NEW YORK — Three polling organizations published results within hours of each other on Tuesday showing the same thing: independent voters in the United States are breaking for Democrats at a level that has historically preceded wave elections, and a near-majority of registered voters say their vote in November is primarily an act of rejection toward Donald Trump.
The convergence is not coincidental. The NPR/PBS NewsHour/Marist poll, conducted September 14–15 among 1,280 registered voters, found Democrats leading the generic congressional ballot 53 to 41 percent. Among independent voters alone, the margin was 57 to 32, a 25-point gap that, if it holds through November, would represent one of the most severe independent defections from a party in power in modern polling history. On a morning when the same president was threatening military action against Iran from the United Nations podium in New York, his standing at home continued its downward arc.
The Marist numbers arrived alongside a Verasight/Strength in Numbers survey showing Trump’s approval at 33 percent and an Economist/YouGov poll pegging it at 35. The consistency across methodologically distinct surveys matters. These are not house effects canceling each other out. They are three different organizations, three different samples, producing the same picture on the same day.
What the numbers show about motivation is perhaps more structurally significant than the raw ballot margins. Forty-five percent of registered voters surveyed by Marist said their congressional vote would be “mostly a vote against Trump.” Only 24 percent said it would be “mostly a vote for Trump.” The remaining respondents described themselves as casting ballots on issues rather than on the president. That 45-to-24 ratio, with the anti-Trump motivation more than doubling the pro-Trump motivation, is a pattern political scientists associate with punishing midterms. It mirrors the 2006 cycle, when disgust with George W. Bush’s Iraq war policy produced a 31-seat Democratic pickup, and the 2018 cycle, when resistance to Trump’s first term delivered a 41-seat swing.
The historical analogues are imperfect. Gerrymandering has reshaped the competitive map significantly since 2006. Republicans drew or controlled the drawing of congressional lines after the 2020 census in enough states to create a structural floor that did not exist under earlier configurations. Forecasters who once calculated Democratic wave conditions translating directly into seat gains now work through a filter of 35 to 40 Republican-held districts where the underlying partisan lean has been hardened by redistricting. Whether a 12-point generic ballot edge produces gains proportional to 2006 or 2018 is genuinely unclear.
The CBS News battleground polling conducted earlier this month projected Democrats at 228 House seats if current trends hold, one seat above the majority threshold. That projection was built on similar independent-voter defection data. The Marist poll’s 57-to-32 independent margin is, if anything, wider than what CBS was measuring.
What the polling cannot tell observers is whether the Democratic structural advantages will survive the ground game. Presidential-year turnout models do not translate directly to off-year elections. The college-educated suburban voters who have been the most consistent source of Democratic gains since 2018 vote reliably in midterms; the younger and lower-income voters who moved toward Democrats in the Marist survey are historically more difficult to turn out in non-presidential years. A 25-point independent margin means very little if the people producing it do not show up.
The Trump campaign and the Republican National Committee have not publicly engaged with the Marist results. A White House spokesperson did not respond to a request for comment by the time of publication. The RNC’s communications office offered no statement.
The press freedom context adds a layer to what the polling captures. Trump’s administration has been locked in federal litigation with CNN, MSNBC’s parent company, and Politico over what those outlets described in a joint lawsuit as unconstitutional exclusion from White House access based on editorial positions. Whether the administration’s assault on independent media coverage shapes the information environment those voters inhabit by November is a question the polls cannot answer.
The polling average aggregated across all public surveys currently shows Democrats at plus-7.4 on the generic congressional ballot. The Marist survey, at plus-12, is at the upper end of that range but not an outlier in direction. No major public poll published in September 2026 shows Republicans ahead on the generic ballot, according to NPR’s polling analysis.
The clearest statement the data permits: the structural conditions for a Democratic wave exist. Whether those conditions produce a wave depends on machinery, including organizing, turnout, and candidate quality in individual districts, that polling cannot measure. November is seven weeks away. The independent voters in the Marist survey have registered their preference. The question, as it always is, is which of them will vote.
