Stable U.S. vote shares conceal the independents deciding Congress

Across six nationally representative surveys of 16,060 registered US voters conducted between November 2020 and December 2025, the estimated Democratic share of the two-party generic congressional ballot remained remarkably stable, fluctuating within a 4.6-percentage-point range. Support was 52.9% in November 2020, fell to 48.3% in December 2023 and rebounded to 52.4% in December 2025, despite a period encompassing the attack on the Capitol, the Supreme Court’s Dobbs decision, conflicts in the Middle East and two presidential transitions. The aggregate figures, however, conceal a more consequential dynamic: a comparatively small group of voters without firm partisan attachments accounted for almost all of the recent movement, while their issue priorities were sharply distinct from those of committed partisans.

The study used repeated cross-sectional YouGov surveys conducted in November 2020, November 2022, June and December 2023, June–July 2024 and December 2025. Samples came from an opt-in online panel and were weighted with propensity scores to national benchmarks, including presidential vote. The same provider, sampling approach, weighting procedures and generic-ballot wording were used throughout, reducing the risk that changes over time resulted from switching polling organizations or question formats. Survey-weighted estimates and confidence intervals were used for party shares, while quasi-binomial regressions expressed the results as average marginal effects. Because the data are repeated cross-sections rather than a panel tracking the same people, the analysis cannot demonstrate that particular individuals changed their minds or establish causal effects.

Researchers defined “persuadable voters” as self-identified partisan independents who described themselves as ideologically moderate or unsure. This operationalization was intended to capture people with weak prior partisan commitments rather than independents who merely avoided selecting a party label. It was also tested against broader and narrower alternatives. The findings were strongest among people with weak partisan ties generally, and less consistent when persuadability was restricted specifically to ideological centrists. That distinction matters because an independent may remain ambivalent about party control while holding firm ideological commitments. The results therefore point toward weak partisanship and political uncertainty as more reliable markers of potential movement than centrism by itself.

Issue priorities showed much larger differences than conventional demographic groupings. In December 2025, 35.2% of all respondents named the economy, inflation or tariffs as their most important issue, while 32.0% prioritized preserving democracy and political violence and 12.5% selected immigration. Democratic support among registered voters was 79.5% in the democracy-and-political-violence group, 42.2% among those prioritizing economic issues and 4.8% among those prioritizing immigration. Among the narrower class of persuadable voters, 38.8% prioritized economic issues, 29.8% democracy and political violence and 8.3% immigration. These are associations in observational survey data, not experimental findings, but their magnitude was striking.

The relationships remained after controls for ideology, party identification, education, race and Hispanic origin. Relative to voters prioritizing economic issues, those prioritizing democracy and political violence were 7.1 percentage points more likely to favor the Democratic candidate in Congress, while those prioritizing immigration were 13.0 points less likely to do so. For a baseline profile of an independent, moderate, low-news-attention, non-college, white, non-Hispanic voter, the model predicted Democratic support of 26% when the top issue was the economy, 67% when democracy and political violence ranked first and 0.6% when immigration ranked first. Liberal identification was associated with a 30.4-point advantage for Democrats, whereas conservative identification was associated with a 46.4-point Republican advantage, illustrating how issue priorities can partly operate through ideological and partisan sorting.

The strongest evidence for movement came from comparing the June–July 2024 and December 2025 surveys. Independents, including partisan leaners, represented 22.1% of two-party respondents in 2024 and accounted for 3.4 points of the 3.6-point increase in Democratic support, or 94.6% of the change. Their Democratic preference rose from 45.1% to 60.7%, while support among voters identifying with a party moved only from 49.7% to 49.9%. True independents without partisan leaners supplied 18.9% of the increase, while their Democratic preference rose from 49.4% to 62.5%. The latter change had a wide 95% confidence interval of minus 1.6 to 27.9 points and a p-value of 0.080, so the evidence was suggestive rather than conclusive at conventional thresholds.

Repeated cross-sections also prevent a clean separation of persuasion from changing group composition. Between 2024 and 2025, Democratic leaners increased from 32.7% to 44.7% of the independent category, while Republican leaners declined from 45.8% to 29.9%. This means some of the apparent movement could reflect voters changing how they described their partisan identity rather than changing their congressional preference. The narrower true-independent result suggests that changing identity cannot explain everything, but the study cannot determine which mechanism predominates for particular respondents. Demographic composition also failed to account for the aggregate stability: trends in education and race groups were comparatively modest, and interactions between survey date, education and race were not statistically significant.

A comparison with historical data indicated that the generic congressional ballot was a much stronger predictor of the national House popular vote than presidential approval, with reported coefficients of determination of 0.77 and 0.05, respectively. The authors also found little relationship between state-level turnout changes from 2020 to 2024 and changes in the two-party presidential vote, with an R-squared of 0.07. Within the survey data, reported prospective and retrospective turnout measures barely altered the two-party Democratic margin: all estimated differences were below one percentage point and each 95% confidence interval included zero. These checks address aggregate turnout concerns, but they do not reproduce voter-level election outcomes or show that turnout was irrelevant in close individual districts.

Historical data from 2020 and 2022 suggested that the Dobbs decision may have altered how independents weighed abortion. The estimated association between rating abortion as important and choosing a Democratic congressional candidate shifted from minus 3.8 percentage points in November 2020 to plus 10.8 points in November 2022. By contrast, violent crime and immigration remained associated with Republican congressional preference in both surveys. The researchers could not compare issue-salience instruments directly across all six waves because the 2020 and 2022 surveys asked respondents to rate each issue’s importance, whereas the December 2025 survey asked for a single most important issue. The findings therefore describe changes in issue profiles at selected points rather than a continuous measure of issue-priority shifts throughout the entire five-year period.

Overall, the evidence suggests that durable partisan coalitions can coexist with rapid movement among voters who lack strong attachments. The analysis is observational, relies on self-reported party identity, issue priority and turnout, and studies registered voters rather than confirmed election participants. It also combines independent moderates with ideologically unsure respondents in its main definition and cannot establish why any respondent changed preference. Even with those limitations, the results identify a concentrated source of electoral volatility: in a closely divided, highly polarized environment, modest shifts among a small group of weakly partisan voters can affect which party controls Congress, even while national vote-share estimates appear remarkably unchanged.

Subject of Research: How U.S. issue priorities and weak partisan attachment relate to changes in generic congressional vote preference from 2020 to 2025.

Article Title: Partisan preferences remain stable as decisive voters remain persuadable: evidence from five years of political turmoil

Article References: Alvarez, R. M., & Ebanks, D. (2026). Partisan preferences remain stable as decisive voters remain persuadable: evidence from five years of political turmoil. Frontiers in Political Science, 8, Article 1954942. https://doi.org/10.3389/fpos.2026.1954942

Image Credits: AI Generated

DOI: 10.3389/fpos.2026.1954942

Keywords: U.S. elections, persuadable voters, partisan independents, issue voting, congressional elections, YouGov surveys, public opinion, political polarization, generic ballot, electoral behavior, Partisan, preferences