Chapters
Shownotes Transcript
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.
Have you heard of the musical Titanic? No. Have you seen the movie The Titanic? Yeah. Do you know who Celine Dion is? Of course. What do you mean? I'm just making sure. I don't want to assume we have the same cultural knowledge. Oh my God. So there's a musical called Titanic the Musical in which Celine Dion hijacks the story of the Titanic. Yeah.
That sounds good. Yeah, I went to see it. Is it like camp? Oh, it's amazing. I mean, they literally have moments in the show where they're like, we know that everyone here is gay. But it was National Coming Out Day. I feel like I haven't been very good at celebrating the Jewish holidays this year, so I might as well be good at celebrating the gay holidays. Are there multiple gay holidays? There's like Pride. National Coming Out Day. I mean, all holidays are kind of gay, if we're being honest. Is there a holiday that's decidedly not gay?
Thanksgiving isn't super gay. Thanksgiving? Cornucopias are gay. Hello and welcome to the FiveThirtyEight Politics Podcast. I'm Galen Druk. I'm Nate Silver. And this is Model Talk. There we go. We're all synced up and in studio. We have a little less than a month until Election Day. How are you feeling? Just planning my November itinerary. Yeah? Yeah. Where are you going to be?
Little Florida, probably. Like late November or on election day? I mean, no, I'll be here in the ABC News building, hopefully on TV on election day. No, no, no, no, no, no. There's like some good, there's like this sequence where you got like a poker tournament in Florida and then Art Basel, right? Oh my God, look at you. I don't even know if we can be friends anymore. You're really like outdoing me in every way.
On the topic of one month until election day, how accurate are the polls one month before election day, the midterm polls? Or predictive, I should say. I don't know if I have a good soundbite for what the exact number is. I mean, let's take a look at an example race, though. In Florida, the Democrat Val Demings trails incumbent Republican Marco Rubio by six percentage points. So let's see how that translates in the polls-only forecast into an actual probability.
The fundamentals think this rate should not be that close. In the polls only forecast, Val Demings has a 20% chance. So you're down by five points. You have a 20% chance. That means a margin of error, which is supposed to span...
95% of outcomes is actually much larger than five points, right? So the polls are still not that precise. And they never will be that precise, right? I mean, average polling error in national elections is four points. Is that how we should think about average polling error? The average polling error in the last three weeks of elections has been larger than that. Isn't it around like five points? It's also different for presidential elections and midterm elections, right? It's different for...
presidential elections and congressional races. Generally, the further you go down ballot, the more error there is in polling. Although recently, the presidential races have sucked so bad that they're starting to catch up. All right. So I should say before we go any further that it is October 12th. Democrats have a 66% chance of keeping control of the Senate. Republicans have a 71% chance of winning the House. It looks like the long days of inverse odds are over.
for now. I think these things can at times be noisy, but there's been about a five percentage point decrease in Democrats' odds in the Senate over the past month. Is that enough to say that we're seeing a trend in favor of Republicans now? Because I know you were somewhat hesitant. You were saying like, oh, it seems like the news cycle might be changing, but it's too soon to really say that the winds have shifted in terms of
I don't know. I mean, there's complicated things going on here. Right. So let me make the following two assertions. Number one, I think things have gotten a little better for Republicans. Number two, I think the narrative has shifted more than is merited by the evidence. Maybe there's more than two things. Number three, to some extent, this is a tightening that like our deluxe model expected. It's sort of priced in.
Wait, but then shouldn't the model not move very much? Well, it hasn't moved very much. Like 71 versus 66 is not really much of a shift in the House. So it already thought the polls would tighten, so therefore the odds aren't going to change much. Well, yeah. And okay, number four, this is not a presidential race. I'm looking directly at the camera. There are different candidates on the ballot in 35 Senate races and 435 congressional House races, I should say, right?
If there is some news development in Georgia or Pennsylvania, which are two newsy races lately, right? It does not necessarily affect the race in Arizona or Colorado or Oregon or whatever else. So you don't see nearly as much movement in congressional elections as you do in presidential elections where you have, number one, the same candidates on the ballot in every state. So everyone's on the same focal point. And number two, every presidential poll is informative of the state of that race. If Trump gains...
three points in a Georgia poll and a Texas poll, it probably tells you something about how he's doing in Florida too. That is not a safe assumption to make in a congressional race. So you're not going to have the kind of wham bam dramatic movement that you have in presidential races. About the most dramatic thing we saw was the Dobbs decision. Even that took a couple of months to get priced into our model. Maybe it should have been bigger, right? Or quicker. That's the kind of
model building question. People's senses are miscalibrated. So in terms of understanding how the forecast works, and especially we haven't talked as much about the House because the Senate is easier to wrap your head around, right? There are four or five races that are really going to decide control of the chamber, whereas in the House, there's approximately 50 districts that are somewhere in the range of competitiveness. And so when folks, I've seen this because we've gotten questions from listeners, folks will go and look at their own district and
and say, I don't really understand why the forecast is quite like this. I don't understand what polling they're using, etc. For the House district forecasts, say in Pennsylvania, for example, Pennsylvania's 7th district is competitive. When a Senate race poll comes out in Pennsylvania, because there are a lot more of those than any House races in Pennsylvania,
Does it also affect the forecast in House districts? No. In principle, it might be fine. We haven't tested that. So it's not, that's not the name. So how are we getting our sense of how candidates are doing in House races across the country? Because in the same way that in the Senate race, there's a different candidate in every state, in the House race, there's a different candidate in every single district, and we're not getting a lot of House polling. I mean, mostly you're using the fundamentals, right? So we have a partisan lean index.
So 538 PLI with a generic ballot and add a couple of points for an incumbent and evaluate fundraising to some degree. Right. And that's that's how it's mostly doing it. The polls in the House are few and far between. A lot of the polling in House races is partisan anyway. Right. So we're relying on on tools apart from district polls. They're not as many in 2018. The upshot did a ton of district polls, which did very well.
Unfortunately, the New York Times is not giving us all this free data anymore. Damn you, New York Times. And so we're left a little bit more to forecast based on kind of fundamentals. Speaking of polls and accuracy in the final month of an election,
Real clear politics, maybe their competitor of ours, has decided to do something. Frenemy. Frenemy has decided to do something a little unique this year, which is for their Senate races, they are calculating the average error this far out in a particular state.
and then showing if you apply that average polling error to the current polling average, what the results would be in that state. So to give folks some sort of example, currently in their polling average in Pennsylvania, John Fetterman leads Mehmet Oz by 3.7 points. Fetterman's doing a little bit better in our polling average. It's more like six points in our polling average.
But it will tell you that polls between 2016 and 2020 underestimated the GOP in the state by six points on average across those three elections. And it therefore, sort of in their mind, corrects the polling average to show that Oz is leading by two and a half points. We have a long history of talking about unskewing the polls on this podcast. And for folks who are maybe new or haven't been following us,
all the way back in 2016 when unskewing the polls became a really popular thing. I think it becomes a popular thing in basically every election. I know that you don't like this, but can we talk a little bit about why without just throwing RCP under the bus immediately? A few things. Number one, it's kind of faux empirical in the sense of if you actually look at is the polling bias in past elections predictive of polling bias in future elections? The answer is
historically is basically no. Or at the very least, you'd want to regress it heavily. Maybe there's some positive correlation. So a six point bias in past elections translates to one points, right? So that's part of it. It's just like, I think it's kind of like, I think it looks like it's being rigorous when it's just kind of not particularly. I mean, it's an interesting thing to look at, right? Number two, I think they're kind of trying to like cover their asses and have it both ways a little bit. And hey, we have three forecasts, right? So you can say we're covering our asses too. But I think there's like,
It's hard to do this. It takes a lot of thought. I just find it very kind of ad hoc. I mean, in general, and that's the third complaint, like RCP, I like some of the people there are a worthy site. I used to go on the site every day and but like their whole process is ad hoc and it has been for years. And I think it's more likely that a process.
is biased when it's ad hoc. We can debate what the term bias means. It's hard to avoid bias. But the reason that you want to have a process that's rigorous and well-defined is because you don't want to have to make judgment calls. What about this is ad hoc? Because it seems from the outside... Their standards for including polls are... There's no standard, right? And it's frankly biased toward tending to
include dubious conservative polls, but not include dubious liberal polls. So that is a critique of their polling average in general. But when it comes to this sort of unskewing feature that they have that will apply an average polling error over three elections to the average this time,
Is that ad hoc? Because to me, I look at this and I'm like, oh, this is interesting information. I know the caveats. I know that polling error isn't predictable, although we are in this sort of period where people are beginning to question whether or not that's still the case.
And to me, this is interesting information. So the New York Times will publish a version of, so it's also stealing the idea from the New York Times, but they would publish a thing where it's like, if the polls were biased in the same direction, what would it look like? This is like the official RCP projection, right? If you go to their Senate map, they have Fetterman losing in Pennsylvania, not based on the current polls, not based on some prior, not based on even some subjective forecast, but based on the fact that
that they are unskewing the polls. It's unskewing the polls, right? That's their official forecast and that's whatever. That's fine. I mean, we could bet on this, right? And would you bet against them? I would bet on the Democrats at 50-50. I mean, they have 52 Republican Senate seats. I'd bet on that. Yeah, for sure. I'm not allowed to. Yeah, yeah, yeah. And I have enough reputational risk on the line that wouldn't make sense to bet, but it's kind of bullshit. Okay, so bad use of polling? Yeah.
It's a non-use of... Well, okay, I guess it's use of bowling. No, it's like, it's just kind of... I don't know, man.
I don't know, man. It's unskewing. We got a bunch of hot topics today. Ready for the next one? Yes. This has been a long time coming. We're going to talk about the Oregon governor's race. Oh, yeah. States along the way. Can I bring you back to the first model talk we did this cycle when I cited the odds in some of the governor's races and I mentioned Oregon and you were like, the fuck are you doing mentioning Oregon? Who cares about that? And I was like, hey, look.
It's looking like it could be competitive. And you're like, yeah, whatever. I didn't say that. I said I don't care about Oregon. I mean, Oregon's a lovely state. I have some family in Portland. I'm looking at our model right now, and it shows that Oregon's election for governor is a 50-50 toss-up between Christine Drazen and Tina Kotek. There's also an independent candidate running, which I think is why our forecast shows this as a 50-50 shot.
So the independent is a Democrat turned independent who is running on sort of this idea that Democrats have lost their way on law and order and things like that, crime and homelessness in the state and so on.
Are we factoring in the possibility that an independent could win in Oregon? Or what are these 50-50 odds? Because it looks like there's no odds. Yeah, we're giving her an explicit probability, which we estimate is less than one chance in 100. It probably is 0.5% or something, right? She's down by 18 points. Usually, independents tend to fade down the stretch. We've actually done a lot of modeling on independents. So this is like not...
This is something we put a lot of effort into, and it's pretty improbable for her to win. She has to make up 18 points against two candidates, which is harder than doing it against one, which would be hard enough. And again, people tend to abandon...
abandon independence down the stretch if it looks like they're not a viable candidate. Okay, but then that would make me think maybe it's not a 50-50 proposition because if people are going to abandon this independent and this independent seems to be pulling more support from Democrats in a state like Oregon,
Wouldn't we expect the Democrat to rise? Yeah, that would be a good reason to expect the Democrat to rise. Let's talk about Oregon, right? Yeah, let's talk about Oregon. There's also a bunch of competitive house races there too. I mean, Oregon is a politically interesting state this cycle, regardless of how blue it may seem. It's kind of WWC. You know what that means? Yeah, white working class. I've been here for the past seven years. Yeah, pretty fucking white Oregon and not that wealthy. Although obviously you have like very liberal culture in Portland, but like Oregon is not
inherently that different than Wisconsin. Okay, how about this comparison? Is Oregon to Republicans what Montana is to Democrats? Well, it's a counter... In some ways, Montana's become... Actually, it hasn't become more... I mean, there are parts of... I would want to check migration patterns right in the West. It's interesting. It's a region where people tend to move around more than on the East Coast, for example. It's a frontier. Are Californian conservatives...
who are tired of the state's high taxes and Gavin Newsom moving to Oregon, there's been a lot of outmigration from California. That could be a factor in some of these races, potentially. Potentially. Although, did you see the article on our website this week? Many Americans say they want to relocate for political reasons. Few actually do. I know we have some data from the pandemic showing that people were moving. Can I just make a comment about Gavin Newsom for being spicy, right? Oh, God. All right. Go for it. How many people left California? Like 600,000 people left California or something? Uh-huh.
That's a really bad argument if you're trying to be president, right? People like literally left your state. I don't think anyone other than Twitter thinks that Gavin Newsom, Twitter and Gavin Newsom think he might be president, but literally no one else. On prediction markets, he is the second or third most likely Democratic nominee. Well, it's because people who bet on prediction markets spend too much time on Twitter. I don't even think they're necessarily, because he kind of caters to like
He's the third most likely nominee after Biden and Harris. And not even that far behind. 9.4%. So look, I think Newsom. And then Pete Buttigieg at 5.2%. I think Newsom has been smart as far as saying, f*** it.
I'm going to like defy convention and kind of start running a not so subtle at all 2024 campaign on the hopes that Biden doesn't run. Right now, he's doing that in a way that's talking about issues like crime and immigration that Raphael Warnock sure as doesn't want to talk about. Right. But in a cynical way, in part because Newsom has like these fanboys among the resistance group.
It's kind of working. So, I mean, we're getting ahead of ourselves. It's not even, we haven't even put 2022 in the books yet. We're already talking about 2024. I promise you the week after 2022, we're going to do our first, no, our second 2024 primary draft. But Gavin Newsom is running for 2024.
de facto in 2022 in a way that at the margin probably hurts Democrats right so it's an issue pertinent to 20 I mean at the margin sure but like there's also a Republican who's running for president fair enough when he's trying to like extend like the Ron DeSantis Martha's Vineyard news cycle by like let's have a national debate it's like you know talk to any Democrat like running in a swing state or swing district they probably don't think that's helpful
I'm going to point that out because I don't see people pointing that out. And it's like, it's kind of weird to run. I mean,
You're making the argument that politicians are kind of selfish, which I think a lot of people point out. Right. But you want deterrence. I mean, you know, they're more selfish if they don't get called out for selfishness. We covered this in our podcast series, The Primaries Project, which is that our system of government is designed to, our system of parties more specifically, is designed to encourage individual politicians to outshine their own party.
And that's what people do. And in like a parliamentary system, in a system where parties have a lot more control and power and are able to do a lot more work in terms of selecting the kinds of candidates who run for office, people are more invested in the party and furthering the party's goals. But it's never really been the case that if you're an ambitious politician in America, you have to stick to the party line. Oftentimes the most ambitious, successful people sort of run in ways against their own party. That's fair. I'm just saying like,
Gavin Newsom is running for 2024. He may or may not run in 2024, but he is running a national campaign trying to rule out his national standing at a time when a Democrat is president and a Democrat from his state is vice president. If people only acted in the interest of their own party, we would have... Well, one, politics would be different. I'm just saying people are... And the situation...
in particular amongst republicans in this midterm cycle would be very different i mean the only reason that democrats republicans and neutral political reporters should like understand that gavin newsom is running for 2024 all right let's move on we kind of started talking about oregon but we really got sidetracked there and talked about california i think you just wanted to talk about california do we have more to say about oregon
Great state, but it wouldn't be that surprising if it became a little bit more purple. If it became a little more purple, how are you thinking about our 50-50 odds at this point for Oregon governor? I think the thing you mentioned earlier where the fact that this candidate is left-leaning the independent candidate and is more likely than not to lose ground, I think in the past we've actually tried to model this and actually tried to predict if the independent candidate loses ground, how does that affect the rest of the race? I think we're not doing that in the current version of the model.
But yeah, that's probably a reason to lean a little Democrat there. But it's close, right? I don't see any... Also, governor's races often are close. I mean, it's becoming less common to have tickets splitting, but like... Yeah. You know, you've had Democratic governors in Kentucky and Louisiana and Republicans in Massachusetts and Maryland. And again, it's become less common, but like...
A Democrat that's seen as too far left in Oregon, that's not crazy to think that, like, in a gubernatorial race, in a way that wouldn't play out in a Senate race, that you'd have an issue there. So what you said there is quantifiable. And we actually have, I think, an article up on the website this week talking about how the ticket splitting between governor and senator has declined significantly.
Over the years, we can see that trend in the data. This cycle, it does seem like there are some pretty high profile splits. I think maybe the highest profile is in Georgia, but we're also seeing splits in Arizona, Pennsylvania. New Hampshire is a big one. New Hampshire is a big one to some extent in Wisconsin. There are the weird places, of course, where like there's a Democrat defending the governor's mansion in Kansas, but we would never expect a Democrat.
Democrat to win a Senate race in Kansas, I don't think. So is this out of the ordinary or defying trends, or is this still what we would expect? Ticket splitting has declined. Is it still declining? We will collect a data point this year. All right. Fair enough. I mean, there can be a bit more ticket splitting in midterms because, like, you know, there are more people who turn out for president and for other races. They may just turn out for president. Like, when you vote in New York, let's be honest, there's a bunch of f***ing
you have to fill out right it's like I don't give a shit about these judges or whatever I'm sorry I'm terrible I'm I feel like very uncivic for saying that but like I feel like I don't want I think more importantly you can't really vote because they're not competitive so I have a long running history of voting for people like
Honey Boo Boo, Mama June. I voted for all kinds. I probably voted for you for one write-in position at some point in time. I appreciate that. But I just don't. I always write in an uncompetitive race. And then like the machine boops. Do you know you have an undervote? It's like, yeah, I want to waste my time filling it out. Wait, the machine boops? In some states, the machine boops at you if you undervote.
Oh, I'm going to pledge from now on. I'm not going to vote direct to camera. I'm not going to vote from now on for any offices that haven't spent time actually understanding who the candidates are. I'm just going to undervote. And if the machine doesn't like it, well, that's not my problem.
That's such rebel behavior, Nate. That makes you like a lot of Americans? No, people don't undervote enough. You think people spend too much time voting in races? Because that's how politics and parties work. Understood. Parties combine a whole bunch of different culture, policy, information so that you can shorthand everything. Because most Americans don't give a shit about all the stuff that we're talking about. They're like too consumed with their daily lives, but they know that they prefer one party in general over the other.
I mean, but it leads to some outcomes, right? I mean, it leads to things like people just fill in the names they like. So in Chicago, like if you have like a Irish name or like a
Polish name. But that's also how politics works, right? Because a lot of it is culture and identity anyway. Let the people who take time to research local offices vote for those offices. And I'm here to say that if you want to use shorthand and partisan cues to vote for a Republican or Democrat, that doesn't make you in any way unique. And in fact, it probably means that you have a more normal life because you're focused on the things that matter, like family, community,
Faith, love, television, children. It takes us time to fill in all the ovals. Going to Titanic the musical. The ovals take a while. That's it with my questions for today. So next you're going to have to do battle with our listeners and I will try to defend their questions as best I can. So let's get to those 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.
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.
All right, Nate, are you ready for some listener questions? I have to listen to them first. Our first question comes from someone who just identifies as a Virginia contrarian. Okay. The question is...
How does the model factor in undecided voters? Does it assume that undecideds are likely to vote GOP based on historical midterm trends? Then someone follows up on this question because I posted it on Twitter so people can respond to each other. And they say, I would like to second this question. Candidates are often effectively tied in many polls, but with a significant percentage of undecideds. Many close races don't necessarily end up that way because of undecideds hurting behavior.
Is the latter a variable in your model? So a couple of things. When there are more undecideds, there's more uncertainty in the forecast. That's a pretty important principle. Does the model make assumptions about which way undecideds break? Not the light version, which is just poll-based, but the fundamentals versions do, right? I mean, I think if it's a really red state and the race is tied, but the forecast is the Republican wins by five,
That's kind of saying that they expect the undecideds to vote to break GOP. Okay. So it depends on the circumstance is what the answer is. I mean, the default is that you shouldn't try to predict how undecideds break, right? But if you have other information apart from
you kind of implicitly are. Is this what people talk about when they talk about momentum sometimes? Because do you remember in 2016 when part of the issue was that a lot of elite deciders broke for Trump in the final week or so of the campaign after the Comey letter? Yeah. I know what to do with momentum, though, per se. But when news events happen late in the campaign, it can create momentum.
a hurting effect amongst undecideds. I know we hate the word momentum. I guess I'm trying, I guess I'm just picking fights with you. I'm trying to come up with a defense for momentum. No, I think that's not quite the right way. I mean, so where there can be momentum in that sense is in primary races, right? Where like the Cox example I give is that if you're someone who preferred Elizabeth Warren to Bernie Sanders, but prefers both of them to like Joe Biden, if you begin to see Warren declining in the polls,
and Sanders rising, then you might say, I want to use my vote well, not for the candidate in third place. So I'm switching to Sanders. Then that becomes self-perpetuating because the next person also sees Sanders gaining. So you can have those weird effects in the primaries. In the general election, I think momentum is a term that you would probably be better off if you never used. All right. Ban momentum. Hashtag ban momentum. Yeah.
The next question is one that we've gotten a decent amount, but now that we're in the final month of the election, we'll come back to it. So basically, the way the forecast works is that generic ballot polling currently shows a one point advantage for Democrats. But the forecast still thinks that this is ultimately going to end up at Republicans having a two point six point advantage in sort of like the national vote. And so the question is, how quickly will the forecast sort of correct to the current polling in these final four weeks?
if it doesn't get to our plus 2.6. So let's talk about the reasons for the 2.6, right? One reason for that gap is that Democrats have more abandoned districts where they're not running a candidate at all. That accounts for like something like one or 1.5 points of that gap. Some of the gap is also the generic ballot we show on the site. Confusingly, we should probably fix this, is not the same as the one the model uses because the model built on a likely voter adjustment. What?
And the third reason is this gap between the prior and the current. So, yeah, I mean, if there is like one point of perceived shift for the GOP, the model kind of tacks that on to the forecast for individual races. This seems like a slightly different question, which is like, does the polling average get more aggressive about adjusting new information later on in the race? The answer is, yeah, it does, right? A three-point shift difference.
Later on in the race, it's more likely to be fully priced in. Also, you have more polls at the end of the race, right? I think kind of what this person is asking in a way is if the election were tomorrow and Democrats had a one-point lead in the generic ballot average, what would the forecast say? Galen, do you know what happened to the Nowcast? Yeah, you murdered it. Okay, yeah.
There's a reason why the Nowcast is dead and this person's asking for a Nowcast and the Nowcast... I mean, but this is Model Talk. We can do whatever we want. Oh, no. We can be whoever we want on Model Talk. We can't revive the dead. We can be the Nowcast. We can't revive the dead. When's Halloween? Halloween's only two weeks away. Maybe on Halloween. For our Halloween episode of Model Talk, can we revive the Nowcast? Someone should dress as the Nowcast. Yeah.
Please, listeners, dress as the Nowcast for Halloween and send us a photo. And if you do that, we will, I don't know, we'll invite you on stage at our live show. It's like kind of a Nick Nolte vibe, I think. What? Am I dating myself? Maybe. Isn't he like always getting in trouble for being extremely, yeah. For dressing up as the Nowcast? No, this is like the Nowcast looks like Nick Nolte on a bender.
I don't know how rude this is because I don't honestly know who he is. Oh, my God. Well, who is he? He's an actor. You don't know who Nick Nolte is? I know, but what does he do? He was famous back in the 90s. Jesus. For what? For being in movies. He's an actor. Which movies? Okay, if he's so famous, name one movie that he's been in. Well, I think he's actually not been in movies lately because he's kind of... Okay, so he was in filmography. Do you know who this person is? Sophia doesn't have this person. Oh, my God.
Cape Fear, The Prince of Tides, Lorenzo's Oil. This is truly a who's who of movies I have never heard of in my life. Mulholland Falls, The Thin Red Line. You've heard of that, right? No. I probably got some...
Oscar consideration? What? Hotel Rwanda. Okay, yes, I have seen that. Huxley on Huxley. No. I'm not convinced that this is a famous person. But I did see the picture that you pulled up when you Googled Nick Nolte on a bender. Blue chips. So I understand the reference that you're making. 48 hours.
North Dallas 40. And I respect that impulse, that if someone is going to dress up as the now cast, they may well be on a bender. Next question. The now cast lives in the moment, Galen. Yeah. YOLO. YOLO. Next question.
Question. Is enthusiasm among voters factored into the model somehow? This is the point in elections where we get just a lot of questions that are like, is this in the model? So yeah, is enthusiasm amongst voters factored into the model? More specifically, they say, if polls show voters from one party are more enthusiastic to vote than voters from another party, does a model factor this in as an indicator of higher turnout from one party's voters? So the way the model works is it looks, it compares likely voter or registered voter polls. And if there's a gap,
then it adjusts toward that gap. Also, it uses historical priors about the fact that the out party, meaning Republicans in this case, are usually more enthusiastic in the midterms. But it's a pollster's job to factor in enthusiasm, right? And you can also over-factor it. And one issue that polls sometimes have is that people who are more enthusiastic about voting are also more enthusiastic about responding to polls. I think Nate Cohn of the New York Times pointed out today that only 0.5% of
Their calls are being completed by the person they want to actually answer the phone. So probably someone who's really jazzed about politics is more likely to answer a poll. Have you ever been polled? Sure, yeah. In what race? It was some poll. It was not in a race. It was some poll about teenage substance use. I was 18. They're like, can we speak to the 18-year-old in the house? And I beep-booped the answers. Boop-boop. Touch tone. They're like...
Hello, 18-year-old Nate Silver. Do you smoke marijuana? And you're like, sure do. No. No. No? I'm not going to answer that question. Oh, okay. Okay. You know that no answer means yes in this situation. No, it's not. That's terrible polling practice. Well, you know, these days everyone's trying to unskew the polls anyway, so I might as well join the herd. Okay.
So I've never been polled, at least not in an election. I got called by some, I think either national or New York State survey asking me about health information. It was a super long survey. Do you have like the list of like circle 24 things that you've done? Have you hang glided? Have you gotten a tattoo? Have you been polled? Have you done marijuana? Have you been polled? You don't have the polled one circle? That's kind of sad. I don't. I don't. I have some other interesting ones circled, but not that. Okay.
Ariel asks, which state has the most predictive power? That is, if you got to choose a single state for which you would get the full election results today, which would you pick to increase the certainty of the model the most? Well, we have this interactive. We can actually go through and see which states explore the way. So let's look at the Senate. Let's look at
I know the likeliest tipping point state. It's Georgia. Breaking news. Yeah? Democrats have moved up in the forecast, no, 67%. Ooh. Not 66. Well, it's kind of f***ed up because like, if you knew that Republicans win the Senate race in Colorado, let's click on that. O'Day wins. They haven't. Oh my gosh, the model is...
82% chance of winning the Senate. All of a sudden jumped up to 82% because that would be such a singularly bad result, right? But the most informative in both directions is probably something like Georgia or Nevada. I mean, I think Nevada is actually... Because Nevada is like the most normal race on the ticket. Georgia, you can imagine Republicans having a bad year. Or excuse me, Republicans having a good year, but just not being willing to close the deal for Herschel Walker because of all the baggage he brings. Nevada is more normal. But I think you still want to know Georgia because...
If Democrats win Georgia and Pennsylvania, they could still keep control of the House and lose Nevada. And so in that case, I would want to know Georgia. I want to know Nevada. Also, the House races, I think, are more interesting in Nevada.
oh, we're saying that you would find out all of the results for... That's what I assumed in the question, yeah. I don't know that that was part of it. But maybe if that's the case, then yes, because there's like four districts there that are within the competitive range, right? Yeah. I mean, it could be some other state where you have a lot of... Pennsylvania has like variation in different house districts, so it might be that too. I mean, Pennsylvania really... I was just making this argument to a friend who lives in Pennsylvania and is doing political reporting. They really span...
the spectrum of American politics from east to west in Pennsylvania. It's the longest pencil in the world. Have you heard that? Is that a comment? Is that like my family thing? That's your family thing. Pennsylvania is the longest pencil in the world? It's really f***ing long. If you have to... We used to drive through a group in Michigan. My grandma lived in Westchester County, New York. You have to drive all the f***ing way through Pennsylvania to...
I've made that drive. I drove from Madison, Wisconsin to New York City in a U-Haul. And I stayed in, what is it called? State College? Is that where Penn State is? Yeah. I stayed in State College in a Best Western with no windows. No, I'm just shouting out to like the Clarion Pennsylvania Days Inn. There were children running through the hallways all night. It was like children on the corner. Okay. So you choose Nevada. I choose Georgia. Okay.
Next question. Can you provide a general ranking of information the model uses that has the biggest impact on a percentage basis to the model output? Things like polls, scandals, general ballot. How do these things compare with one another? And the answer is that it depends on the race that we're talking about. Yeah, I mean, if you have robust polling in a race, then that is what the model defaults to, with the exception of the deluxe model, also still kind of just a little bit based on what external forecasters say. Yeah, it's kind of like
Polls is the number one thing of that race. If you don't have polls of that race, you default to kind of like the generic ballot plus partisan lean index, right? Of the things next on the docket, fundraising is probably the thing that influences the model the most.
Next question. This is from Tanner. Why are polls in deep red states like Nebraska, South Dakota, and Oklahoma showing that the governor races are competitive? And why is the model not really changing to reflect those polls? Slash when will it change? Well, partly, sometimes what happens, it's like you will see a poll showing like the Democrat behind by like four points in one of these states. That just happened in South Dakota. It's actually not that great for the Democrat. The Democrat's behind in that poll. It's just one poll, right? Like, let's look at South Dakota.
So South Dakota, you have one poll showing the Republican ahead by four points. The prior is that the Republican wins by like 20 points. One poll showing the Republican actually winning isn't going to shift things that much. If you have multiple polls, it's a different issue. If you have a poll showing the Democrat leading by four points,
It's a different issue, right? But like, it's not really telling you that much. So the answer, which will be a surprise to long time listeners is that you have to wait for more data. It's also a registered voter poll, right? How many people, you know, 565 registered South Dakota voters from a polling outfit that has not released a lot of polls historically, right? It's good people are pulling South Dakota, right? There's not a lot of confidence you can have in that data point to override
The very strong prior you might have that in a Republican, well, maybe neutral year that South Dakota is a very red state. This is something of an existential question from David, which I like. Why does it seem that the North Carolina Senate race has a lot less coverage than other competitive states? That's a great question. I think in part because North Carolina is seen as like a tease state where Democrats often get within a couple of points of
And lose. Because the polling there is very close. I mean, I think Sherry Beasley, the Democrat, is trailing Ted Budd, the Republican, by a half a point to a point currently. Yeah, and also the candidates aren't as interesting narratively as like Fetterman and Oz or Warnock and Walker. But no, I think people think North Carolina is like this fake... Fake battleground state? Yeah. But you know, Cal Cunningham...
Might have won if he could have gone back and not had an affair. Not had an affair, right? Another notch on the belt of scandals mattering. Hashtag scandals matter. You know, Kay Hagan was elected from North Carolina. Barack Obama won it once, I think, or was it twice? Once, only once. And barely, though. And barely. But, you know, in general, I think people, I think it's kind of superstition, right? Nevada was a state, I said both Nevada and Nevada at this point. It's Nevada. Nevada. Nevada.
Is a state, I'm just gonna have my own, Nevada. I think it's Nevada. Nevada. Nevada. Pennsylvania is a state that was never gonna go Republican until it did, right? Arizona and Georgia were states that were considered tease states until they weren't and Democrats broke through, right? So you're making the case for paid more attention to North Carolina. Yeah, I think it's superstition basically. So you call it a fake battleground state? What are other fake battleground states? And is Florida one now?
I think Florida is unique because there is a lot of migration to Florida, and I think some of that is politically motivated. I mean, people don't realize how weird Florida is, right? If you describe the population of Florida, it's like a coastal state that's extremely racially mixed. But also extremely old. It doesn't matter that much. What do you mean? It doesn't matter that much. Why not? In terms of how people vote? Yeah. Age is kind of a rate as a vector. It matters to some degree. Like, some reason why...
If you do a big regression analysis... There are pretty reliable patterns, aren't there? In that usually states that are racially diverse are pretty young, right? Okay, but also younger people are much more likely to be college educated and are much more likely to be non-white. And those things are more predictive. So once you account for those, that you have a college educated...
very racially mixed younger generation and an old white non-college older generation, then age doesn't really account for that much. I mean, older white voters are pretty swingy. Like Maine and Florida are two of the oldest states that have lots of old white people in them. I'm telling you, if you put in a big regression analysis, like age doesn't predict. Age doesn't matter. It matters some, but it's like overrated. Okay, go ahead. I still think that makes...
Florida unique. It's like a very racially diverse state that has a lot of old people in it and like the old people are more reliable voters than the younger voters. It's unique because people move to Florida in part because they're selecting for the politics in the same way they moved to Vermont. Vermont is a
rural white working class evidence to back that up at this point because like I know that people very publicly do that you know like conservative personalities during the pandemic were like I'm moving to the free state of Florida but when it comes to actual like regular people who don't have a platform or whose jobs aren't based on making political statements with their lives like are those people also moving for political reasons I think in increasing years jobs is a different thing like if you're moving for a job that's common you know people move to Florida or Texas or Georgia all the time for jobs I mean people's
Political views are so predictable based on their kind of other lifestyle characteristics anyway that like you want it with a lot of migration that at least affects political trends. But like people don't understand. I agree with that. That I fully agree with. I think the whole like people are so partisan. They're like moving to here or there or wherever based on their partisanship overstates American partisanship.
People move to when they retire, particularly move to be in communities where they have other people get along with. They move for financial reasons. So Florida is a state where there's no state income tax, for example. I mean, does that count as a political reason? Not having a state tax? Yeah. We did an interview on this podcast that showed that somewhere between 15 and 20 percent of Americans are like your stereotypical American.
partisan actors who like see the world through a partisan lens and would like talk about politics and the country in those terms and that like 80 to 85 percent of americans aren't really viewing politics that way okay so why is florida gone from but like you don't need 82 percent right right 15 or 18 percent would you agree with that analysis though in general that like
To pretend that America is just split in two is overwrought and that most people aren't really viewing politics and making decisions with their lives based on politics the way that pundits often talk about it. I think the average pundit overrates how partisan people are, but I think people migrating is understudied. And Florida is a case where I think people move to on the libs. All right. Well, when you have more data, bring it to me.
The next question, we're down to the final questions here. This is a fun one. Dave asks,
They all have pretty much the same odds. So how would you personally rank Barnes, Beasley, Ryan, and Oz in likelihood to win? That is Mandela Barnes, the Democratic Senate candidate in Wisconsin. That's Sherry Beasley, the Democratic Senate candidate in North Carolina. Tim Ryan, Democratic Senate candidate in Ohio. And Mehmet Oz, the Republican Senate candidate in Pennsylvania. Let me, let me, uh,
Put some numbers in my spreadsheet here. There's a new poll that came out that was bad for Barnes, by the way, so that affects things a little bit. As a former Wisconsinite, I wasn't buying that race. I'm going to go Oz. For likelihood of winning, Oz is number one. Beasley, Barnes. Ryan? Ryan. Ooh.
I want to come up with a reason to disagree with that. But maybe I'll go Beasley-Oz. I think I'll go Beasley-Oz. What's the model actually say? I should probably look at that. No, the reason that they're asking this question is that they all approximately have the same odds of winning. So, yeah. I mean, so really the empirical answer is this is unanswerable because they all have the same odds of winning. But I think I would go Beasley-Oz, Ryan Barnes.
I mean, there is the Fetterman thing. It's a little spice for this late in the podcast. Which you think that people reacting to his seeming inability to answer questions that are asked vocally without the help of transcription, that people will see that and be like less inclined to vote for him.
I mean, for a long time, the news cycle in Pennsylvania was about crudite in New Jersey. And now there is this fairly sustained focus on Fetterman, including difficulties from his stroke, including maybe some of his positions on crime. You know, that seems like a shift that may not be fully priced into the polls yet. I think people are probably overdoing it. I mean, I think it's perfectly appropriate to look at a candidate's fitness for office.
I mean everything about that little news cycle last night I thought was kind of annoying. Like on the one hand I thought he seemed like he was, you know, actually doing fine in the interview, right? On the other hand, like I totally disagree with any idea that like it's inappropriate for the media to be probative of this question.
Yeah, I mean, I think, to your point, that this is oftentimes used for partisan purposes, that people will try to shut down conversation about fitness for office, whether it's Trump or Biden or Fetterman, by saying that it's like an inappropriate question. I've said this many times on this podcast. There should have been more open discussion about Ruth Bader Ginsburg's health, right? Yeah.
And I think the media was irresponsible not to talk about it more. Ultimately, at the end of the day, voters have no control over that. So, okay, are you sticking with, what was it, Oz, Beasley, Barnes, Ryan? Because you think there's upside for Oz, right?
That is yet unregistered in the polls in Pennsylvania. Not a lot, but I think a little bit. According to the polling averages, I think... Isn't Beasley currently doing the best? Or is Ryan still leading in Ohio? Yeah, and polling average, Ryan, well, leads, quote-unquote, by 0.3 percentage points. I don't know. By the actual polling averages, then, it's Ryan, Beasley, Barnes, Oz. But, in fact, you're going...
Oz, Beasley, Barnes, Ryan. Sounds like you're doing battle with the data. You're unskewing poles. And we've come full circle. Final question from Bart.
How often does 5e Fox call home to his mother, a lovable though neurotic, chain-smoking fox who has a nice condo on Miami Beach? I mean, 5e is a little estranged from his mom. What? Yeah. 5e's not a family guy? He's not a mama's boy? You know, foxes are actually somewhat solo creatures. They're not super sociable. Is that why the fox symbolizes 538?
A sort of individualistic curiosity? Maybe individualistic curiosity. Yeah. I mean, I think foxes are like... Let me look at Google. Well, the expression is like, be a fox, not a hedgehog, right? Like, first of all, what's the term...
Okay, well actually, according to LiveScience.com, foxes are various social creatures that live in packs. Wow. Wow. Okay. Sounds like data and evidence really doing you dirty there. Look, 5E had a rough upbringing, okay? I didn't want to admit it, but it's true. What made it so rough? Well, now we have countries, unlike the relatives, foxes are not pack animals.
Wait, so we're getting conflicting signs. Foxes live in small groups or alone. See, I don't know. Yeah, I thought foxes were kind of not super sociable. Why did Fivie have a rough upbringing? Look, I don't want to get too personal with Fivie, but like, you know, grew up in a den. What's it? What's it? You know, where do foxes live in a den or a warren? That's not part of my responsibility to know. Yeah. A fox hive. A fox hive. Like it's a hive.
No, 5E had some issues growing up. Yeah, he was kind of a nerdy fox. When the pollsters called 18-year-old 5E to ask 5E about substance abuse, what were his answers? I mean, probably taking some catnip or whatever. Foxnip. Some foxnip? Yeah. We're entertaining ourselves too much with this. So you're saying 5E doesn't call his chain-smoking, neurotic fox...
mother in Miami Beach. I mean, the other thing is like Miami Beach is not a very hospitable place for foxes. It's outside of their range. Aren't you the one who put them there? No, 5E, no, 5E was in Vermont. Oh, 5E was in Vermont, the forecast. This is a really complicated family. 5E is very reluctant to travel to non-forested places. Okay, makes sense.
All right. Well, before we go to housekeeping matters, one, as usual, we have a live show in Washington, D.C. on October 25th. Get your tickets. It's going to be fun. Also, we are hiring a freelance audio editor for the Politics Podcast. So if you want to come work with us, you can find that job posting on the FiveThirtyEight website. I've also posted it on my Twitter feed and I'll drop it in the show notes as well. We're pretty fun to work with, right, Nate? I think we're fun to work with. All right.
We're going to leave it there. Thank you. It's fun. It's fun, everybody. My name is Galen Druk. Sophia Leibovitz and Kevin Ryder are in the control room. Chadwick Matlin is our editorial director. He also yesterday welcomed a new baby into the world. So we're very excited for him. Congratulations, Chad, to you and your whole family.
and Ellen Ganeski is our intern. You can get in touch by emailing us at podcast at 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. Thanks for listening, and we'll see you soon.
