Check the Count

When the national mood shifts, almost every county moves the same way. It happened toward Democrats in 2008 and toward Republicans in 2012, 2016 and 2024. In 2024, 88.6% of counties moved toward Republicans and not one county flipped to the Democrat. A one-sided sweep is what a wave looks like. Tampering would look different: a shift packed into certain counties, one brand of voting machine, or certain batches of ballots, that demographics and history can't explain.

In the 9 states that report results by type of vote, mail and absentee ballots leaned more Democratic than Election Day votes in 707 of 708 counties in 2020. The single exception was Chicot County, Arkansas. This is why leads can move late on election night in states that count mail last.

Texas publishes results for every precinct, along with registered voters, ballots cast and the share of voters with a Spanish surname. That lets us run tighter tests than anywhere else.

2026 Texas, preliminary
About 9.8 million Texans voted in 2026, the most in any Texas midterm since at least 1970, from a record 19.0 million registered voters (51.6% turnout). In 2022, 8,143,047 voted (46.0%). These 2026 figures are preliminary; certified county results are due after the state canvass, between 18 November and 3 December. Source: Texas Tribune, from Texas Secretary of State preliminary data.

Democrats won about 254 House seats and 54 Senate seats, according to news-network projections. Counting is not finished, and no state has certified its county results yet.

1. Uniform or concentrated? We will compare each county's 2026 result with 2022 and 2024. A wave moves nearly every county the same way, like 2008 and 2024. Tampering would show up as a shift concentrated in certain counties, machine brands or batches.
2. Ballots against voters. Every counted ballot should match a voter recorded as having voted. Texas, Georgia and other states publish voter-history files a few weeks after an election, and we will compare them county by county.
3. Did each state follow its own audit law? We will check whether each state ran the post-election audit its law requires, and published the result.

Almost every state requires a check after the election, usually a hand count of paper ballots from a sample of precincts or machines, or a risk-limiting audit that keeps counting until the result is confirmed. The table lists each state's audit type and the law that sets it.

• 7Risk-limiting
• 37Traditional hand count
• 3Procedural
• 2None required
• 2Other or unclear

As summarized by the National Conference of State Legislatures, updated 24 September 2026. We are checking each row against the statute itself. NCSL: Post-election audits.

1. A flag is a question, not proof. Results are marked Review (unusual, worth asking about) or Strong (impossible on its face).
2. Every flag lists the ordinary explanations to rule out and the official record that would settle it.
3. We account for demographics, county size and the previous result before calling a change unexplained.
4. We count against chance. With 3,100 counties, some will look extreme by luck alone.
5. We use totals only: no individual voter records, and no names of election workers.
6. Nothing is published automatically. A person checks every finding against official records first.
7. Anyone can rerun the tests and get the same numbers.

• MIT Election Data + Science Lab, county presidential returns 2000–2024, CC0.SHA-256 9299583148f5f8264baf50b5d929c260699c46d915909b13ea3c3850918946f8
• Texas Legislative Council, precinct election data 2012–2024, CC BY.SHA-256 ed6956085e80d8153adce0829c279c8915c05f1867b2004b8b0988336469ff56
• NCSL post-election audits.
• Checks between sources: The two sources agree on Texas 2024 county totals to within 708 votes out of 11.2 million.

npm test
npm run check:tx
npm run check:us -- --tlc "data/raw/tlc/extracted"