Skip to content
The Sweep: Two Weeks' Notice
Go to my account

The Sweep: Two Weeks’ Notice

You've got polling questions; Sarah's got polling answers.

Let’s be honest: There was not a lot of “must read” campaign news this week. A record amount of money and early voting continue to make headlines, but I didn’t think any of the specifics were worth our time to highlight. 

And the dueling town halls epitomized their respective stars—empathetic and evasive vs. confrontational and meandering. No surprise there. Perhaps the most interesting thing to come out of the town halls was the ratings. Nielsen reported that “Biden’s town hall on ABC averaged 14.1 million viewers, compared to about 13.5 million for Trump’s NBC town hall,” even though Trump’s event also appeared on MSNBC and CNBC. (And it looks like ABC was better able to monetize their event.)

Does this mean that enthusiasm for Biden’s campaign has overtaken Trump’s ride-or-die MAGA base? Maybe. Another possible explanation, though, is that NBC intentionally scuttled their own event after receiving mounds of internal and external criticism for counterprogramming the already-announced Biden event. In a less-than-subtle choice, MSNBC showed a scheduling graphic that night that didn’t even include the 8pm hour. 

Either way, it almost certainly makes this week’s debate Trump’s last opportunity to get in front a substantial number of registered voters who aren’t already voting for him. 

So in lieu of our usual Campaign Quick(ish) Hits, let’s practice our rank punditry.

What does Trump need to do at the debate?

Ok, look, are there things Trump could do to “win” this debate? Of course there are. He could focus on the economy, reminding some of the voters who have abandoned him this year of the one area in which they approve of his performance as president. And he could stay away from his increasingly long list of personal grievances that are only meaningful (or even explainable) to his base. 

He could insist on keeping the focus on Biden’s refusal to answer whether he would pack the court. After the moderator asks and Biden dodges once again, Trump could use his next answer to say, “I won’t answer any more questions until the Vice President tells the American people whether he will fundamentally alter the third branch of our government. So I’m going to give my time to him to try again—yes or no, Mr. Vice President?” And when Biden dodges again, he would turn his time back to Biden again. And again. Until the moderator was put in the unenviable position of keeping the focus on Biden’s evasion or insisting that she, as a journalist, didn’t want the answer to her question anymore. But this would require the president to stop talking and accept that that awkward silence would play to his benefit.

But is he capable of doing any of these things? Nope. So let’s move on.

Are the polls accurate this time around?

Let’s put it this way: At this point, Biden has such a massive polling lead that the only path for Trump to win reelection is if polling itself is fundamentally flawed in a way that is novel and specific to this cycle. Otherwise, even if the polls are as far off as they were in 2016, Biden’s lead is still insurmountable. 

But 2020 is different. So how might the polls be wrong?

First, perhaps Trump is a uniquely divisive candidate this time around, which means that so-called ‘shy Trump voters’ are lying to pollsters and artificially inflating support for Biden. This theory is particularly unlikely. For this to be true, we’d expect to see unusually high numbers of registered Republicans defecting from Trump, but that’s not the case. In fact, Trump’s strength with Republicans remains both high and consistent. 

Second, polling is as much art as science, based on the pollsters’ magic formula of who they believe will actually vote. This means that they don’t just call a bunch of people and report their findings. Instead, respondents are “weighted” to match something like the “likely electorate.” This is what went wrong in 2016: Pollsters under-weighted non-college educated white voters in the Midwest, setting the stage for big surprises in Wisconsin, Michigan, and Pennsylvania.

For the very reason that this is what caused the Great Polling Meltdown of 2016, it is unlikely to happen again this time around. If anything, we might expect pollsters to have overcorrected and be underestimating Biden’s support.

Third—and I hear this one a lot—is the possibility that Trump voters are uniquely unlikely to respond to polls, distorting the weighted results. This theory requires that Trump’s attacks on polls as “fake news” have caused his voters—distinct from registered Republicans as a whole—to refuse to answer polls. This is not an unreasonable theory to me, but it’s unlikely that it could create more than a one or two point distortion without pollsters seeing something wobbly in their numbers. 

Here’s how it would work: If a pollster has around a 5 percent response rate in a national poll, they are assuming that the 5 percent of people who responded to their questions are not fundamentally different than the 95 percent of people who didn’t. Thus far, believe it or not, that assumption has been accurate. 

But here’s the problem: it would be very hard for this to distort every national poll without seeing an overall dropoff in registered Republican answer rates. In talking to pollsters, I am told that they have actually seen the opposite: Registered Republicans are slightly more likely to answer polls right now. So the only way for this distortion to be happening is if it is happening among Trump voters separate and apart from registered Republicans. Again, it’s possible that this could account for a couple-point swing, but it’s unlikely to be enough to shake the final result.

Are you worried about problems on Election Day?

Yes. But not for the same reasons as everyone else, it seems.

As I mentioned last week, Florida, Arizona, and North Carolina will all have a good chunk of their absentee ballots tallied on Election Day—unlike Pennsylvania, Wisconsin, or Michigan. And a week later, Biden is still leading in all three of those states by more than 3 points. So the worst-case scenario—that Trump wins on election night but the millions of absentee ballots that will overwhelmingly lean toward Biden result in endless litigation around signature matching—is increasingly unlikely.

Instead, I’m far more concerned when I talk to Trump voters, almost all of whom are convinced that Trump is about to win in a landslide, and Biden voters, who claim that they still have 2016 PTSD but can’t actually imagine a world in which Trump can win again. Given the current mood of the country, which one could optimistically describe as “tense,” what happens when 70 million people don’t actually believe that their guy lost? 

And I don’t mean the hyperventilating scenarios in which Trump won’t leave the White House or Biden refuses to concede. (David and I have answered the question of why Trump can’t and won’t just hang around the White House several times on Advisory Opinions—no, he doesn’t need to be evicted; no, the military wouldn’t be involved in a coup; no, no, no. Under the Constitution, he would cease to be the President when the new guy is inaugurated on January 20, 2021. Then, legally speaking, he’s just a trespasser on the White House grounds.)  

What I am concerned about is adding fuel to the not-so-fringe-anymore stuff we’re seeing around the country already. But even so, I believe the majority of Americans—regardless of who they’re supporting—are looking forward to having this election over and done with and that that sentiment will prevail even among the shocked and disappointed. 

Is turnout going to be sky high?

We don’t know yet. There are 158 million registered voters, and 138 million of them voted in 2016. Of those, more than 50 million people—41 percent—voted early, including 17 percent that voted early in-person and 24 percent who voted by mail. 

As of this weekend, 26 million people had already voted early—”more than six times the number of votes cast by the same point in 2016.”

But there’s good reasons for this. There’s a worldwide pandemic, voters have been hearing a lot more about early voting opportunities, and states have relaxed some of their restrictions on who can vote early and how to do it. So the real question is how many of the early voters this time around are just the same voters that would have voted later-but-still-early or voted on Election Day in a previous cycle. 

So, sure, there’s some evidence in the polls that turnout may be higher based on self-reported likelihood to vote and enthusiasm. But as of now, the hype over these early vote numbers is just that. 

Do you think Republicans lose the Senate?

Yes. Once again, unless the polls are just wrong and polling as a profession is about to die in a fiery blaze, Republicans will lose the Senate. But what about anti-Trump Republicans who will split their ballots? First, these strategic, divided government voters are pretty rare to begin with. Second, they’re getting a lot more rare as time goes by. 

In 2016, Pew noted that “all 34 Senate contests tracked the presidential vote in their respective states.”

Once again, 2020 could just be special, and there’s no doubt Trump provides a unique circumstance that might inspire such split ticketing. But there’s no evidence for that yet.

What states will you be watching on Election Night?

  1. Florida

  2. Florida 

  3. Florida

Florida has backed the winning presidential candidate since 1996 and Trump is down 3.9 points in the FiveThirtyEight polling average. Sure, Trump’s campaign manager has pitched reporters on winning maps for the President that don’t include Florida, but none of them make any logical sense. He loses Florida but wins Michigan? Huh? More realistically, they argue that their most likely path to 270 is Ohio, Florida, Georgia, Iowa, Maine’s second congressional district, Arizona, North Carolina, and Pennsylvania. We’ll know Florida early because their polls close at 7 p.m. EST (except on the western edge of the panhandle) and they start counting their absentee votes three weeks ahead of time. 

(Ditto North Carolina, by the way, which will be our first bellwether to see how accurate the state polls were this time around. Trump is currently down 3.2 points.)

The truth is that if Trump loses Florida, the polls were right or underestimating Biden’s support. And that means it’ll be an early night for all of us.

Photograph by Mario Tama/Getty Images.

Sarah Isgur is a senior editor at The Dispatch and is based in northern Virginia. Prior to joining the company in 2019, she had worked in every branch of the federal government and on three presidential campaigns. When Sarah is not hosting podcasts or writing newsletters, she’s probably sending uplifting stories about spiders to Jonah, who only pretends to love all animals.