Arizona Public Radio | Your Source for NPR News
Play Live Radio
Next Up:
0:00 0:00
Available On Air Stations

Evaluating The State Of Polling After The 2016 Presidential Election


The polls leading up to the 2016 presidential election were wrong, wrong, wrong. Pollsters were wrong. Reporters who cited those polls were wrong on a scale that makes history.

But how can that be in a time when there's so many avenues into the minds and hearts of Americans? Aren't polls these days supposed to be scientific? Don't pollsters earn lots of money on the assurance that they're accurate?

Sean Trende is senior elections analyst for RealClearPolitics. He joins us now from Columbus, Ohio. Thanks so much for being with us.

SEAN TRENDE: Thanks for having me.

SIMON: Why did so many polls go wrong?

TRENDE: Well - so I'm going to disagree with you a little bit that the polls were historically wrong. As a matter of fact, at RealClearPolitics, we had all the states as toss-ups that ended up being toss-ups. And we meant it that they could go either way.

If you look at the national polls, our average said that Hillary Clinton would win by three. It looks like she's going to win the popular vote by a point, which means we had her a point too high. And Donald Trump was a point too low.

States like Pennsylvania - we had a two-point margin for Clinton. And Trump ended up winning by a point, which is, you know, within the normal sampling error of 2016. What people were wrong about was being way too confident...

SIMON: Well, let me just point out...


SIMON: ...I mean, missing the winner is a big - I mean, you know, it's like if the Indians said after the World Series, but we scored more runs. The object is to win.

TRENDE: Well - and there's nothing you can do in a close election. You're going to miss some of them. Polls come with error margins. If a race is really close, like the one-point race in Pennsylvania, you're just going to get polls that show Hillary Clinton ahead. That's just how the probability works.

SIMON: But, of course, we had all these - there were all these polls that said she - there was a 90-percent chance or 80-percent chance that she would win.

TRENDE: Now, that's right. And that's - that goes to the pundits, OK? And the pundits were terrible. I will 100 percent give you that because they were overconfident in what the polls were saying. Nate Silver's FiveThirtyEight gave Donald Trump a 1 in 3 chance of winning. And I think some people forgot that things with 1 in 3 chances of happening happen all the time.

SIMON: Yeah. So you don't think there's much fixing to be done. Or do you?

TRENDE: Well, I think there are some things that pollsters can fix. I think one thing they got wrong - the errors - the poll errors were correlated with the white, working-class share of the electorate. I think they didn't get a good handle on the white, working-class vote. So I think they're either getting too few white, working-class voters, or they aren't getting a wide enough sample of them.

SIMON: Why is it in this day and age, when there are so many ways - I mean, you tweet something about bonsai. And the next time you log in to Amazon Prime, you're pelted with bonsai offers. There's so many different ways, some of them a little spooky, that algorithms and corporations and the Russian government (laughter), for that matter...

TRENDE: (Laughter).

SIMON: ...Have about knowing about our deepest, innermost thoughts. Why can't pollsters and pundits approach it?

TRENDE: Well, again, I think the pollsters did OK. But for the pundits, you know, part of the problem was that you would be hard-pressed to find a pundit on the left that wasn't terrified of Donald Trump. So you've got this groupthink going where everyone was looking at the polls and saying Donald Trump was going to lose because that's what most of them wanted.

SIMON: Do you have any concern, Mr. Trende, about the misuse of polls when people, myself included, sometimes don't understand the particular distinctions that ought to be made?

TRENDE: If I had one wish, it would be world peace or feed everyone. But if I had, like, 50 wishes, one of them would be for people to understand probabilities a little bit better. The polls are a good tool. They're never off - they're rarely off, I should say, by more than a couple points.

And so if you have a bunch of polls showing Hillary Clinton up 10 points, she's ahead at that point. It's just once you get down to a two or three-point race, these errors pop up all the time.

SIMON: Sean Trende of RealClearPolitics, thanks so much for being with us.

TRENDE: Hey, thank you for having me. Transcript provided by NPR, Copyright NPR.