Welcome to FiveThirtyEight’s weekly politics chat. The transcript below has been lightly edited.
sarahf (Sarah Frostenson, politics editor): FiveThirtyEight alum Harry Enten found that former Vice President Joe Biden’s average lead of 6 percentage points over President Trump is the steadiest lead in a race with an incumbent running since at least 1944. In other words, Biden has led, on average, by 6 points since the beginning of 2020, but he’s also led by 6 points since the beginning of 2019.
Biden also has a decent lead in a number of swing states according to some early polls, yet many Americans don’t believe Trump will lose despite the many, many polls showing Biden ahead. How do we explain this disconnect? Are voters overreacting to Trump’s win in 2016? Then, let’s discuss how to make sense of the polls at this point in the cycle.
OK, first: What’s up with voters giving Trump such high odds of beating Biden — 55 percent said they think Trump will win in November according to an average of recent polls. That’s been relatively consistent over time, too. Do you think that will change, though, if Biden remains in the lead?
perry (Perry Bacon Jr., senior writer): Trump’s standing might be stronger in perception than in reality for many voters for two reasons. First, the polls had Clinton as the likely winner in 2016 and Trump won — so, of course, some people are wary of the polls again. Second, I think the possibility that Trump might take extralegal, unethical or norm-violating steps (like trying to have the Ukrainian government investigate the son of one of his rivals) to win the election makes people a little less confident that the election will take place in a traditional way. (Basically, if some people already think Trump will cheat to win, they will likely not find the polls much comfort.)
nrakich (Nathaniel Rakich, elections analyst): I think people are clearly overcorrecting for what happened in 2016. There’s a natural human bias to not want to make the same mistake twice, and for a lot of voters, that mistake in 2016 was, “Donald Trump can’t win.”
I think there’s also a sense that Trump is Teflon politically.
And I get that, since opinions of him are so baked in. But it’s also not reallllly true, since some events (like the government shutdown) have hurt his popularity. And regardless, he is still quite unpopular, so even if those attitudes are baked in, they’re not good for Trump.
That said, people are correct to be uncertain about the election. We are still six months out.
geoffrey.skelley (Geoffrey Skelley, elections analyst): These numbers on how many Americans think Trump could win reelection could also still shift — for instance, they already have in Fox News’s polling over time. After the 2018 Democratic wave in the House, only 39 percent said Trump would be reelected compared with 52 percent who said he wouldn’t. Since then, Trump’s standing has improved, but it’s bounced around some too. How things play out in the next few months and what coverage of the race looks like will shape public opinion.
perry: Another problem is the reporting that Trump staffers X or Y or various senators or GOP strategists think Trump is in trouble tells us little. I read those stories from May to October 2016, and they often turn out to be wrong. The best information we have is what we have already mentioned: 1. Trump’s consistently weak approval ratings; 2. the fact that those weak ratings correctly foreshadowed a huge Democratic wave in 2018; 3. the head-to-head polls showing him behind Biden.
nrakich: Yeah, totally, Perry. Political insiders are subject to the same kind of bad punditry that the rest of us are.
sarah: But, as Nathaniel said, the idea that nothing Trump does will negatively impact him does seem, at this point, to be baked in. I’m curious, though, what can we point to as signs that Trump might be in trouble — aside from Biden’s 6-point lead?
One thing that’s stood out to me is Trump’s lost points among voters who don’t like either candidate running. In 2016, Trump won these voters by a large margin, but here in 2020, they’re breaking overwhelmingly for Biden.
geoffrey.skelley: For me, it’s hard not to connect Trump’s standing in the polls with his approval rating. It’s been consistently stuck between roughly 40 and 45 percent in most polls. And, overall, he’s at about 45 percent in FiveThirtyEight’s tracker of registered or likely voter polls, with about 52 percent of voters who disapprove of him (his approval has generally been a little better among voters than adults, and voters will determine his future).
I’d argue that this had real repercussions in the 2018 midterms, too. Just look at the exit polls — 88 percent of voters who approved of Trump voted for the GOP, but 90 percent of those who disapproved of him voted Democratic. So, if many more people continue to disapprove than approve of Trump, I think it’ll be challenging for him to win.
nrakich: Sarah, to your point about the “haters” bloc, I agree that that seems important. But I also am wary of that turning into the new “working class voters in a diner,” a.k.a. the type of voter who dominates coverage of the election to the exclusion of the (equally important!) voters who don’t fit that description.
sarah: For sure. I’m just curious how much evidence we have that Trump is in trouble and how that reconciles (or doesn’t) with voters’ belief that we shouldn’t underestimate him here in 2020, and I think if he continues to do poorly with this group of voters that could be telling.
Of course, one point in Trump’s favor is that there’s little evidence that Biden has made progress in winning back the white voters without a college degree who supported Obama in 2008 and Trump in 2016.
perry: I’d argue that Biden’s lead is not particularly solid — so voters thinking he will not win makes a lot of sense to me. Basically, I agree with the voters.
nrakich: Wait, really, Perry? Not to put words in your mouth, but from my conversations with you, I thought you believed that the election would be competitive — that either side could win. But I think voters think Trump will definitely win.
perry: Fifty-five percent of voters thinking Trump will win is not 90 percent.
Voters saying they think Trump has a very good chance of winning reelection seems entirely consistent with the data, too.
Say the election were today — I would expect Biden to win Arizona, Michigan, Pennsylvania and no other states Trump won in 2016. Yes, Biden would win the election, but this is not exactly a landslide.
sarah: Ah, so could it also be that we (the media) are overestimating how much voters are overcorrecting for 2016?
sarah: That is, a majority think Trump will win, but to your point, Perry — it’s 55 percent of voters saying this, not an overwhelming 90 percent majority.
nrakich: I think that is fair. I have probably been guilty of that so far in this chat.
geoffrey.skelley: Pundits don’t want to look stupid — and after 2016, they probably felt pretty foolish, so they may be doing more overcorrecting than the public is.
Then again, there are weird, small disconnects among election watchers — PredictIt’s markets, for instance, make Biden a favorite in enough states to win a majority in the Electoral College, but the overall market for the election winner still favors Trump. 🤷♂️
sarah: Yeah, it’s an unsatisfying answer, but I think it’s a combination of the two. That is, we definitely risk overestimating how much voters think Trump is “Teflon Don” and can therefore weather every scandal. But, at the same time, I think there’s a real phenomenon happening here, where voters overall are less likely to take the polls at face value.
This was another stat in Enten’s article, but it stood out to me that in 2018, despite many indicators pointing toward a Democratic wave, many voters didn’t think Democrats would take back the House. According to Gallup polls, 50 percent of Americans said they thought Republicans would win the House compared with 44 percent who said Democrats would. What stood out to me, though, was that this apparently was the first time in Gallup’s polling when Americans incorrectly forecasted who would win the House!!
geoffrey.skelley: I guess people had to see it to believe it. It might also help explain why Fox News’s polling on whether voters thought Trump would win reelection took a sharp tumble right after that election — because voters saw the GOP in the Trump era actually lose the House.
perry: People are wary of polls after 2016.
nrakich: With little justification, it should be said. The national polls were spot-on in 2016 (it’s just that the national popular vote doesn’t pick the winner of the election). And the polls were great in 2018.
perry: Nathaniel’s right that the polls were actually pretty good — but it’s also the case that media coverage didn’t seriously entertain the possibility of Trump winning.
So, you have conservatives who are generally distrustful of the media and are always disinclined to believe polls covered in the media and, as is often the case, conducted by media organizations. Combine that with liberals who felt the polls gave them a false sense of security in 2016 and you have a sizable bloc of people who are going to be skeptical of 2020 polling.
sarah: Right — and also to be clear, even though the overall accuracy of the polls in 2016 was only slightly below average by historical standards, the state-level polling wasn’t as accurate. And because it was off in a few key swing states, I think that colored the overall accuracy of the polls for many folks.
nrakich: Yeah, I think you have a combination of Democrats having very low self-confidence in their ability to win an election due to post-2016 trauma and Republicans being overconfident because they keep getting told not to listen to the media or polls.
perry: News consumers in America weren’t really presented with an accurate assessment of Trump’s chances of winning in 2016 — and I say that as someone who was covering the election in 2016.
FiveThirtyEight did better than others, but FiveThirtyEight is not the entire news media.
geoffrey.skelley: And, as Nate Cohn at The New York Times wrote on Tuesday, there may be reasons to remain cautious of state-level polls this year, too. On the one hand, there are fewer undecideds and more pollsters are weighting by education, but on the other hand, there are more online state-level polls with questionable methodologies, including some that weight by how people said they voted in 2016 — a suspect choice partly because people’s memories are fallible.
perry: Are we even sure the markets are wrong, though, and that Trump is an underdog?
geoffrey.skelley: Trump has consistently trailed Biden by 4 to 6 points nationally, so I would say yes. Obviously, the smaller that edge is by Election Day, the greater the possibility for Trump to lose the popular vote but still win the Electoral College, if he still has an advantage there (and early evidence suggests he may).
perry: But couldn’t Trump lose nationally by 6 points and still win? Or say the polls are a bit off and Biden is really leading by only about 4 points nationally, isn’t that about even odds if Trump’s Electoral College advantage holds?
geoffrey.skelley: If he loses the popular vote by 6 points, I would be surprised if he still won.
perry: Sure. What about 4?
geoffrey.skelley: It’s possible, but again, this is assuming he still has a notable edge in key states in the Electoral College.
perry: Isn’t it obvious that Trump has an Electoral College edge again?
He is stronger in Wisconsin and Florida than he is nationally, for example. He is stronger among white working-class voters than among other blocs.
geoffrey.skelley: Right, but I think the size of that advantage is debatable — some claim it’s bigger than in 2016, but I don’t think we know that yet.
nrakich: Let me see if I can thread this needle… Yes, I strongly suspect Trump will have the advantage in the Electoral College again for the reasons you cite. But past attempts to predict who has the Electoral College edge before the election have very frequently gotten it 180 degrees backward. And who has the Electoral College advantage bounces like a ping-pong ball between the parties from cycle to cycle.
geoffrey.skelley: Also, while the national polls could be off by a couple of points again, it’s worth remembering that the direction of that error varies from cycle to cycle. So it’s also possible for that error to go in Biden’s direction too, in which case the race might not be that close.
Campaigns are also dynamic. Just ask Clinton — there’s reason to think that the release of James Comey’s letter less than two weeks before the election may have moved the needle just enough for her to lose. And, remember, we’re still roughly six months out — so things can and probably will change in different ways. What if the economy continues to go sideways over the next few months? Voters may start to say, well, Trump is more likely to lose than not.
perry: I totally agree with that. But the tricky thing is that while campaigns are dynamic, views on Trump are very undynamic.
So that’s why I tend to think that events don’t really matter this cycle. My main question is really, “Are the swing state polls accurate this cycle?” And I don’t feel confident I know the answer to that question, so I am as confused as the voters are. So, like voters, I might be overstating Trump’s chances because I am overcorrecting based on what happened in 2016 — and I am probably overstating Trump’s chances based on perhaps also overcorrecting for what happened in 2016.
geoffrey.skelley: Maybe it will boil down to how voters view Biden as an alternative to Trump. If the election is truly a referendum, that’s not good for Trump. But if there’s increasing doubt about Biden’s candidacy, it could reduce Democratic turnout or maybe even push some of those “hater” voters into Trump’s camp. After all, Biden’s lead isn’t insurmountable. As Enten recently noted, there’s even been a downtick in Biden’s lead across national polls from about 6 points to 4 points, which line up with the allegation of sexual assault made by former Biden Senate staffer Tara Reade gaining traction.
In other words, events still matter at the margins, and the margins could be decisive, considering Trump won because of about 78,000 votes in three states in 2016.