The anatomy of a resurging rumor stemming from peer-reviewed research that non-citizens vote in U.S. elections

Jun 20, 2024

A sign that says "Vote Here" with an arrow.

2024 U.S. ELECTIONS RAPID RESEARCH BLOG

By Mert Can Bayar, Ashlyn B. Aske, Danielle Lee Tomson, Kate Starbird, and Jevin West
Center for an Informed Public
University of Washington

This is part of an ongoing series of rapid research blog posts and rapid research analysis about the 2024 U.S. elections from the University of Washington’s Center for an Informed Public.

  • A 2016 election rumor using peer-reviewed research has resurged, this time claiming that 10 to 27% of non-citizens vote in U.S. elections. The research had previously been instrumentalized by then-candidate Donald Trump but had been highly criticized by the academic community both before and after Trump cited it.
  • This blog post reviews the history of the rumor and the anatomy of its spread. It does not offer new data analysis about the peer-reviewed research itself. Instead, it details the history of the claim, records several critical points of the scholarly discussion in the last decade, and tracks the social media spread of this particular claim.
  • This blog post was written to assist journalists, fact-checkers, and the broader public in tracing the rumor’s origins and contextualizing its place in current debates about the 2024 U.S. election.
  • This rumor and tactic of using peer-reviewed research in misleading ways will likely re-emerge in the 2024 election or future election cycles. Our purpose here is to provide a comprehensive review of the rumor and highlight its broader implications for using peer-reviewed research and scientific methods to endorse and validate false rumors about integrity in the U.S. elections.

On May 13, a website called Just Facts published an article that renewed attention to an old claim that non-citizens were illegally voting in U.S. elections. This rumor originates from a 2014 peer-reviewed paper and initially spread in 2016 when President-elect Trump voiced it both before and after the U.S. election. It has resurfaced again in 2024, amassing 26 million views on X over two days.[1] Though the spread of the rumor on social media has since waned, the underlying claim is important to study because it leverages peer-reviewed scientific research in misleading ways, which can contribute to diminishing trust in academic knowledge and institutions as well as elections. We expect to see this rumor and this tactic of leveraging science and statistics in misleading ways to be repeated this election cycle. 

Claims about non-citizen voting are common in election rumoring. Significant portions of the U.S. public already believe in similar false rumors and conspiracy theories about non-citizens’ involvement in U.S. elections. In the past decade, the Just Facts website has posted multiple articles endorsing false claims of election fraud and non-citizen voting. In their recent article, they claim that 10% to 27% of non-citizens, including documented and undocumented immigrants, vote in U.S. elections. This claim extends the findings of a 2014 paper, which was met with heavy criticism from the academic community. In said paper, political scientist Jesse Richman and his co-authors analyzed 2008 and 2010 CES survey data, an established dataset in political science that samples a representative group of American adults for social scientific research. According to their analysis, Richman and his co-authors concluded that 6.4% of noncitizens voted in the 2008 elections, and 2.2% voted in the 2010 elections. They suggested that these numbers could have had a significant impact on the results of the 2008 presidential elections and key U.S. House and Senate races. However, this study was widely refuted in public and academic discussions, critiquing the sample size, data, and methods, among other points. 

In this rapid research post, our team at the University of Washington’s Center for an Informed Public explains the history of this claim, the scholarly discussion surrounding the rumor in the last decade, and the spread of the rumor on several social media platforms in mid-May 2024 following the publication of the Just Facts article.

Significance of This Tactic

This rumor is worth our attention due to its validation from a highly contested peer-reviewed study, which was deployed to add credibility to the false claims of the scale of non-citizen voting. This credibility or epistemic authority, combined with the misuse of scientific methods and data, is a typical pattern in the spread of misinformation about science, public health, and elections. Political elites may inappropriately use scientific articles to evidence narratives in their interest, even if authors did not intend their work to be used as such. 

Online influencers, newsbrokers, and elites can use such articles to validate their claims and fit them into their narratives. If the scientific article is discredited or contested in the scientific community, other influencer or newsbroker communities can use the same case to discredit the scientific peer-review process and feed larger mistrust in the scientific method and established institutions of scientific inquiry (e.g., peer-review process, scientific journals, researchers or research institutions).

The Present and Past Controversy Around the Peer-Reviewed Article 

The Just Facts article was written by James Agresti, who identifies as an independent researcher. In the article, Agresti claims to have conducted a new and improved data analysis of the non-citizen voting phenomenon and claims that 10 to 27 percent of non-citizens — again, undocumented immigrants and documented immigrants without citizenship — vote in the U.S. elections. According to fact-checkers and academic researchers, this claim is false. Agresti’s study is based on Richman’s findings, which are flawed and unreliable. 

On May 26, Alex Kasprak from Snopes published a fact check on Agresti’s recent findings and Richman’s 2014 article. Snopes’ fact check includes email interviews with the author of the Just Facts article, James Agresti; the leading author of the 2014 article, Jesse Richman; and Brian Shaffner, who was one of the primary investigators of CES data that have been used in the 2014 article and also one of the authors of a 2015 article that addressed the claims made by Richman and his colleagues. This is not Agresti’s first claim about non-citizen voting. He made a similar claim regarding the 2016 elections. Alex Kasprak did a similar fact-check on the issue in 2017. 

The 2014 article that started this rumor, “Do non-citizens vote in U.S. elections?,” was written by three political scientists, Jesse Richman, Gulshan Chatta, and David Earnest, and published in Electoral Studies, a well-known peer-reviewed journal in quantitative political science. The article used an established dataset, CCES (now known as CES). The authors estimated that 6.4 percent of non-citizens voted in the 2008 elections and 2.2 percent voted in the 2010 elections. These percentage estimates are based on a total of 29 survey respondents: 21 respondents (6.4%) in 2008 indicated they were non-citizens and reported voting, while 328 indicated they were non-citizens; 8 respondents (2.2%) in 2010 indicated they were non-citizens and reported voting, while 363 indicated they were non-citizens. The respondents were sampled from a larger population from the CES survey of more than 30,000 people. The authors claimed that these percentages were enough to change the outcomes of the 2008 elections, including Electoral College votes for the presidency and the composition of Congress.

The political science community has heavily disputed and discredited the 2014 article and its findings–well before the findings were politicized and amplified in claims made by then-candidate Donald Trump. In fact, the scholars who ran the 2008 and 2010 CCES surveys — the dataset Richman and colleagues used to write the 2014 article — disputed the article’s findings. They published a short response article in 2015 stating that measurement error (people who accidentally reported that they are non-citizens although they are citizens) in the 2008 and 2010 CCES surveys explain the findings of non-citizen voting. They state that the percentage of non-citizen votes in the U.S. elections is likely zero.  

The claim that non-citizens vote in high numbers in U.S. elections got traction when Donald Trump voiced it after he lost the popular vote to Hillary Clinton in the 2016 presidential elections. Trump attributed this loss to “illegal votes” and referred to the 2014 article as evidence for his claims. It is important to note that the leading author of the 2014 article, Jesse Richman, has been interviewed multiple times defending his research findings, even if he has refuted that his study supports Trump’s claims of widespread non-citizen voting.  In an interview with Wired, Richman defended their 2014 findings but stated, “Trump and others have been misreading our research and exaggerating our results to make claims we don’t think our research supports.” 

He further noted that the partisan nature of the discussion is similar to “a two-front war.” He stated that people on the left aim to discredit their study on often flimsy grounds, while people on the right want to pretend this study is much more than it is or says much more than it does. That said, Richman has been chosen as an expert witness in several lawsuits related to non-citizen voting. His latest estimates of non-citizen voter registration are between 0 and 1 percent, drastically lower than the estimates they presented in the 2014 paper. 

Following Trump’s statements and growing attention to the 2014 paper in the media, in 2017, 200 political scientists signed a letter contesting the findings of the 2014 article. In this letter, they concluded that “the scholarly political science community has generally rejected the findings in the Richman et al. study, and we believe it should not be cited or used in any debate over fraudulent voting.”[2] Yet the scientific community had critiqued this article long before Trump had deployed it to credential his claims. 

Academics’ Rebuttals of the Peer-Reviewed Article Before It Was Used in a Rumor 

The 2014 article received public and scholarly attention right after it was accepted for publication, especially after the authors Richman and Earnest wrote a summary of their findings on The Washington Post’s Monkey Cage blog on October 21, 2014. In this piece, titled “Could non-citizens decide the November Election?” Richman repeated their claims and noted the limitations of their work. The Washington Post wrote a note at the beginning of this article citing and linking to several other rebuttals of Richman and Earnest’s piece as well as Richman and Earnest’s response to the rebuttals.  

One of these rebuttals came from Michael Tesler, a political scientist who wrote a response piece titled “Methodological challenges affect study of non-citizens’ voting” for the Monkey Cage blog on October 28, 2014. The piece raised possible methodological concerns about the 2014 article and summarized its traction in conservative media. 

Political scientists James McCann and Michael Jones-Correa wrote a response piece titled “Are non-citizens following American election laws?” In this piece, they analyzed survey data asking about campaign donation behavior on Latino immigrants from the 2012 Latino Immigrant National Election Study (LINES), a nationally representative survey of immigrants from the Spanish-speaking countries of Latin America. While the dataset does not include questions on voting or registration, the authors looked at survey respondents’ answers on campaign donations. They found that almost none of the non-citizens sampled in the dataset donated money to candidates in the 2012 elections, indicating that the non-citizens’ engagement in election-related activities is likely zero.

In another rebuttal, John Ahlquist and Scott Gehlbach, also political scientists, titled “What can we learn about the electoral behavior of non-citizens from a survey designed to learn about citizens?” pointed out that the CCES survey question on voter registration asked, “are you registered to vote” but did not specify in which country. They stated that might have confused some survey respondents (who may have been registered to vote in another country) and caused a few incorrect answers. 

On November 2, 2014, Richman and Earnest responded to the critics on the same platform. In their response, titled “Do non-citizens vote in U.S. elections? A reply to our critics,” the co-authors defended their article and its findings on the extent to which non-citizens vote in the U.S. elections, which they estimated to be 6.4 percent in the 2008 elections and 2.2 percent in the 2010 elections.  They responded to Tesler’s criticism of measurement error (citizens who mistakenly chose the non-citizen box in the survey), stating that both in 2010 and 2012, 85 respondents chose the non-citizen box (twice), and 10 of these 85 declared voting. Even though Richman and Earnest accept that CES does not provide a representative sample of non-citizens, they argue that they conducted their analyses in an appropriate manner. 

On the other hand, Richman’s recent estimates of non-citizen voter registration (not actual voting) stand between 0 and 1 percent. These estimates are based on Richman’s analysis as an expert witness in an Arizona investigation on non-citizen voter registration and the recent CES dataset, where he concluded that non-citizens did not decide the Arizona Presidential election. Glenn Kessler from The Washington Post requested these court documents be made public and reported on March 2024, almost two months before Agresti’s article was published.

The Resurgence of the Rumor in 2024

We analyzed the re-emergence of this rumor on several social media platforms, including Twitter/X, TikTok, Gettr, YouTube, Truth Social, Facebook, Telegram, and Gab. However, the restrictions on researchers’ data access on many of these platforms limit our abilities to conduct quantitative analyses on the dynamics of the spread and prevent us from mapping its engagement network. Therefore, please consider this analysis exploratory only, not a comprehensive analysis of the rumor. 

Overall, the false claim that “10 to 27 percent of non-citizens vote in U.S. elections” received a contained spread on Twitter/X, receiving attention from a few newsbroker accounts and a small number of elite replies that boosted its visibility on the platform. Its spread on other platforms as of the first week of June 2024 remains relatively low. The spread is largely contained between May 15 and May 17, 2024.

Figure 1 illustrates the frequency of posts containing false claims across various social media platforms. The peak activity occurred over two days (May 15-17), primarily on Twitter/X.

Figure 1 illustrates the frequency of posts containing false claims across various social media platforms. The peak activity occurred over two days (May 15-17), primarily on Twitter/X.

On X, what appears to be the account of the Just Facts website tweeted about Agresti’s article on May 13, 2024. This tweet received little attention; its views remain just over 1400. The article and the rumor started receiving more attention when an account named @LeadingReport tweeted about it as “breaking” news on May 15, 2024. Leading Report appears to follow a similar tweet-sharing structure to many “newsbroker” accounts we follow, in general, sharing information with a breaking news framing. In particular, the style of Leading Reports’ tweets reminds us of @spectatorindex, another account we covered in a previous blog post.[3]

Leading Report’s tweet significantly increased the visibility of the Just Facts article with its 1.8 million views. In the next two days, the false rumor received more attention, adding up to a total of 26 million views, with several tweets and replies from newsbrokers and elite accounts. The most viewed tweets were from @Not_the_Bee, with 24.4 million views, and @kylenabecker, with 466 thousand views.

Elon Musk interacted with both of these accounts’ tweets, including a reply to Kyle Becker and a quoted tweet of Not The Bee’s tweet. In both instances, Musk mentioned Community Notes to fact-check the accuracy of the claim. We have not seen any Community Notes on either of these tweets as of the publication of this blog post. Musk’s reply to Kyle Becker has been viewed 93,000 times, while his quoted tweet of the Not the Bee has over 24 million views.

We suspect that the 24 million views of Musk’s quote of the Not the Bee’s tweet drive this rumor’s high visibility. A quick glance at Not the Bee’s tweets and their views reveals that its views are often lower than 100 thousand. For instance, a similar tweet posted on May 20 about immigration got 47 thousand views. Therefore, in our estimates of visibility, we do not add Musk’s quoted tweet into our calculations since it is likely that 24 million views of Not the Bee’s own tweet and 24 million views of Musk’s quoted tweet share a highly overlapping viewership. Therefore, we conclude that the total viewership of this rumor on X is close to 26 million rather than 50 million.

Figure 2 shows the two influential tweets in this contained spread and Musk’s interactions with them. @kylenabecker’s reply to Musk involves a link to the 2014 article.

Figure 2 shows the two influential tweets in this contained spread and Musk’s interactions with them. @kylenabecker’s reply to Musk involves a link to the 2014 article.

The spread’s anatomy: Evidence is produced by a peer-reviewed study, newsbrokers post it to resonate with existing narratives, and elite accounts boost its visibility. 

We have seen a similar trend of spread in this false rumor about non-citizen voting that we saw and analyzed before. In this case, evidence (of non-citizen voting) is produced by a heavily disputed academic article and a new article that claims to replicate and extend its findings. Then, the newsbroker accounts such as @LeadingReport use this evidence to advance existing narratives such as “millions of illegal immigrants vote” or “Democrats cheat.” The newsbrokers’ visibility is boosted even more when an elite account — in this case, Elon Musk’s account — interacts with them. It is important to note that Elon Musk’s account has not endorsed the claim but rather boosted its visibility.

We will likely see some variation of this rumor’s return before the 2024 U.S. elections. This particular rumor is valuable to pay attention to because its validation (and/or evidence) comes from a highly contested peer-reviewed study. This scientific validation opens up another level of epistemic authority to endorse the false claims of non-citizen voting in the public space. 

Previously, we have noted that many of these election rumors are inherently conspiratorial since they assume malevolent intent (of political actors) with a secret plot (e.g., mobilizing non-citizens to vote and alter election results). Since non-citizen voting became a salient issue in the 2024 elections, it will be crucial to understand its causes (e.g., the migrant crisis at the border), its implications for U.S. public opinion, and the pattern of its reemergence and spread on social media platforms. 

For further reading about misinformation about science: 

Author Information

  • Mert Can Bayar is a postdoctoral scholar at the University of Washington’s Center for an Informed Public.
  • Ashlyn B. Aske is a CIP graduate research assistant and a Master of Jurisprudence student at the UW School of Law.
  • Danielle Lee Tomson is a CIP research manager. 
  • Kate Starbird is an associate professor in the UW Department of Human Centered Design & Engineering and a CIP co-founder.
  • Jevin West is an associate professor in the UW Information School and a CIP co-founder.

References

  • [1] In the current landscape of X, a tweet with 24 million views can be considered a viral tweet. For a reference point, the highest viewed tweet of BBC News (World) in the first three days of the Hamas Israel Conflict received 25 million views.

  • [2] Given the relatively small number of quantitative political scientists across all U.S., having 200 political scientists is a significant number that can be indicative of a scholarly consensus. We think that their letter reflects the prevailing view on the 2014 piece by Richman, Chatta, and Earnest within the discipline.

  • [3] Both Leading Report and Spectator Index tweets convey neutral tone and minimal sentences, maintaining an objective breaking news impression. Leading Report uses more visuals and video content while Spectator index does not. Leading Report’s content is partisan and pro-Trump whereas Spectator index generally maintains a non-partisan tone in its coverage.

Other News