Fact Check: The Truth Behind Claim '10 to 27%' of Noncitizens in US Are Illegally Registered to Vote
Claim:
A new study found that "10 to 27%" of noncitizens in the U.S. are registered voters.
Rating:
No "new study" concluded that 10 to 27% of noncitizens in the U.S. are registered to vote. Instead, one independent researcher, James Agresti, published a re-interpretation of a widely discredited 2014 paper to make untenable conclusions about non-citizen voting behavior in 2024.
The paper on which this claim is based produced estimates of noncitizen voter participation that even its author, Jesse Richman, no longer agrees with. The scientists responsible for collecting the data Richman used say it was a wildly inappropriate use of their data.
Despite these new developments, Agresti still relies on outdated estimates from that paper that are widely considered to be useless. Nonetheless, Agresti told Snopes he stands by his work.
In mid-May 2024, several conservative influencers began posting a new voter fraud talking point on social media platforms: "A new study shows that between 10% and 27% of noncitizens are illegally registered to vote."
The assertion was amplified by the conservative Heartland Institute and reshared by many conservative news and conspiracy theory websites. For example, Not The Bee, the right-leaning news arm of the satirical website Babylon Bee, posted a story based on these claims with the headline "up to a third of illegal immigrants are registered to vote, according to new study."
That headline garnered a repost from tech billonaire Elon Musk, who asked if the claim was accurate:
This claim was not accurate. It was merely the latest in a long line of mathematically misguided, headline-generating assertions originating from an organization named Just Facts, which describes itself as "a research and educational institute dedicated to publishing facts about public policies and teaching research skills." President and Co-Founder James Agresti authors most of the content on the site.
Since 2017, Agresti has repurposed the findings of a controversial 2014 study about noncitizen voting, conducted by Old Dominion University political scientist Jesse Richman, to argue the plausibility that noncitizens illegally participate in U.S. elections and that noncitizen votes have swayed past elections in favor of Democrats. Agresti uses Richman's research as the primary academic foundation for his claims.
That 2014 study used as its primary dataset a survey on voter behavior carried out every two years by polling firms on behalf of a Harvard University-led initiative now known as the Cooperative Election Survey (CES). The CES asks respondents "more than 100 questions regarding electoral participation, issue preferences, and candidate choices."
In 2017, over 200 political scientists and statisticians signed an open letter rejecting the reliability of Richman's findings. "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," they wrote. In 2018, Harvard's Stephen Ansolabehere, one of the principle scientists behind CES, testified in a trial challenging the constitutionality of a Kansas voter-registration law that Richman's estimates of noncitizen voter participation were "not statistically different than zero."
In this piece, Snopes explains why Agresti's claim that "between 10% and 27% of noncitizens are illegally registered to vote" is unsupported by the evidence he cites, and we place this new claim in the context of the long-standing, misinformation-generating debate from which it originated.
For this story, Snopes spoke by email with Agresti, Richman, and Brian Schaffner — a Tufts political scientist and another primary investigator on CES. Schaffner rejects Richman and Agresti's use of the CES to study noncitizens entirely, and Richman disputes some of Agresti's claims. Agresti told Snopes he stands by his work.
A Decade of Dubious Deductions
Through online surveys, CES generates a dataset of tens of thousands of voting-age Americans every two years. This large dataset is then adjusted to match a random sample of voting-age Americans based on census data.
In his 2014 paper, Richman studied a small subset of individuals from CES surveys for the 2008 and 2010 elections — the people who said in a voluntary survey regarding U.S. voting behavior that they were noncitizens. In 2008, from a pool of 32,800 CES respondents, 339 people indicated that they were noncitizens. Out of 55,400 respondents in 2010, 489 people indicated they were noncitizens.
Richman's conclusion that about a quarter of noncitizens are registered to vote stemmed, in 2014, principally from these two pools of self-described noncitizens. As described in the 2014 paper, his best "adjusted estimate" came from those pools of people and from a database, Catalyst, that CES uses to match respondents to voter records. The paper said in part (emphasis ours):
Our best guess at the true percentage of noncitizens registered uses the 94 (weighted) noncitizens from [the 2008 CES] for whom Catalyst obtained a match to commercial and/or voter databases to estimate the portion of noncitizens who either claim to be registered when they are not (35%) or claim not to be registered when they are (18%). We then use these numbers to extrapolate for the entire sample of noncitizens in 2008 and 2010.
Because most noncitizens who said they were registered were in fact registered, and quite a few who said they were not were actually registered, the adjusted estimate is the highest of the three estimates, indicating that roughly one quarter of noncitizens were likely registered to vote.
Much of the debate surrounding the 2014 Richman study, which Snopes covered in detail in 2017, focused on the fact that this sample size was so small relative to the larger pool that it could be accounted for, in large part, by actual citizens who did indeed register to vote but hit the wrong box on the citizenship question. A 2015 study by Shaffner and other CES colleagues published in 2015 raised this issue.
Schaffner and his colleagues identified a significant number of citizens who did incorrectly state they were noncitizens in those CES surveys, as he told Snopes in 2017:
In 2012, we re-interviewed 19,000 people who had been respondents for the 2010 [CES]. We asked them the same question about citizenship status as we had asked them in 2010. Of these 19,000, 121 had claimed to be noncitizens in 2010. In 2012, 36 of the 121 had changed their response to "citizen."
Additionally, 20 people who had clicked on the "citizen" option in 2010 changed to "non-citizen" in 2012. Thus, it is clearly the case that a small share of respondents were misclicking on response options to that question in at least one of the two surveys (about 0.3%).
Since 2016, when Agresti's claims about massive noncitizen election participation first went viral, the author says he has used an improved version of the method employed by the Richman study to calculate noncitizen voter registration rates in recent years.
In his most recent post, Agresti argued that "the latest available data and an enhanced version of a stress-tested methodology from a scholarly journal," — i.e., the Richman study — "has found that about 10% to 27% of non-citizen adults in the U.S. are now illegally registered to vote:
The U.S. Census recorded more than 19 million adult noncitizens living in the U.S. during 2022. Given their voter registration rates, this means that about two million to five million of them are illegally registered to vote. These figures are potentially high enough to overturn the will of the American people in major elections, including congressional seats and the presidency.
While this latest version of the "enhanced" Just Facts method does factor in CES data from 2022, it still fundamentally relies on the analyses of datasets from 2008 and 2010. Agresti's use of this old data is crucial, because Richman himself has largely moved on from the more sensational claims made in that study.
As an expert witness in several voter fraud-related cases since 2018, Richman has stated under oath that he believes the best estimate of the registration rate by noncitizens in the U.S. is below 1 percent.
An Expert Report Complicates Agresti's Claim
Richman's 2014 paper re-emerged as a topic of discussion in March 2024 — before Agresti made his most recent claim — thanks to documents produced in a 2023 case challenging the constitutionality of recently enacted voter ID laws in Arizona. As an expert witness, Richman was given access to Arizona Department of Motor Vehicle (DMV) records and voter rolls. He conducted two studies to identify noncitizens who registered to vote in Arizona using that data and an additional study with more recent CES data.
Those reports were initially filed under seal, but they were unearthed thanks to a public records request by Glenn Kessler of The Washington Post. In a March 2024 story, Kessler described what Richman found (emphasis ours):
Richman said he found [using Arizona DMV and voter records] that 1,934 voters (0.43 percent of Arizona voting-age noncitizens) had records that indicated they were not citizens at the time they registered or after registering to vote. …
Richman also examined 2022 data from the Cooperative Election Study, which now has a more precise way of asking a question about a voter's citizenship and a larger sample size than the surveys he studied for his 2014 paper.
He found that just under 1 percent of noncitizens indicated they had registered to vote — again, much smaller than his previous findings.
Via email, Snopes asked Richman if he felt as though these new analyses validated his critics' concerns about his earlier work. In response, Richman argued that the analyses lent some credence to the criticism, but also ameliorated other specific concerns with the work:
In terms of credence, the fact that the estimates are lower supports the possibility that there was some bias generated by response error on the citizenship status question as the critics alleged. …
[But] there is no evidence in the recent [CES] data of the kind of inconsistent responses to the citizenship question that critics hypothesized, so there isn't direct evidence for the critics' main claim [of response error].
There may have been no "direct evidence" for error on the citizenship question in the 2022 CES dataset, as Richman argues, but it would be misleading to suggest this silenced critics about the possibility of response error in earlier CES datasets, on other questions within that dataset, or with clerical errors in the datasets against which they were validated. Agresti, in an email to Snopes, falsely claimed that Richman's expert report "completely debunked" the notion of incorrect answers to the citizenship question.
Factually, there is evidence of citizens incorrectly stating they were noncitizens in those earlier CES datasets. And those datasets are the ones that provided the estimates with much higher figures that Agresti continues to cite. Agresti dismissed Richman's new numbers as merely "lowball estimates" that unfairly undercount noncitizens. Even Richman told Snopes such a conclusion is in error:
I think the 10 percent figure has a foundation in my work, though it needs to be caveated more clearly to indicate the risks of being biased upwards, and I think it is absolutely inappropriate as a lower bound.
The appropriate lower bound should be based upon individuals with validated registration status who said they were registered (i.e. less than one percent in 2022).
I don't think the 27 percent figure is reasonable, partly because of changes in the methodology of the CES, and partly because I don't fully understand how it was calculated.
Agresti, in an email to Snopes, argued that Richman was wrong, in part because "Richman used and defended the same methodology in 2016 and 2017." The inexorable forward march of time, such as it is, often produces new information that challenges previous work. Thanks to new data and enhanced methodologies, Richman does not agree with the high-end conclusions he produced a decade ago, and Agresti continues to rely on them in spite of that fact.
'Squeezing Blood from a Turnip'
It might be easy to get lost in the minutiae of Richman's work and Agresti's apparent misuse of it. A much more simple rebuttal to the claim that up to 27% of the noncitizen voting population is registered to vote comes from the fact that the CES survey underlying the claim is fundamentally unsuited to answer any question about noncitizens. Schaffner told Snopes:
The CES … is designed to sample American adults. Once the weights are applied, it is therefore a representative sample of American adults on a wide variety of demographic, socioeconomic, and political variables. It is not designed to be a sample of noncitizen adults and therefore it is not fit for the purpose of studying that subset of respondents.
Agresti, in a Just Facts post and in discussion with Snopes, employed a rhetorical sleight of hand to justify his use of the purported noncitizen data pool in CES in spite of this criticism. In essence, Agresti argues that because the CES itself used modeling to make its sample representative, he, too, can weight his noncitizen data pool — in his case to match the demographics of the noncitizen population.
"The data used for the 2014 Electoral Studies paper and the 2017 Just Facts study were both weighted to make the results nationally representative of the non-citizen population," Agresti explained to Snopes in an email.
That's not how any of this works, according to Schaffner. There are some things you simply cannot model away. One situation in which such modeling fails is when the pool of survey respondents you draw your study group from was collected in such a way that would exclude or undercount the very group you are studying, something known as a "noncoverage bias."
"No amount of adjustment or modeling like what Agresti describes is going to actually fix the issue that the CES is never intended to include a representative sample of noncitizens," Schaffner told Snopes. "Many or even most noncitizens are not going to be eager to sign up to take a survey that requires them to provide identifying information," he argued.
"There are much better ways to analyze whether noncitizens register to vote," Schaffner told Snopes. These methods, some of which Richman used in his recent expert reports, involve looking voter rolls and other state records to identify any individuals who appear to be noncitizens. Studies like these, including the Richman expert reports, "overwhelmingly resulted in finding very few noncitizens registered to vote," Schaffner told Snopes.
"The fact that people are still trying to use the CES to do a study like this reflects that they don't like the result one gets from using a much more valid approach and therefore they are trying to squeeze blood from a turnip instead," Schaffner argued.
The Bottom Line
The claim that 10 to 27% of noncitizens in America are registered to vote is not rooted in an academic, peer-reviewed study. It is, instead, based on one person's interpretation of a controversial, decade-old paper that the political science field has since overwhelmingly rejected.
Because no "new study" has come to this conclusion, because the "groundbreaking study" from 2014 on which the claim is based has been widely rejected, and because the author of that (discredited) study rejects its present use, we rate the claim as "False."
Sources:
Ansolabehere, Stephen, et al. "The Perils of Cherry Picking Low Frequency Events in Large Sample Surveys." Electoral Studies, vol. 40, Dec. 2015, pp. 409–10. ScienceDirect, https://doi.org/10.1016/j.electstud.2015.07.002.
Cooperative Election Study. https://cces.gov.harvard.edu/home. Accessed 23 May 2024.
Just Facts - About Us. 25 Nov. 2023, https://www.justfacts.com/aboutus.asp.
Just Facts - Immigration. 23 May 2024, https://www.justfacts.com/immigration#electoral.
Kessler, Glenn. "Analysis | The Truth about Noncitizen Voting in Federal Elections." Washington Post, 9 Mar. 2024. www.washingtonpost.com, https://www.washingtonpost.com/politics/2024/03/06/truth-about-noncitizen-voting-federal-elections/.
Mueller, Zachary. "Fish v. Kobach Trial - Day 6 ⋆ Institute for Research and Education on Human Rights." Institute for Research and Education on Human Rights, 14 Mar. 2018, https://irehr.org/2018/03/14/fish-v-kobach-trial-day-6/.
Office of Public Affairs | Court Finds That Arizona Voter Registration Provisions Violate Federal Law | United States Department of Justice. 1 Mar. 2024, https://www.justice.gov/opa/pr/court-finds-arizona-voter-registration-provisions-violate-federal-law.
"Peter Schweizer: Border Crisis Could Increase Number of Illegal Ballots Cast in 2024 — up to 2.7M." Breitbart, 15 May 2024, https://www.breitbart.com/politics/2024/05/15/peter-schweizer-biden-border-crisis-could-increase-illegal-ballots-cast-in-2024-up-to-2-7-million/.
"Quantifying Illegal Votes Cast by noncitizens in the Battleground States of the 2020 Presidential Election." Just Facts, 8 Nov. 2020, https://www.justfactsdaily.com/quantifying-illegal-votes-cast-by-noncitizens-in-the-battleground-states-of-the-2020-presidential-election.
Richman, Jesse T., et al. "Do noncitizens Vote in U.S. Elections?" Electoral Studies, vol. 36, Dec. 2014, pp. 149–57. ScienceDirect, https://doi.org/10.1016/j.electstud.2014.09.001.
"Study: 10% to 27% of noncitizens Are Illegally Registered to Vote." The Heartland Institute, 14 May 2024, https://heartland.org/opinion/study-10-to-27-of-noncitizens-are-illegally-registered-to-vote/.
Study Finds Up to a Third of All noncitizens in the United States Are Illegally Registered to Vote | The Ohio Star. 15 May 2024, https://theohiostar.com/news/study-finds-up-to-a-third-of-all-noncitizens-in-the-united-states-are-illegally-registered-to-vote/ohstarstaff/2024/05/15/.
"Study: 'noncitizens' Are Registered to Vote, Could Decide next President | WND | by WND Staff." WND, 15 May 2024, https://www.wnd.com/2024/05/study-noncitizens-registered-vote-decide-next-president/.
"Substantial Numbers of noncitizens Vote Illegally in U.S. Elections." Just Facts, 15 Dec. 2016, https://www.justfactsdaily.com/substantial-numbers-of-noncitizens-vote-illegally-in-u-s-elections.
Tesler, Michael. "Methodological Challenges Affect Study of noncitizens' Voting." Washington Post, 7 Dec. 2021. www.washingtonpost.com, https://www.washingtonpost.com/news/monkey-cage/wp/2014/10/27/methodological-challenges-affect-study-of-noncitizens-voting/.
Up to a Quarter of Illegal Immigrants Are Registered to Vote, According to New Study: "Potentially High Enough to Overturn the Will of the American People." https://notthebee.com/article/up-to-a-third-of-all-illegal-immigrants-in-the-us-are-registered-to-vote-just-in-case-you-wanted-to-know-how-2024-is-going-to-go. Accessed 23 May 2024.
