Monday, August 24, 2026

CheckDeezOut logoCheckDeezOut

A 0.02% Statistic Became 'I Won the Election'

A Census Bureau report flags 24,000 possible noncitizen voters from 2020 — 0.02% of ballots cast. Trump calls it proof. Reporters found an authorship problem.

By CheckDeezOut EditorialPublished
A 0.02% Statistic Became 'I Won the Election'
Image credit: Element5 Digital / Pexels

On Tuesday, the U.S. Census Bureau posted a seven-page document to its website. It had no listed author. Within hours, President Trump had turned it into a verdict.

"The Census Bureau has begun checking the Voter Records from 2020 against their Citizenship Records," he wrote on Truth Social, saying the first 128 million records checked had turned up "over 24,000 Noncitizens" who "voted illegally." Then the line that mattered more than the statistic: "I WON THE ELECTION!"

Commerce Secretary Howard Lutnick, whose department oversees the bureau, called it "just the beginning" and predicted the next batch of records would show "tens of thousands" more.

What actually sits inside that seven-page document, and who put it there, tells a much messier story than either man's post.

What the Report Says, and Who Wrote It

The document — titled "Noncitizen Voting in the 2020 Election, A Beginning Analysis" — describes matching a commercial voter file against federal citizenship records from the IRS, Social Security Administration, State Department and Department of Homeland Security, according to CNN. The voter data came from Data-Clear, a Virginia firm paid $649,500 in 2024 to supply it.

Of 128 million voter records checked, roughly 24,000 — about 0.019% — matched someone flagged as a noncitizen in one of those federal databases.

The bigger issue for former Census officials wasn't the number. It was who produced it. NPR reported that the analysis wasn't done by career civil servants, but by a team that included people affiliated with the America First Policy Institute, a think tank founded by veterans of Trump's first administration. Robert Santos, who ran the Census Bureau until early 2025, told CNN he was "shocked and dismayed" to see it posted at all: "I refuse to call it a report... It is clearly a product of White House and Department of Commerce analysts."

Misty Heggeness, a former bureau economist whose own research is cited in the document, pointed to the missing byline as the tell. "There are usually layers and layers of rigor and testing," she told NPR. "The fact that there's no author on this report leads me to wonder whether or not that did happen."

The Math That Didn't Make the Post

Even taking the 24,000 figure at face value, it's a rounding error against the scale of the 2020 election. Joe Biden won the national popular vote by more than 7 million votes. CNN laid out the state-level comparison that never showed up in Trump's post:

  • Arizona: 1,100 flagged records; Biden's margin was 10,457
  • Georgia: 400 flagged records; Biden's margin was 11,779
  • Wisconsin: 500 flagged records; Biden's margin was 20,682

The report also can't say who those 24,000 people voted for, because ballots are secret. And several statisticians argue the number itself is likely inflated. John Abowd, the bureau's chief scientist through 2022, told CNN a match rate that small is "too small to distinguish from routine matching errors" and that no career researcher he'd worked with would present it as proof. David Becker of the Center for Election Innovation & Research put it more bluntly to both outlets: "This alleged analysis... fails the laugh test. Census data cannot be reliably matched to public voter data."

Two Very Different Stories About the Same Seven Pages

Here's where the coverage splits, and it's not a subtle split. NPR and CNN's reporting centers the authorship question, the AFPI staffing, and the statistical caveats — treating the 24,000 figure as a claim to be tested, not a finding to be repeated.

Pro-Trump outlets ran the opposite framing. WLT Report headlined its piece "BOMBSHELL," led with Trump's "I WON THE ELECTION" line, and presented the country-of-birth breakdown from the White House's own graphic (Mexico: 3,800 records; Canada: 950; the Philippines and Jamaica: 900 each) without engaging any of the methodology criticism NPR and CNN had already published by the time that piece ran. The Gateway Pundit used nearly identical framing. Neither outlet mentioned the AFPI staffing question or the unsigned-report red flag that anchors the mainstream coverage.

That's not two outlets disagreeing about facts — it's two outlets deciding which facts belong in the story at all.

Why the Staffing Detail Outweighs the Number

The number was always going to shrink under scrutiny; these kinds of Census-DHS matching claims have before. What's different this time is the process. Andrew Walchuk of Common Cause called the report "a farce" intended to "generate a storyline the president will like." That's an advocacy group's characterization, but it lines up with a pattern reporters have flagged for months: political appointees inserting themselves into work that's supposed to run through career statisticians precisely because it's supposed to be insulated from the White House.

It's a similar dynamic to the fight over Federal Reserve Governor Lisa Cook, where the administration built a paper trail against an official at an agency designed to operate independently of it. The Census Bureau is heading into its own high-stakes moment — the 2030 census — and Amy O'Hara, a former bureau official now at Georgetown, told NPR that anything casting doubt on the bureau's "motives or operations or transparency" lands at the worst possible time for an agency that needs the public to trust it enough to fill out a form.

Trump is using the report to push the SAVE America Act, which would require documentary proof of citizenship to register to vote nationally. That's a real legislative fight with real stakes, separate from whether this particular document proves what he says it proves. The Census Bureau, the Commerce Department and the White House did not respond to reporters' requests for comment on the authorship question. Until they do, the 24,000 will stay exactly what David Becker called it: an allegation being treated, by half the coverage, as a verdict.

Sources & Further Reading

Related stories