Research conducted during the past two decades has concluded that there is no evidence of significant election fraud in contemporary American national elections (Alvarez and Katz, 2008; Mebane, 2008; Herron, 2019; Eggers et al., 2021). This reality has not prevented questions about the integrity of the electoral process in the United States. Sometimes those questions arise due to misunderstandings or misperceptions about how elections are conducted in the United States. A good example is the “blue shift” in reporting of election results after Election day; observers sometimes may wonder why reported election results shift from one party to another in the days or weeks after an election. But researchers have shown that these shifts are not due to election fraud, rather they arise from distinct patterns in the timing of ballot return, processing, and tabulation (Foley and Stewart, 2020; Li et al., 2022).

But sometimes these questions about the integrity of American elections persist, and take on a life of their own, even after concerns about election outcomes have been demonstrated to have legitimate and valid explanations. One of the most prominent examples of this occurred after the 2020 election, when then President Trump, his supporters, and some Republican officials, continued to press claims of election fraud despite evidence to the contrary. Situations where prominent political elites continue to press false claims of election fraud can be very problematic, as millions of American voters may continue to believe that election fraud exists in American national elections, which might affect their confidence in election administration and their willingness to participate in the electoral process (Green et al., 2022; Berlinski et al., 2023).

In this paper, we analyze data collected in June and July of 2024, from a sample of 2,211 American registered voters weighted to represent the electorate, to document the extent to which concerns about various types of election fraud, or types of election fraud, persist from 2020. We analyze in detail who among this sample believes that ten different types of types of election fraud are common or not common. This allows us to evaluate the salience of types of election fraud as America heads into a critical presidential election as well as for whom they are salient.

Twenty-four years ago, many researchers and scholars were shocked by the 2000 U.S. Presidential election which exposed problems with voting technology and administrative practices. The election was very close, with only a few hundred votes separating winner from loser in important battleground states like Florida. The closeness of the election brought a great deal of scrutiny of election administration and voting technologies, as prior to the election there was little research on U.S. election administration and technologies (Caltech/MIT Voting Technology Project, 2001). This attention to the details of election procedures and technologies helped develop new research areas in law, social science, and computer science (Alvarez et al., 2012).

The research agenda immediately after the 2000 U.S. presidential election focused heavily on election technology. As states and counties in the U.S. rushed to replace old lever and punchcard voting systems, some of them began acquiring and using new electronic voting machines. By 2004, concerns began to emerge about the security of electronic voting systems (Kohno et al., 2004). Others began to question whether procedural changes, like requiring government-issued photo identification in order to vote, might effect voter perceptions about election integrity, election confidence, and opinions about the extent to which types of election fraud are perpetrated in the U.S. This was a central argument in the United States Supreme Court's decision in the 2008 Crawford v. Marion County Election Board decision on requiring government-issued photo identification for voting (Ansolabehere and Persily, 2007; Atkeson et al., 2014). All of these questions and concerns led scholars to start studying what voters themselves thought about voting system security and election fraud (Alvarez and Hall, 2008b,a). This direction of research has become especially salient following the 2020 election and the events on January 6th, 2021.

Despite the scholarly interest in voter perceptions of election fraud, there has not been a great deal of quantitative research on the topic, nor has there been much theoretical development of a model of perceptions of election fraud. Most of the attention of scholars has been on measuring and analyzing the cross-sectional variance in survey measures of voter confidence and trust in elections (Atkeson and Saunders, 2007; Alvarez et al., 2008; Atkeson et al., 2015). In the literature, some studies have looked at the association between survey measures of election fraud and voter confidence, typically finding that voters more concerned about fraud have lower confidence and trust in the election process (Alvarez et al., 2021; Berlinski et al., 2023).

That said, the limited literature on the correlates of perceptions of election fraud have provided some important guidance for our study. First, and not surprising in a highly partisan and polarized environment, opinions about election fraud are rooted in voter partisanship and in how political parties frame discussions about election fraud (Beaulieu, 2014). Indeed, elite messaging plays an important role in forming public opinions, and shaping polarized partisan views (Bowler et al., 2015; Bowler and Donovan, 2016, 2024). Relatedly, it makes sense that the same “loser effect” seen in studies of voter confidence will be seen in perceptions of election fraud, with partisans of parties losing recent elections more likely to express concerns about election fraud (Anderson et al., 2005; Sances and Stewart, 2015; Sinclair et al., 2018), and more likely to support electoral reform (Hood and McKee, 2022, 2023, 2024). A particularly salient example of this effect followed the 2020 election when, after his electoral loss, Donald Trump claimed that Joe Biden's victory was fraudulent. Though untrue, a large majority of Republicans said they agree with his claim (Jacobson, 2023).

Some have studied perceptions about election fraud over election cycles, noting that while partisanship is correlated with opinions about election fraud, the relationship seems modest (Enders et al., 2021). Importantly these same studies have confirmed other research results, showing that conspiratorial thinking is correlated with perceptions about election fraud (Enders et al., 2021; Edelson et al., 2017). This underscores the general hypothesis that voters respond to their information environment, in particular what trusted elites say about election fraud (Cottrell et al., 2018). These studies provide the basic framework for our subsequent analysis of our recently collected 2024 dataset, which we describe in the next section.

We conducted an online survey of 2021 U.S. registered voters, which was fielded by YouGov June 26 to July 3, 2024. YouGov aims to produce a dataset that would approximate a representative sample of this population. They weighted the responses (as described in more detail in the Supplementary material), and in the analyses reported below, we use the weights provided by YouGov. The margin of error for the survey is approximately 2.4%.

In our survey, we asked subjects to tell us how commonly they believe that ten examples of election fraud occur in the United States. Subjects could indicate that they believed for each of these types of election fraud that “it is very common”, “it occurs occassionally”, “it occurs infrequently”, “it almost never occurs”, or “I'm not sure.” We provide in the Supplementary material the complete data for all response options to each of the ten election fraud questions. However, for analytical purposes and ease of exposition, in the paper we focus on the responses aggregated to common (it is very common or it occurs occasionally) or not common (it occurs infrequently or it almost never occurs). We provide additional methodological details about the survey in the Supplementary Sections 1, 2.

These are similar to, and in some cases identical to, questions that have been included in past surveys, in particular the The Survey of the Performance of American Elections (Stewart, 2022). This allows us to compare our responses from the summer of 2024 to those from previous surveys, specifically those from 2012, 2014, 2016, 2020, and 2022 (Stewart, 2013, 2015, 2017, 2021, 2023). The ten election fraud questions (and the labels we use for them) are:

Our analysis of these data proceeds as follows. We first show graphically the responses we received in our survey, and we also compare ours to those from the SPAE. We then discuss the simple bivariate associations we have in our data, between each of the election fraud questions, and our demographic and attitudinal features (the detailed tables are provided in the Supplementary material). Next, we use multivariate regression models to examine the covariates from our survey which most strongly associate with belief in election fraud. In the main text below, we focus on a summary scale of the number of types of election fraud that each subject in our sample endorses; we use this summary regression for ease of presentation and interpretation. In the main text we show marginal effect plots to clarify the associations between each covariate and the summary scale of the types of election fraud. We provide regression results for each of the election fraud beliefs in the Supplementary Section 4. Our survey contains quite a few covariates of interest: age, gender, race and ethnicity, education, region of residence, urban or rural residence, partisanship, ideology, political interest.

The survey also contains batteries of questions designed to produce measures of beliefs in conspiratorial thinking and populism, the wordings of these questions are provided in the Supplementary Section 1. To form conspiracy and populism scores for each participant, we ask six conspiracy and populism questions, rated on a five-point scale, which we then sum to form each of these measures. These measures thus range from a minimum of 6 to a maximum of 30. Finally, we bin these scores in 5-point intervals to present the results.

Our analysis of perceptions of types of election fraud among American registered voters in 2024 has several findings with implications for democratic governance and electoral integrity in the United States. First and foremost, our results demonstrate that beliefs in various forms of election fraud remain widespread, despite a lack of evidence for their occurrence at any meaningful scale. This persistence of election fraud beliefs, even years after the contentious 2020 election, suggests that these perceptions may have become deeply entrenched in the American political psyche.

One of the most striking findings is the extent to which party affiliation correlates with beliefs in types of election fraud. Republicans consistently express higher levels of belief in various forms of election fraud compared to Democrats, even when controlling for other factors such as demographics, political engagement, and general conspiracy thinking. This partisan divide in perceptions of electoral integrity is concerning, as it may contribute to further polarization and undermine the legitimacy of election outcomes for a significant portion of the electorate. Whether this is evidence of a “loser effect” or might be a more lasting reflection of the partisan polarization in the United States today will require additional research after the 2024 election.

The strong association between belief in non-election conspiracies and acceptance of election fraud narratives is also noteworthy. This relationship suggests that susceptibility to election fraud beliefs may be part of a broader pattern of conspiratorial thinking. As such, efforts to address election fraud myths may need to consider broader strategies for improving critical thinking and media literacy among the general public.

Interestingly, our findings challenge some common assumptions about the rural-urban divide in political trust. Contrary to expectations, we found that rural respondents, particularly rural conservatives, were less likely to believe in types of election fraud compared to their urban counterparts. This unexpected result warrants further investigation and may have implications for how we understand the geographic distribution of political attitudes and beliefs.

Another counterintuitive finding is the relationship between political engagement and belief in types of election fraud. Our results indicate that those who follow politics less frequently are actually less likely to believe in election fraud. This suggests that increased exposure to political information, particularly among Republicans, may paradoxically increase susceptibility to election fraud narratives. This finding raises important questions about the quality and sources of political information consumed by engaged citizens and the potential echo chamber effects in partisan media ecosystems.

Our findings underscore the need for new approaches to address entrenched beliefs in election fraud. Experimental studies should explore multifaceted interventions that combine information provision with techniques drawn from research on countering conspiracy theories. For instance, researchers should continue to test the effectiveness of “inoculation” methods, exposing individuals to weakened forms of misinformation to build resistance against future encounters (see, for example van der Linden et al., 2017, 2021, 2022; Lewandowsky et al., 2023; Carey et al., 2024). Additionally, longitudinal studies tracking changes in election fraud beliefs over time, particularly before and after major elections, could provide valuable insights into the durability of these perceptions and the factors that influence their evolution.

This study has provided a comprehensive snapshot of American voters' perceptions of types of election fraud in 2024, revealing the persistent and partisan nature of these beliefs. By examining a wide range of election fraud narratives and their correlates, we have illuminated the interplay between partisanship, conspiracy thinking, political engagement, and geographic factors in shaping these perceptions. Our analysis reveals that despite the passage of time since the contentious 2020 election, beliefs in various forms of election fraud remain widespread, challenging assumptions about rural-urban divides and the effects of political engagement. These findings highlight the challenges facing American democracy as it contends with pervasive skepticism about electoral integrity, even in the absence of evidence for significant fraud.

As we approach future elections, the task of rebuilding trust in the American electoral system is both urgent and formidable. Yet, throughout its history, American democracy has faced and overcome significant challenges. By deepening our understanding of the roots of election fraud beliefs and developing evidence-based strategies to address them, we can work toward a future where faith in the democratic process transcends partisan divides.