A GOP Pollster Checked 3,312 Senate Polls. Democrats Were Overstated in Every Single Cycle.
- 3,312 Polls the number of U.S. Senate polls across the 2018, 2020, 2022, and 2024 cycles that Patrick Ruffini analyzed for polling error — Patrick Ruffini, Aug. 18, 2026
- 3.7 Points the average margin by which polls overstated the Democratic candidate across all four election cycles studied — Patrick Ruffini
- 73% the share of individual polls in the dataset that overstated the Democratic candidate relative to the eventual certified result — Patrick Ruffini; Breitbart
- 5.9 Points the peak overstatement in late July, before narrowing to roughly 2.6 points by Election Day across the cycles studied — Patrick Ruffini
Echelon Insights co-founder Patrick Ruffini set out on August 18, 2026 to test a recurring impression: that Democrats look competitive in red-state Senate races over the summer, then fade by Election Day. He pulled every Senate poll from the old FiveThirtyEight archive across four cycles — 3,312 polls in all — and measured each one against the eventual certified result.
The pattern held in all four cycles. Polls overstated the Democratic candidate by an average of 3.7 points, in 73 percent of individual polls, peaking around 5.9 points in late July before narrowing to roughly 2.6 points by Election Day.
“I wanted to test if this was a consistent pattern, not just the selective memory of a Republican,” Ruffini wrote.
Ruffini’s dataset includes some of the decade’s most-watched Senate polling misses. In 2020, Quinnipiac had South Carolina’s Senate race tied; Lindsey Graham (R) won by 10 points. In 2022, Tim Ryan (D) led most Ohio polls through the summer; JD Vance (R) won by 6. In 2024, Sherrod Brown (D) led all but two pre-October Ohio polls; Bernie Moreno (R) still won.
“The average polling error across all polls was D+3.7, and 73% of polls released overstated the Democrat relative to the election result,” Ruffini wrote. “Again, that’s across more than 3,000 polls, so that’s hardly just statistical noise.”
Talking Points Memo editor Josh Marshall, a Democratic-aligned analyst, reviewed Ruffini’s underlying data and called it “broadly reliable,” while noting the pattern holds in roughly 70 percent of races rather than literally every single one, and offering his own theories — differential response rates by voter trust and engagement, and a late-cycle shift toward likely-voter screens — that Ruffini’s own piece does not itself advance.
“Ryan led in the vast majority of polls conducted in Ohio over the summer until being bested by JD Vance by 6 points. The same pattern repeated itself in Ohio in 2024 when Brown led in all but two pre-October polls in his race against Bernie Moreno.”
Patrick Ruffini, Echelon Insights
Independent AAPOR task-force reports corroborate a related trend at the presidential level: national polls underestimated Trump or Republican support in both 2020 and 2024. (AAPOR’s own 2016 review found the miss that cycle was concentrated in state-level polling — national topline polls were within normal historical error.) One important scoping caveat: FiveThirtyEight’s own review of 2022 found that, in the aggregate across House, Senate, and governor races nationally, polling actually slightly overestimated Republicans — a reminder that Ruffini’s finding is specific to competitive Senate contests and their summer-to-Election-Day trajectory, not a claim that every poll everywhere skews one direction.
The analysis lands as this year’s midterm polling has already drawn its own credibility questions. GOP pollster Frank Luntz called for a “total and complete shutdown” of Midwest primary polling after high-profile misses in Wisconsin and Michigan this August.
I am calling for a total and complete shutdown of primary polling in the Midwestern United States until we can figure out what the hell is going on.
Ruffini’s own piece stops short of explaining why the pattern exists — it documents the trend without assigning a mechanism. Readers should treat the “why” as an open, actively debated question among pollsters and analysts, not a settled finding alongside the “what.”
Patrick Ruffini’s analysis of 3,312 Senate polls across four election cycles found Democrats overstated by an average of 3.7 points, in 73% of individual polls, peaking near 5.9 points in late July before narrowing by Election Day. Democratic-aligned analyst Josh Marshall calls the underlying data “broadly reliable” while noting the pattern isn’t universal. AAPOR’s own task-force reports corroborate a related presidential-level pattern in 2020 and 2024. FiveThirtyEight’s 2022 aggregate data is an important scoping caveat: this is a finding about competitive Senate races specifically, not all polling everywhere.
