This has been a year of unpredictable elections.
This summer, the United Kingdom shocked the world with its vote to leave the European Union (aka Brexit). And this Tuesday, the United States will hold the long-awaited election between Hillary Clinton and Donald Trump, a man whose nomination and success as the Republican candidate many pundits failed to predict.
Which presidential candidate you think will win depends on where you go for predictions. But beware—politicians, journalists, and pollsters have incorrectly predicted many contests. Here are some of the most famously wrong political predictions in history.
Napoleon the Underdog
When the French Second Republic held its first presidential elections in 1848, elites thought General Louis-Eugène Cavaignac would easily defeat his rival, Louis-Napoleon Bonaparte, who was the nephew of Napoleon I.
To their surprise, the victor was Bonaparte, “whose populist instincts and defence of law and order, combined with his second-hand charisma as nephew of Napoleon I, played well in the countryside,” write the authors of The Wiley-Blackwell Dictionary of Modern European History Since 1789.
Bonaparte was both the first and last president of the republic. That’s because four years later, he had himself declared emperor.
Roosevelt Wins Again
In 1936, the American weekly Literary Digest confidently predicted that Republican Alf Landon would defeat the Democratic incumbent Franklin D. Roosevelt in a landslide. Instead, FDR won all but two states, and went on to win two more presidential elections.
The problem with Literary Digest’s prediction was that it was based on telephone and postcard surveys. At the time, wealthier homes were more likely to have telephones, and wealthier people were more likely to support Landon. Postcards, meanwhile, only captured the opinions of people who felt strongly enough to respond.
“It was not a scientific poll,” says Allan Lichtman, a distinguished history professor at American University who has correctly predicted every U.S. presidential election since 1984. “It didn’t have a scientific sample and it relied on self-selective returns. It’s similar to the Internet polls you see today.”
Interestingly, 1936 was also the first year the Gallup company conducted its famous presidential polls. Through its scientific poll, it correctly predicted a Roosevelt victory. “This was the birth of modern polling and the end of old-fashioned straw polling—until we got the Internet,” Lichtman says.
Churchill Wins War, Loses Election
In the summer of 1945, British Prime Minister Winston Churchill’s approval rating was understandably high. He’d just led the Allies to victory in Europe, and “politicians and commentators confidently predicted that he would lead the Conservatives to victory at the forthcoming general election,” writes Paul Addison, author of Churchill: The Unexpected Hero, for BBC.
Yet those who anticipated Conservative victory were shocked when the Labour Party won the House of Commons in a landslide, costing Churchill his office.
Churchill was popular as a military leader, Addison writes, but voters favored the Labour Party in part because, while Churchill had focused on winning the war from 1940 to 1945, the Labour Party helped rule the home front.
Although the Conservatives lost in 1945, they won a majority in the early 1950s, making Churchill the prime minister again.
“Dewey Defeats Truman”
One of the most famously bad predictions is captured in an iconic 1948 photograph of newly reelected President Harry Truman holding up a Chicago Tribune that falsely declares “Dewey Defeats Truman.”
“All of the polls were pointing to a Dewey victory, but they stopped polling a few weeks before the election,” Lichtman says. When early election day results seemed to suggest that Thomas Dewey would win, the Tribune went ahead and set the headline for the next day.
Although Truman, the Democratic incumbent, was an unpopular president, Lichtman says he still had appeal.
“He had established America as the new leader of the free world, the country was relatively tranquil, [and] he wasn’t up against a particularly charismatic candidate in Dewey,” he says. “So he didn’t have everything going for him, but he had quite a bit.”
Everyone’s a Pundit
Today, voters are besieged by multiple polls a day—both scientific ones like Gallup’s, and straw polls like the one that embarrassed the Literary Digest.
The plethora of predictions means that at the end of this week, a lot of people will be wrong, and a lot of people will be right—in some cases, like Nate Silver’s, that could impact a person’s career.
As for what the bad predictions of the future will look like, the New Yorker may have given us an idea. When Kanye West declared last year that he would for president in 2020, the magazine ran a cover cartoon featuring a triumphant West holding a newspaper that read “Trump Defeats Kanye.”