Are Democrats Really Going To Win In Ohio And Wisconsin? 52:37

Testing 1, 2. Testing 1, 2, 3, 4, 5. What did you eat for breakfast? I haven't eaten a thing all day. Are you serious? Yeah. You arrived to the podcast at 1.35 hangry? Yeah, I don't know, man. I don't know. I should have eaten something, shouldn't I? I had food late last night. Do you want some almonds? Uh, no, I'm good. Are you sure? Yeah. Okay, no complaining about, like, being hungry when the podcast takes an hour. Okay, have I complained about being hungry? There you go. Okay. God bless. Okay.

Hello and welcome to the FiveThirtyEight Politics Podcast. I'm Galen Druk. I'm Nathaniel Silver. And this is Model Talk. It's still so nice to do that in person after all these years. So it's Model Talk. Nate, how are you feeling about the model? That's a good model. Yeah? Yeah. I don't think...

My opinion about how good the model is changes that much from week to week. No? Not at all? Not really. You don't have like a temperamental relationship at all? I'm kind of chill about the model at this point, man. I know. Well, this is well established at this point on Model Talk. Yeah. The model's still in Miami. I mean, in August it actually went north for the... Where does the model decamp to? The Hamptons. The Hamptons? Yes. Again, I don't share the purpose. I don't personally like the Hamptons that much, but the model is the model. The model is in the Hamptons.

Why wouldn't the model go join 5E in Vermont? The model and 5E don't really get along. At all? Like they're very different characters? They're entirely different personality types, yeah. How did that happen? I mean, I don't know how 5E was chosen to represent the model, but he's been great as a spokesperson for it, but they're very different. They're not friends. Oh, that's so interesting. Yeah. You know, sometimes rivals can make great companions. They can, right? But the personality types are just kind of very...

Oil and water. Interesting. Does the model vote Republican and 5e votes Democrat? Like is that putting too much of a finger on the scale if we reveal that? 5e is nonpartisan. 5e is nonpartisan? Scrupulously nonpartisan. And the model isn't? Or the model also is? I mean the model is kind of – I don't want to get too much into the model's politics. Yeah. OK. All right. All right. We'll save that for another day. OK.

That was a tangent, but I should be transparent here. We are recording this model talk again a few days early. Folks will notice that this is airing on a Monday, which is, again, a bit rare. I'm on vacation again. Which is not rare. Come on. It's been rare for the past seven years. I just happened to take off two weeks close to each other.

This is what the model currently shows on Wednesday, August 10th. Republicans have a 79% chance of winning the House and Democrats have a 60% chance of keeping the Senate. The last time we recorded a model talk, we remarked on how little changed everything was. That is no longer the case this time. The model has moved quite a bit, particularly when we're talking about the Senate in Democrats' favor.

And we've seen that last week, you know, I'm saying this as if it's Monday, August 15th, last week's special election in Minnesota's first congressional district, the week before that, the Kansas vote on abortion, that it seems like the environment has changed and in Democrats' favor. How would you describe what you're seeing in the national environment today? Yeah, we should say, by the way, I mean,

It's all arbitrary thresholds, but at 60%, which is what the model is at as we're taping this, that's our threshold for what we call lean Democrat instead of toss-up, right? 60 or higher is a lean and not a toss-up. Arbitrary, but for what it's worth, that's one heuristic for when you can say something's changed.

But now clearly in the real world, you're seeing lots of evidence of this. I mean, Kansas was one big sign. And in the special election in Minnesota where the Republican won by only four points, four or five points. I don't know if votes are counted yet in a district that Trump won by double digits. We should say this is historically a district that has been more Democratic in congressional races than presidential races. It is a legacy Democratic district. But still, when the incumbent retires – and it was a Republican incumbent who died actually –

There's no carryover effect from like four elections ago. It's truly shifted and races for Congress. And so I think like that's a bad sign for the GOP and the Kansas thing is a bad sign, too. You know, if you look out kind of turnout in different races, like in Washington state, for example, where you have this blanket nonpartisan primary election.

I mean, everything would point toward the fact that something has changed in the environment. And is that something the Dobbs decision? Because, you know, it's a big country with lots of things happening. We've also seen gas prices fall over the same period of time, basically, as the reaction to the Dobbs decision. How is it? Is it possible to disentangle what has caused the shift in environment? And if so, what do you chalk it up to? It's, uh,

not easily doable. That's always a problem in political analysis is that we're in a new cycle where many things are happening at once. Yeah, I would think Dobbs is the most important component, but the economy is probably 1B where we saw a report coming out today saying that inflation had not worsened anymore over the past month, right? Over the past year, it's still up 8.5% year over year. But the past month,

there was no further growth in inflation, at least economy-wide. I've seen that number in particular spun lots of different ways today, right? People focusing on the July numbers in particular, right? Prices didn't rise from June to July. But if you compare this July to last July, prices are still 8.5% higher. We're not going to dwell on this too much because we should talk about the forecast, but you spent a lot of time digging into economic numbers today.

From your experience, what do you think this means politically? I mean, it's a bit like if you're down 21-0 in a football game and then you score a touchdown and you're down 21-7. It's both true that the recent play went well and that you're still down over the longer term, right? A lot of sports analogy. So if you look at consumer expectations about inflation, which is also done in different surveys, those have declined too. I mean, the good thing about inflation is that

None of this bulls**t matters as much as it does for other things because people see prices like literally...

all the time. You tend to notice when gas increases or decreases, when groceries increase or decrease, if you're planning a summer vacation, how bad are the airfares and the hotels, right? Back to school supplies. You notice all that stuff. And so you are sensitive to prices. And so, you know, people have some awareness that things are bad, but getting better in the survey data. And that's also true in the actual hard data. So I think the spin is kind of not as

important as usual. I mean, the question is like, what does this pretend for going forward? If we have 0% monthly inflation growth going forward, then eventually the year-over-year figure will decline to zero, right? I frankly don't know exactly kind of what the consensus of economic expectations is. I mean, I think the view is that you will see a probably gradual decline, but voters do seem to have noticed to some degree.

Okay, so here's our first listener question, which gets at the changes that we've seen in the forecast and in the national environment overall. I should say for folks who are interested, we have not just the forecast on the website, but you can also look at generic ballot polling averages online.

on the website, which will show that while Republicans used to have a two-point advantage, it's now pulled even with basically 44% of Americans saying that they would support Republicans for Congress and 44% of Americans saying that they would support Democrats for Congress. You can also check out Biden's approval rating, which has ticked up slightly, maybe just two points as well in recent weeks.

But here's the question from Mo. Since June 1, Democrats' probability of holding the Senate increased by 20 percentage points, whereas chances of holding the House increased by only six percentage points. Why the big difference? Have Democrats' chances in the Senate been driven more by individual candidates? There's kind of a tactical answer to this first that's a little bit boring but important, which is that when you're nearer to 50-50,

the odds tend to swing more because you like a bell curve, right? And on the tail of the bell curve, you have to kind of travel a long way for you to get an increase is like the metaphor, right? And so when Dems chances of winning a Senate were so, or let me kind of flip this, right? You could say Democrats virtually had like what, like a 13% chance of winning the house and now it's up to 21%. So it's almost,

50% higher than it was before, right? And in the Senate, it's gone from 40 to 60, so it's also 50% higher. So in that sense, it kind of is more similar than you would think, but the absolute changes are lower when you're kind of on the tail end of the probability curve, right? So that's one thing. The other thing is that there is a lot of more polling in individual Senate races. So in states like

Georgia, the model, or Pennsylvania, right? The model thinks the fundamentals should favor Republicans in those states, at least moderately, but you keep getting polls showing Warnock and Fetterman ahead. And in Ohio, you keep getting polls saying that Tim Ryan is better.

I'm not sure if I believe that, but the model is trying to have an algorithm that we design that is trying to remove my subjective judgment about that. Whereas in the House, polls seem to be few and far between. I don't think you're going to see the rich and robust House polling that we had four years ago. So it's working more off the generic ballot and kind of fundamentals and those have shifted, but there's not the reinforcement from –

So on that point, exactly. Listeners have noticed, as have I, that in the Ohio Senate race in particular, we've seen a divergence between the light version of the model and the deluxe version of the model. And I think your explanation of what the polling shows is the reason, you know, in the light version of the model, we've seen a divergence between the light version of the model and the deluxe version of the model.

version of the model, it says that that race leans Democratic. In the deluxe version, it says likely Republican. And in the polling average, we publish polling averages for individual Senate races as well. Ryan leads Tim Ryan, the Democrat, leads J.D. Vance by four points. What should we make of all, you know, like that's a lot of different data points to try to decipher. What should we make of all that?

Yeah, no, I said polls showing Ryan competitive, and that's kind of a little bit misleading. I'll admit. I mean, he's actually been ahead in most polls. I mean, this is one where, you know, is Ohio still purple enough where candidate quality can make the difference? I mean, maybe. Sherrod Brown was reelected in Ohio in 2018, but that was in an environment that was... Like lean Democrat by six points or something. More than that, right? It was like eight or nine or something, right? And this is an environment that is...

Democrats hope it's neutral. Neutral might be optimistic for Democrats. So it's a heavier lift, but I think you can say that, you know, candidate effects go two ways, right? It's one thing if you have like a bad GOP candidate or a good Democratic candidate. You might have both in this case. You have a inexperienced Republican candidate who kind of, in some sense, embodies, I mean, when J.D. Vance won the nomination, there were like some headline and I don't want to

criticize for Axios or Politico or something like he embodies the best of both worlds, right? He comes from like a rural working class background, but now he's kind of a venture capitalist. And so he has, you know, those ties to and Yale, et cetera, right? Again, anytime someone says the best of both worlds, it could also be the worst of both worlds, right? Maybe he seems like a phony to working class voters, but maybe he doesn't really kind of have like that. I mean, I don't know. I mean, a lot of his campaign has seemed

pretty forced, right? It's like if you took a bunch of people from Yale and elite institutions and beam them down to rural Ohio with all their biases about how rural Ohioans are, I don't know. I don't think his campaign is doing a very good job. He's not raising very much money. And Tim Ryan's a guy who has been representative for many years, comes across as kind of working class, and probably has shared brown-white politics and has been a successful model in Ohio for Democrats. So you have both a

Pretty good Democratic candidate and a pretty bad, I think, Republican candidate. Do they cross enough where that would be sustained through November? I mean the deluxe model, as you guessed, is probably not. Right. So let's talk about what's going on in the deluxe model. It basically assumes, I think, that as we get closer to Election Day –

Republicans' advantage in the generic ballot polling will pick up. You know, polls will switch to a likely voter model instead of a registered voter model. Usually that advantages Republicans. That's already baked into the current forecast. We already adjust for... I haven't looked recently at what that adjustment is. I mean, clearly, like... For sure. But what I'm saying is, yes, in the deluxe model, it's thinking about that. It's thinking a Republican will ultimately win this election because of these, you know, X, Y, Z things, even though the polls right now consistently show Tim Ryan leading. Yeah.

Is there any reason to believe that that won't be the case? Well, I should say with a deluxe model, we're also counting for these expert forecasts, right? And so the experts are skeptical of Tim Ryan's ability to win in Ohio. If you look at the classic model, which is the in-between version that uses polls but hedges them against kind of fundamentals, that's more of a toss-up as Vance, 57% to win Ryan, 43%. So we would call that a toss-up tilting Republican model.

I mean I think there have been all these cases in states like Missouri with Jason Kender and stuff like that and various races like Tennessee, right? I forget the Democratic candidate was in Tennessee. But you will see in these red states where the polls show Democrat winning and it's a mirage and it's based on faulty polling or lack of name recognition for the Republican. I mean Ohio is not as red as those states though.

It's somewhere between red and purple. If, by the way, Democrats, I mean, because some polls show that Biden is losing more ground with minority voters than with white voters, right? If that's true, then a state like Ohio, which is pretty white, might actually be relatively good for Democrats. We got a lot of questions that go something like this.

There's plenty of divergence between the Lite, Classic, and Deluxe models. We've said that the Deluxe model long-term is more accurate. What's the point in publishing all three of these models when, especially in a case like this, it just leaves you somewhat confused about what the forecast even thinks will happen? What is the forecast anyway, Galen? Personally, I think the...

I think the deluxe forecast is cheating, really, right? I think it's like cheating to like use someone else's predictions and merge them with your own. Wait, so why do we even use it? Good question. Should we get rid of it? Should we just decide right now to get rid of it? Well, because our incentives are just to be accurate, right? I feel like people are going to be like angry when they hear this. Like what? You don't even think you should use the deluxe model, but the website automatically takes you to the deluxe model? I mean, if I were betting my own money, I would bet on the deluxe forecast.

And which is probably true in general. If you talk to like sports bettors, they'll say, I have my model, but then I kind of hedge and blend it with other people's stuff or the Vegas lines. And that's why I use the basis for like actually betting. Because anytime you design a model, you're not the only person in the world who has a smart way to forecast something. And the consensus is almost always right.

Better than the best individual forecast So we average all the polls to make a forecast and then we average all the forecasts to make a forecast sort of I mean, we don't it's okay. So you want to get a little bit more technical? We don't actually trust so the experts are quick political inside politics and the Sabato crystal ball We don't actually trust their judgment about the macro environment. This gets a little complicated and

So if they are systematically different in the macro environment, right, if they think it's going to be a really good Republican year and the objective indicators say, no, it's become more purple, they're actually not very good at that. The objective indicators are better. The reason why I think, and again, we like these guys a lot. We're friendly with them. They tend to be lagging when human beings are putting out subjective forecasts. They tend to react too slowly because people are idiots and people are

Think that if you change your mind, you're an idiot. When in fact, if you don't change your mind, you usually wind up being an idiot, right? But your incentives are kind of, you're going to slowly kind of ease into something even when there's been more profound shifts. So they're often lagging about kind of where the environment is, but they're good about pigeonholing individual factors and individual races, right? You can tell when there's a Republican candidate or Democrat candidate that's just completely non-viable some of the time, right? You also have intel in terms of talking to people

on the ground in that state, you may have access to internal polls or campaign strategy understandings that the public wouldn't have, or at least that the model wouldn't have, right? And again,

You can say, well, what does subjectivity add? Well, I think if you were rigorous about your subjectivity, it can add a lot. And these people who make forecasts of every congressional race every year and done it for decades, and it's verifiable in public, right? If you do that, then you can make subjective forecasts and actually get pretty good at that and calibrate those pretty well. So they add a lot of value for individual races. What I find a bit weird, though, is that –

They are supposed to add a lot of information about like candidate quality, right? Saying J.D. Vance maybe looks good on paper, but he's not really selling, right? And they seem to have gone like more macro almost, right? They seem to have been convinced that like that it's all about kind of baseline partisanship and that candidate quality factors don't matter as much. And I'm not

I'm not sure that's right. Well, I think that that's perhaps a lesson that people took away from 2018. We were looking at all of these Senate races that seemed competitive and that at the end of the day, every single state basically voted for Senate the way that you would expect it to for president or what have you, right? Like a lot of these Democrats in reddish territory lost re-election. Would you expect Joe Manchin to win re-election?

I mean that's the rare exception. John Tester? And for that matter, Florida was a bit of a surprise in 2018, right? I mean Florida is more red than blue but it was such a blue environment. You have Democratic incumbent Bill Nelson ran a poor race I guess. People would not have expected Susan Collins to win in 2020 when Joe Biden actually ran very well in Maine in 2020, right? I think people have maybe overcorrected for not taking that candidate quality seriously.

as much. So to this point, let's get to a listener question on this. Jacob asks, I'm curious how the Deluxe incorporates expert race ratings. How many ratings are incorporated? You mentioned what they are, three. Is each expert weighed based on their past results, like pollsters? How much of an influence do they have in the actual Deluxe model? It's just, no, it's just they're all weighted equally. And we translate their ratings

words right into what that means probabilistically or in terms of like a voting margin so what does like lean democratic mean we've looked at historically kind of what that translates to in terms of vote margin all right let's talk a little bit more about how we expect the environment to evolve going forward

You're a podcast listener, and this is a podcast ad. Reach great listeners like yourself with podcast advertising from Lipson Ads. Choose from hundreds of top podcasts offering host endorsements, or run a reproduced ad like this one across thousands of shows to reach your target audience with Lipson Ads. Go to LipsonAds.com now. That's L-I-B-S-Y-N-Ads.com.

I started to bring this up a little bit earlier on, but the expectation of the forecast is that the environment will shift in favor of Republicans as we get closer to Election Day. That's based on history. That's based on what we know about how the likely voter model changes, what the polls will say. We also know that usually the out party gains in generic ballot polling as we get closer to Election Day. The out party is obviously Republicans in this case.

Is there any reason to believe that history won't hold in this case? That, you know, the likely voter model won't show that Republicans are likelier to turn out and Republicans won't get more motivated to vote against the incumbent president's party as we get closer to election day? Sure, there are reasons, right? A couple of reasons. One being that Democratic voters are very concerned about

about the Dobbs decision and also about January 6th, about the possibility of further Supreme Court decisions that go against traditional liberal interests like gay marriage. So Democrats have reasons to turn out that I think are different than usual. Usually when your party is in power, then you get a little complacent. Here, Democrats have seen things that threaten their interests even when they are ostensibly in power because of the power of the Supreme Court, right? It's kind of one bucket.

The other bucket is that it used to be that the GOP had higher turnout in the midterms of things being equal because its voters were older, more affluent, more educated. That's flipped at least a little bit where Democrats are more educated voters on average. More educated people tend to be more politically engaged and turn out more at the midterms. Trump did not really have –

the sort of get out the vote program that republicans had in the kind of carl rove years he relied on his personal brand a lot um but when he's not on the ballot himself to get people to turn out for a midterm for a jd vance or somebody like that might be a much tougher sell and also when trump gets more involved it also motivates democrats to vote further and so um so yeah i'm not sure that you're gonna see the traditional enthusiasm advantage

Based on that, if the generic ballot numbers that we see today, which is basically a neutral environment tied 44-44, were to hold until Election Day, what would the forecast look like? I don't know. I mean, if they were – I mean, because they're also – this gets complicated. They're also factors apart from the generic ballot that the model looks at to make a forecast, right? It's not assuming the generic ballot is the only assessment of the popular vote. If the popular vote is tied on Election Day, then the House is –

50-50, roughly. And in that circumstance, then Democrats would be pretty heavy favorites in the Senate, I would think. Jeremy asked this question a different way, which is, what would the generic ballot need to be today for the House odds to be 50-50? I mean, again, there are different things going on, right? Partly because, like, the generic ballot is an approximation of the House popular vote, but an imperfect one in part because the House popular vote, for example, is

There are not candidates from both parties running in every district, right? There are non-competitive races, and that can lead to lower turnout. We try to model all of that, so it's a bit of a complicated calculation. I mean, for the time being, until we get more polls that directly compare likely voters and registered voters, I think the model is still assuming for now that the GOP gets like two points from the switch to likely voters. So if we have the general ballot tied in all polls, it probably means like a two-point, roughly one-and-a-half-point GOP lead

among likely voters and you assume another point or two of erosion between now and election day for Democrats, right? So like working backward, if Democrats had a lead of three or four on the generic ballot instead of tie, then it might begin to show like the house tied. Can I be totally honest here? Yeah. This conversation almost makes me a little wary, right? Because some of these states that we're talking about, Wisconsin, Ohio, Pennsylvania,

are states where we have seen big polling misses in the past two presidential elections. And so I do, to be totally honest, worry that we're having this conversation about what polls are showing today. They're showing Democrats up in these states that have shown Democrats up, you know, before in two election cycles at least.

Only to have like very big polling misses. I mean, the polling miss in Wisconsin was huge. Right now, I think it shows Mandela Barnes up by two points against Ron Johnson or something like that. The most recent polling that we have from Marquette Law. And it just makes me think, like, should we be talking about these races this way? Should we imbue all of this with a little more skepticism? Like, I know what the answer is, that like we can't predict the polling error and that in 2018, the polls were really accurate and

but I can't help but be wary about this conversation. Yeah. I'm not sure if that's a good instinct or not though, right? Clearly, well, no, right? Yeah, no, I get it. I mean, I get it. You can't know what you don't know. It's a very natural instinct. I'm not sure that, I mean, the prior that polls should not have a predictable bias, I think is a well-founded assumption in part because again, polls are supposed to say why the, why Wisconsin polls so bad last time, right?

I'm going to look under the hood and try to correct any problems that I had last time. I mean, yeah, I don't know. I mean, would it shock me if the polls are wrong in the same direction again? No, it obviously wouldn't shock me, right? But we designed the model with... But then where are we, right? Then people just at a certain point start to say like, okay, fool me once, shame on you. We do assume the polls are error prone. Fool me twice, shame on me. Well, okay. But we're still talking about small sample sizes. I mean, look, I don't know. I think one thing that's interesting is...

is that the same polls that show Democrats doing fairly well on races for Congress show Joe Biden being really unpopular. And so that suggests to me that like the polls are picking up on something and it's that voters are unhappy with the direction of the country but don't think Republicans provide solutions or maybe the reason in part why they're dissatisfied. I mean they're not – don't think Biden is doing a great job although he has had some important wins lately. Yeah.

But like that gap should kind of make you more confident in the polls. People are saying things are going to hell. I don't like Biden, but I'm still voting Democratic because Republicans are unacceptable in terms of the candidates they're offering or the policies they offer or January 6th or Dobbs or whatever else, right? And so it's a sophisticated story we get from polls. It's not just people say, oh, yes, it's a progressive majority if everyone turns out, right? It's not that. It's people are capable of differentiating and –

you know, and seeing some very unappealing Republican candidates and rhetoric and policies. Yeah, I mean, to put a finer point on what you're saying, the Marquette University law poll that I mentioned earlier, the same poll that shows Mandela Barnes leading by two against Ron Johnson. Again, you know, I'm wary about how we talk about that, but so be it. It shows Biden 15 points underwater in Wisconsin. It also shows Tony Evers at about even.

So clearly, yes, people are making a distinction between how they feel about Biden and how they feel about local politicians, which would be sort of diverging from a trend that we've seen in recent decades, which is people feel increasingly about local politics the way that they do about national politics, even if they still haven't totally converged.

This strikes me as different from what we saw in 2010, right? People were mad at Barack Obama. They didn't like, you know, that we were in a recession. They had negative views of him. His approval rating was quite low. And it felt like there was very much a relationship between what they saw out of Washington and the Obama administration and how they voted in the 2010 midterms. What is different here? I mean, I kind of think a lot of things. I mean, so there are, first of all, there have been

where the president's party gains seats. They're few and far between, but the most recent ones are 1998 and 2002. 2002 is normally attributed to September 11th. 98 is attributed to Monica Lewinsky and GOP overreach. And like, in some ways, I'm not sure that this is necessarily true

Different than that, right? That, yes, Democrats were in power, at least to the presidency. They did not control the House after 96, I don't believe, right? They lost it in 94. Right, right. Right. But the GAP was showing kind of how much power it had and how far it would go given the joy it did have in the House. Here you're kind of seeing it through the courts and maybe through state courts.

and maybe through rhetoric and threats in January 6th and extracurricular stuff about denying election results. So the normal kind of power balancing that occurs where people say, Democrats, Obama was elected and I thought he was going to moderate, but instead we got Obamacare and he tried to do cap and trade and blah, blah, blah, right? Both parties have been trying to like make substantial changes. If anything, the Democrats have made more modest changes. The climate bill is...

I'm always worried about these big omnibus bills and how they pull. I wouldn't put a lot of stock in it, right? But like, it's not going to cause big changes to people's bottom lines in the short term. It'll cause more tax audits, which will maybe make people annoyed down the line, right? But like, but it's not something that's like unearthing things as much as like Obamacare did, for example, or major changes to individual tax rates, even, or to welfare programs. And so the Biden administration has done things that are

fairly modest in their ambitions in some ways, which is not to downplay. I mean, climate people think the climate bill is really good. It'll have a big impact, but like it doesn't have the most visible impact. It's not scary to people. Well, some of the most impactful bills probably have been bipartisan actually. So there's no one really campaigning heavily against them. Biden has, I mean, in some ways we kind of all thought that, okay, well he has this 50-50 Senate. Isn't that going to make him impossible for him to do anything when especially two of the centers are Sinema and Manchin, right? And in fact,

It kind of serves as a good filter because to get stuff through, right, it's kind of impossible for Biden to like pass unpopular legislation. His majorities are not big enough. So you're not going to get stuff that could cause a big voter backlash.

I should mention before we forget that in listeners today, which is five days from now, August 15th or later, there is a big primary election tomorrow, Tuesday, August 16th, which will feature Representative Liz Cheney, her attempt to be re-nominated in Wyoming. I think the expectation is that Harriet Hageman will win the Republican nomination in that state. I don't think there's that many people saying otherwise.

But it's going to be in the news when people are listening to this. So let's talk about it. There's also, I should say, a special election in Alaska on Tuesday, August 16th, which features Sarah Palin. What do you have to say about both of those elections? I don't know, man. I'm not paying that much attention to them, really. Really? Yeah. I'm focusing on November, Galen, man. We have other people who are focusing on these primaries. I mean, we should leave people with something. We're going to be...

Yeah, look, I mean, it feels weird. I didn't even realize there was a special election in Alaska on Tuesday. I know where they are, but we have like 35 different people that are blogging about these constantly, right? And I'm just, macro man, she'll want the model. Hamptons. She'll want the model in the Hamptons. Look at you. Look at her, folks. Look at her. It feels weird to talk about a race that will have already happened by the time. No, it's about to happen. If people listen to this on the 15th, the primary is on the 16th.

Well, Wyoming will happen by the time you're hearing this. No, it won't. Wait, you said it was Thursday. Tuesday. It's happening on Tuesday the 16th. So it's happening tomorrow.

It's August 10th. You don't even know what day it is, Nate. Get the hell out of here. I thought you said it was Thursday. And I was like, why was it on a Thursday? No. I think, let's rewind the tape. Guys, this is truly, this is the behind the scenes look at how little Nate is paying attention to these election. You are totally wrong. You confused me because I thought you said Thursday. I want to go to the tape. And see what I said. I think there was a Thursday in there. And so that threw me off. We're in this studio with four other people. Did I say Tuesday or Thursday?

I think I heard it Thursday. Anna says I said Tuesday. Anyway, anyway, what do you got for me? I mean, my sense has been that some of the Trump-backed candidates have underperformed their polls a little bit, but she's down by like 25 points. And so you can underperform a little bit and still not overcome a 25-point deficit. I should caution that primary polling is—

in congressional races as well as presidential primaries is not very accurate, but like, that's a pretty big margin of error to overcome. Alaska will give us kind of meaningful information about November, really. I mean, who holds that seat for the next few months is not that interesting. But they have a new system this year, which is a top four system. So the top four finishers in the general election primary, which I think is in September, will advance the November ballot. And then you have

instant runoff or ranked choice voting thereafter, right? So I think there has been some Alaska polling of the general election showing that the Democrat actually will be maybe ahead after the first round, but then they have two or three GOP candidates. They'll consolidate the second choice vote and then probably win. But like Alaska is like a little bit of a wild card because, you know, we don't know how we used to

What will be the system? We know in New York, for example, that a lot of people didn't complete their ballots, right? They didn't rank candidates one through five, I think it was. They just kind of listed their first choice and made their first two. And then, so there's a lot of wasted vote in the end. I mean, Alaska is like a little bit of a wild card. And tomorrow, from the listener's point of view, will be kind of a dry heat for the somewhat interesting Alaska elections in November. I'm not sure I particularly have a strong feeling about Palin one way or the other. She has been...

you know, performing a little better than you might think, given that she kind of left Alaska not being a very popular figure. People were angry that she quit as governor. She's also someone who might kind of do well on the first choice vote because she has a loyal audience, but she may turn other people off and so you might have someone overtake her on second round. So it'll be good data for how to navigate this new system later on is my main takeaway. There you have it. Let's move on to some rapid fire listener questions.

You're a podcast listener, and this is a podcast ad. Reach great listeners like yourself with podcast advertising from Lipson Ads. Choose from hundreds of top podcasts offering host endorsements, or run a reproduced ad like this one across thousands of shows to reach your target audience with Lipson Ads. Go to LipsonAds.com now. That's L-I-B-S-Y-N-Ads.com.

I love sports. I love them so much, I never want them to stop. But as the playoffs wind down, we get fewer games, and the sports aren't sporting like I want them to. But FanDuel lets me keep the sports going whenever I want. All I have to do is open the app and dream up bets anytime I'm in the mood.

And this summer, FanDuel is hooking up all customers with a boost or a bonus daily. That's right, there's something for everyone, every day, all summer long. So head over to FanDuel.com slash sportsfan and start making the most out of your summer. FanDuel, official sports betting partner of Major League Baseball.

All right, Nate. Some of these questions are not quite straightforward, and I think we might be inclined to give long-form answers. We're going to try to move through, since we took listener questions, we're going to try to move through as many of these as we can. I think we got...

For this model talk, somewhere in the range of like 90 questions. And I'm not going to keep you here till tomorrow. So first question in the rapid fire. Benjamin asks, in general, how much space in the model is allotted for Deus Ex Machina events that are often not predictable but can and do affect public opinion? For example, the Mar-a-Lago raid, October surprises, et cetera. I mean, that's...

Yeah, it's kind of baked into the uncertainty about the model. I mean, for the presidential model, we make more sophisticated sets of assumptions about kind of when there's more uncertainty or when there isn't. We don't have that like uncertainty index in the congressional model. But yeah, I mean, you know, if you did, by principle, you might assume there's more uncertainty than usual because there's a lot going on in the news cycle. And so that could maybe make the model, maybe that makes the model a little bit overconfident. I mean, I don't know. We build...

Well, how much uncertainty is there in the model? There's a lot of uncertainty in the model, man. There's a lot of them. From zero to 10, how much uncertainty is there? 8.8. 8.8? No, I mean, look, in general, when people build models, they underplay the uncertainty for various technical and non-technical reasons, right? I think we're pretty kind of like conscientious of not doing that. Maybe you can argue that we go...

too far, right? There's like this, I heard Larry Summers doing an interview the other day, the kind of economist who is a little bit overweight, right? And he's like, you know... I hope this has to do with the story. No, but he's like, yeah, it's like sometimes things directionally don't worry about the direct, right? It's like, you know, Larry, you should lose weight. He's telling himself that would be good, right? And yes, there is at some point at which you lose so much weight,

That you're malnourished, but he's probably never going to get to that point, right? It's like people err so much on the side of underestimating uncertainty that like, yes, it certainly is possible that you have a model that's underconfident and that overestimates uncertainty. But it's just not the mistake that people make very often, right? And so I'm not that worried if we're like underconfident, right?

And I think this is like a year where you do not want to be particularly overconfident. There are lots of news stories that are very fluid and dynamic. You could have a new COVID variant, for example. You could have Trump charged with crimes, right? You could have lots of different things happen. You have an elderly president who could get ill. You have lots of things that could happen. And so it's – I think the uncertain is probably pretty appropriate. Yeah.

All right. Raid asks, what are the odds of me, a random American who doesn't answer the phone if it's an unfamiliar number being included in a poll? Are you asking internet polls? If not, then you're not. If you're not asking polls, then the odds are zero. Yeah, I was thinking about this. Like the odds because we don't only do telephone polls now are not zero. But I don't know. How do you end up in an environment where you will be polled online?

Some of them are opt-in that you sign up for emails and then they will periodically choose to contact you. You could also get a letter in the mail that will invite you to join a poll. I think people have tried that method. Some states have emails associated with voter registrations. People will email you, cold email you.

At random, that's kind of closer to like a random digit dial almost. So the answer is not zero, but I'm not going to try to come to a specific answer on that question. Pavel asks, when the 538 midterm model was published, prediction markets prices moved towards the model forecasts. Seems that the Scottish teens, haha, inside joke, that's what we refer to the betters that predict it. That's how we refer to them.

Seems like the Scottish teens are reading 538. Would you comment on that? Do you see any notable patterns for the House versus Senate models, for example? I don't know exactly what Pablo's talking about there at the end. But basically, the question is, hey, what do you think about this? A couple of things. I mean, I get a lot of iteration of the question, like, you know, how do you compare betting markets to forecasts that you might put out, right? And the answer is that

For politics, so there's a contrast between politics and sports, right? In sports, or like the stock market, you have a very mature industry full of people who spend every waking hour trying to make the best forecast that they can, right? So when you have an individual model...

Competing against the consensus, it usually struggles to do better than the Vegas lines, right? Maybe it breaks even after the rig or the vig that the House takes, the House cut. Exceptions tend to be events with a lot of public interest. And so the World Cup, the Super Bowl, a big –

UFC fight, a big boxing fight. Those are cases where there's so much public interest that the market may not kind of correct to equilibrium example, but like an ordinary- Maybe because there's so many amateurs getting involved. Yeah. That there's so much dumb money that like, yes, in theory, you could like capitalize things well enough, right? That's not true in politics. You, there are not

elections every day like there are baseball games every day, right? It's frankly a very specialized skill set that like a handful of people like literally across the world have tried to do on their own, right? So very few people are like have really taken the effort necessary to build a proper politics model and there are lots of people with very dumb opinions about

Right. Lots of people who kind of overrate their ability. And so so the politics market is lots and lots of dumb money. And therefore, when you have like a forecast that is smart, like you would just kind of probably trust that before your own in 538 does have like a does have like a historical track record where like directionally we've kind of done very well against the markets historically. Right. At the same time, it has to be people trying to anticipate what the market price is. Right. And so, yeah.

If you know the 538 model comes out, then even if you think it's bullshit, it's probably going to influence the market, right? So that will shift the price kind of just because it like is an anchor. But yeah. Have you ever bet on a political event? My interpretation would be that this would be unethical given that we have market moving power and I think would be frowned upon by my employer.

If I did bet minus, it would be kind of strange though, right? I would probably want to hedge my risk, right? And bet against the model because I do think that like I probably have a reasonably large change in future income based on my forecast being correct. So I'd want to hedge that actually. So sneaky, Nate. It's not sneaky at all. No, I know, I know, I know. Yeah. Okay. Let's get into maybe a little bit wonkier questions here. Which toss-up Senate candidates does Nate think he could beat in poker?

All of them, I guess, is maybe the answer. Yeah, which candidates would be good at poker? I'm trying to... JD Vance might be good at poker. You know what? Usually when you see a VC or hedge fund person at the poker table... You're like, oh, I'm going to take all their money. Pumped about that. Them and doctors and film producers, right? Those are like the... Wait, what? I'm just saying empirically, when you get the hedge fund guy, it's someone who has a lot of money...

I'm friends with people in finance, so this is not meant to be disparaging, right? They have a lot of money, a lot of self-confidence, a lot of mathematically adjacent skills, but like, if you don't actually know how to apply those to poker, you're gonna get your ass kicked, right? And so that's the kind of Dunning-Kriger, what's it called? The effect where people overestimate their own ability, which is actually people debate whether it's a real thing or not, right? But like, they overestimate their skill set relative to their actual skill in poker, and that's

That's someone you want at your table. Is there someone you don't think you could be at poker? Not – so let's broaden the field, not just toss up races. Anyone in federal government in an elected position. So the House, the Senate, the White House. There's a whole tradition of presidents. Obama, I'm sure, was pretty good at poker, right? Nixon, I'm sure, was good at poker. I don't know like kind of which members of Congress are – who thinks well strategically? I don't know. I mean I bet –

I bet Mitch McConnell would be pretty good at poker if he applied himself. I don't know. Do you think he'd know? But like you said, you might be strategic and you might have mathematical skills, but you have to know how to apply them to poker. I'm just looking at the list of candidates here. I mean, no one strikes me as being self-evidently awesome at poker. John Fetterman, maybe? I don't know. Next question from Zach. Is the correct phrase the data is or the data are? Uh,

The data is. I thought this was. So according to FiveThirtyEight's copy desk, we now consider data to be singular. Yeah. I think there are still a lot of people on the internet who consider data to be plural. I know. It sounds pretentious, though. I just have to be honest. The data are. It probably is like a little better grammatically, but like it's kind of pretentious. You heard it here, folks. It's kind of pretentious when you say the data are. You know, speak generously. You got to bring people along when you try to communicate with them.

So where should we leave things? We'll leave it on an actual practical question. When do we start getting more polls? It's a desert out here. After Labor Day? After Labor Day is the traditional answer. I don't think we're going to get the volume of polling that we had in 2018. I mean, first of all, in 2018, there was a lot of focus on individual House races, which required more polling. I mean, I think we'll get decent amounts of polling in the

six or seven or eight most competitive Senate races. But, you know, polling is getting more expensive. News agencies are more concerned about polls being wrong and therefore polling back in some respects, right? The news industry itself has been in kind of a recession for forever, basically, right? Well, since the 90s. And so I don't know that you're going to have, I think you're going to, you're seeing probably a secular decline in the volume of polling that may become apparent. You know, one thing that helps a little bit is that

Most of the key races are in big traditional purple states that often have local universities or often have a lot of polling, right? You'd rather that than have a bunch of races where you need to rely on polls in Alaska or West Virginia or Maine or Tennessee or something, right? And so these are states that people are used to polling and will get –

And we'll get some polling there more than if you had a bunch of like, you know, whereas like when Joe Manchin's up for reelect in two years, like that'll be a nightmare trying to get good polls of West Virginia because there aren't a lot of competitive races in West Virginia. Here, North Carolina, Georgia, Wisconsin, although polls have been terrible in Wisconsin, Nevada. These are states that pollsters are used to handling. All right. Well, any parting thoughts other than those?

I think I should have something witty to say, but I don't know. Yeah. I'm on. I'm already ready for vacation. I don't have anything more. No. What I should say is we've already recorded one podcast because we recorded the primary reaction podcast this morning. I was up late last night for the actual. If you're in Vermont and see 5E, say hi. If you're in the Hamptons and see the model. Also say, or if you're.

on Far Island and you see me, you can also say hi. Okay, yeah. Let's leave it there. Thank you, Nate. Thank you, Galen. My name is Galen Druke. Anna Rothschild, Kevin Ryder, Michael Tabb, and Sophia Leibowitz are in the control room. Sophia Leibowitz is also on video editing. Chadwick Matlin is our editorial director and Emily Vanesky is our intern.

You can get in touch by emailing us at podcast.538.com. You can also, of course, tweet at us with any questions or comments. If you're a fan of the show, leave us a rating or review in the Apple Podcast Store or tell someone about us. And if you're watching this on YouTube, make sure to subscribe to the FiveThirtyEight YouTube channel. Thanks for listening, and we will see you soon.