A 2025 review by the American Association for Public Opinion Research (AAPOR) found U.S. polls moved closer to 2024 results but continued to overstate Democratic margins, keeping voter representation concerns alive ahead of 2026.
The review covered 611 polls of presidential, Senate, gubernatorial and House races conducted in the campaign’s final two weeks. Their average absolute miss on the two-party margin was 3.3 percentage points. The comparable averages were 5.3 in 2020 and 5.2 in 2016. Absolute error measures the size of the gap between a projected margin and the final one; it does not show which party the poll favored. National presidential polls had a 2.6-point average miss, while state presidential polls averaged 3.0 points.
On direction, polls placed Democratic margins 2.7 points above the results across offices in 2024, after overstatements of 4.6 points in 2020 and 3.1 in 2016. In each of the past three presidential elections, Republican support came in stronger relative to Democrats than pre-election polling had suggested, AAPOR found.
Federal Election Commission (FEC) results compiled from state election offices show Trump received 77,302,580 votes, or 49.80%, while Kamala Harris received 75,017,613, or 48.32%. Trump won 312 Electoral College votes to Harris’ 226.
AAPOR identified several groups and places that polls had difficulty capturing: Republicans in GOP-leaning areas, Hispanic voters whose support for Harris was overstated, and voters who skipped the 2020 election but leaned Republican. Polls also projected less turnout in counties that backed Trump in 2020 than occurred, while turnout fell in Biden-supporting counties. AAPOR said the turnout modeling gap accounted for a moderate portion of the remaining directional error.
The review did not find one polling approach that repeatedly produced more accurate estimates. Most major methodological choices had little relationship to error size, and the value of using past voting behavior depended on how pollsters measured and applied it. AAPOR found only minor average directional misses in both the 2018 and 2022 midterms.
A January 2025 Silver Bulletin report described disagreement among pollsters over whether recurring misses were specific to Trump or reflected broader difficulty reaching less-educated, infrequent or Republican voters. Natalie Jackson, vice president at GQR Insights, told Silver Bulletin, “The not-so-good thing is that we still underestimated Trump, which is concerning.” The outlet reported that pollsters viewed 2026 Senate results as one indicator of whether polling misses also occur outside Trump-led presidential races.
The next federal general election is scheduled for Nov. 3, 2026.
