H eading into Wednesday night’s third and final presidential debate, Republican candidate Donald Trump continues to warn his supporters that the election on Nov. 8 could be “rigged”—a claim he has made with no specific evidence, and that has been rebutted by everyone from President Obama to Speaker of the House Paul Ryan to experts like the commissioner of the federal Election Assistance Commission.
Those refutations hinge largely on the fact that there is no evidence that rigging the election is even a possibility. And it goes even deeper than that: Experts say that in all of American history, there has not been a single instance of a presidential election being truly rigged.
That doesn’t mean there has never been a rigged election, period. It has happened—but, despite Trump’s rhetoric, those who study the history say it is less likely to happen in 2016 than before.
In order to understand this point, the first thing to understand is what it means for an election to be “rigged.”
An election can have many problems—technical mistakes, poor administration, discriminatory laws and even individual voter fraud—without being rigged. In order for an election to be truly rigged, there has to be a top-down and intentional manipulation of the result away from what it would have been if the votes had been counted according to whatever the law is at the time, says Pippa Norris, Director of the Electoral Integrity Project. “You hear [about rigged elections] a lot in autocracies, you hear it in countries in Africa where for example the president doesn’t want to give up power or where the sore losers say the election was stolen, and in some cases that’s true because there have been real manipulations of the outcome,” she says, “but it’s not a claim which is heard in established democracies like the United States.”
So what does it look like when an American election is really rigged?
One modern example, says Edward B. Foley, author of Ballot Battles: The History of Disputed Elections in the United States, can be found in Lyndon B. Johnson’s race for the Senate in Texas in 1948.
According to the account by Johnson biographer Robert Caro, the future president won a runoff primary by a mere 87 votes, after 200 extra votes were added to what Foley calls “the infamous Ballot Box 13.” That election may well have seen manipulation beyond that instance too, and on both sides. As Foley points out, though there’s no demonstrable evidence that Johnson himself participated in what happened, there were certainly people in the area who had an interest in his victory and who would have been able to make sure it happened. That sort of planned and intentional manipulation qualifies, for Foley, as rigging an election.
It makes sense that any real examples of rigging would happen in something like a Senate primary rather than a national general election; decentralization is cited as one of the main reasons why it would be nigh impossible to rig a presidential election in the United States. The more local you get the more a single authority can control an outcome.
But most examples of demonstrable election interference happened in the 19th century or earlier; for example, a 1792 congressional race in Georgia was found to be corrupt, leading to a decision to leave the seat vacant.
“We’re improving in our ability to avoid this problem,” Foley says.
And even what happened in 1948 would be hard to pull off these days. For one thing, even local elections are under far more scrutiny than they were at the time.
“The mechanism is fairly simplistic of what happened in 1948,” Foley says. “At the end of the day you look at the polling list and see who hasn’t voted, and then you just pretend to vote extra votes on behalf of the people who never showed up that day. But the only way you can do that is in a polling place that’s essentially under the control of one party, that doesn’t have bipartisan observers or media observers.”
For another thing, his research has found that “cultural attitudes about democracy and expectations about the voting process” have changed significantly. First during the Progressive Era and again during the Civil Rights Era in the 1960s, Americans saw a radical shift in the kind of behavior that was considered acceptable during an election. And, though he says it’s impossible to prove causation, Foley points to the coming of women’s suffrage as a third milestone, after which society moved away from seeing elections as a militaristic contest, and began seeing them as a civilized event that ought to be safe for voters.
Finally, another important change came in the form of technology. Early ballots were distributed by political parties and could look very different depending on who did the printing. That meant it was obvious at a glance which party would be getting a particular vote. That lack of secrecy opened the doors to intimidation or enticement of individual voters. In the years since the advent of secret ballots, election officials have made a concerted effort (though not always successfully) to keep voting secure and private.
In short, no presidential election in American history can be shown to have been rigged, and it has been getting steadily more difficult to rig any elections.
So how did we get to a point where a significant portion of the electorate believes that the 2016 just might be the year a presidential election actually does get rigged?
To Norris, the perceived feasibility of a rigged election is closely tied to the events of 2000, when voting problems and a close race led George W. Bush and Al Gore all the way to the Supreme Court. In the years since, the public has become more aware of the possibility of something going wrong with an election. Meanwhile, that same time period corresponds to an increase in the polarization of the electorate. That becomes a problem because local election officials are often partisan, unlike the nonpartisan or bipartisan civil servants who usually hold such positions in other established democracies. (Not to mention the fact that the Supreme Court’s current incomplete state would make it an awkward place to ultimately resolve such a dispute.)
“We haven’t invested sufficient resources to make sure our elections run smoothly,” Norris says. “Then you add the polarization to those technical errors, and you can immediately see how you can have problems.”
History indicates that, rather than worrying about rigged elections, voters would spend their anxiety more wisely on what might happen after a close election. Simple human error is more likely to cause a problem than a nefarious plot, but the ways to resolve those errors are not necessarily adequate and nonpartisan.
Foley points out that “the structural problem with an election is that you have a winner and a loser and nobody likes to lose,” and that the stakes have become even higher as the power of the presidency has increased. That means there’s more incentive to question a result if it’s close. As early as the 1820s, as Foley discusses in his book, it was clear to individuals like James Madison that the Constitution had not provided an adequate fail-safe in such a case.
One of the elections that sometimes comes up in discussions of possibly rigged presidential contests is 1960, as some claim there is evidence that John F. Kennedy’s victories in Illinois and Texas—and thus in the Electoral College as a whole—were illegitimate. In fact, Foley says it’s impossible to know either way. Kennedy’s opponent Richard Nixon knew that he would have to overturn both results to win the election, and that Texas was a one-party state (and home of JFK’s running-mate Lyndon Johnson) where there was no fair process for a recount, so there was no point in pursuing any such charge. That lack of a good recount system may have led Nixon to give up when he could have won. Or, on the other hand, it may have left an undeserved mark of speculation on Kennedy’s record, which would have been cleared by a recount.
The real lesson of 1960, the same lesson that was seen in the 1820s and in 2000, is that it’s important to have a reliable system for verifying an election result.
Without such a system, it can be hard to convince voters that their voices will be fairly heard—even when history shows that they have little cause for concern.
“It’s easy to lose trust in the system. It’s very difficult to reassemble it and put it back together again,” Norris says. “You can think of it as a bit like Humpty Dumpty.”
Write to Lily Rothman at lily.rothman@time.com