A government study claiming that tens of thousands of noncitizens were listed on voter rolls during the 2020 US presidential election has come under severe scrutiny over major analytical flaws and irregular internal review procedures. While President Donald Trump and Commerce Secretary Howard Lutnick publicly praised the study on Truth Social and X, an examination of the document's methodologies reveals that its conclusions rely on mismatched administrative datasets rather than verified illegal votes. Furthermore, the report was authored by political appointees and researchers with direct ties to the Trump administration and Republican organizations, including a former official from Elon Musk's Department of Government Efficiency (DOGE) and researchers from the America First Policy Institute (AFPI).
Political Origins and Involvement of Republican-Aligned Personnel
The study, initiated inside the Center for Economic Studies—a Census Bureau division that uses confidential administrative records for research—was produced largely outside standard scientific workflows. Records show that key contributor Dan Sweeney serves as deputy general counsel at the Department of Commerce overseeing the Census Bureau, having previously served as the DOGE legal liaison and deregulatory attorney at the Environmental Protection Agency. Another core contributor, Michael Lachanski, was appointed just two months prior as deputy director for data, policy, and science at the Census Bureau following a fellowship at the Claremont Institute.
Two additional authors, Matthew Jensen and Jeff McCrea, were brought into the agency from the America First Policy Institute under the Intergovernment Personnel Act (IPA). The IPA framework allows temporary two-year appointments for specialists from universities and nonprofits to assist government agencies. McCrea previously managed the campaign for former Washington Congresswoman Cathy McMorris Rodgers, who signed an amicus brief contesting 2020 election results. Jensen previously worked at the American Enterprise Institute, where he contributed to drafting the Tax Cuts and Jobs Act during the first Trump administration.
Danah Boyd, a professor at Cornell University studying census methodology, noted that while the Census Bureau routinely uses IPAs for scientific tasks, those appointees normally follow the direction of career civil servants. In this case, Boyd observed that the authors functioned as political operatives acting under the umbrella of a non-profit organization to execute work serving partisan interests rather than objective agency research.
Numerical Breakdown and Flawed Data Linkage Methodology
Titled 'Noncitizen Voting in the 2020 Election: A Beginning Analysis', the publication claims that out of more than 128 million examined 2020 voter records, approximately 24,000 were linked to administrative records indicating noncitizen status at the time of the vote. State-level figures ranged from 4,300 in California to fewer than 15 in Wyoming—totals far too small to have altered the outcome of the 2020 presidential race. The authors noted that an additional 32 million voter records remain unanalyzed.
To reach these numbers, the authors merged commercial and federal files, including the DataClear 2020 voting file, United States Citizenship and Immigration Services (USCIS) records from 2025 and 2026, DACA approvals, State Department visa records from 2017 to 2023, Immigration and Customs Enforcement (ICE) records, 2020 IRS Form 1040 files, Social Security Numident records from 2019 and 2020, and State Department passport records spanning 50 vintages between 1978 and 2025.
Former agency officials highlighted that merging files across widely separated timeframes creates massive technical obstacles. Amy O'Hara, a former official who led the administrative data linkage unit at the Census Bureau, explained that administrative records frequently fail to update citizenship status dynamically, making cross-year file deduplication and record matching highly unreliable.
Statistical Validity and False Positive Rates
To cross-reference records while preserving privacy, the Census Bureau relies on Protected Identification Keys (PIKs), which replace personal identifying details with unique numeric identifiers. However, a January 2026 Census Bureau study demonstrated that PIK assignments are susceptible to errors, causing unassigned or misassigned records that introduce systematic selection bias into linked datasets.
John Abowd, former chief scientist and associate director for research and methodology at the Census Bureau, described the publication as 'statistical nonsense' and highlighted on LinkedIn that publishing Census research without explicitly naming authors is highly irregular. Abowd explained that when commercial voting files are merged with immigration databases, false match rates increase significantly. Because the false positive rate in data matching is higher than the estimated rate of noncitizen voting, the flagged 24,000 records overwhelmingly reflect data matching errors rather than noncitizens casting ballots.
An internal Census Bureau source stated that while the report claims 'high confidence', it omitted standard confidence intervals from its statistical tables. The minuscule ratio of alleged noncitizen records relative to state populations indicates that the findings represent PIK matching failures. Boyd added that releasing unverified analysis under the Census Bureau's official authority risks eroding public and academic trust in official government data products.
Accelerated Approval Process and Institutional Protocol Deviations
The report's internal authorization bypassed routine review protocols. On August 7, the Census Bureau's Disclosure Review Board approved the publication via an expedited email vote—a procedure typically reserved for tight congressional deadlines or emergency executive requests. Board members who cast votes were shown only the three summary data tables included in the final release, without receiving the accompanying methodology or narrative analysis.
When asked for clarification, the Census Bureau declined to comment. The White House referred inquiries to social media posts by President Donald Trump on Truth Social and Commerce Secretary Howard Lutnick on X praising the findings. Authors Sweeney, Lachanski, Jensen, and McCrea did not respond to requests for comment.
Historical Context and State-Level Voting Record Audits
Extensive historical audits across US states have consistently demonstrated that noncitizen voting is virtually non-existent. In Utah, Lieutenant Governor Deidre Henderson conducted a comprehensive review of the state's 2.1 million registered voters, identifying 27 confirmed noncitizens, none of whom had cast a ballot. Similar state-level reviews in Pennsylvania and Idaho yielded matching conclusions.
Despite empirical evidence, claims of widespread noncitizen voting remain a focal point of political debate. Earlier policy proposals, including an internal Department of Commerce memo suggesting the exclusion of noncitizens from population totals used for congressional apportionment, alongside prior Republican challenges to census privacy algorithms, underscore ongoing efforts to alter federal election administration and census data practices.
