The Most Valuable Data Point
33:42 Share

- As a pollster, it is so impossible to narrow it down to one data point. It's almost cruel and unusual on your part to even ask.

Hello and welcome to the FiveThirtyEight Politics Podcast. I'm Galen Druk. It has been 1,451 days since the last presidential election. And in that time, there's been an attack by Trump supporters on the Capitol, a botched withdrawal from a 20-year war, the end of Roe v. Wade, decades-high levels of inflation, a growing crisis at the border, war in Europe and the Middle East.

multiple assassination attempts, and the decision of the incumbent president to withdraw from the campaign. Now, in 11 days, Americans will select a president for the first time since all of that happened. I say all of that not to rewrite We Didn't Start the Fire for 2024, but to say this.

a lot has happened to scramble Americans' perceptions of their country and their leadership. And today, we have a lot of data to tell that story, maybe too much data, especially right now as high-profile final days polling begins to pile up.

So today we've assembled a crew of election data nerds. I hope no one minds being called that. And we asked them what they're looking at. Specifically, we asked them to choose one data point that they think is the most valuable for understanding the election in these final angsty days before November 5th. Okay, let's get started.

We've talked a lot about changes in voting patterns amongst different demographic groups, in particular, how Black and Hispanic voters might be cooling on Democrats. But there's another group that Democrats are increasingly concerned about, and that group is the focus of our first Data Point. And it comes from Republican pollster Adam Geller.

If I had to boil it down to one data point, I'm going to look at Gen Z voters, and I'm really specifically going to look at male Gen Z voters, otherwise known as Trump bros. The polling that we're seeing out there, and it's not just the private campaign polling that we're doing, but also the public polls. I think that there was a Harvard Youth Poll and

Gallup and other public organizations, but certainly the private ones that I'm seeing are this enormous gender gap that exists regardless of age. But it's particularly acute among younger voters.

younger women are really identifying as liberal and identifying as Democrat. And of course, I think the conclusion is in part driven by reproductive rights, not necessarily exclusively, but certainly that has a lot to do with it. On the other hand,

Young Gen Z men, Trumpros, who are more likely to be fans of guys like Joe Rogan and Dave Portnoy, be fans of UFC, are attracted to President Trump. There's a personality attraction there. If I look at exit polls from 2020 and I look at Gen Z men, according to the CNN 2020 exit poll,

52% of young men in 2020 voted for Joe Biden. 41% voted for President Trump. That is a number that I think, and a demo that I think is going to fundamentally flip this year. And that would be a big deal in some of these battleground states that are as tight as they appear to be in the polling. On the other hand, in that same exit poll, 67%

of young women 18 to 29 voted for Biden. Well, 67% doesn't give much more growth for Kamala Harris. You know, there's only so much more growth you're going to get out of young women. You're close to topping out. Maybe you can get another 10%, but you have the potential for 30 or 40%

of young Gen Z men to flip over to President Trump in this cycle. And that can be a really significant number in some of these razor-tight states.

When it comes to just how much young men might be swinging from Democrats to Republicans, there's a significant amount of disagreement in the polls. Nate Cohn at The New York Times recently took a look at all of this, and the range is from

Pew showing Republicans winning young men by 26 percentage points to the Harvard Youth Poll showing Democrats, Harris, winning young men by 17 percentage points. For their part, the New York Times-Siena College Poll shows Trump winning them by 17 points.

We won't know until the votes are actually cast. But what we can say is that fewer young men are identifying with the Democratic Party when you look at recent voter registration numbers.

The trouble for Republicans is that that growing strength does not extend to young women. In fact, we are seeing the opposite trend amongst young women. They are growing increasingly liberal. And in fact, that Harvard Youth Poll found Harris leading Trump amongst young women by 47%.

percentage points. And I should say here that if you did have to choose between having an advantage with women and having an advantage with men, you might choose having an advantage with women because, after all, women make up a majority of voters in the United States of America. Now, we are going to hear plenty more about the gender gap over the coming weeks. And on the topic of

that gender gap, polls have shown Republicans making some inroads with Black men and Black voters a little bit overall. Now, we next talk to Terrence Woodbury, who's a Democratic pollster at Hit Strategies, a firm that works with the Harris campaign and focuses on surveying voters of color. He says he's asking Black voters how much they think their vote matters.

Hit Strategies is paying very close attention to a metric that we call perceptions of power. You know, a lot of pollsters use metrics like enthusiasm or motivation to predict who's actually going to vote. What we've discovered over time is that not all black folks are voting enthusiastically.

And so instead of measuring enthusiasm, we measure perceptions of power, which has a more direct correlation to turnout. And what does that mean? On every poll, we ask the question, regardless of how often you vote, how much power does your vote have to make a difference in your life?

the higher they rate those perceptions of power, the more likely they are to vote. And what we're seeing is some very interesting trends here that perceptions of power have increased significantly since Kamala Harris became the nominee. We have seen her increased perceptions of power amongst Black voters by almost 12 points since becoming the nominee. But the challenge here is that

perceptions of power ebb and flow in the Black community, right? They get very, very high during election season, and then they drop very, very low during off-season. And what we've learned from Black voters about why that is, is during election season, they're being courted by candidates on both sides. And so today we see perceptions of power in the Black community right around 52%.

That is not as high as the high watermark of power that we saw in 2020, which is 73 percent. But it is higher than the perceptions of power that we saw in 2022 of just 44 percent. So we are recovering, improving those perceptions of power amongst Black voters. I think the vice president at the top of the ticket has gone a long way at improving those perceptions of power, but we still have some ways to go in the last stretch of the election.

And the Democratic ticket standing with Black voters is falling short of its high watermark overall. So earlier this month, the New York Times-Siena College polled roughly 600 Black voters and found that amongst registered voters, Harris led Trump 71% to 18%. That 53-point margin, of course, sounds like a lot, and it is the clear majority, but it's a setback compared to Biden's final margin in 2020, which was roughly 81 percentage points.

On the other hand, Democrats have made up a good amount of ground with college-educated voters, and in particular, college-educated women. Though, we heard from one Republican pollster, Brent Buchanan, who said that he's starting to see some data that complicates that picture.

Earlier in the cycle, the whole conversation was around this shift of non-college educated voters towards Republicans and on the inverse, college educated voters moving away from Republicans.

And one really interesting thing that we have seen in our data, and this is across tens of thousands of interviews, especially in the swing states, college-educated women are actually moving towards Republicans, and college-educated men are moving away from Republicans. And when you dig into what is driving that, I really think the broadening conversation around immigration, the consequences of it, especially as it relates to physical safety,

is driving college-educated women towards Republicans. And I haven't really been able to put my finger on yet what has caused college-educated men to shift away from Republicans more towards Democrats. And all of this really occurred after the swap out of Harris for Biden at the top of the Democratic ticket. And so as you look at some of these swing states where we're talking about everything's happening on the margins,

this weird inverse of gender roles within partisanship and the swap of educational attainment within partisanship, but opposite of the gender roles, is really, really fascinating. And even though these swing states have lower educational attainment than, you know, a Colorado or Virginia, I do think it's going to make a difference on the margins.

That runs counter to much of the narrative about the gender gap, but perhaps a reminder to keep an open mind until we see the actual results of the election.

We don't have a ton of polling on women with college degrees, but the polling we do have right now shows a mixed story. So polling from Fox News of white women with college degrees does in fact show Trump gaining on Harris. Her 25-point lead in September was down to six points in mid-October. Marist, meanwhile, found the opposite. Among white college-educated women, Harris held leads of 17 and 31 in two September polls, and in their October poll, she was up 42%.

Of course, these are crosstabs, so we're talking about relatively small samples and we should take them with many grains of salt. We will have to wait and see what actually happens. And that's the case with all of the trend lines below the surface. I also want to add that we may not even know on election day based on the exit polls what the gender gap actually looks like.

Some of the most reliable data on these subgroups comes from verified voter surveys that will come out months after the election, and oftentimes months after the narrative has already been baked in. Take, for example, the gender gap in 2020. The exit polls showed a really wide gap on par with 2016. Months later, though, those verified voter surveys showed a much narrower gap

gap, with Biden pulling almost even amongst men and winning women by about 11 percentage points.

But one thing we can say with perhaps some confidence is what the big issues of this election are going to be. As always, the economy, also immigration, abortion, and a few others. And how voters feel that the candidates are addressing those issues in a personal way is the focus of our next data point, courtesy of our friend, New York Times polling editor Ruth Agelnik.

I think the single most valuable data point that I've been following is this question of policy impact. How much will Trump's policies help or hurt you or people like you? And how much will Harris's policies help or hurt you or people like you? And there's just so many fascinating things under the surface there. We've been asking this question a couple of times, and we consistently see more people saying that Trump's policies helped them or helped people like them.

and that Harris's policies would hurt them or hurt people like them. And I think, to me, it aligns closely with the horse race, and you start to understand why some of these groups who might have been moving away from Democrats and towards Republicans are doing that. These are the same groups that are more likely to say that Trump's policies helped them.

Some of these Latino voters that are, you know, as we see in polling, moving a little bit more towards Republicans, Black voters who are moving towards Republicans, younger voters. A lot of this is aligned with this idea that they think Trump's policies when he was president really helped people like them and that they worry that Harris's policies as president would hurt people like them.

The New York Times fielded this help or hurt question back in September in a poll of Arizona, Georgia, and North Carolina. And they found that Trump was net positive on the question. So 45% helped to 35% hurt, plus 10% net. On a similar question, Harris was in the negative. So 37% help to 42% hurt, net negative five.

Other polls have found similarly rosy retrospective views of the Trump administration, especially on how good the economy was in those years. For more on that, here's Julia Azari, professor of political science at Marquette University. Which candidate voters are trusting on the economy? I still think there's a significant amount of kind of retrospective voting baked in.

And in particular, I think the Republican Party has tried and the Trump ticket has tried to make this a retrospective election and remind people of what they liked during the Trump years and what they haven't liked during the Biden years. And I think that one thing that we'll sort of know is if Harris continues to kind of gain on trust in the economy right now, I believe she's tied with Trump as we're having this conversation.

If Harris's numbers continue to be more favorable in that direction, then I think what we'll see is that that retrospective narrative for Republicans loses a bit of its power.

Both candidates have been campaigning hard on their economic policy platform. And as Julia noted, some surveys do show Harris having pulled even with Trump on trust to handle the economy, in particular on specific issues like grocery prices or housing. But more often, Trump still holds a lead on the issue. A recent ABC News Ipsos poll showed him up eight percentage points on the economy overall and seven percentage points on inflation, which is in line with other polls.

I should say here, though, that Harris is doing better than Biden was doing against Trump earlier this year, where he trailed by 10 percentage points or more. Next up, we're going to turn from the issues and demographic trends to the all-important question, who will actually show up? And how can we predict that? But first, a break.

That's amazon.com slash adfreenews to catch up on the latest episodes without the ads.

It's cliche and frankly, a little self-evident. But in this case, the old saying will probably hold true. It all comes down to turnout. Who makes up the electorate will determine who wins the election. We had a few respondents focus on turnout, and one was Democratic pollster Celinda Lake. She was one of Biden's two main pollsters in 2020, and her firm is Lake Research Partners.

I think the most valuable data point for understanding this election is the turnout number, which is also the most mystifying. I think there's very likely to be a surge of Trump voters that we don't know about. And we know that among people who were registered in 2020 but did not vote in 2020, he is actually tied or winning that vote.

And then I think there will be a surprise or a very hard earned surge on the Harris side among young people, among people of color, among younger women, among younger pro-choice women, among those single ladies, whether they have cats or not.

So if anybody can guess the turnout number, they're going to guess the winner because I think this race is coming down to a very dead even race and it's just a question of who is going to get their base out the best.

Of course, figuring out who will actually show up is one of the biggest challenges pollsters face. Ann Seltzer has a lot of practice with that. She is the legendary pollster who oversees the Des Moines Register's Iowa poll, and she made headlines last month for a poll that found Harris trailing Trump by just four percentage points in Iowa, 47 to 43. Trump won the Hawkeye State by eight points in 2020.

That finding, Ann says, it has less to do with voters changing their mind about Donald Trump and more to do with who says they're definitely going to show up.

In my polls, what I want to look at is the incidence of likely voters. That is, what proportion of the people I talk to qualify as a likely voter? In our most recent poll, we were able to see a big uptick in people saying that they were definitely going to vote, and that's our definition. And that gave us the opportunity to dig in deeper and find out that the demographics of

of who said they were definitely going to vote had changed since our June poll, where Donald Trump had an 18-point lead. In our September poll that had just a four-point lead for Donald Trump, it could largely be attributed to more women, more younger people, more college-educated, more suburban people showing up.

as likely voters. It isn't that Donald Trump lost ground. It's not that there was conversion from people supporting Trump before changing their mind to vote for Harris. He got roughly the same number of people out of our total 800 samples saying they would definitely vote and their first choice was Donald Trump. But now there were more definite voters and

And so the proportion saying the Democrat, it had been Joe Biden in June and saying Kamala Harris in September, she got a bigger share of that. So that tells me what's going to matter in this game is turnout. A lot of likely voters are what we call high propensity. That means they almost always vote. You can pretty much count on them to show up because they've done it before.

On the other hand, you have low propensity voters. Their history of showing up is spotty, if it exists at all. A lot of these voters are blessed with having other interests or, well, maybe stressors besides politics. But they also make elections harder to predict. We don't know whether or how a lot of these voters will vote. A lot of them don't know whether or how they'll vote themselves.

A Pew study from 2023 found that 12% of adults voted in 2020, but not in 2018 or 2022. 16% of adults voted in 2020 and in one midterm year, either 18 or 24. But most Americans fall into the extremes of very engaged or not engaged at all.

37% of Americans voted in all three of the past three national elections, and 30% voted in none. Meanwhile, 5% say they only voted in off-year elections, so in either 2018 or 2022. Shout out to those people who really feel passionately about the midterms. I would enjoy talking to you.

That is all to say, a lot of low-propensity voters will show up, and how they vote is what interests Lakshya Jain. Lakshya does election modeling and other data analysis at SplitTicket.org.

I think one of the most critical data points is how people who didn't vote in 2020 intend to vote in 2024. The thing that people don't really remember is that it's not so cut and dry that if you voted in 2024, which is likely to be a lower turnout election than 2020, then it means you must have voted in 2020. In fact, 20% of the electorate, 15 to 20% of the electorate may not even have voted in the last election.

as is pretty standard for most presidential elections. You have 15 to 20% of churn cycle over cycle. What that means is that although Kamala Harris has held a remarkable amount of ground and possibly even gained ground with people that vote in past elections, who have voted in midterm elections, who vote in special elections and primaries, how is she doing among that slice of voters that is more disengaged and does not vote?

in special elections, in primary elections, doesn't show up generally and doesn't really tune into the news. These types of voters are not moved by the traditional anti-Trump appeals. They're pretty disengaged. They're pretty skeptical of institutions as a whole. Democrats used to win this constituency quite handily.

That advantage has been eroding. So can Donald Trump make up the gains that he has lost with the more engaged demographic by gaining more among these disengaged voters? To me, that is the most critical point that I'll be watching in polls and in crosstabs to see broadly how has that picture changed. And it's important to note Kamala Harris has been making significant gains over Joe Biden 2024 wise in that regard. Biden was losing these

marginal voters by much more than Harris appears to be losing them in polls. So whether that actually holds or not as the election nears is something that is going to be fascinating to watch, especially because these voters tend to tune in very late.

That disengaged voter who didn't vote in 2020 shares some similarities with maybe another group that we talked a lot about earlier this year, the double haters. That is, voters who hold negative views of both candidates. Now, when we last talked to our panel of experts for an episode like this back in February, this was a very common answer.

the double haters. That was when it was just becoming clear that Donald Trump would win the Republican primary and Americans would get a rematch of the 2020 election. It would pit Trump, who a majority of Americans have viewed unfavorably for years, against Joe Biden, whose approval rating was in the basement. And polls showed that perhaps one-fifth of the electorate disliked

both candidates. So the double haters were going to be pivotal. A lot has happened since then. The race has changed, but maybe it hasn't changed so much that we should overlook those haters altogether. Here's Republican Ramesh Panuru. He's the editor of the National Review. I'm keeping an eye on the so-called double haters. That is the people who have an unfavorable impression of both of the major party candidates. They're

These people were decisive in 2016 when they broke heavily for Trump over Clinton. They broke a little less heavily for Biden over Trump in 2020. There are fewer of them this time because Harris's approval rating is higher than Biden's was when he dropped out of the race. But there are still enough of them that they can be decisive again.

in a tight election, and it's going to be very interesting to find out which of them they decide they dislike least.

For context, Joe Biden's net favorability in October 2020 was about positive 5. Hillary Clinton's in October 2016 was negative 10. At last check, Harris is about negative 1, and Trump is net negative 10. So two candidates who are technically underwater, although Harris just slightly, and not nearly as bad as in 2016.

Although to add some context to how voters are thinking about candidates that they don't like, in a recent ABC News Ipsos poll, 20% of Donald Trump's supporters had an unfavorable view of him. So even though there may not be quite as many double haters, it does still seem like a core part of Trump's coalition is a group of people that does not actually like him.

One dynamic that also defined the final weeks of the 2016 race was bombshell stories that damaged both candidates. For Trump, it was the Access Hollywood tape, and for Clinton, it was the FBI Director James Comey's letter to Congress about her emails. What stories get attention in the final days is the focus of our next expert, Patrick Ruffini. He's a pollster at Echelon Insights.

So the data point I'm paying the most attention to in the final weeks is a question us pollsters refer to as information flow. Are the things voters are seeing, reading, or hearing about the candidates making them feel more positive or negative towards those candidates?

With the race so close, it's more important than ever to understand if the dynamics of the race might be changing. And InfoFlow is very sensitive to changes in momentum and media coverage.

So in our last national poll after the debate, Kamala Harris had a lead on the full ballot of five points, and she was net positive on information flow by 11 points. So that means that voters who reported seeing mostly positive news about her outnumbered those who saw mostly negative news by 11 points.

And Donald Trump was negative 15 points on information flow for a gap of 26 points favoring Harris.

In August, when Harris was still on her honeymoon, but the candidates in our poll were tied, she had net positive information flow of nine points, while Trump had net negative information flow of eight points, a gap of 17 points. Now, this doesn't perfectly predict the margins, but the poll-to-poll changes can be revealing.

So, for instance, in the month before, right before Joe Biden dropped out of the race and after the first Trump assassination attempt, Biden had a net negative information flow of 35 points, while Trump had positive information flow of four points, which is a pretty good reflection of what was happening in the media environment at that point.

So I like this data point because it moves a lot from survey to survey to a much lesser extent than other questions we ask. People aren't just trying to answer in a positive way for their candidate, but seem to be giving a more unvarnished assessment.

As late in the game as we are, a lot can happen in 11 days. I mentioned the Comey letter that was sent to Congress on October 28th. The information flow can change in an instant. Okay, so it's time for our final data point. And finally, we have one that mentions races for the House and Senate. Glad we can show at least some respect to our friends down ballot. This one comes from Lenny Bronner, a senior data scientist at the Washington Post.

What Lenny is watching is how votes are distributed across states and how high-profile Senate races might impact that. The thing I'm really keeping a close eye on in this election is the difference between the polling in the presidential election and the Senate elections, specifically in the Senate elections that are most likely to decide control of the Senate. Democrats are generally polling a lot stronger than

in Senate elections in Pennsylvania and in Michigan and Wisconsin and Arizona and Nevada than Vice President Harris's. There are a few reasons that this is a particularly interesting phenomenon. The first of all is that it's quite different from what we've seen in recent years where Republicans

have generally been polling ahead of Donald Trump in key elections. But it may also be an indication that we are going to be seeing an election that is more similar to the 2022 midterms rather than the 2020 presidential election, which would obviously be a good thing for Democrats. Wait, wait, wait. But can you elaborate on that? Why is the Senate gap suggests that it would be more like the 2022 midterms?

Well, one of the key parts of the 2022 midterms was that there were basically two separate elections that were happening simultaneously. There was one election happening basically across the country, and then there was another election happening in the most important states.

So Republicans, you know, obviously had this turnout advantage, but Democrats won in the key states. And that had something to do with obviously Republicans in those states being more likely to vote for Democrats than other places, but also because Democrats had less of a turnout gap in the most important states. People who lived in swing states and key states obviously were aware of this and were sort of choosing to vote differently than people who are living in states that were less contested.

And so it is possible that we're seeing something similar again, where people that live in states that are most likely to contest the election are voting differently than people in states, say, like New York, Florida, California, Texas, that are just large, have a lot of people in them, and are less likely to decide the election.

And ultimately, what Lenny's talking about here will play an important role in determining the popular vote electoral college gap, which I'm sure is a data point that will get plenty of attention after the election. But we're going to leave things there for now. Thank you to Adam Geller, Terrence Woodbury, Brent Buchanan, Ruth Egelnik, Julia Azari, Patrick Ruffini, Celinda Lake, Anne Selzer, Lakshya Jain, Ramesh Panuru, and Lenny Bronner.

My name is Galen Druke. Our producers are Shane McKeon and Cameron Tretavian, and our intern is Jayla Everett. You can get in touch by emailing us at podcasts 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 will see you soon.