Census Bureau report claims 24,000 noncitizens voted in 2020, but experts question how it got there

A seven-page Census Bureau report claims that more than 24,000 noncitizens voted in the 2020 presidential election. The report carries no author's name, does not detail its methodology, and was not produced by the agency's career civil servants, according to NPR.

President Trump cited the analysis in a social media post this week, saying it "proves that over 24,000 Noncitizens voted illegally" and that the number will "explode" once the bureau checks another 32 million voter records.

The team behind the analysis included people affiliated with the America First Policy Institute, a think tank founded by officials from Trump's first administration, a Census Bureau employee told NPR. NPR granted the employee anonymity because of fear of retaliation at work. The bureau, the institute, and the White House did not respond to NPR's requests for comment. The institute promoted the analysis on social media, saying it "maintains that maximizing transparency is a critical step to restoring confidence in our elections."

What 24,000 votes represents

The analysis examined more than 128 million voter records. Against that total, 24,000 works out to 0.019 percent, or roughly one in every 5,300 votes.

David Becker, executive director of the Center for Election Innovation and Research, made that point to NPR while dismissing the analysis. "This analysis is laughable. Census data cannot reliably be matched to voter data," Becker said. "That said, it's good to see the president admit these numbers are so small that they couldn't have changed the outcome in 2020."

Georgia is the clearest test

Georgia offers a direct comparison, because the state already ran its own investigation.

Secretary of State Brad Raffensperger, a Republican who has prioritized preventing noncitizens from voting, conducted an intensive audit in 2024. The audit identified fewer than a dozen noncitizens who had previously voted in the state. The Census Bureau report estimates Georgia had 400 noncitizen voters in 2020 alone, more than 30 times as many.

Raffensperger has written to acting bureau director George Cook asking for the underlying data. "I am asking for all the identifying information utilized by your agency to detect the alleged 400 non-citizens who would have voted in Georgia in 2020," Raffensperger wrote in the letter, which was shared with NPR.

Nevada's secretary of state, Democrat Francisco Aguilar, raised a similar objection after the report claimed 250 noncitizens voted there. "The federal government has once again made sweeping claims regarding noncitizen voting allegations, without showing the public the proof behind the numbers," Aguilar said in a statement. "These kinds of claims, without fully explaining how they came to these determinations, just cause chaos and fear ahead of a critical midterm election."

It is the second such federal claim about Nevada this summer. In July, the Department of Homeland Security said it had found 250,000 noncitizens on voter rolls in California, Nevada, New Jersey, and Pennsylvania, without disclosing how it reached that figure. Election officials in those states questioned its accuracy.

The methodology objections

The technical criticism centers on record matching. Linking a voter file to federal citizenship records means deciding when two entries describe the same person, and voter files generally lack unique identifiers that would make that certain.

Amy O'Hara, a former chief of the bureau's Center for Administrative Records Research and Applications who now directs Georgetown University's Massive Data Institute, said the report gives readers no way to judge that risk. "This brief lacks any sort of measures of variance or discussion of bias," O'Hara said, questioning whether matched records represented the same people "or do you have people with the same names potentially living in the same state and representing a mismatch?"

O'Hara also disputed the report's assertion that the bureau is uniquely capable of publishing statistics on election crime, noting that irregularities in data are not proof of fraud. "Even if they observed anomalies in the data, those anomalies aren't necessarily fraud," O'Hara said.

Michael Morse, a University of Pennsylvania law professor who has researched voter registration files, told NPR the estimated rate of apparent noncitizen voting is "both a tiny fraction of total votes and entirely consistent with both types of error."

Misty Heggeness, a former bureau economist whose research the report cites, focused on the missing byline. "My biggest concern would be there's no author attached to the report, which is not common practice within the Census Bureau," Heggeness said. "Usually there are layers and layers of rigor and testing because the Census Bureau is a neutral statistical agency."

Why it lands on the bureau

The report arrives during a hard stretch for the agency. The bureau is operating without a permanent director, after losing statisticians and economists in the federal workforce cuts we covered in the staffing crisis at federal statistical agencies. It is also carrying out a Commerce Department order that changed how it protects published data, and facing a draft rule that would change who gets counted in 2030.

O'Hara framed the risk in terms of the count ahead. "We are drawing in on the 2030 census that really requires the participation of everybody in the country. Anything that casts a negative light on the Census Bureau's motives or operations or transparency or lack of adherence to policy could be detrimental," O'Hara said.

Heggeness said the missing sourcing is itself a signal for readers. "This document really does have a lack of a source and citations and explanation about methods," Heggeness said. "If the public sees a document like this that lacks all of those really integral pieces that are generally put out with official statistics, that's a really good sign to question the validity of the document."

The report's title, "Noncitizen Voting in the 2020 Election, A Beginning Analysis," suggests more are on the way.

Sources

Reporting, quotes, and the account of the report's origins are from NPR, by Hansi Lo Wang, Jude Joffe-Block, and Miles Parks, published August 19, 2026 and updated August 20. The percentage and per-vote figures are CensusEasy calculations from the report's own numbers as described by NPR: 24,000 out of more than 128 million records examined. NPR reports that many studies and audits have found noncitizen voting in federal elections to be rare.