With the midterm election approaching, 49% of adults in a new UC San Diego survey said they expected the contest to be free and fair. Twenty-three percent said they did not, and 28% were unsure. The result describes public confidence, not the security or administration of the election itself, but that perception can shape whether voters participate and accept outcomes.
The UC San Diego Center for Electoral Innovation and Engagement surveyed 5,016 adults online from Sept. 1 through Sept. 9. The Forthright sample was matched to the U.S. adult population and included attention checks. Researchers report a margin of error of plus or minus 1.4 percentage points for the full sample. Estimates for parties or other subgroups carry larger uncertainty because those groups contain fewer respondents.
Confidence in accurate vote counting has declined across three measurements. Seventy-seven percent expressed confidence after the 2024 election, compared with 60% in January 2026 and 55% this month. The latest figure included 63% of Democrats, 58% of Republicans and 49% of independents. Those gaps are narrower than a simple partisan story would suggest; skepticism is not concentrated entirely in the party that lost the last national election.
The survey also tested specific concerns. Artificial-intelligence deepfakes worried 76% of respondents, more than redistricting at 44%, foreign interference at 41% or cyberattacks at 39%. A deepfake question measures anticipated risk, not documented exposure. The distinction matters because vivid synthetic media can raise anxiety even when a respondent has not encountered a verified example tied to a ballot or election office.
One proposed rule produced broad support: 71% favored banning election officials and poll workers from wagering on election results. Betting creates a conflict concern because an official may have access to nonpublic information or influence an administrative decision. The survey did not ask respondents to design enforcement, define which markets count or distinguish a small local wager from a financial position on a national outcome.
Mail voting remained more polarized. The full report found that 61% of Republicans supported limits on mail ballots, while 70% of Democrats opposed them. Policy preferences can reflect both confidence and convenience. A voter may trust counting procedures yet favor in-person voting, or rely on mail while worrying about misinformation. A single support-or-oppose item cannot fully separate those reasons.
Online panels can approximate population characteristics, but weighting does not remove every difference between people who join survey panels and those who do not. Answers can also shift with question wording, news events and whether an election has recently produced a preferred result. The reported margin addresses sampling uncertainty; it does not capture every source of measurement error or prove that attitudes will remain fixed through Election Day.
The most consequential finding may be the size of the undecided group. More than one in four adults did not yet say the midterms would or would not be fair. Election agencies, campaigns and platforms will compete to fill that uncertainty with procedures, claims and images over the coming weeks. The survey supplies a baseline for belief. Audits, transparent rules and the conduct of the election will supply the evidence against which that belief should be tested.