The 2022 midterm elections are underway, with many Americans, in particular Republicans, increasingly fearful that the outcome will be affected or even determined by election fraud.
A Fox News poll conducted in early October found that 55 percent of those surveyed were extremely or very concerned about voter fraud. Among Republicans, nearly three-quarters said they were that concerned.
These fears have been further fanned by the constant refrain by former president Donald Trump and his allies that Joe Biden triumphed in the 2020 race only because he stole the election — a lie that has been debunked over and over again.
Here’s the truth:
— By every single metric, election fraud is rare in the United States.
— Almost no elections in the past 50 years have been flipped because of documented voter fraud, with occasional exceptions at the local level.
There are logical reasons for this.
The decentralized system of American elections — where elections are run by more than 8,000 local governments and almost 90 percent of Americans vote on paper ballots, according to the U.S. Election Assistance Commission — make it impossible to steal a nationwide election through voter fraud.
It is true that election results can be manipulated. Politicians, for instance, might draw election district boundaries to make it all but certain that one political party will win. Trump allies sought to undermine the 2020 results with schemes involving fake electors, an effort to block Congress from affirming the outcome.
But that’s different from ballot stuffing, illegal voting and other types of behavior that would directly affect the outcome at the ballot box and have become the obsession of many election deniers.
Indeed, false claims about the prevalence of voter fraud are nothing new.
Long before Trump sought to undercut the legitimacy of the 2020 election, Republican officials at both the state and federal levels often made dubious claims of election fraud to justify new and potentially burdensome rules such as Voter ID laws. When Donald Trump won the presidency in 2016 but lost the popular vote by more than 3 million votes, he embraced false claims that the margin was the result of fraudulent votes cast by undocumented immigrants. Trump even assembled a voting integrity commission to validate such claims — but it was disbanded after it uncovered no evidence of fraudulent votes being cast.
Let’s unpack the details.
Few cases of documented fraud
Whenever experts and reporters have tried to tally cases of election fraud, the numbers remain minuscule.
Nearly 160 million votes were cast in the 2020 election. By the popular vote count, Biden beat Trump by 7 million votes. But the presidency goes to the person who wins the most electoral college votes, which are awarded to a state’s victor. That can magnify the importance of votes cast in states where the margins are tight.
But an Associated Press review of every potential case of voter fraud in six battleground states found fewer than 475 out of more than 25 million votes cast in those states — Arizona, Georgia, Michigan, Nevada, Pennsylvania and Wisconsin. The disputed ballots amounted to just 0.15 percent of Biden’s victory margin.
In Pennsylvania, for instance, 26 votes were flagged by election officials as potentially fraudulent — and Biden won there by more than 80,000 votes. In one case, a man went twice to the polls, voting once on his own behalf and once for his son.
The state with the most cases is Arizona, with 198 potentially fraudulent votes — such as a woman suspected of sending in a ballot for her dead mother — but Biden still won the state by more than 10,000 votes.
In the 1960 race between Republican Richard M. Nixon and Democrat John F. Kennedy, 10 states were decided by fewer than 10,000 votes. Republicans charged that Kennedy won Illinois because of voter fraud committed by the Democratic Party in the state. But academic research later found election irregularities would not have changed enough votes to alter the outcome.
The conservative Heritage Foundation maintains a database, dating back to 1979, that it says includes a “sampling” of election-fraud cases brought by prosecutors. “The database is up to 1,384 proven cases and we are following dozens of other prosecutions that are ongoing,” says Hans von Spakovsky, a Heritage senior fellow. That’s an average of 32 examples per year that the group has documented.
Von Spakovsky says the number would be much higher except that local prosecutors are not always interested in pursuing election-fraud cases. He cited a report from the conservative Public Interest Legal Foundation that found more than 150 referrals of potential criminal violations of election law in 2020 had been ignored in Florida.
Still, these numbers — the most comprehensive account of U.S. election-fraud cases in the past four decades — are just a fraction of the 2 billion votes cast in federal elections in that same time-period, according to a calculation for The Fact Checker by the Brennan Center for Justice. In a critical review of the Heritage numbers, the center said the database “confirms that widespread voter fraud does not exist.”
Fraud involving mailed ballots is rare
During the 2020 election, many states expanded options for voters, such as vote by mail, because of the coronavirus pandemic. Trump repeatedly said that these changes would lead to fraud — and some states have since added new restrictions on mail voting.
But a Washington Post analysis in 2020 of data collected by three vote-by-mail states with help from the nonprofit Electronic Registration Information Center (ERIC) found that officials identified just 372 possible cases of double voting or voting on behalf of deceased people out of about 14.6 million votes cast by mail in the 2016 and 2018 general elections, or 0.0025 percent.
Five states — Colorado, Hawaii, Oregon, Utah and Washington — hold elections almost entirely by mail and have reported virtually no fraud, according to a Brookings Institution analysis of the Heritage database.
Voter impersonation is also very rare, according to the data. Writing in The Post in 2014, Loyola Law School professor Justin Levitt — now a White House adviser — said he had found only 31 credible cases of people voting in the name of someone else between 2000 and 2014, out of 1 billion votes being cast.
Some cases of alleged voter fraud turn out not to be intentional. A Texas man, Hervis Rogers, waited in line for more than six hours to vote in 2020, not realizing that he was not eligible under state law because he was on parole for a felony. Texas Attorney General Ken Paxton ordered his arrest, but a judge dismissed the charges last month.
Fraud rarely affects the outcome
In the last half-century, there are only scattered examples of where election fraud appeared to have made a difference in the outcome. They often take place in races that attract relatively few voters and thus the impact of fraud could be greater.
In a 2003 race for the Democratic nomination for mayor of East Chicago, Indiana, for instance, one candidate eked out a 278-vote victory, out of more than 8,000 votes cast, mostly on the strength of absentee ballots. But a judge determined the evidence showed “pervasive fraud, illegal conduct, and violations of elections law” in the gathering of those ballots. The state Supreme Court ordered a new election.
More recently, in 2018, North Carolina ordered a new congressional election after an absentee-ballot scheme was discovered in one county that helped to narrowly tip the election to the Republican. Suspicions rose after 61 percent of the vote-by-mail ballots in the county were cast for the Republican candidate, despite the fact that only 16 percent of them were registered Republicans.
The most prominent case of a major race likely being decided by election fraud took place 75 years ago. In 1948, Lyndon B. Johnson, then a member of Congress, won a Democratic runoff Senate primary against former governor Coke Stevenson by just 87 votes. Historian Robert A. Caro documented in 1990 that Johnson gained hundreds of votes through fraud and malfeasance — a victory that eventually took Johnson to the presidency.
2020 fraud claims were disproven
Since losing the election, Trump has held rallies across the country where he devotes a long section of his speech to repeating debunked claims about the election. Trump’s technique has been to overwhelm his listeners with details — usually irrelevant details — to leave an impression of an election system that is highly suspicious and fraudulent.
But Trump ignores that at least 86 judges, including Trump appointees, rejected at least one post-election lawsuit filed by Trump or his supporters and that they consistently found there was no substantive evidence to support claims of fraud and irregularities. “Calling an election unfair does not make it so,” wrote Trump federal appeals court nominee Stephanos Bibas in one opinion. “Charges require specific allegations and then proof. We have neither here.”
Then audit after audit pushed by Trump’s supporters in many battleground states again have failed to show that fraudulent votes were cast.
In Arizona, for instance, a Republican-commissioned review of nearly 2.1 million ballots in Arizona’s largest county actually added to the margin of Biden’s narrow victory in the state. Biden gained 99 votes while Trump lost 261 votes. “Truth is truth and numbers are numbers,” said Karen Fann, the Republican Senate president who commissioned the vote review.
Beyond ground-level claims about dead voters and absentee ballot fraud, Trump and his allies have also sought to bolster their case by citing statistical anomalies that they say show the election was stolen. For instance, they have pointed to suspiciously high turnout in Democratic strongholds, and the failure of Biden to win “bellwether counties” — counties that supposedly vote in alignment with the final result.
In a report for the Proceedings of the National Academy of Sciences (PNAS), a peer-reviewed journal of the National Academy of Sciences, a trio of political scientists in 2021 debunked every one of these statistical claims.
“The common logic behind these claims is that, if the election were fairly conducted, some feature of the observed 2020 election result would be unlikely or impossible. In each case, we find that the purportedly anomalous fact is either not a fact or not anomalous,” wrote Justin Grimmer of Stanford University’s Hoover Institution, Andrew C. Eggers of the University of Chicago, and Haritz Garro of Stanford.
For instance, statistical analysis showed that bellwether counties in fact had “no special prognostic value” and that Biden’s performance was not much different from the 2016 Democratic presidential candidate, Hillary Clinton.
As for claims of high turnout in counties where Republicans made fraud accusations, it turns out the differences were small or insignificant. “There is no evidence that turnout was unusually high in the suspicious counties, let alone that turnout was inflated in these counties by fraud,” they wrote.
Another frequent claim was that Biden outperformed expectations in counties that used Dominion voting machines, suggesting votes had been switched as Trump’s legal team had claimed. Again, that proved to be false: “Using the most rigorous specifications we find no evidence that Biden outperformed expectations in counties where Dominion machines were used.”
Send us facts to check by filling out this form
Sign up for The Fact Checker weekly newsletter
The Fact Checker is a verified signatory to the International Fact-Checking Network code of principles