Were The Polls Wrong? A Look At The Future Of The Polling Industry
MARY LOUISE KELLY, HOST:
Speaking of polls, longtime Republican pollster and strategist Frank Luntz made a prediction on CNBC yesterday.
(SOUNDBITE OF ARCHIVED RECORDING)
FRANK LUNTZ: At this point, if Trump does pull it out - and there are people who think that he does have a chance - it would be the complete and utter destruction of the polling industry, at least in politics.
KELLY: Well, there still is not a clear winner, though former Vice President Joe Biden has continued to pick up electoral votes throughout today. It's safe to say, though, the polls were off. And we're going to talk about what went wrong with Nathan Gonzales. He's editor and publisher of Inside Elections.
Hey there. Welcome.
NATHAN GONZALES: Thank you for having me.
KELLY: So you have been writing about this today, and it looks like your big takeaway is that once again, polls underestimated the support for President Trump. How off do you think they were?
GONZALES: Well, I mean, it depends on the state and the district, but, you know, it looks like at least anywhere from five to seven, eight points or more. I do think that there might - with Frank's comment about the destruction of the polling industry - might be a little bit of hyperbole because I think there's a distinction between polls and elections where President Trump or candidate Trump is on the ballot and other races. You know, for - I've been doing this for almost 20 years, and looking at polling data usually leads you pretty close, if not right at right at, the results. And for example, in 2018 in the midterm elections, when President Trump wasn't on the ballot, using our same methods of analysis, we were pretty darn close in our House and Senate projections. But...
KELLY: Although let me jump in and take you two years before that - 2016, when a lot of polls led us to believe it was going to be a sweep for Hillary Clinton. Is this the same problems at work this year, or what - how did this happen again?
GONZALES: I mean, I think it is the same problem in that there's a continuing to underestimate President Trump's support in elections when he's on the ballot. And that - but what's, you know, interesting about where we are today is that the majority of the data showed that Joe Biden was more likely than not to win the presidential race. Here today, with votes still being counted, he's still more likely than not to win the presidential race. But President Trump's - the underestimation of his support had a tremendous impact on the House and Senate races, as Sue laid out. And that's where, because Trump was able to meet or exceed his 2016 performance, that allowed Republicans to have a very good night.
KELLY: So talk to me about the mechanics of what's going on here. Is this outdated methodology? Is that playing a role? You know, there's been a lot of focus on - not that many people have landlines anymore, and the polling industry needs to way update its methods. Is that what's going on, or is there something separate?
GONZALES: I think there's a combination of factors. It's not just landline and cellphone. It might be, is the telephone, any phone - is that the best way to reach the most broad - the - reach all the people that you need to reach? Do you need to be reaching people by text and online? Are you getting the right sample, the right makeup of men and women, people with a college degree, without a college degree? Are you getting not just the right number of Hispanic voters, but are those Hispanic voters - are they similar to the rest of the population in that state or district? I think it's clear that President Trump is overperforming from four years ago with Hispanic voters, particularly Hispanic men, and that is accounting for some of the underestimation of his support.
So pollsters had - they were having challenges before 2016. 2016 pushed it along. This is only going to really cause a lot of soul-searching in the polling industry, particularly on the Democratic side.
KELLY: Well, stay with that soul-searching. I mean, what is the answer here as pollsters are trying to figure out, how do we do this better going forward?
GONZALES: Well, I think the first is once we have results, we can really match up how far-off the polls were, and then there will - do a deep dive into what happened. And I think what's maybe missed about polling is that particularly partisan pollsters are often discounted because there's an assumption that their data is wrong and tainted. But I actually think there's a very different view that partisan pollsters use - they have to have good polling data in order to make millions of dollars' worth of strategic decisions. So they have a vested interest in solving this puzzle. But at least temporarily, Democrats definitely took a step back in evaluating what electorate was going to show up in these elections.
GONZALES: And they're going to have to look at it going forward.
KELLY: All right. Thank you, Nathan Gonzales.
GONZALES: Thank you.
KELLY: He is editor and publisher of Inside Elections.
(SOUNDBITE OF MUSIC)
NPR transcripts are created on a rush deadline by an NPR contractor. This text may not be in its final form and may be updated or revised in the future. Accuracy and availability may vary. The authoritative record of NPR’s programming is the audio record.