A brief, unauthored Census Bureau analysis that President Trump has cited to renew claims of widespread illegal voting by noncitizens is drawing sharp criticism from data and election experts, and it has connections to a Trump-aligned think tank, according to people familiar with the work.
The seven-page document, titled ‘Noncitizen Voting in the 2020 Election, A Beginning Analysis,’ reports an initial match of more than 128 million commercial voter records against federal administrative data and concludes that roughly 24,000 apparent noncitizen votes were cast in the 2020 presidential contest. The bureau said it is continuing to review an additional 32 million records. The president posted about the analysis on social media, saying the final figure will ‘explode.’
But former Census employees and independent experts say the brief is methodologically opaque, lacks the usual authorship and documentation expected from the bureau, and may overstate problems because of flawed matching techniques.
People familiar with the analysis told NPR that the work was not carried out by the bureau’s usual career statisticians and that individuals connected to the America First Policy Institute — a think tank founded by officials from Trump’s first administration — were involved in the team behind the analysis. The Census Bureau, the think tank and the White House did not respond to requests for comment. AFPI promoted the analysis on social media, saying transparency is necessary to restore confidence in elections.
Experts say the report’s approach — matching commercial voter files to citizenship records compiled from federal agencies — is inherently risky without careful controls. Amy O’Hara, a former head of the bureau’s Center for Administrative Records Research and Applications, warned that matching errors can create false positives: people with the same or similar names may be conflated, or naturalization records may be incomplete or out of date. She criticized the brief for offering no measures of uncertainty or assessment of bias and for suggesting the bureau is ‘uniquely capable’ of compiling statistics on election crime, a claim she called surprising.
David Becker, executive director of the Center for Election Innovation & Research, said the analysis is ‘laughable’ because Census data are difficult to match reliably to voter files. He also noted that the scale of the alleged problem is very small relative to total votes: 24,000 apparent noncitizen votes out of about 128 million ballots is about 0.02 percent — a rate that could easily be explained by data error and could not have changed the election outcome, he said.
Michael Morse, a University of Pennsylvania law professor who studies voter registration files, echoed concerns about missing unique identifiers in voter records and incomplete naturalization records. He said the brief’s lack of transparency makes it hard to evaluate the methodology and that the estimated rate of apparent noncitizen voting is consistent with routine data errors.
Former bureau economist Misty Heggeness, whose work is cited in the report, said the absence of an identified author is notable and worrying because it departs from the Census Bureau’s usual practices of layered review and documentation. She said a document lacking method details, sources and citations should prompt skepticism from the public.
The report arrives amid other recent administration claims about noncitizen registration and voting. Last month, the Department of Homeland Security announced it had found 250,000 noncitizens on voter rolls in California, Nevada, New Jersey and Pennsylvania; DHS has not publicly disclosed how those calculations were made, and state election officials have questioned the accuracy of the finding.
Critics are also concerned about possible political consequences. The Trump campaign and allied lawmakers have used claims of noncitizen voting to press for stricter voting laws, and a high-profile Census Bureau analysis that appears partisan or methodologically weak could damage the bureau’s reputation as an independent statistical agency. That reputation is already under strain after a wave of departures by career statisticians and economists following workforce reductions under the administration.
Observers warn that undermining trust in the bureau could have broader effects on participation in the 2030 census and other national data efforts. O’Hara and Heggeness both stressed that official statistical work should include transparent methods, documentation and clear measures of uncertainty so the public can assess credibility.
With the brief labeled ‘A Beginning Analysis,’ authors or sponsors may release further work. For now, experts urge caution: the apparent number of noncitizen votes reported so far is tiny, the methodology is not publicly documented, and matching administrative and commercial datasets without rigorous safeguards can produce misleading conclusions.