Polls missed the scale of Democratic gains in midterms
Democratic pollster Anna Greenberg says new voters and a bias in who answers surveys caused public polls to understate the margin of victory in several key races.
By Rebecca Svoboda ·
The 2026 midterm elections produced a level of Democratic success that many analysts did not fully foresee. While pre-election surveys correctly identified a strong showing for Democrats, they underestimated the magnitude of the results in specific regions and demographics.
Democrats flipped at least seven Senate seats, including in traditionally Republican states such as South Carolina, Kansas, and Texas. In many of these races, the final margins were wide, and counties that voted for Donald Trump in the 2024 election recorded double-digit swings toward Democratic candidates in 2026.
Anna Greenberg, a Democratic pollster with GQR Insights and Action, assessed the performance of the polling industry in an interview with NPR on Tuesday. She stated that while the polls were directionally correct, they failed to capture the full extent of the Democratic surge.
Greenberg identified two primary factors for the discrepancy. First, she noted that Democrats are historically more likely to participate in surveys than Republicans. To account for this bias, many pollsters adjust their models to be more conservative, which can result in lower projected Democratic vote shares in public data.
Second, Greenberg pointed to the high volume of new voters in the 2026 cycle. She explained that polling requires a clear understanding of the electorate to predict outcomes accurately. When a large number of first-time or returning voters participate, as occurred in 2016, 2020, and 2024, predictions become less reliable. She said the scale of new voter participation in the recent primaries provided early signals, but the final turnout still exceeded expectations.
Greenberg compared the 2026 midterm to the 2018 election in terms of turnout size but noted it was more impactful, resembling the dynamics of a 1976 presidential election. She suggested the results may represent a realigning election, where voters shift permanently toward the Democratic party and the Republican party must redefine its future direction.
A notable demographic shift occurred among young men. Greenberg observed that this group moved substantially toward the Democratic Party in 2026. This trend contrasted with the 2024 presidential campaign, where a focus on male-centric messaging and podcast culture was perceived to resonate with young men. She stated that the strategies that worked in 2024 had little impact on this demographic in the midterms.
The polling industry has faced criticism for recent inaccuracies, particularly in primary elections where races appeared closer than the final results showed. Greenberg distinguished between campaign polling and public polling. She argued that many of the errors cited by the public stem from cheap, online polls generated for news coverage rather than for strategic campaign advice. She maintained that professional campaign pollsters maintained a higher level of accuracy in their private models.
When asked whether the 2026 results constitute a permanent political realignment or a temporary reaction to low presidential approval ratings, Greenberg noted that realigning elections typically follow a major external event. She acknowledged that Donald Trump has significantly impacted the political system over the last decade but stated that historians usually identify realignments years after they occur. She said it remains necessary to observe whether the new voting coalitions hold firm in future elections.
