10
Loading video...
By Sam Stevenson
Associate News Editor
0
Share
Newsweek is a Trust Project member
See more of our trusted coverage when you search.Prefer Newsweek on Googleto see more of our trusted coverage when you search.
Final polling averages underestimated Democratic support in several of the most consequential races of the 2026 midterm elections, including contests for the U.S. Senate and state governorships.
To identify the biggest misses, Newsweek compared RealClearPolling’s final head-to-head averages with results reported by the Associated Press as of 2 p.m. ET Wednesday.
The clearest wrong-leader misses came in the Texas governor’s race, where RealClearPolling’s average favored Republican Greg Abbott before Democrat Gina Hinojosa won, and South Carolina’s Senate election, where Republican Darline Graham Nordone led the average before Democrat Annie Andrews prevailed.
Read More on Politics
...
Other averages identified the Democratic winner but substantially understated the final margin, including in Georgia’s governor’s race.
Costas Panagopoulos, a distinguished professor of political science at Northeastern University, told Newsweek in a phone call before the election that a discrepancy between polling and the result can arise from more than one source.
“Whenever there is a discrepancy between the outcome and what the polls were suggesting, there is a possibility for late-breaking developments that account for that,” he said. “There’s also the possibility that the polls missed the mark in some way.”
The comparisons remain preliminary while final ballots are counted in several states.
They nevertheless show that the polling story of 2026 cannot be reduced to whether an average placed the winning candidate narrowly ahead.
Why It Matters
Polls helped determine where parties spent money, which candidates received national attention, and how forecasters assessed control of Congress before Election Day.
If surveys misjudge turnout or voter preferences in the same direction across multiple states, the resulting error can alter expectations for control of an entire chamber.
Democrats ultimately won the Senate after gaining seats in Texas, Kansas, Ohio, South Carolina, Iowa, Maine and North Carolina.
The party also retained competitive seats in Michigan and Georgia, securing at least 54 Senate seats while Alaska and Mississippi remained unresolved on Wednesday morning.
The House also changed hands, confirming a national Democratic victory that reached states where Republicans had recently enjoyed substantial advantages.
Texas Senate
Democrat James Talarico led Republican Ken Paxton by 2.6 points in RealClearPolling’s final head-to-head average, 49.1 percent to 46.5 percent.
AP called the race for Talarico, putting him ahead by 12.1 points, 55.1 percent to 43 percent.
The difference produced a 9.5-point polling error – the biggest overall miss of the races in Newsweek’s analysis.
RONALDO SCHEMIDT / AFP via Getty Images
Texas Governor
RealClearPolling’s final head-to-head average placed Abbott 2.3 points ahead of Hinojosa, 49.7 percent to 47.4 percent.
AP called the race for Hinojosa, who led by 7.1 points in the latest count, 52.8 percent to 45.7 percent.
The movement from Republican plus 2.3 to Democrat plus 7.1 produced a 9.4-point polling error – the biggest wrong-leader error in Newsweek’s analysis.
...
South Carolina Senate
South Carolina’s Senate contest produced the second-biggest wrong-leader polling miss of the cycle.
The final RealClearPolling head-to-head average gave Republican Darline Graham Nordone a 2.6-point advantage over Democrat Annie Andrews, 48 percent to 45.4 percent.
AP called the race for Andrews, who led by 4.4 points in the latest result, 51.1 percent to 46.7 percent.
Moving from Republican plus 2.6 to Democrat plus 4.4 produced a 7-point error.
...
Georgia Senate
Georgia’s Senate polling correctly favored Democratic Senator Jon Ossoff, but substantially understated the size of his victory.
The final RealClearPolling average placed Ossoff 8.6 points ahead of Republican Mike Collins, 51.9 percent to 43.3 percent.
AP called the race for Ossoff, whose reported margin was 18 points, 59 percent to 41 percent.
That also produced a 9.4-point error, despite the average identifying the winner. It is one of the clearest examples of why polling accuracy cannot be assessed solely by whether the leading candidate went on to win.
...
North Carolina Senate
The final RealClearPolling average placed Democrat Roy Cooper 11 points ahead of Republican Michael Whatley, 52.8 percent to 41.8 percent.
AP called the race for Cooper, whose reported margin reached 19.6 points, 58.7 percent to 39.1 percent.
That produced an 8.6-point error.
...
Georgia Governor
RealClearPolling’s final average placed Democrat Keisha Lance Bottoms narrowly ahead of Republican Rick Jackson, 49 percent to 48.7 percent. The 0.3-point advantage effectively presented the race as tied.
AP called the race for Bottoms, who won by 8.6 points, 54.3 percent to 45.7 percent.
The resulting error was 8.3 points.
...
Iowa Senate
Iowa’s Senate average was tied, although Josh Turek ultimately defeated Ashley Hinson.
RealClearPolling’s final head-to-head average showed Turek and Hinson at 47.1 percent each.
AP called the race for Turek, who led by 4.1 points, 50.5 percent to 46.4 percent.
That resulted in a 4.1-point error.
...
Ohio Governor
The RealClearPolling average placed Democrat Amy Acton 4.5 points ahead of Republican Vivek Ramaswamy, 49.6 percent to 45.1 percent.
AP called the race for Acton, whose reported margin was 7.9 points, 52.3 percent to 44.4 percent.
The average therefore understated the Democratic margin by 3.4 points.
...
Maine Senate
The final RealClearPolling average gave Democrat Troy Jackson a 3.6-point lead over Republican Senator Susan Collins, 50.2 percent to 46.6 percent.
AP called the race for Jackson, with him leading by 7.3 points, 53.6 percent to 46.3 percent.
That produced a preliminary error of 3.7 points. Because Jackson was already above 50 percent, Maine’s ranked-choice process was not expected to reverse the outcome.
...
Ohio Senate
The final RealClearPolling average showed Democrat Sherrod Brown leading Republican Senator Jon Husted by 2.8 points, 48.5 percent to 45.7 percent.
AP called the race for Brown, giving him a 4.8-point advantage, 50 percent to 45.2 percent.
The difference produced a 2-point error, meaning the average captured both the winner and the general competitiveness of the race.
...
Why Polls Can Miss
Panagopoulos cautioned before the election against treating a numerical lead as proof that a candidate was genuinely ahead when the candidates’ possible ranges overlapped.
“A 48-to-44 race that’s within the margin of error is not necessarily implying a lead for one of the candidates,” he told Newsweek. “Statistically, it’s a dead heat.”
That distinction is particularly relevant in races such as Iowa, where the final head-to-head average was tied.
A nominal advantage inside the margin of error carries less analytical weight than an eight- or nine-point discrepancy between the average and the result.
Likely-voter modeling creates another source of uncertainty. Pollsters must determine not only which candidates respondents support, but which respondents are likely to cast ballots.
“Deciphering who the likely voters are and being accurate about that for Election Day is as much an art as it is a science,” Panagopoulos said.
...
Polling organizations use different combinations of voting history, registration records, self-reported enthusiasm and certainty of participation.
If those screens underestimate one party’s turnout, several otherwise accurate interviews can still produce an inaccurate estimate of the electorate.
The toplines alone cannot distinguish late movement from a turnout or weighting error.
That will require pollsters to examine their field dates, undecided voters, weighting decisions, and likely-voter screens against the electorate that participated.
Newsweek’s reporters and editors used Martyn, our AI assistant, to produce this story. Learn more about Martyn here.
Register for our November 10 discussion, 'Newsweek Webinar: How the Politics of AI Could Shape Midterms,' which will examine how AI policy played in key races, data centers, lobbying and the political lines that emerged.
Contact Newsweek editors on this story: Sam Stevenson, Gray R. Thomas
View 0 Comments
Submit Correction
View Editorial & AI Guidelines