How to Make Sense of the Polls

People's shadows reflect onto a large U.S. flag.
“When you trust the polls, you need to trust them for what they are,” Sean Trende says, “which is not oracular, perfect insights of wisdom but, rather, a good metric of where things stand right now.”Photograph by Matt Mills McKnight / Getty

With less than a month to go until Election Day, former Vice-President Joe Biden leads President Donald Trump by an?average?of more than nine percentage points in national polls. Biden has been ahead for the entire year, and he also received something of a polling bounce after Trump’s chaotic performance in the first Presidential debate.

Still, a?plurality?of Americans believe Trump will win the election, and the President’s approval rating has remained in the low-to-mid-forties all year, despite the fact that a clear majority of Americans disapprove of his handling of the coronavirus pandemic. There also remains considerable skepticism about the accuracy of polls, which missed Trump’s strength in the key Midwestern states that delivered him the White House in 2016. Moreover, Trump is likely to perform better in the Midwest and Florida than he will nationally, which means that he could still win the election despite losing the popular vote by several percentage points. (Currently, Biden leads narrowly in Florida, and by slightly larger margins in Pennsylvania, Wisconsin, and Michigan.) The numbers place Biden in the lead, but much uncertainty remains, with record numbers of voters submitting ballots by mail, a coronavirus outbreak in the White House, and the continued effects of the pandemic across the country.

I recently spoke by phone with Sean Trende, the senior elections analyst at RealClearPolitics, and the co-author of the Almanac of American Politics 2014. During our conversation, which has been edited for length and clarity, we discussed whether pollsters have addressed the mistakes of 2016, whether the Biden campaign should go all out to try to win Texas, and how Trump might perform differently now that he is running as an incumbent.

Why should we trust the polls?

Generally speaking, the polls have been pretty good over the years, and given enough time you'll do better with polls than just your gut instinct. But polls are imperfect instruments. When you trust the polls, you need to trust them for what they are, which is not oracular, perfect insights of wisdom but, rather, a good metric of where things stand right now.

How confident are you that pollsters have fixed the problems we saw in 2016, especially in the Midwest?

I’m still concerned. If you look at 2018, the polling in the Midwest wasn't that great. We were supposed to have a Democratic governor in Ohio. There was supposed to be a Democratic governor in Iowa. There was supposed to be a Democratic senator from Indiana. The Democratic senators in Ohio and Michigan both underperformed the polls.

Florida was off a little, too, right?

Florida was very off. I think there were, like, one or two polls that showed Ron DeSantis winning the entire way. And people forget that in 2012 there was a big polling miss, but it just happened to be in the Democrats’ favor, so it didn’t affect the outcome. It’s good that people are starting to weight [results] by college education, because that’s a salient factor, but in terms of treating [polls] as things that can’t be off by two, three, four points? No, I wouldn’t do that.

Is there anything in the polls that you are concerned about this time? Anything along the lines of not weighting by college education?

There’s a possibility that the problem wasn’t, in fact, that pollsters weren’t weighting by college education. That was the big, easy fix. But maybe it was something else. We don’t know what it is. The 2018 results give me some pause. I look at the polls showing Biden with twenty-per-cent-plus leads among older voters, with boomers voting roughly the same as millennials, and Gen Z, and that makes me nervous. I can tell myself a story about why that would happen, but I can also tell myself a story about conservative Trump voters hearing, “Hi, I’m from NBC/Wall Street Journal polling,” and hanging up the phone because they don’t trust the mainstream media. That makes me very nervous.

What you’re talking about is not exactly the same, but it seems connected to this idea of the “shy Trump voter,” which is this theory that there is social stigma attached to voting for Trump, so voters would not tell pollsters that they were going to do so. But we don’t really have any evidence that shy Trump voters actually exist, correct?

It’s a story you can tell yourself, but I don’t think the evidence is that good for it.

Why do you think the evidence isn’t very good for it?

In Internet polling, you just click on your keyboard, and no one really knows how you answered. When you get a robo poll, it’s just that automated voice, and you’re pressing buttons. But when it’s a live person on the other end, there’s maybe a one- or two-point dropoff in Trump’s performance from the live polls, and you might ascribe that to shy Trumpers.

There’s also this question of Trump’s job approval, which is running two or three points ahead of his vote share. And, going back in time, Presidential job approval is one of the strongest indicators we have of how the President is going to fare. Who are these people that approve of the job he’s doing but aren’t going to vote for him? Again, you can tell yourself a story. There are people who like his policies, but hate his persona. But it could also be that these are voters who are saying they’re undecided, but they really aren’t. But, again, this is just kind of conjecture, it’s not really good evidence.

So you’re saying we have reasons to think the polls might be off for various reasons, but there’s not great evidence for any of them?

Yeah, there aren’t, like, flashing red lights, but there are things where, if Trump does end up winning, or coming very close, we’re going to look at these things and say, “Ah, there it was all along.”

It seems like the best case for Biden right now is not just that he’s ahead and that there are fewer undecided voters than in 2016 but that he’s been ahead so consistently. It’s not clear that there’s been any point in the race, going back to the beginning of the year, where Trump actually was leading him in enough states to get to two hundred and sixty-nine electoral votes. Is that your sense, too?

I think that is probably the single best argument why we wouldn’t see something like 2016. If you go back to 2016, that race wanted to be a two-point race. It would be close to two points, and then Trump would do something stupid like talk about the judge who couldn’t rule for him because he was “Mexican.” And Clinton would blow up a lead, and then it would tighten back to two points, and then he would insult Khizr Khan and his family, and the lead would blow open. This is different. This is a race where it’s actually been kind of boring. Polls usually have Biden up six, seven points.

It seems like that’s the number that the polls want to be at. You have events like the murder of George Floyd, and Trump’s response to that, and then you had the debate, and Trump’s positive test last week, which pushed them a little higher. But, basically, since the pandemic started, it hovers around seven.

Yeah. I think that’s right.

So, then, what would need to change for Trump to win? What, in addition to the polls being wrong, are we going to look back at in a month and say, “Boy, we were sure wrong when we talked in October”?

The story I have mentally pre-written is that job-approval thing. He had gotten up to like forty-six per cent [among likely voters], which means that a lot of these undecideds are probably going to break for him, because they ultimately approve of the job he’s doing.

If what we’re seeing right now is just differential response, which is a well-established phenomenon. When there’s a big blowup in the news cycles, one party won’t answer polls. That could be what we’re seeing right now. And so if, in a couple of weeks, this has gone back to him being at forty-five or forty-six per cent, and he’s leading Biden on the economy by eight, nine points, that’s the story for why he wins if he does.

Advertisement

I was going to say a version of that, which is if Biden has a bad debate, or seems a little bit out of it, and the Electoral College/popular-vote gap is bigger than we think it’s going to be, there’s a little bit of mail-voting trouble for Democrats, and then Trump somehow ekes out a Midwestern state and Florida.

I think you’re focussing more on the type of effect we’d see, and I'm looking more at why I think it would come across. I think it’s the same story. I think the mail thing is an underrated potential headache in this election. You hear the stories, like out of Pennsylvania, where they had the issue with the naked ballots, right? And the Supreme Court of Pennsylvania shocked me and ruled that those don’t count. There are just a lot of things you can do wrong on your absentee ballot—you move and you forget to put the updated address on it, you get your birthday wrong, or you just have a brain fart. And so with Democrats really pushing this mail-voting option, it just increases the possibility that in a close enough race, it could be meaningful.

If Joe Biden were winning this election by breaking through in Florida, Texas, Arizona, and maybe North Carolina or Georgia, and he was going to lose the Midwest again, the way Hillary Clinton did, that would suggest to me that her loss was more inevitable than perhaps we thought at the time. But if Joe Biden is going to be very strong in the Midwest, that suggests to me that she really ran an atrocious campaign, and that Trump’s win had more to do with her weakness as a candidate than it did working-class white voters trending away from Democrats. Do you agree?

I think that’s a useful framework. I think you have to keep that framework flexible, because we could end up with something somewhere in the middle, but there’s no doubt that the upper Midwest is trending toward Republicans. Take a year like 2018—twenty years ago, Republicans wouldn’t have kept the governorships in Iowa and Ohio, and Sherrod Brown wouldn’t have had a seven-point race. And there’s no doubt that the Sun Belt is trending toward the Democrats. And so you can tell a decent story about how Hillary Clinton just got stuck in the middle of this transition.

Can you talk about Florida, specifically? Why is Florida trending Republican, or at least staying out of reach for Democrats, even if Biden ends up winning?

You know, I don’t like to do this in interviews, but I have no idea. If you get a good answer from someone, let me know, because I don’t know why. I mean, you can tell us the obvious race story about Ron DeSantis in 2018, but why do Ron DeSantis and Rick Scott win?

You’re implying DeSantis won, in part, because people didn’t want to vote for Andrew Gillum, who’s Black?

Yeah. I would be shocked if that wasn’t at least part of it. But I don’t know. Someone made the point that you can go back to 1992 and add up all the votes for Republican and Democratic Presidential candidates in the state, it’s almost an exact tie. It is a very rigid state, which makes no sense to me, given its shifting demographics and the complexity of the demographics there. I don’t know if it’s just a bunch of things that cancel each other out in the big picture, but it’s weird. That’s an unsatisfying answer.

I know the Hispanic vote in Florida is different than the Hispanic vote elsewhere in the country, but it seems like Trump has held up relatively well with Hispanics, and may, in fact, do better than last time. How do you understand that? Is it the educational divide in our politics manifesting itself across racial lines?

Trump is doing reasonably well with Hispanics, and the polling suggests he’s doing better with Cubans than he did last time around. But, like I said, there’s just a lot of moving parts in Florida, because Miami has substantial suburbs, and you’ve got these areas around Tampa and around Orlando that are shifting blue, you have got a growing Puerto Rican population, which is more Democratic than the Cuban population, for sure. And there has probably been some offset with rural whites continuing to trend Republican, and there is a substantial rural white population in Florida, but at a certain point Republicans start to max out on that. I’ve been surprised by how stable it’s been.

And what about Trump and Hispanic voters generally?

I’m not as surprised about that. I think people have gotten the Hispanic vote wrong, first off. As we’ve suggested in this call, there is no such thing as a Hispanic vote. You really have Cubans and Puerto Ricans, and the Puerto Ricans are U.S. citizens. The citizenship battle doesn’t play the same way it does with Mexican-Americans. Mexican-Americans in Texas have historically been a lot more conservative than Mexican-Americans in California, and that has something to do with the migration patterns. It is a very diverse community. And, going back, there has always been a substantial minority of Mexican-American voters who have supported hard-line immigration measures. You usually get somewhere between a quarter and a third of Hispanics supporting these things. It’s not a monolith. And so the fact that Trump didn’t fall to ten or twenty per cent, given his hard-line stances on immigration, I don’t think should have been that big of a surprise.

Do you think the Biden campaign should go all in on Texas?

Yes, absolutely. It might be a cycle or two away, but something like eighty per cent of its vote is cast in the big cities. There’s almost no rural or small-town vote cast in Texas anymore. And so when you’re in a universe where the suburbs are trending so hard against Republicans, it makes a sea change in Texas politics. And that’s what you've seen there. I don’t think Ted Cruz winning by three points was just because people don’t like Ted Cruz. I think there are fundamental shifts going on in that state that are reflective of national politics.

Other experts I have talked to look at these polls that seem more optimistic about Texas than other states polling similarly, like Ohio, if the race is tied going into Election Day.

Yeah. Ohio is a state where a large portion of the vote is cast in rural and small-town areas. It’s a big part of why the upper Midwest is trending Republican. There aren’t enough big cities to offset that. But Texas is all big cities. It is all Dallas, Fort Worth, Houston, San Antonio, increasingly Austin. So as big cities and suburban areas go blue, Texas could get very blue, very fast.

If this election plays out the way many people expect it to, with Biden up by seven nationally and winning two hundred and ninety electoral votes or something, what will that tell us about 2016, if anything? Was Trump always overrated at politics because he was running against Hillary Clinton and because the Electoral College was favorable to him?

Advertisement

I think it is something different. I think Trump actually has some good, shrewd, very base political instincts, combined with an absolute lack of shame. Like, an absolute lack of caring about human beings, or telling the truth, or anything like that. When he was up against Hillary Clinton, who tried very hard to play by the rules, and make that part of her shtick, it was just a horrible matchup. You couldn’t really paint him as the religious conservative.

I accept your statement there, Sean.

Exactly. But you think about the lines of attack, Democrats did well in the past. Trump wasn’t a religious conservative, and he wasn’t a Koch-brother conservative. It made him a very difficult target for Hillary Clinton. But I think elections with incumbents are fundamentally referendums. So, while I do think Trump has some decent political instincts, he’s just incompetent as a President. I mean, if you want to be kind to him and draw big-picture things, he’s a walking embodiment of why the Presidency is not an entry-level job.

I think that, for various reasons, both hard-core Trump supporters and terrified “resistance” Democrats have embraced the idea that he’s a bit of a political magician. But the longer this goes on, it seems clear that he can’t stay out of his own way.

I think one of the big failings of Trump’s Presidency—and there are many—is that he never made the transition from an insurgent candidate to a President. I think he had a very effective insurgent campaign, and it was hard for Hillary Clinton. But, once he became President, he had no reason to listen to people who actually know stuff about politics, who would tell him, “Hey, you’re the President now. You need to put down that Twitter thing.” That’s great for your initial election campaign, but people don’t want their Presidents yelling at the Prime Minister of Denmark because she won’t sell him Greenland. COVID was absolutely a layup for him.

For people who doubt that, look at Andrew Cuomo’s approval ratings after mishandling COVID-19 in many different ways. He seems like he cares. He takes it seriously. And his approval ratings went up. It’s not that hard.

We can debate policy choices and what should be done because people get at the fundamental level that this is “act of God” territory, using that term broadly. People want to hear Presidents give moving speeches, even if what they do isn’t that effective. And he just couldn’t do it.

I think most Americans just want him to just treat this like it’s serious. And because of his own personal issues, his psychology, he just can’t do it. I don’t think there’s a much more complicated explanation than that.

Yeah. Joe Biden had a tweet about how, when he becomes President, there’s going to be a mask mandate and we’re going to accelerate vaccines. And I went through all those things like, O.K., you’re not going to be able to enforce the mask mandate federally. And we’re already doing the moonshot on the vaccine. But then I realized, you’re missing the point, Sean. That’s what people want. To just say you’re going to do something to make it better and look like you care. And Trump’s just incapable of that.

Do you have advice for people about how they should read the polls the next few weeks?

I would pay attention to his job approval. See if it starts to get back up into that forty-five-, forty-six-per-cent range. And just remember, even at an aggregated level, polls are off, and when FiveThirtyEight tells you there’s a twenty-per-cent chance of something happening, you don’t get to round that down to zero.


Read More About the 2020 Election