Both final 2026 national polls with Verasight data came within about 2 points of the House popular vote margin, while the polling averages missed it by more than 4 points. Verasight collected the data for both polls, working with Strength In Numbers and The Argument. The polls had Democrats ahead by 12 and 11 points among likely voters, respectively. Both are within the sampling margin of error of the current House popular vote margin of 13.2 percentage points, with more than 115 million votes counted. The polling average, by contrast, had Democrats ahead by 8.8 points. That 4.4-point miss was at least double the miss of either poll with Verasight data.
This post covers public national pre-election polls with Verasight data from the final three weeks of the campaign. The chart above compares those polls with the polling averages and the vote count so far. Votes are still being counted in some states.
National House vote polling accuracy
Two national pre-election polls with Verasight data were fielded in the final three weeks of the campaign. Both reported the U.S. House vote among likely voters. The October Strength In Numbers/Verasight poll, fielded Oct. 13-19 among 2,028 adults, found Democrats ahead 54% to 42% among likely voters. The Argument's poll, fielded Oct. 21-23 among 1,549 registered voters, found Democrats ahead by 11 points among likely voters.
With more than 115 million votes counted, Democrats lead the national House vote 55.9% to 42.7%, a margin of 13.2 points, according to Cook Political Report's vote tracker. That is in line with The New York Times projection of about 13 points. The Strength In Numbers/Verasight poll is about 1 point from that margin, and The Argument's poll is about 2 points from it. The final averages from FiftyPlusOne, RealClearPolling and Silver Bulletin had Democrats ahead by 8.7, 8.8 and 8.8 points, so each missed by more than 4 points. On average, the two polls missed by 1.7 points, less than half the 4.4-point miss of the composite average.
G. Elliott Morris, who publishes Strength In Numbers and co-founded FiftyPlusOne, compared national likely-voter polls from the final three weeks of the campaign against a D+13 margin. Of the 20 other entries in his comparison, 19 underestimated the Democratic margin, and 13 missed it by more than 4 points. Both polls with Verasight data missed by about 2 points or less.
More on the polls and Verasight's methods
Both polls drew on the Verasight panel, which Verasight recruits directly and verifies before panelists take a survey. Every panelist completes SMS authentication through a mobile phone registered with a major U.S. carrier, and each survey checks for non-human responses, speeding and straight-lining, removing respondents who fail. Verasight also weighted the Strength In Numbers poll to the Census Bureau's Current Population Survey, a running three-year average of party identification from Pew Research Center and the 2024 vote. The Argument weighted its own data.
“We use multi-mode outreach to recruit voters (address-based sampling, text message recruitment, online targeting), meaning that we capture a more comprehensive slice of the population than methods that rely on a single source of data,” Jake Rothschild, Director of Data Science at Verasight. “We also use up-to-date weighting benchmarks, including rolling averages of partisan identification, to make sure our polls reflect the changing electorate.”
A poll is included if it is a national pre-election poll, the midpoint of its field period fell on or after Oct. 13, three weeks before Election Day, and its topline results were released publicly. The polling average is a composite, the mean of the final generic-ballot averages from FiftyPlusOne, RealClearPolling and Silver Bulletin (D+8.8). The result is the House popular vote margin in Cook Political Report's vote tracker on Nov. 10 (D+13.2, with more than 115 million votes counted). Margins are in percentage points, with each poll's margin as published.
