Consider, for a moment, the biggest upset in American political history.
For many of us, the 2016 presidential election immediately comes to mind. But well before Donald Trump beat Hillary Clinton, Harry Truman pulled off his own shocker in 1948.
In the months leading up to that election, Truman’s opponent, New York Gov. Thomas Dewey, was comfortably ahead in the polls. But, as we know, polls don’t always get it right. Truman defeated Dewey — despite what the Chicago Daily Tribune’s front page said the next day — and he became the 33rd president of the United States.
His win also rattled the polling world. Twelve years earlier, George Gallup had founded the American Institute of Public Opinion, and since that time, pollsters had successfully predicted the outcomes of the 1936, 1940 and 1944 elections.
The polling community got ‘48 wrong for a few reasons. George Gallup noted that pollsters thought Dewey had the election in the bag and stopped polling too close to the election. In the years following the election, the pollsters would make major changes to their polling practices.
But 2016's political upset was different. Pollsters haven’t really changed their methods. And Frank Yang says that’s because national polls were mostly accurate.
“There was a lot of hand wringing about the death of polling,” says Yang, a Democratic pollster and partner with Hart Research Associates. “Could people ever trust pollsters again? Even though, ironically, national polling was pretty accurate in guessing the Clinton/Trump margin. We’re in this weird, Orwellian world where people said the polls were wrong. And, actually, the national polls were right, except we don’t elect our presidents on a national popular vote, we elect them by the electoral college.”
But there are some changes afoot in the polling world, though the changes are due to new technology, not spurred by backlash. Some of these advancements might even be making polling more accurate, according to Courtney Kennedy, The Pew Research Center’s director of survey research.
“Cellphones were super effective in solving some of the really critical problems we were running into with landline [telephones] in the late 1990s and early 2000s,” Kennedy says. “We were really running into problems reaching young adults, reaching non-whites. Well, guess who we reach on cellphones?”
However, Kennedy also warns that not all the effects of technology are good. Because of widespread — though not necessarily accurate — online polling tools, anyone can be a pollster, whether they have expertise in the field or not. And Kennedy says because of that, more untrustworthy information can influence voters.
“It used to be that you had to have a lot of capital and a lot of knowledge to be a pollster. It was a small group of people who knew what they were doing, by and large,” Kennedy says. “[Now], the barriers to being a pollster have kind of disappeared. And that, for some people, is a good thing. But it brings a lot of concern, as well, because you have a lot of people putting out quote-unquote facts about these races, and it’s not really based on good methodology.”
