WASHINGTON, Aug 22: A preliminary U.S. Census Bureau analysis cited by President Donald Trump does not establish his claim that 24,000 noncitizens illegally voted in the 2020 presidential election, experts say, pointing to significant gaps in the report’s methodology.
Trump referred to the seven-page document in a Truth Social post this week, claiming it had compared voter records with citizenship information and identified thousands of noncitizens who cast ballots.
The report, however, describes its findings as a “beginning analysis” and says more than 32 million voter records remain to be examined. It does not provide enough information to establish that the people identified were actually noncitizens when they voted or that they unlawfully participated in the election.
Experts question accuracy of matches
According to the analysis, nearly 160 million voter records from the 2020 election have been reviewed so far. More than 24,000 records were linked to administrative information indicating, with a high degree of confidence, that the individuals were noncitizens at the time of the election.
Election and data experts caution that such a match does not automatically prove that a person voted illegally.
Mark Lindeman, policy and strategy director at Verified Voting, said the report provides insufficient information about the databases used and how potential errors were addressed.
Justin Levitt, a Loyola Marymount University law professor and former Biden administration voting-rights adviser, identified several possible sources of error. These include inaccuracies in the private voter database, mistakes in federal records and problems occurring when records from different systems are matched.
Data matching can produce false positives
Experts said a variety of circumstances could result in a U.S. citizen being incorrectly flagged as a noncitizen.
For instance, two people with similar names or identifying information could be mistakenly linked to the same record. Address information may also be outdated or incorrectly associated with a voter. In addition, citizenship records may not have been updated when a person’s status changed.
Barry Burden, director of the University of Wisconsin-Madison’s Elections Research Center, also questioned the report’s transparency.
He noted that the document is unsigned, identifies itself as an initial analysis and relies on information supplied by an unfamiliar private company. It also does not clearly explain how the report determined that a particular data match was sufficiently reliable.
Without those details, experts say the reported figure cannot be treated as a verified count of illegal votes.
Figure would not alter 2020 election outcome
The reported number also represents only a tiny fraction of the votes cast in the 2020 presidential election.
Federal Election Commission figures show that 158,429,631 votes were cast for president that year. The 24,000 records cited by the Census analysis amount to about 0.015% of the total vote.
Joe Biden defeated Trump by 7,059,526 votes nationally. Election expert David Becker said that even if every one of the 24,000 records represented an actual noncitizen vote cast for someone other than Trump, the number would not have been sufficient to change the winner in any state.
Trump nevertheless continues to maintain that he won the 2020 election and has used the report to renew calls for stricter voting requirements.
SAVE America Act remains contested
Trump has urged Congress to pass the SAVE America Act, legislation that would require documentary proof of citizenship for voter registration and establish nationwide voter identification requirements.
The measure has passed the House of Representatives but has faced opposition in the Senate, including from some Republicans.
The White House has defended Trump’s position, arguing that noncitizens do register and vote in U.S. elections. However, the Census Bureau’s preliminary analysis does not by itself verify that the more than 24,000 matched records represent illegal votes.
Experts therefore say the report should not be interpreted as evidence that the 2020 presidential election result was affected by widespread noncitizen voting.