A comprehensive analysis conducted by the Pew Research Center indicates that while preelection polls correctly predicted the general direction of the 2026 U.S. elections, they systematically understated the magnitude of the Democratic Party's advantage. Based on unofficial vote counts as of November 11, 2026, the research suggests that polls failed to fully capture the extent of the Democratic victory in numerous major statewide contests and the overall U.S. House of Representatives vote.
The study examined final preelection polling averages from three primary sources—RealClearPolitics, Silver Bulletin, and FiftyPlusOne—comparing them against unofficial results from 27 U.S. Senate and gubernatorial races, as well as estimated House totals. The findings show that in 21 of the 27 analyzed governor and Senate races, polling underestimated the Democratic margin by at least 2 percentage points. Across all 27 races, the mean understatement was approximately 4 points.
Significant disparities were observed in specific high-profile races. In Texas and Georgia, polls understated the Democratic margin by 8.9 percentage points in both the U.S. Senate and gubernatorial contests. Similarly, the North Carolina Senate race saw an understatement of 8.5 percentage points. Conversely, some races showed higher accuracy, such as the gubernatorial and Senate contests in Kansas and the governor's race in Minnesota, where the error was less than 1 percentage point. A rare instance of overestimating the Democratic lead occurred in the Pennsylvania governor's race, where polls overstated the margin by slightly more than 2 points, although incumbent Governor Josh Shapiro still secured a victory by approximately 22 points.
Regarding the U.S. House of Representatives, the 'generic House ballot'—which measures general partisan sentiment—showed a Democratic advantage of nearly 9 percentage points according to RealClearPolitics. However, the actual combined House vote as of November 11 reflected a Democratic lead of just under 14 points, representing a polling error of nearly 5 points.
Pew Research notes that if the average error remains between 4 and 5 points, it would exceed the errors recorded for similar races in the 2018 and 2022 midterms, as documented by the American Association for Public Opinion Research. This level of error is comparable to state-level presidential polling errors seen in 2020, where Joe Biden's margin was similarly overstated, and aligns with gubernatorial polling errors from 2024.
The analysis attributes part of the election's trajectory to a pervasive 'dark mood' among the electorate, characterized by deep dissatisfaction with President Donald Trump's administration, particularly regarding his economic policies. Polls had consistently highlighted a significant enthusiasm gap, indicating that Democratic supporters were more engaged and more likely to vote—and participate in preelection surveys—than their Republican counterparts.
While it is too early to definitively isolate the causes of the polling discrepancies, researchers are examining whether weighting procedures and 'likely voter' models failed to account for shifts between the 2024 and 2026 electorates. Further analysis of panel surveys and turnout data is expected to clarify whether the outcomes resulted from changes in voter choice or unexpected shifts in turnout dynamics.
- Categories:
- Politics
