Trust in elections: why confidence and accuracy are different problems

Trust in American elections is low and deeply split by party, and it rises or falls mainly with whether a voter’s candidate won. In a September 2024 Gallup survey, 84% of Democrats but only 28% of Republicans were confident votes would be accurately cast and counted, a record 56-point gap; after Republicans won that November, confidence among their voters rose sharply. The problem is not only whether elections are accurate, but whether the evidence of accuracy is visible and believed.

In plain words: an election can be counted correctly and still be doubted. Most people never see how ballots are checked, so they rely on what leaders and news sources tell them. When your side loses, you are more likely to doubt the result. When your side wins, you are more likely to trust it. Fixing this means showing people the checks, not just telling them everything is fine.

The problem of election trust, defined precisely

Election trust is several different beliefs, and surveys measure them with different questions.

Voter confidence. A belief that votes were, or will be, counted as cast. The MIT Election Data and Science Lab notes that researchers usually ask whether votes “were counted as cast,” which is narrower than asking whether an election was fair.⁠³

Scope of confidence. Confidence in one’s own vote, in one’s county, in one’s state, and in the nation as a whole are distinct. They routinely differ by large margins.⁠³

Winner-loser effect. The tendency of supporters of winning candidates to report more confidence than supporters of losing candidates. MIT’s researchers describe it as the strongest single influence on survey confidence.⁠³

Election integrity. The actual accuracy, security, and lawfulness of an election: eligible voters can vote, ineligible ballots are excluded, and ballots are counted as cast. Integrity is a fact about the election; confidence is a fact about the public.

Verification. Evidence, such as a paper record checked by an audit, that lets someone other than the official in charge confirm the outcome. See Audits and verification.

Reassurance. A statement by an authority that the election was sound. Reassurance can be true and still fail to persuade people who distrust the authority.

The central confusion this page addresses is treating low confidence as proof of low integrity, or treating high integrity as proof that confidence should follow.

Scale and evidence of distrust in elections

Confidence in American elections, measured before the 2024 presidential election by Gallup in a telephone survey of 1,007 adults conducted September 3 to 15, 2024, stood at 57% overall. Democrats were at 84% and Republicans at 28%, a record 56-point partisan gap. Republican confidence had been 55% in 2016 and 44% in 2020.⁠¹ The pattern has reversed before: in 2004, under a Republican president, Republicans were more confident than Democrats.⁠¹

After the 2024 election, the picture changed. In a Pew Research Center survey conducted November 12 to 17, 2024, of 9,609 adults including 8,072 citizens who said they voted, 88% of voters said elections across the country had been run at least somewhat well, up from 59% after the 2020 election. Among Trump voters, the share rose from 21% in 2020 to 93% in 2024. Confidence that mail and absentee ballots were counted as intended rose among Trump voters from 19% in 2020 to 72% in 2024, while it fell among supporters of the losing candidate: 78% of Harris voters in 2024, compared with 95% of Biden voters in 2020.⁠²

These numbers measure different things. Gallup asked before the election about votes that would be cast and counted; Pew asked voters afterward about an election whose winner was known. The swing between them is consistent with the winner-loser effect rather than with any change in how ballots were counted.

Scope matters as much as timing. In MIT’s 2020 module of the Cooperative Election Study, 38% were very confident before the election that their own vote would be counted as intended, falling to 35% for their county, 29% for their state, and 15% for the nation. After the election, confidence in one’s own vote rose to 61%. In the 2024 Survey of the Performance of American Elections, nearly 70% were very confident their own vote was counted as intended, with lower figures for county, state, and nation.⁠³ People tend to trust what they can see most directly.

MIT’s analysis of Survey of the Performance of American Elections data found that the partisan gap in 2020 was 32 points, the largest it recorded, and narrowed in 2024 as Republican confidence rose and Democratic confidence fell. It also found that “the overall average level of voter confidence has not changed very much across the past six elections”.⁠³ The aggregate is stable because the two parties’ confidence moves in opposite directions.

Causes and competing explanations

Distrust in elections has several proposed causes. They are not mutually exclusive, and each is supported by different evidence.

Winning and losing. MIT’s researchers state that “the strongest influence on voter confidence in such surveys, regardless of how the question is asked, is whether the candidate one supports has won or lost an election”.⁠³ Close elections lower confidence further: in 2024 data, the closer a state’s result, the less confident its voters were, whichever candidate won.⁠³

Elite rhetoric. In an experiment published in 2021, Clayton and colleagues found that exposure to statements questioning the legitimacy of the 2020 election “erodes trust and confidence in elections” and increased belief that the election was rigged among people who approved of the president’s job performance.⁠¹¹ Leaders can move their supporters’ confidence in either direction.

Real vulnerabilities and real cases. Fraud and error are not imaginary. The bipartisan Commission on Federal Election Reform, led by former President Jimmy Carter and former Secretary of State James A. Baker III, reported in 2005: “There is no evidence of extensive fraud in U.S. elections or of multiple voting, but both occur, and it could affect the outcome of a close election.” It added that “the electoral system cannot inspire public confidence if no safeguards exist to deter or detect fraud or to confirm the identity of voters”.⁠⁵ People who distrust elections can point to documented cases; people who trust them can point to their rarity relative to the number of ballots.

Rule changes. The rapid expansion of mail voting in 2020 coincided with the deepest partisan gap in mail ballot confidence: 19% of Trump voters were confident mail ballots were counted as intended that year.⁠² Whether distrust followed the change in method itself or the political argument about it cannot be separated from these data.

Invisible verification. Most voters never observe an audit, a canvass, or a chain of custody log. When evidence of accuracy is technical and remote, belief rests on trust in whoever reports it. That is an interpretation, but it is consistent with MIT’s finding that confidence falls as the scope moves away from what a voter personally experiences.⁠³

Who is harmed when trust in elections falls

The costs of distrust fall on specific people and institutions.

• Election officials and workers. In a Brennan Center survey of 858 local election officials conducted April 15 to May 17, 2025, 38% reported experiencing threats, harassment, or abuse because of their job; 52% were concerned about the safety of colleagues and staff; 59% worried about political leaders interfering with how they do their jobs; and 81% were concerned about false information about elections spreading on social media.⁠¹²
• The losing side’s voters. A voter who believes the count was wrong loses the reason to accept an outcome they dislike. Peaceful transfer of power depends on losers’ acceptance.
• The winning side’s voters. A result doubted by roughly half the country is a weaker mandate, and a later loss can be framed the same way.
• Election administration. Officials must spend time and money answering challenges, and the Brennan survey found 75% of officials saying their annual budgets need to grow in the next five years to meet administration and security needs.⁠¹²
• Every voter. When confidence becomes a partisan scoreboard, it stops working as a signal of actual problems. A real failure can be dismissed as sour grapes, and a sound result can be dismissed as a cover-up.

What has been tried to build trust in elections

Efforts to raise election confidence fall into three groups: producing evidence, tightening rules, and issuing authoritative statements. Each has a mixed record.

Paper records and audits. According to the National Conference of State Legislatures, all 50 states and Washington, D.C. conduct some type of post-election audit. Thirty-six states plus D.C. require a traditional tabulation audit, and seven states, Georgia, Colorado, Maine, Maryland, Nevada, Rhode Island, and Virginia, have a statutory requirement for risk-limiting audits, which use statistical methods “designed to limit the risk of certifying an incorrect election outcome”.⁠⁸ After the 2020 election, federal and state election officials stated that all states with close presidential results had paper records of each vote.⁠⁴ Advocates such as the Rhode Island working group that piloted risk-limiting audits describe them as “one of the most important tools states can use to bolster voter confidence”.⁠¹⁰

The evidence that audits raise confidence is thinner than the evidence that they catch errors. In a 2021 user study of 105 U.S. residents, Dalela, Kulyk, and Schürmann found that participants “felt less confident” when told how many ballots a risk-limiting audit had checked, apparently because a statistically sufficient sample can look small.⁠⁹ The study is small, but it shows that a sound method can fail as communication.

Identification rules. The Carter-Baker Commission proposed a uniform system of voter identification based on the REAL ID card or an equivalent for people without a driver’s license, and recommended that states provide photo IDs free of charge.⁠⁵ In Crawford v. Marion County Election Board (2008), the Supreme Court upheld Indiana’s photo identification law. The lead opinion noted that the record contained “no evidence of any such fraud actually occurring in Indiana at any time in its history,” but held that “public confidence in the integrity of the electoral process has independent significance” because it encourages participation.⁠⁷ Whether identification laws actually raise confidence, and by how much, is not established by the sources reviewed for this page and remains an open question. See Identity and registration.

Authoritative statements. On November 12, 2020, a joint statement by the Election Infrastructure Government Coordinating Council executive committee and the Sector Coordinating Council, including the Cybersecurity and Infrastructure Security Agency, the U.S. Election Assistance Commission, and associations of state election officials, called the November 3 election “the most secure in American history” and said there was “no evidence that any voting system deleted or lost votes, changed votes, or was in any way compromised”.⁠⁴ Four years later, Gallup measured Republican pre-election confidence at 28%.⁠¹ Reassurance from officials did not settle the question for people who distrusted those officials.

The strongest arguments about election trust across the political spectrum

Election trust is argued from several directions, each with a serious case.

Safeguards build confidence. Many conservatives argue, as the Carter-Baker Commission did, that confidence requires visible safeguards that deter and detect fraud, and that rare does not mean absent.⁠⁵ The Supreme Court’s recognition that confidence has “independent significance”.⁠⁷ supports treating perceived integrity as a legitimate policy goal, not only measured fraud rates.

Access is part of integrity. Many progressives argue that rules justified by rare fraud impose real burdens on eligible voters. The Brennan Center’s 2007 report argued that a person is more likely to be struck by lightning than to impersonate another voter at the polls⁠⁶ On this view, an election that excludes eligible voters lacks integrity even if no ineligible ballot is counted. See Access and exclusion.

Trust should be earned, not requested. A libertarian argument holds that citizens are right to withhold trust from any authority that asks for it, and that the burden is on governments to make processes independently checkable. On this view, the November 2020 statement,⁠⁴ and the 2007 report⁠⁶ are both forms of reassurance, and the remedy is verification any citizen can inspect, such as public audit procedures and observable chain of custody.

Confidence is a partisan barometer. Some election scholars argue that because confidence moves mainly with winning and losing⁠³ survey confidence should not drive election policy at all. Rules should be judged by measured accuracy and access, and confidence numbers read as political mood.

Institutions must protect their workers. Election administrators and institutionalists argue that the most urgent harm is to the people who run elections, citing threat and harassment rates,⁠¹² and that losing experienced officials degrades the very accuracy that critics care about.

These positions overlap. A person can favor identification requirements, broad access, public audits, and protection for election workers at once.

Unresolved questions about trust in elections

• What raises confidence durably? Audits produce evidence of accuracy, but the one controlled study found disclosure of audit sample size lowered confidence.⁠⁹ How should audit results be communicated?
• Do identification rules change confidence? The legal and policy record treats confidence as a goal,⁠⁷ but the size of any effect is not established in the sources reviewed here.
• Can elite rhetoric be countered? If statements from trusted leaders reduce confidence,⁠¹¹ are there messengers or forms of evidence that restore it across party lines?
• Which measure should guide policy? Should lawmakers respond to confidence in one’s own vote, which is high, or confidence in the nation, which is low.⁠³?
• How should officials be protected without reducing transparency? Public observation builds trust, but officials report threats in person⁠¹²
• Application to this framework. Can teaching the difference between verification and reassurance change how people interpret results? No evaluation exists. See Evaluation and Questions.