No, a Census Bureau "Beginning Analysis" Didn't Prove Trump Won the 2020 Election

The Census Bureau report, titled "Noncitizen Voting in the 2020 Election, A Beginning Analysis," compared a commercial voter list against federal databases from agencies including ICE, USCIS, SSA and the IRS. Of roughly 160 million ballots cast, it found more than 128 million records matched to citizens "with high confidence," over 24,000 records matched to noncitizen immigration or visa records, and more than 32 million records the Bureau says it has not yet analyzed ^[3].

"Matched" is not "proven." A voter record was flagged as a probable noncitizen when the Bureau found no proof of naturalization in the databases it checked — not because it found direct evidence the person voted while ineligible. Lead Stories rated the viral claim "Overstated," noting the absence of citizenship paperwork in a database doesn't mean someone isn't a citizen, only that the Bureau couldn't confirm it was ^[4].

The Bureau's own title calls this a "beginning analysis." Election law and demography experts flagged that the report lacks standard Census safeguards: no listed authors, no margin of error, and no discussion of data-matching bias — unusual for a Bureau publication. Georgetown demographer Beth Jarosz said, "A properly done Census report would have had that information and the fact that it is missing is glaring" ^[2]. Former Census Bureau chief Amy O'Hara said the report lacks "any sort of measures of variance or discussion of bias," and questioned whether database matches reflect real duplicate individuals or simply people who share a name ^[2].

David Becker of the Center for Election Innovation & Research put a number on the uncertainty: the 24,000 figure represents about 0.02% of all votes cast, well within the kind of noise that database-matching error alone can produce. "Census data cannot reliably be matched to voter data," he said ^[2].

Even taking the 24,000 figure at face value, it is not close to enough to flip a state, let alone the national outcome. In Wisconsin, the report identified about 500 potential noncitizen voters — Trump lost the state by more than 20,000 votes. In Arizona, the figure was roughly 1,100, against a margin of loss over 10,000 votes ^[2]. The nationwide margin Biden won by in 2020 dwarfs the entire 24,000-record total, and the report contains no data on which candidate any of these voters supported ^[4].

Loyola Law School's Justin Levitt noted the report omits standard statistical caveats that would let readers judge whether the 24,000 figure is even reliably above zero once matching error is accounted for ^[2]. Multiple prior state and federal audits have found noncitizen voting in U.S. elections is exceedingly rare ^[3].

FALSE. The Census Bureau report does not prove more than 24,000 noncitizens voted illegally in 2020, and it does not — and cannot — show that the election outcome would have been different. The document itself is labeled a preliminary "beginning analysis" lacking standard reliability disclosures that outside experts say are necessary to trust the number at all. Even accepting the figure as accurate, it is orders of magnitude smaller than the margin by which Trump lost in 2020 ^[2][4].