Why Experts Believe We Can Trust the Polls *This* Time

Joe Biden holds up a finger as he speaks in front of an American flag
Joe Biden speaks at the United Auto Workers headquarters in Warren, Michigan, on Thursday. Jim Watson/AFP via Getty Images

Subscribe to What Next on Apple Podcasts for the full episode.

We all know what happened last time we tried to predict who would be president. In October of 2016, 7 in 10 voters said they thought Hillary Clinton would be moving into the White House. Among Clinton supporters, 93 percent expected her to win. But there is a key difference in what the polls looked like four years ago, which was like the two candidates playing tag: Clinton would have a lead, then Donald Trump would close the gap, then she would lead again, and that lead would shrink sharply, et cetera, et cetera. But it’s not like that now. As Slate’s senior politics writer, Jim Newell, says: “Trump has never come anywhere close to catching Biden. His lead will vary, but he’s never really come within breathing distance of catching Biden so far.” Many of us still feel burned by the last election’s polls, but maybe—just maybe—the numbers are a little more reliable this time around. For Wednesday’s episode of What Next, I spoke with Newell about what the data shows and whether you can trust it. Our conversation has been edited and condensed for clarity.

Mary Harris: With so many national polls showing Biden in the lead, politicos know what everyone’s thinking: I’ve been to this rodeo before. So all summer long, pollsters have been laying out their case. Part of the reason these folks are so confident is that in the past few years they’ve changed the way they do their work. They’ve started making sure their samples include people who they assumed might not vote in the past: non-college-educated white people. Many of these people turned out not to be nonvoters, but Trump voters.

Jim Newell: In 2016, this whole split between college-educated whites and non-college-educated whites was a pretty new development, to have this massive a gap between these two demographics. Analysts hadn’t really thought to weigh all their polls by education. But most of the good pollsters are now weighting by education in an attempt to capture those who are less likely to respond. I think that’s still a little bit of a problem that pollsters are aware of, where the people most likely to respond are higher-educated, higher-income. I do not know why. Maybe wealthier people like to talk to pollsters on the phone more. But it’s a real thing. Another reason for the big polling miss in 2016 was that late breakers went for Trump so decisively.

People who were undecided.

Yeah. If you thought about people who made a decision in 2016, a lot of them just loathed Hillary Clinton, loathed Donald Trump, were putting off making a decision for as long as possible. And then in the end, they broke for Trump. It doesn’t seem like Biden is as loathed by the opposition as Clinton was, fairly or not. But also, in some of these polls of people who don’t like either Biden or Trump, Biden has been doing pretty well.

What do we know about how Biden is doing in this demographic that broke for Trump, these white, non-college-educated folks? Because the argument for Biden was always that he’ll connect with those people better.

Trump still has huge margins among white, non-college-educated voters. But Biden’s position is a little better relative to Clinton’s. Biden’s doing better with white voters across the board. That’s, to me, an interesting story of what’s going on and what kind of coalition Biden is putting together. He’s not just strictly recreating the Obama coalition, which had really strong margins and turnout from voters of color and younger voters. This is one where it’s … a little whiter than either of those. White Obama voters with college degrees were not a strong demographic, especially in 2012.

Another demographic that seems to be breaking for Biden is seniors.

Older voters were Trump’s best age demographic in 2016. He won them by about 10 points or so. This time, Biden’s been leading among them by 12 points in some polls, though not in all of them. If this holds, Biden would have a chance to be the first Democrat since Al Gore to win senior voters. That’s real trouble for Trump if that materializes. He won’t have anywhere else to hide, given his deficits among so many other groups. And it really could have a big effect when you look at pretty much all of the swing states: Arizona, Florida, Michigan, Ohio, Pennsylvania, Wisconsin. They all have higher-than-average proportions of older voters as part of their electorate. So it’s a real softness for Trump right now.

There are other things complicating the polls for Trump: This year there isn’t a strong third-party candidate who could, in a tight race, give him an advantage. And while undecideds broke for him last time, there just aren’t that many of them this year—they make up about 5 to 10 percent of the electorate, instead of the about 20 percent they did back in 2016. So the president is trying to make inroads with other, surprising demographics.

Relative to 2016, he does seem to be performing better among Black and Latino voters, specifically Black and Latino men. But it’s not like he’s winning. He’s going from, like, 5 percent to 8 or 9 percent among Black voters. And Latino voters, from 28 to 32 percent or something like that. I mean, he can clip a couple of the Democratic margins here and there, but his real base is still non-college-educated white voters. And the one thing he has going for him is that even with all the support he got from that demographic in 2016, they’re still a pretty untapped group: There are a ton of non-college-educated white voters in swing states who have not historically turned out to vote. If he can register a lot more of those voters than we’ve seen before, that seems to be his best opportunity. Not saying it’s a great one—they’re also low-propensity to vote in the past, so you can’t just double that overnight.

Part of what makes the polling more interesting to watch now is that after Labor Day is when state data begins to improve. This is important in a country where some states matter more than others. And in Florida, a poll this week seemed to show the race tightening, with Trump and Biden tied. But Arizona looks like it’s headed in a different direction.

Four years ago, Arizona was considered kind of a reach for the Clinton campaign. She could go for it so long as she had the rest locked down. Instead, she visited without having the rest locked down. That was a bit of a problem. But in polling averages in Arizona, Biden’s been up in just about every poll. It’s a combination of the electorate there becoming more Latino in its composition as well as some suburban decay for Trump, which you’re seeing everywhere. I think it’s going to be pretty close, but you would have to call Biden the favorite there just based off the polling we’re seeing.

Well, Trump pulled a bunch of ads in Arizona recently, and it was seen as a sign that maybe the campaign was giving up on the state to some extent. But then that means they’re relying on Midwestern states, which are pretty newly red if you’re talking about Michigan and Wisconsin. So that seems like a bit of a risky strategy.

I don’t think he’s giving up in Arizona. They were pulling ads from a lot of places around the conventions—this is partially related to how they’ve blown $800 million already and are somehow having to tighten their belts a little bit. I think it’ll be back up on the air in Arizona because if they lose there, it really puts all the stress on holding Michigan, Pennsylvania, and Wisconsin when they’re not in a great position in any of those states right now. And they’ll still have to protect Florida because if they lose Florida and Arizona, the election’s over.

I think it’s worth talking about those states specifically—Wisconsin, Michigan, Pennsylvania—because there were such thin margins there in 2016. How much faith should we put in these polls even when they’re breaking it down locally, which is so important in our particular system?

I think one thing that’s a little different is some of those states did not really have strong mail-in voting apparatuses in 2016. It was mostly in-person right around Election Day. But in Michigan and Pennsylvania, they’ve now passed legislative changes to allow for more mail-in voting, and this was done even before the pandemic.

What does that functionally mean for the election? Does it mean that processing the votes will take more time, or something else?

Well, it will take more time because in Pennsylvania they can’t start counting any ballots at any point until polls close. So that means if we have an election coming down to Pennsylvania—say Democrats have lost Florida and Arizona and they’ve won Wisconsin and Michigan—that could be a nightmare. It really could be just having to wait for the count. And then there could be lawsuits over provisional ballots. That sounds terrifying, but it is a little terrifying.

Just because they’re new systems, because this is the opportunity to kick the tires?

They’re new systems, and states just have not had to handle the coronavirus and this much mail before. And now, mail-in ballots themselves are highly politicized— you have all state officials from each party trying to make it easier to vote by mail or harder to vote by mail through all these little mechanics. I just really hope there aren’t too many ballots thrown away because they were not properly done or they were spoiled or something. I hope that how Americans voted in the election is captured in the election results. That’s the stuff that makes me worried.

In our particular system, individual states are really important because of the Electoral College. And where you’re voting really matters because you’re voting for representation to do the actual vote. I was struck by the fact that there was reporting this weekend that said allies of Trump think there’s virtually no chance that he’ll win the popular vote, like in 2016. That’s a striking admission, and it makes me wonder how much we can say in advance about who is ahead and who isn’t because so much comes down to this ancient system of democracy.

A couple of polling analysts were showing Biden’s percentage of winning the Electoral College. Depending on his popular vote margin, if, say, Biden won the popular vote by 3 points, Trump’s still got a really good chance of winning the Electoral College that way. I think up to about a 5-point Biden win, Trump could actually win the Electoral College. There are a lot of ways to describe the system where you can win a popular vote for a presidency 5 points and still lose the presidency. I’m not quite sure what the exact word for it is, but I know it’s something I would think a bit about.

I keep thinking about the polling from 2016, and it makes me wonder whom we’re leaving out this time.

The boat vote. The boaters.

In all seriousness, I think that’s a really good point. And naturally, I can’t say who it is because, you know, they’re missing. But that feeling of despair—that people think, despite the polling being very good for Biden, Trump is just going to win—I think is really interesting. If people are so despairing that they’re not even registering to vote, that’s a big problem for Biden. But I also think it can be kind of useful. I remember, in 2016, running around the sidewalk and screaming at strangers that Trump could win and you’re not paying attention enough, and people just looked at me like, Of course Trump’s not going to win. This time I feel like it’s the opposite, and I think that could make sure that people actually do get out to vote. If you think Trump’s going to find some magic trick to try to steal this, then make sure you’re not complacent.

Subscribe to What Next on Apple Podcasts

Get more news from Mary Harris every weekday.