Kristen Soltis Anderson is a Republican pollster and author of The Selfie Vote: Where Millennials are Leading America (and How Republicans Can Keep Up). She has written for Politico and The Huffington Post, as well as appeared on CNN’s State of the Union and The O’Reilly Factor. The interview was conducted on February 22nd, 2017.

HPR: What mistakes do you think pollsters made in this past election?

Kristen Soltis Anderson: Oh, this is a great question. I think, it’s the same way that people criticize the media, and there’s a lot that goes into the bucket of the media. You know, Sean Hannity is in the media and so is Marty Baron at the Washington Post. But there’s not a lot similar between them. The same thing with pollsters. There are pollsters who are public pollsters. Their job is to produce a lot of data that will get news coverage. It’s media organizations. I think a challenge that media organizations have is that they’re so wrapped up in having to cover the horse race: cover the ‘who’s up, who’s down.’ ‘Clinton was up six last week, now she’s only up by four. Why is she slipping?’ And really, it’s the coverage of the polls that I think really misstates what polls are supposed to be doing.

This is not to say ‘it’s not the pollsters’ fault, it’s the media’s fault.’ But to say, if you’re a pollster you sort of know if one day somebody’s up six and the next day they’re only up four, that’s not anything to really freak out about. That’s margin of error. That could just be noise, it might not mean anything. But the media treats that as, ‘Clinton has slipped two points. More at eleven.’ It becomes a big headline and a big thing. I think part of what led everyone astray was that a Clinton national popular vote lead in the polls of three points should not have been suggesting she has a 99 percent chance of winning, right? Models that say that she has a 99 percent chance of winning just were insanely over-confident about the idea that there wouldn’t be some kind of polling error. That was piece number one.

I think the things that pollsters need to do are, one, emphasize things beyond just that ballot test question. I did an interview here with Tony Fabrizio, who is Trump’s pollster. He told me that one of the reasons why they had an inkling that Trump could win—they didn’t think he was a favorite, but that they thought he had a better chance than people were giving him credit for—was that they saw the undecideds in the poll. On other questions in the survey, those undecideds were expressing views that sounded very Trumpian: ‘I want wrecking ball style change.’ So you could infer, if these folks actually show up to the polls, this will not be undecideds breaking fifty-fifty like they did Obama-Romney. This will be undecideds breaking eighty-twenty, and that will put Trump over the edge. And I think one of the things that media pollsters get wrong is we get so tied up in that horse race number and having to make that the centerpiece of everything. There’s never really an interesting ‘let’s dig into those undecideds, let’s figure out what they’re all about.’ I think that’s a part of how this all got missed.

HPR: How accurate do you think Trump’s approval ratings are right now?

KSA: Reasonably so. In all of survey research, there’s only really one time—and think, the market research industry is huge, we use market research to determine who’s going to buy what products, what ads should go on TV where, etcetera, etcetera—the only time a pollster is ever proven right or wrong is election polling on the ballot test. There is no way to ever prove if your market research firm says 60 percent of America likes Coca-Cola. There’s no way to prove if that’s true or false. There’s not going to be any embarrassing moment for that pollster, where ‘ooh, it turns out only 57 percent of Americans like Coca-Cola.’

Campaign polling is actually insanely high-stakes. ‘Did Trump get 47, or 44 percent of the vote:’ there were huge consequences to getting that right or wrong. Approval rating is a little less high stakes. Is the political environment dramatically different if Trump’s approval is actually 45 percent instead of 42 percent? Not really. That’s kind of the same sort of political environment. When we see these numbers about Trump, they might not be perfect; no polls are perfect. But the consequences of being off by two or three percent is much less when you’re talking about a metric that is one of the atmospherics, rather than what’s the ballot test number.

The only other thing that I’ll say about the ballot test question and Trump is there is a chance—and this is something that really concerns me in the polling industry—that there is now this kind of broken trust. It used to be the case that lots of people didn’t take polls. But they didn’t take polls because they were too busy, they couldn’t be bothered, they didn’t want to pick up a number they didn’t know. And luckily for pollsters, that was relatively evenly distributed across the population. People who took surveys were not actually that different from people who didn’t take surveys. The biggest difference was that survey takers were slightly more civically engaged. So it’s a bias that actually helps you if you’re a political pollster. The problem now is Donald Trump could tweet tomorrow, that ‘@KSoltisAnderson is an idiot pollster who knows nothing.’ There’s a chance that relatives of mine would go ‘yep, idiot pollster.’ The polling industry now has become a part of partisan warfare.

Now, there is a big problem if you have a group of people who are ideologically or in a partisan way, or in a ‘how-you-feel-about-the-president’ sort of way, less inclined to take polls. Does that actually exacerbate the problem? They were already maybe a little less inclined to confess that they were Trump supporters, or to take polls before the election. Now, has that been exacerbated to where we are undercounting Trump’s support by five or more percent? I don’t know if that is the case. The problem is that it’s sort of unknowable.

I don’t think that the magnitude of the polling miss in the 2016 election suggests that there were huge, huge, huge numbers of people that were getting missed. Again, these were in many cases Pennsylvania, Michigan, and Wisconsin. The polls were off by three, four, five percent. In some cases, that’s still margin of error. It’s not a great excuse, but I think the idea that all of the polls were horribly wrong is not the case. But I do think that the coverage now after the fact has really made pollsters a hot topic in a way that it makes me worried if Trump supporters think they never want to talk to pollsters. That makes my job very hard.

HPR: In your 2015 book, The Selfie Vote, you wrote that “It’s no secret that the GOP has had a hard time winning over the millennial generation.” How do you interpret Republican gains in this past election cycle?

KSA: They were done without young voters. You did see an uptick in turnout among young Republicans if you look at the voter file. If you look at the exit polls, Donald Trump only wins about 36 percent of young voters, which is about the same percent that Mitt Romney won. On the one hand, that should be great news for Republicans: ‘hey, Donald Trump was supposed to be this whole huge disaster, and he held serve; he did just as well as Mitt Romney did with young voters.’ On the other hand, he was not running against Barack Obama. He was running against a much weaker opponent, one who saw a lot of support go to third party candidates. I don’t think that the problem was solved for Republicans at all.

I think the challenge is that this has always been a slower burning problem, and that eventually Republicans’ luck was going to run out. But everyone’s feeling very satisfied with themselves now that they were able to win the presidency while only winning just barely over a third of young voters.

Here’s the bigger problem for Republicans: take a look at how people in their thirties voted. People in their thirties were not very positive about Trump at all. Voters in their thirties in the past have leaned a little more toward the right, or they’ve been split pretty evenly. Here, that wasn’t really the case. I forget what the exact numbers were, but part of my thesis has been that as voters who were eighteen, nineteen, twenty during the Obama years get older, they will not just magically become Republicans. I think this election’s results still bear that out. Trump ran up the numbers with white voters, men, rural voters, but the ‘selfie vote’ is demographically not those places. Republicans have still not solved their problem. What concerns me even more is that they may not think they have to worry about solving that problem anymore.

HPR: What specific policies do you think the Republican Party should focus on to ensure that they keep winning in the future?

KSA: Economic growth that everybody gets to partake in is number one. If people feel like their bank accounts are a little bit healthier and they feel like they’ve got a little more money in their pocket, and maybe they can go on a vacation for an extra day or two this year. That’s the kind of stuff, more than anything else, that will affect things. If Trump can deliver on that—he may say offensive things, he may turn people off—but if people are feeling like they’re actually better off, they’ll give him a lot more runway. If that doesn’t happen, I think his leash is pretty short.

Image Source: Flickr/AdvertisingWeek