Skip to main content
What Could The Polls Be Missing?

In this week’s politics chat, we weighed reasons the polls could be underestimating Hillary Clinton or Donald Trump. The transcript below has been lightly edited.

micah (Micah Cohen, politics editor): The first presidential debate is behind us, and by most metrics, Hillary Clinton had the better night. But we’ll have to wait a few days to see how/whether the polls move as a result. So, in the meantime, I’d like us to ponder the road ahead. Specifically, whether there are factors not captured by the polls that will help Trump or Clinton.

natesilver (Nate Silver, editor in chief): Did you see the Drudge poll? Trump won 80-20.

micah: I don’t trust the Drudge poll; I always go to the Patch poll first.

clare.malone (Clare Malone, senior political writer): I like to ask my cab drivers what they thought.

natesilver: I’m a big fan of the poll, but it’s a little Trump-leaning. The raw numbers were 99-1 Trump, but our model adjusts it to 97-3.

micah: Anyway, the goal here is to consider factors that might be working in either candidate’s favor that the polls are missing.

clare.malone: Astrological signs.

micah: So, let’s run through some.

Hypothesis #1: The polls are underestimating Clinton because they don’t factor in her superior ground game.

Most reports (and we’ll have an article with some extensive data on this soon) suggest that Clinton is far better organized than Trump, with more field offices, for example, and a better get-out-the-vote (GOTV) operation. Isn’t that reason to think the polls are underestimating Clinton by a bit?

natesilver: That’s one of the better arguments, yeah.

harry (Harry Enten, senior political writer): Well, the question you have to ask yourself is whether or not the field game is being picked up by the polls. In a wonderful, perfect universe, it would be. In reality, I have my doubts. That’s why I’m not entirely sold on polls showing a large gap (in Trump’s favor) between registered voter results and likely voter results.

natesilver: The empirical literature on the ground game is messy because there are so many conflating variables. But what would worry me if I were Trump is that the race isn’t that close if everyone turns out — and I have the worse turnout operation.

micah: How would the polls pick up on a better ground game?

harry: A ground game is supposed to get people to vote. So, if Clinton’s campaign is getting more people engaged, then polls that are aimed at identifying likely voters should pick up on that. More of those voters should make it through the likely voter screen.

natesilver: Well, maybe they’ll make it through the likely voter screen, but a lot of likely voter screens rely on past voting history, and may or may not have adequate provisions in place to capture new voters.

harry: Right.

natesilver: Also, a lot of likely voter screens do something that’s probably dumb, which is to set a hard cut-off for voting propensity instead of doing it probabilistically. Clinton has a “long tail” of semi-likely voters that she could potentially draw from.

harry: Also correct.

micah: Clare, have you seen evidence of a superior Clinton ground game on your travels?

clare.malone: Clinton and the Democrats certainly have a more traditional, organized field operation. This is not to say that Trump doesn’t have some of the basics — the campaign has, for instance, phone-banking apps that their volunteers can use from home, along with walking apps for door-to-door stuff. But Trump is still substantively relying on the Republican National Committee’s operation, and they’ve been suffering from a lack of enthusiasm in some of the management-level types who usually run campaigns.

I would say Trump is still mostly counting on his earned media strategy.

We also know that the Trump team is trying to turn out people who DON’T usually vote, and who probably don’t answer polls.

micah: So Clare, couldn’t Trump’s reliance on the RNC for GOTV hurt him? Down-ballot Republicans might want to turn out slightly different voters than Trump would want.

harry: We saw very little proof during the primaries that Trump turned out untraditional voters.

clare.malone: Right. I think the RNC is a little nervous that Trump people might not vote Republican all the way down ticket, so in some of their door-to-door activities, they might be targeting the more traditional Republicans who are turned off by the presidential race. I.e., looking to target some college-educated whites for Senate races in key states.

natesilver: Yeah, Trump’s performance was pretty meh against the polls in the primaries. And he didn’t do a good job of winning over late-deciding voters, although that’s a slightly different issue than not having a good turnout operation.

harry: Well, we have more evidence than just the polls. We have Catalist data on who turned out to vote: Most Trump primary voters also turned out in the 2012 general election. The only candidate who had a large share of untraditional voters was Bernie Sanders.

micah: OK, so… Clinton could add 20 percentage points to her margin over Trump thanks to a superior GOTV operation?

clare.malone: 20 points seems like quite a lot.

harry: We’re looking at, at most, a few percentage points. It could add maybe a point or two to Clinton’s margin.

micah: NEXT!


Trump is a Gemini, Clinton’s a Scorpio. I invite readers to tell me what that means and I will spin that shit into golden analysis.

harry: What is a Gemini?

natesilver: This is the dawning of the Age of Trumparius


Hypothesis #2: The polls are underestimating Trump because of shy Trump voters.

We’ve covered this a bit, so let’s do it quickly, but the argument is that there’s a social stigma in some areas attached to supporting Trump, so the polls are underestimating his support because some of his supporters are unwilling to say they back him.

Any takers?

harry: It is quite funny to think Trump supporters are shy when they all seem to take the time to vote in these stupid non-scientific online polls after debates.

clare.malone: In the primaries Trump voters weren’t shy, were they? Does that change for the general?

harry: Right, there is no proof they were shy in the primaries. Moreover, Trump supporters are more likely than Clinton voters to make it through the likely voter screens, indicating they are more vocal and enthusiastic in their support. The polls could still underestimate Trump, but they could also underestimate Clinton.

natesilver: In the abstract, it’s a plausible theory. There’s a fair bit of evidence on the impact of social desirability bias. I just don’t know if it fits the facts of the case all that well. Trump voters aren’t shy, for one thing. They’re actually more demonstrative than Clinton voters, in many, many respects.

clare.malone: They’re huggers, you mean?

natesilver: The original “shy” term comes from “shy Tories,” and the idea was that British voters didn’t want to admit to voting for a party that was plodding and un-hip, or at least they were less enthusiastic about responding to pollsters.

So that sounds more like Clinton’s voters than Trump’s, if anything.

The social desirability idea is a separate one that’s sort of gotten lumped together with it. It’s a plausible idea, but “shy Trump” probably isn’t a good name for it. Instead, it’s more like the Bradley Effect.

harry: You know I heard of this one before. I heard it back in 1991, when David Duke was running for governor of Louisiana. Pollsters were all concerned that white voters were afraid to admit they were voting for the former KKK member. Did the polls underestimate Duke? Actually, they overestimated him. The polls underestimated the opposition to Duke from black voters who turned out in large numbers to vote against him.

clare.malone: It seems unlikely to me that there are a lot of shy Trumps, if only because I think he’s become pretty normalized now.

micah: Totally agree ^^^.

clare.malone: If you’re a Republican in a swing state who’s on the fence, for instance, you can say you don’t like the guy, but stand behind the protection of “I care about Supreme Court nominees.” That’s where we are these days.

natesilver: One more thought on shy Trump and social desirability: Pollsters sometimes try to get around asking people sensitive questions by asking about how their neighbors behave. They’ll ask “are people in your community against interracial marriage” for instance, instead of asking directly about the respondent’s view.

What’s interesting is that if you ask voters who they think is going to win the election, Clinton does very well on that question (although less so lately).


clare.malone: That has always seemed like a really weird way to do it, to me at least. It just seems really fraught. What if you hate your neighbors?!

harry: I don’t even know my neighbors.

micah: NEXT!

Hypothesis #3: The polls are underestimating Clinton because she has a lot more money than Trump and will blitz the airwaves in the last few weeks of the campaign.

harry: I don’t buy this one. If the ads were working so well and could blow Trump out of the water, then why isn’t she crushing him yet? She has already been outspending him by a ton.

micah: But maybe people are only tuning in now?

clare.malone: The ad stuff is so hard for me to tell what’s happening — maybe we are reaching the end of an era of effective television ads, right?

I think the best ones she’s had are the ones with little kids listening to all the terrible things Trump says, that kind of thing strikes me as persuasive to people.

harry: I’m not saying ads do or don’t work. I’m just saying we shouldn’t expect them to start working in ways they haven’t so far. At least not to so great a degree to cause a large polling miss.

clare.malone: That’s fair enough.

micah: Do ads work?

natesilver: In down-ballot races, sure — but the impact of ads is pretty minor as compared to the amount of free media that these candidates are getting.

clare.malone: Mentos ads definitely worked on me in my teen years.

micah: Back-to-school ads worked on me for sure. I used to think, “If I only had that cool backpack, I’d be popular.” Or that awesome pencil case.

clare.malone: Those still work on me. I want a pair of saddle shoes every Sept. 1.

[Stage note: Now we’re all talking about how many shoes we own.]

natesilver: One thought, btw, is that maybe Clinton could use her ads to try to improve her favorables instead of trying to tear down Trump. The negative attacks tend to get a lot more amplification from the media, so perhaps there’s less need to advertise on that basis.

micah: I agree with that. She should go all self-promotion.

Hypothesis #4 (and this one I buy): The polls are underestimating Clinton because the remaining set pieces of the campaign — the things we know will happen — play to Clinton’s strengths, all else being equal. The remaining debates, mostly.

So Clinton is running slightly downhill from here on out.

natesilver: That metaphor is confusing! It’s like positive and negative feedback!

Negative feedbacks are usually good — they keep us all from frying or freezing to death, for instance!

micah: She’s running slightly downhill so her polls are more likely to go up. What’s so confusing about that?

clare.malone: I would agree that she does better in forums where she can showcase her competence. I think the next debate, which is a town hall, will be really interesting — seeing Clinton/Trump interact with voters.

natesilver: Obviously (?) there are two theories about the remaining debates. One is that Trump will benefit from even lower expectations and/or Clinton complacency. He can’t help but be a little better, right?

The other theory is that Trump is arrogant enough to think he can win debates without preparing for them — and will use those Drudge polls to convince himself that he actually won last night!

micah: BTW: Clare said a while ago that she thought the debates would clearly be good for Clinton because they would necessarily showcase her competence, and I was skeptical just because debates can be so focused on performance. So I just want to say for the record: Clare was right, at least about debate No. 1.

natesilver: One way to put it is that Clinton is good with the set-pieces of the campaign, while Trump excels in open play — the random chaos of the day-to-day.

harry: The thing that was so funny about that first debate is that people had convinced themselves that Clinton and Trump were on mostly equal ground heading in. The share of people who expected Clinton to win was right near the lower end of what we’ve seen historically. What I’m interested in is whether or not people might convince themselves again that Trump will somehow do better. The question I have to ask is why?! What has he done to make us believe that he can improve?

natesilver: Another question is whether there’s time for a Clinton comeback narrative, and then a Trump comeback narrative, to the extent that the media framing of these things matters.

Basically what’s happened so far:

  1. May was good for Trump.
  2. June was good for Clinton.
  3. July was good for Trump.
  4. August was good for Clinton.
  5. September was good for Trump.

It looks as though — maybe, maybe — October might be good for Clinton. Is there time for November to be good for Trump, or does it not count because it’s not a full month? [Editor’s Note: It’s eight days.]

clare.malone: hah #datajournalism

micah: Last one!

clare.malone: Mercury in retrograde??

harry: I prefer Pluto. (It’s still a planet.)


Hypothesis No. 5: The polls are underestimating ______ because _______.

This is for wildcards! Wikileaks! Clinton’s better surrogates. An economic recession. What else could matter?

natesilver: There’s not much chance of a recession now. Or at least, if there was one, it wouldn’t be declared until after the election. But y’all haven’t heard my self-reinforcing financial panic theory, have you?

harry: No, but I’m sure you’ll tell us.

clare.malone: Regale us.

natesilver: It goes like this: Financial markets start to panic because they worry about a Trump presidency. The panic makes a Trump presidency more likely. Vicious cycle. President Trump!

clare.malone: That seems … not likely.

harry: What a theory.

clare.malone: But on the other hand, I graduated right as the financial crisis hit, so part of me is letting that sink in and panicking — what if I never get a job?? I’m an English major, for god’s sake!

Wikileaks is a good wildcard, by the by. They’ve been eerily silent over the past two months.

harry: I tend to doubt anything is going to happen.

natesilver: I dunno. I think Wikileaks has shown that its judgment about what’s actually newsworthy or not is suspect. Lots of false alarms.

clare.malone: Still think that Wikileaks and terror attacks (god forbid) are the deus-ex-machina possible actors in this election.

natesilver: Our editor has left the slack chat, by the way, so it’s like lord of the flies in here. I’m going to use a sentence with an oxford comma in a moment!

harry: I love the oxford comma.

clare.malone: I love it too.

natesilver: Oh wow, not me.

clare.malone: I think I write with them in most of my drafts and Micah has to change them all. I also use a different kind of em-dash formatting that he changes every time. Sorry, Micah.

micah: I’m back, and apology accepted.

harry: To get back to the point: There rarely are October surprises. Maybe there will be one of this year. But I’ll bet on no October surprise.

natesilver: Well, it wouldn’t be a surprise if you could bet on it, now would it Harry?

Nate Silver is the founder and editor in chief of FiveThirtyEight.

Harry Enten is a senior political writer and analyst for FiveThirtyEight.

Clare Malone is a senior political writer for FiveThirtyEight.

Micah Cohen is the politics editor.