicts who will vote, and is (misleadingly) unbiased. American Politics Research, 42(3), 503–528.

43.

Root

Barclay

(2018, November 28). Voter suppression during the 2018 midterm elections. Center for American Progress. https://www.americanprogress.org/article/voter-suppression-2018-midterm-elections/

44.

Sances

M. W.

Stewart

III . (2015). Partisanship and confidence in the vote count: Evidence from US national elections since 2000. Electoral Studies, 40, 176–188.

45.

Scacco

J. M.

Lawrence

R. G.

Tenenboim

(2016). The “documented voter”: Voter Id Messaging in the 2014 Texas Midterm Election. In Hendricks

J. A.

Schill

(Eds.), Communication and midterm elections: Media, message, and mobilization (pp. 43–56). Springer.

46.

Scott

(2018, July 24). The odds that democrats will take the house in the 2018 midterms keep looking better. Vox. https://www.vox.com/policy-and-politics/2018/7/24/17607138/2018-midterm-elections-predictions-generic-ballot-democrats

47.

Sheagley

Udani

(2021). Multiple meanings? The link between partisanship and definitions of voter fraud. Electoral Studies, 69, 102244.

48.

Sinclair

Smith

S. S.

Tucker

P. D.

(2018). “It’s largely a rigged system”: Voter confidence and the winner effect in 2016. Political Research Quarterly, 71(4), 854–868.

49.

Singh

Carter

(2023). States have added nearly 100 restrictive laws since SCOTUS gutted the Voting Rights Act 10 years ago. Brennan Center for Justice. https://www.brennancenter.org/our-work/analysis-opinion/states-have-added-nearly-100-restrictive-laws-scotus-gutted-voting-rights

50.

Strömbäck

(2005). In search of a standard: Four models of democracy and their normative implications for journalism. Journalism Studies, 6(3), 331–345.

51.

Sullivan

(2018). Voter suppression is a crucial story in America, but broadcast news mostly shrugs. The Washington Post. https://www.washingtonpost.com/lifestyle/style/voter-suppression-is-a-crucial-story-in-america-but-broadcast-news-mostly-shrugs/2018/10/29/4f610768-db74-11e8-b3f0-62607289efee_story.html?utm_term=.8763a7acd580

52.

Tyson

(2018). The 2018 midterm vote: Divisions by race, gender, education. Pew Research Center. https://www.pewresearch.org/fact-tank/2018/11/08/the-2018-midterm-vote-divisions-by-race-gender-education/

53.

usa.gov. (n.d.). Voter fraud, voter suppression, and other election crimes. https://www.usa.gov/voter-fraud

54.

Valentino

N. A.

Neuner

F. G.

(2017). Why the sky didn’t fall: Mobilizing anger in reaction to voter ID laws. Political Psychology, 38(2), 331–350.

55.

Vonnahme

Miller

(2013). Candidate cues and voter confidence in American elections. Journal of Elections, Public Opinion & Parties, 23(2), 223–239.

56.

Wilder

(2021). Voter Suppression in 2020. Brennan Center for Justice. https://www.brennancenter.org/our-work/research-reports/voter-suppression-2020

57.

Wilder

Baum

(2022). 5 egregious voter suppression laws from 2021. Brennan Center for Justice. https://www.brennancenter.org/our-work/analysis-opinion/5-egregious-voter-suppression-laws-2021

58.

Wilson

D. C.

Brewer

P. R.

(2013). The foundations of public opinion on voter ID laws: Political predispositions, racial resentment, and information effects. Public Opinion Quarterly, 77(4), 962–984.

59.

Wines

(2017, March 23). One rationale for voter ID debunked, G.O.P. has another. The New York Times. https://www.nytimes.com/2017/03/23/us/election-fraud-voter-ids.html

60.

Zaller

J. R.

(1992). The nature and origins of mass opinion. Cambridge University Press.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

4.26 MB