Donald Trump’s presidential victory may have come as a surprise to many, but some pollsters have long been warning about the need to change their methods.
Florida State University Political Scientist Lance DeHaven Smith noted during the summer that traditional polling methods may no longer apply after a slew of survey’s completely misread the Republican primary field.
“Trump does better in internet surveys than standard telephone surveys and the reason for that is apparently people are reluctant to say they support Trump even though they do," he said during WFSU's It's About Florida show.
"When they’re talking to a person they are embarrassed, but when they’re talking to a computer they are not embarrassed.”
He’s alluding to what Mr. Trump and his call the silent majority: people supporting Trump at the polls but were missing from surveys. But DeHaven-Smith even internet polls can be tricky.
“People will register for political parties or independents when they actually have a different orientation. But you don’t know that when you do the survey. A lot of things can get in a way. It’s become a lot harder to get a random sample from the population. Do you sample likely voters? Do you sample registered voters? A separate sample for each party? There’s a lot of choices to make and each choice can create a bias," he said.
It's called the Bradley Effect, a disputed theory that proposes some people will opt to say something else, when asked about their preferences due to fears of being labeled. On Election Day, WFSU Reporter Jim Ash ran into a couple leaving the polling booth.
“Would you mind sharing who you’re voting for today or would you rather not? " Ash asked Tallahassee resident Sylvia Rousseau.
“I’d rather not. I’d lose too many friends," she said. But her partner Monticello resident Earl Irvin was clear about his choice.
“Well, I voted for Trump," he said, "and I am not ashamed of it... I think he’s a little tired of the way things are going in this country. And he’s made a lot of money. And we need a money person up there instead of a politician. That’s the way I look at it.”
Irvin is a lifelong Democrat, but he’s voted Republican in the past three presidential election cycles.
“I’m a Georgia boy," he said. "So I am a little prejudiced.”
Irvin says he was never polled, but has been voting since he was 21. One survey that did get it right came out of the University of Southern California and the Los Angeles Times. The poll had been widely maligned throughout the election cycle—it has largely shown Trump leading in the race for a while. So how did the LA Times get it right, when almost everyone else was wrong? Instead of asking who respondents will vote for, the poll asks them to estimate the likelihood they’ll vote for a candidate. They also ask respondents about the likelihood they’ll cast a ballot. This, say pollsters, better captures nuances in voter attitudes, and can also help capture people who may not be included in more traditional sampling techniques.
