Administration’s Claims of Non-citizen Voter Registrations Face Data Gaps and Expert Pushback

The Department of Homeland Security (DHS) announced it had identified 250,000 non-citizens registered to vote in California, Nevada, New Jersey and Pennsylvania. The claim was presented as evidence supporting President Donald Trump’s assertions of voter fraud. A seven-page analysis posted on the U.S. Census Bureau’s website alleged that 24,000 non-citizens voted in the 2020 presidential election after matching a commercial voter file with immigration records. Both releases arrived as the administration pushes tighter voting rules ahead of the 2024 midterms.

CNN

President Trump has framed non-citizen voting as a systemic threat, urging Congress to pass the SAVE America Act, which would require proof of citizenship for voter registration. State officials in the four targeted states have expressed skepticism about the methodology behind the DHS data-matching effort.

CNN

• DHS estimate: 250,000 potential non-citizen registrations across four states.

CNN

• Nevada focus: 15,903 names appeared on the preliminary list; only 185 were confirmed as non-citizens after manual review (recorded on August 13). The department also flagged 6,200 “higher-confidence” matches that remain unverified.

CNN

• Census Bureau analysis: 24,000 alleged non-citizen votes out of 128 million voter records examined (? 0.019%). The report notes an additional 32 million records pending review, which officials say could increase the total count.

CNN

Homeland Security Secretary Markwayne Mullin called the DHS numbers “hard evidence” of illegal voting. DHS fraud-division lead Kimberley Vogt said the agency would continue manual reviews to ensure matches are not, for example, naturalized citizens whose records have not been updated.

CNN

Election-integrity experts have questioned the reliability of both the DHS and Census findings. David Becker, executive director of the Center for Election Innovation & Research, called the Census analysis “laughable” and warned that census data lack unique identifiers for reliable matching. John M. Abowd, former chief scientist of the Census Bureau, noted that the 24,000 matches are statistically indistinguishable from routine matching errors. Amy O’Hara, director of Georgetown University’s Massive Data Institute, highlighted the report’s lack of variance measures and bias discussion, calling it a “fatal flaw.”

NPR

• DHS’s aggregate figure of 250,000 contrasts sharply with Nevada officials’ verification of only 185 confirmed cases.

CNN

• The Census analysis presents a 24,000-vote estimate but offers no author, variance metrics, or transparent methodology, prompting former bureau officials to label it “unusual” and “unreliable.”

NPR

• Both agencies have withheld full data sets from state officials, citing concerns about “inaccurate records” and “preliminary confidence levels,” leaving a substantial portion of the alleged matches unexamined.

NPR

• “We’ve identified 250,000 noncitizens registered to vote in just four states,” — Markwayne Mullin

CNN

• “One thing that I love about numbers, and I love about facts, is they don’t lie,” — Markwayne Mullin

CNN

• “We’re going to manually review to be sure that it is not someone’s dad or someone’s son – that it is that person,” — Kimberley Vogt

CNN

• “This analysis is laughable. Census data cannot reliably be matched to voter data,” — David Becker

NPR

DHS plans to continue manual verification of the 6,200 higher-confidence Nevada matches before the November midterms. The Census Bureau intends to process the remaining 32 million voter records, a step that could alter the 24,000-vote estimate. State election officials will await completed reviews before taking action on voter-registration lists.

CNN