In this week’s politics chat, we discuss FiveThirtyEight’s general election forecast for 2016, which launched Wednesday and will be updated through Election Day. The transcript below has been lightly edited.
David (Firestone, managing editor): After a lot of work, our general election forecast has gone live! But not everyone fully understands what a forecast model is. Some people think it’s a poll or are incorrectly saying, “Nate calls the race for Clinton.” Can we start by explaining at a basic level what a forecast model tells us?
Clare (Malone, senior political writer): Say it very, very slowly and use small words, please.
Nate (Silver, editor in chief): Pro buh bil it eez.
Harry (Enten, senior political writer): Looks like some artisanal tea that I’d buy in hipster Brooklyn.
Nate: Not everything 0 percent or 100 percent! Some things in between! Cookie Monster like numbers in between!
Harry: I believe Cookie Monster now eats vegetables.
David: And that in-between number gives Hillary Clinton about an 80 percent chance of winning, which obviously doesn’t mean it’s over.
Clare: Did Cookie vote Trump? Or is he a Bernie Bro?
Nate: I’m sort of annoyed by it being 80 percent, because I feel like that’s the number people most misinterpret. When you say 80 percent, people take that to mean “really, really certain.” It’s not, particularly.
David: I liked your ballgame analogy, Nate, in the article you wrote to accompany the forecast. Teams come back from 20-percent-win situations frequently. In fact, about 20 percent of the time!
Nate: Absolutely amazing how that works!
Clare: You’re annoyed that it’s a high number because people are going to glom onto that and think it holds for the whole election? Not realizing that this is where things stand as of June 29 and that it’ll change as things go on and polls come in?
Nate: It can change, sure. But let’s be clear — 80 percent is the forecast Clinton has to win on Nov. 8. That’s our best estimate of her chances, accounting for the uncertainty between now and then, based on the historical accuracy of presidential polling. If the election were held today instead, she’d be a safer bet still.
The polls can change a lot between now and Nov. 8. And they probably will. But there’s a chance those changes benefit Clinton, and not Donald Trump. And since she’s up by about 7 points now, there’s the chance they help Trump … but not enough to allow him to win.
And that’s the thing. Of the 80 percent of the time Clinton wins — PLENTY of those times are going to involve her sweating. Either because Trump makes it very close at the end or because there are some periods in which things look very tight along the way, as they did for Obama against McCain and against Romney.
But Clinton will win a lot of those close calls, along with her share of landslides.
Clare: So the Clinton campaign should not change its warm-up song to “Landslide” just yet? (The Fleetwood Mac version, obvs, not the Dixie Chicks cover.)
David: Because it’s a model, we’ll be feeding new polls into it as they come out every day, or whenever we have them. And the polls-plus version also changes with economic performance. So we can expect to see fluctuations in the numbers regularly, and sometimes those can be serious changes.
Clare: This model eats!
Harry: Only the best food, no fast food.
Clare: Yeah, can you explain what the “pluses” are in polls plus? The sauce on our model’s low-fat, totally organic polls, if you will.
Nate: The “plus” is the economy, basically. Which isn’t so good right now, but also not so bad. But certainly, in the abstract, you’d expect this to be a close election. With no incumbent and an average economy, that should mean a level playing field.
Harry: I should point out that our model does not take into account something like the president’s approval rating, which isn’t strongly tied to the final result without an incumbent but isn’t nothing either. And the president’s approval rating isn’t terrible right now.
David: If this were a normal election, or even a closer one, what patterns in the polls could we expect to see after the primaries? Bounces after the conventions, movements based on big speeches or ads?
Harry: After the conventions, you may not see big bounces. You sort of saw one last time after the first presidential debate, but that wasn’t a big one in the state polls. I think the clearest case of movement was probably 1948, and that was when the economy was getting better and better. (Truman, of course, closed a big gap in the polls and shocked everyone by winning the election. There were no public polls conducted in the final weeks of that campaign.) It’s not that the polls cannot move. It’s just that they usually move to where the economy might direct it to — not because of a big speech.
David: So it sounds like the biggest changes in the polls could come from factors the candidates themselves can’t control? Like major news events or economic changes?
Nate: I dunno. We’ve had a couple of major news events, and they didn’t seem to move the needle as much toward Trump as he might have hoped. In fact, they didn’t seem to help him at all.
Harry: Wasn’t it just a few weeks ago that some people were saying the attack in Orlando could help Trump?
Clare: I think few people expected his tactless, self-congratulatory response.
David: Yeah, he seems to step on his opportunities whenever they happen, including suggesting that the Brexit vote was great for his Scottish golf course.
Clare: But are there things outside of the economy and a terrorist attack that could hurt Clinton a lot, if this is hers to lose? Email shenanigans, perhaps? Is that just no longer that big of a factor? We’ve all talked about it ad nauseam, it seems. I wonder if peoples’ opinions have changed much on the subject.
David: A lot of Republicans are hoping for an indictment, which seems highly unlikely. Hard to see how the email thing will get worse for her if that doesn’t happen.
Harry: What’s so great about modeling, though, as Nate has mentioned to me, is it can give you an insight into something you cannot see yourself. I don’t know what event can turn Trump’s campaign around, but I know it is possible.
Nate: Yeah, might I suggest that trying to come up with the scenario by which Trump wins is exactly what gets you in trouble? There are known unknowns, but a lot of the 20 percent is unknown unknowns.
Harry: Thanks, Rumsfeld.
Nate: Or it’s a case where several little things go wrong for Clinton, instead of one big thing.
Say a lot of the Bernie Sanders vote goes to Gary Johnson or Jill Stein, and the economy is taking some hits, and Trump’s voters turn out at greater rates than polls expect them to. No one of those factors is enough to overcome a 7-point deficit. But collectively, they could be enough.
Harry: I think one of the great polling questions is: How do you account for third-party candidates? Some pollsters just ignore them. Some aggregators ignore them. But I tend to think you should include them when they are probably going to be on the ballot in all 50 states and the candidate is regularly getting above 5 percent in the polls.
Nate: And most polls are including Johnson, these days. That’s generally been the rule of thumb. Once a candidate gets into the high single digits or low double digits, most polls will include him. And Johnson probably will be on the ballot in 50 states, or pretty close to it.
Clare: Do we anticipate Johnson’s support growing over time, being affected by various hits that candidates take throughout the campaign? Or could his support just fade out as we head into the fall? Did Perot, for instance, see a rise or a fall as the … fall came?
Harry: It could go either way. I hate to use this word, but third-party candidacies are often about “momentum.” Can the candidate catch on and be a viable choice? Perot was, in 1992, and his support held at near 20 percent after re-entering the race. (He was regularly in the 30s before exiting the first time.) Then you have someone like John Anderson, who in 1980 started in the 20s and slowly fell to just above 5 percent.
David: The most striking thing about the model to me is when you get to the state level, which of course is where the electoral battle is really fought. There are some unexpected colors there. Georgia isn’t dark red — it’s pink. In our polls-only model, Arizona is a light blue, as is North Carolina. Clinton even has a shot in my home state of Missouri, which I would never have predicted. Harry, how surprised were you by the roll of states?
Harry: I’m not surprised based on the polls I’ve been seeing from the states. The model puts a nice number on it. I’m more surprised by the results from states where we don’t have a lot of polling. I’m talking about places like Mississippi and South Carolina.
Clare: Yeah, Mississippi being so light red in our model really struck me! Texas too!
David: There’s only one poll in from Mississippi, and it shows Trump ahead — but only by 3 points.
Harry: We have had two recent Texas polls showing Trump up by less than 10 percentage points.
Nate: So the reason the model has states like Mississippi and Texas kinda close is because that’s where the polls have it. It’s been a pretty consistent pattern. It’s hard — not impossible, but hard — to find polls where Trump leads by double digits, even in the reddest states.
Harry: How about that one Kansas poll that had Clinton ahead!
Clare: Was there a state or group of states that surprised you the most as you were going through all this, Nate?
Nate: See, I thought it was quirky things about Kansas or Utah or whatever. But it’s really across the board. There have been Mississippi, Alaska and Texas polls that also show a pretty close race, and those states don’t have a lot in common with one another.
Harry: They all have vowels in them!
Clare: You are excused from this chat now, Harry.
Harry: Is that nice? Is that nice? To think I introduced you to Colin Quinn.
Nate: OMG Hollywood Harry. Getting all name-droppy.
Clare: It’s true. I will always be in Harry’s debt for that one.
Nate: We’re going to trade you to the “Keepin’ It 1600” podcast for a reporter to be named later
Harry: A friend of mine loves that podcast and keeps mentioning it to me.
Clare: Wait, so what’s the matter with Kansas? And Alaska and Utah? Educated guesses?
Nate: I guess what I’m saying, Clare, is that people are maybe overinterpreting why exactly it’s Utah and Kansas where Trump is doing badly, as opposed to Idaho and Nebraska.
Harry: Now as for Kansas, we see Trump doing worse in more traditionally Republican libertarian and religious states.
Nate: Part of it is just because there aren’t a ton of polls in any of these states.
David: In both Kansas and Missouri, Clinton has won a poll. OK, they’re B- or C- polls in our pollster ratings, but it’s a little hard to imagine, and seems to be a reflection of the disarray with the GOP. The hard-line conservatives and evangelicals in those states just aren’t checking the Trump box in the surveys. Yet.
Harry: I should note our polls-only model gives Clinton a 51 percent chance of carrying Nebraska’s 2nd congressional district and its one electoral vote.
Nate: Trump didn’t do all that well in the Midwest in the primaries. And I could come up with an argument for why the temperament there doesn’t suit him, but I’ll just say it’s totally reasonable to think he’ll have trouble in that Nebraska district given the polls in nearby states.
Clare: This is a good chance for you to talk about regionalism, no? States as blocs.
Nate: Maybe, but maybe that’s getting too granular? Trump’s getting only 37 percent of the vote or so right now. It’s been a long time since a major-party candidate did so poorly. Where are his “missing” voters? A lot of them are in red states, as best as we can figure. Which is good news for Trump, in a way. A lot of those voters were superfluous. Say you win South Carolina by 5 points instead of 12? Not that big a deal.
Harry: The two Texas polls had him at 37 percent and 39 percent.
Nate: On the flip side, it might mean that the low-hanging gains for Trump to make are mostly in red states and that if he gains 3 points on Clinton nationally, maybe he only gains half as much in swing states.
David: Clare, you’re going to be hitting some of these regions as the campaign goes on. Which ones are you most interested in visiting to learn some on-the-ground answers to these questions?
Clare: I’m pretty interested in the South and the voters who Nate says are missing, but maybe it’s not that big of a deal they’re missing. Regardless, I’m wondering if some of these people are those who haven’t voted in a while and whether there is a turnout strategy for them from the Republican National Committee, which seems like it’s going to be by default running Trump’s on-the-ground game.
David: It’s not clear anyone has explained the ground-game concept to Trump.
Clare: But also, obviously, places like Arizona that are surprisingly swingin’ this year.
Harry: Arizona is a state where our polls-only model says Clinton is a slight favorite. The reason? The polls.
Nate: I guess I’d just say the map looks strange in part because the map has been so steady over the past four or five elections. Over the long term, that’s actually pretty anomalous. Instead, the swing states shift around, sometimes gradually and sometimes more suddenly.
Harry: If you look at the 1996 election, Arkansas and Louisiana were heavily blue states. No Democratic presidential candidate has won them since. States do and will change.
Nate: To me, the notion that Arizona is a tossup or that Maine might be pretty competitive if Trump’s position improves — those aren’t even particularly radical changes. They’d be considered quite normal, by historical standards.
Harry: If you look at the 2012 campaign, the Republican candidate for Senate in Arizona, Jeff Flake, barely won.
David: All this is why readers should tune into this model frequently over the next four months. States can and will change colors, and if you wake up early one morning, you might be the first to catch it on our site.
Nate: And we have the two different versions of the model too. The map in polls-plus looks more familiar, but it gets there by making some stronger assumptions. Polls-only is more open-minded about what might happen.
Remember that when Obama won in 2008, he won by 7 points, which is exactly the margin by which Clinton is ahead now in the polling average. And he flipped a bunch of states to the blue column — Virginia, North Carolina, Indiana — that people thought of as being pretty surprising at the time.
Those states are in different categories. The demographics have trended Democratic in Virginia. Indiana was more of a one-off, because Obama contested it and McCain blew it off. North Carolina is still red-leaning but — well, you win in a near-landslide, and you win most of the pink states in addition to the purple ones.
David: We’ll be providing frequent updates to the forecast by Nate, Harry, Clare and anyone else who cares to chime in. Please come back regularly and watch the numbers move.
Clare: