Can Google searches tell us something the polls can't?| FiveThirtyEight

The team debates whether tracking Google search terms over time is a better barometer than traditional polling.
42:54 | 09/27/22

Coming up in the next {{countdown}} {{countdownlbl}}

Coming up next:



Skip to this video now

Now Playing:


Related Extras
Related Videos
Video Transcript
Transcript for Can Google searches tell us something the polls can't?| FiveThirtyEight
- Hello and welcome to the FiveThirtyEight Politics Podcast. I'm Galen Druke. There are now only six weeks until election day, six weeks before we start to figure out just how much President Biden has hurt or helped his party in the midterms. Traditionally, a president's approval rating is pretty closely tied to how his party performs. But Democrats have been overperforming Biden in the polls for months. And last week, Biden demurred slightly about his intentions to run for re-election in 2024. So is this an intentional strategy? And how much will the midterms affect Biden's thinking? We're going to discuss. We'll also focus on how demographic and geographic changes are shaping one of the most competitive races in the country this fall, the Georgia Senate election. Georgia's Black electorate has grown faster than any other state in the country over the past 20 years. We're going to look at how that happened and what it means for the state's politics. We'll also dive into whether or not Google search trends have anything to tell us about what's on voters' minds that the polls don't. Here with me to discuss it all is editor-in-chief Nate Silver. Hey, Nate. How's it going? - Hey, everybody. - Also here with us is politics reporter Alex Samuels. Hey, Alex, how are you doing? - Hey, Galen. I'm well. How are you? - I'm pretty good. I'm pretty good. It's Monday, but I'm fully caffeinated and we're going to do this. Also with us is elections analyst Geoffrey Skelly. Hey, Geoff, how are you? - Hey, Galen. It's Monday. So I'm doing. - All, right there we go. Nate, before we dive into our good use of data or bad use of data, I think listeners might already know because I announced it last week, but we have a live show in Washington, DC on October 25. It's our first live show in a very long time. Are you excited? - Yeah, I mean, I love Washington, DC. It's my favorite city in the entire United States. [BLEEP] [LAUGHTER] No, I love doing live shows. And DC is a great venue for a live show. So please come. - It is. It's really fun. It's going to be close to the mid-terms, so on October 25. We're going to do a little bit of model, talk we're going to play some games, we're going to take listener questions. It will be very fun. So if you have not gotten your tickets yet, there will be a link to get those tickets in the show notes of this podcast. You can also find a link on my Twitter account, Nate's, or on the FiveThirtyEight website. So we look forward to seeing you there. But with that, let's move on to today's good use of data or bad use of data. Today's example comes from Axios. In the run-up to the midterms, Axios has been tracking Google Trends data to see which political issues Americans are most likely to search for. Last week, they published a piece highlighting how searches for abortion have fallen as interest in the border has risen. Abortion peaked as the second most searched issue in late-June after the Dobbs decision. At that time, the border and immigration ranked 18th. The latter now ranks 10th and abortion ranks 18th. Jobs has been the number one issue the entire time, and nothing has supplanted it. The increase in searches about immigration come as Republican governors have sent migrants to blue locales like DC, New York, and Martha's Vineyard. So here is the question-- pollsters regularly ask Americans what the most important issues to them are when considering how to vote in the midterms. The rankings vary from one poll to the next. But in the last issue polling we conducted with Ipsos, the priority ranking was the following-- one, inflation, two, crime and gun violence, three, political extremism, four, climate change, five, immigration, six, government debt, then seven, abortion. Is Axios' use of Google Trends data good or bad use of data in the sense that, is it a better barometer than the actual polling that we have? Is it really adding anything? Nate, I'll let you take this first. - So there are a couple of separate questions here, one of which is the use of Google Trends relative to polls, and another of which is how you create these different categories of issues, right? I'm generally a fan of Google Trends data for a couple of reasons. One, it reveals reveal preference, as an economist would call it, meaning what people are actually spending their time searching for and not what they may abstractly say is important, right? You may abstractly say, oh, well, I think it's really important-- my family and health is really important. If you spend every Sunday watching football with your buddies, then maybe that's not true as much in terms of how you actually spending your time, right? So what people search for kind of reveals, in some, ways their true identity or true selves. I mean, there are also things like you may search more for a topic in the news a lot instead of one that is important, but, like, there's not as much, like, global warming news from day to day as there might be for some other topics. It's a long-term kind of thing. But in general, I'm a fan. I do think, though, like, creating these categories is kind of problematic in that pollsters maybe don't spend enough time thinking about it, right? Whether you have a separate category for the economy and inflation is important, for example, right? You kind of say, oh, social issues and abortion might overlap. Also, sometimes the questions they use make assumptions about, like-- they'll say, like, oh, the crisis at the border, the threat to democracy. So they kind of presume the answer in the question. It's actually kind of an example of begging the question, as that term means formally. So you have to be very careful with kind of what your categories are. And it's probably best looked at if you look at several of these polls or metrics that, like, look at them in combination and not treat anyone as being gospel. - All right, OK. It's a pretty comprehensive answer. Alex, is this is good use of data or bad use of data in your book? - I said bad use of data. In my research, I just looked at the Google Trends help page, which says pretty plainly that Google Trends is not a scientific poll and should not be treated as such. As Nate got at, it simply reflects the specific interest of a specific topic at a specific point in time. And a spike in a topic doesn't mean that that topic is winning, it doesn't mean that it's popular, even. What it does convey, though, is that users were searching about a topic for some unspecified reason. And until we know what that reason is, which we can't glean from Google Trends data alone, I think polling is always going to be a superior barometer when considering Americans' feelings toward something. - All right, so we have two different takes so far. Geoffrey, where do you fall on this? - Yeah, I mean, I think I'm partial to bad use of polling. One of my initial problems with it was that they just don't have any data from August 8 to September 4 as a part of their analysis, which is, like, four weeks of data. So who knows exactly where things were shifting during that period? Seems like an important question-- or part of the question in terms of a recent trend. On top of that, I think my other concern is that they framed it very much with, like, the immigration-- like the focus on immigration, especially in light of Ron DeSantis and Greg Abbott, these Republican governors and, like, sending migrants, either via bus or by flight, to, like, liberal places in the country. And I think the inherent problem with sort of framing it around that is that plenty of people who do not agree with the choice to do that could be searching-- you know, hate searching, almost-- about this particular issue. So what I'm saying is it doesn't necessarily reveal any sort of preference, sort of to Alex's point about what the attitudes of the people who are searching this actually are. And so that, I think, makes it tough to really make much of this from that point of view. - Doesn't issue polling have the same challenge in that when you ask voters, you know, what's the most important issue to you in this election-- when someone says border and immigration, you don't necessarily know what they're saying. When someone says abortion, you don't necessarily know what they're saying either. It could be, I really care about abortion and I want abortion to be banned nationally. Or it could be I really care about abortion and I want abortion to be legal nationally. - Yeah, but the problem there is, I mean, the comparison you're making-- the thing is with Google Trends, I can't get cross tab of what Democratic search-- people who lean Democratic search and people who lean Republican search, which can tell you a lot. It's like, oh, if Republicans say they care about abortion a lot, it's not easy-- it's not hard because of other information we have about Republican positions on abortion to figure out, like, what their attitudes might be-- or, obviously, in the wake of the Dobbs decision, the fact that, oh, Democrats in polls are more worried about abortion than they were previously. It's not real hard to intuit what's going on there. But with Google Trends, can't do that. I don't have any follow-ups to figure out some more nuanced aspects of issue positions and whatnot. It's like, that's just not possible with that data-- or at least not with what Google is willing to share. - Yeah. If I could just add something, I'd say if I were to guess, I would say people were probably panic Googling post-Dobbs, and now they have a better sense of the landscape there. Stuff regarding abortion, at least at the statewide level, isn't changing as fast now. And for most of the summer, laws were actively going into effect. And unless you were following the fallout from Dobbs pretty closely, it probably wasn't clear where abortion was legal. And now, I think things have settled down now as the Google Trends data sort of got out. But that doesn't mean that it doesn't matter politically or that it won't be top of mind for voters heading into the midterms. - Nate, so some challenges to your good use of data allocation. - There are different ways for people to express preferences, and attitudes, and beliefs. And you want to look at more ways rather than fewer, right? The fact that people might be searching for abortion because they're concerned about, what if I get pregnant unexpectedly, what can I do? That certainly seems to me like it's revealed preference or people actually do care a lot about abortion. I mean, I agree in the abstract that maybe people are, like-- you don't know their motivations for searching. But you also don't know, like, when you ask in a poll-- like, an NBC poll said, oh, threats to democracy, right? And this is interpreted as, oh, there's all these-- liberals are finally coming around and expressing, which I think are valid, concerns about Republicans invalidating elections after the fact, right, election denialism. However, a lot of Republicans picked that answer because they think the election was stolen in 2020, right? They are wrong about that, but, like, that's where that answer reveals their preference too. So I don't know. I mean, like, immigration is an issue where if you look at the balance of different surveys, I think people are more negative on immigration and more concerned about it than you'd get by reading, like, the elite prestige media, right? So to me, that's, like, not a very good example of a case where these polls are off. To me, like, that would reveal-- I mean, inflation was like this when there were people denying the importance of inflation before it became so obvious, right? So these polls can actually express cases where people care more about an issue than it might get covered in, like, kind of elite liberal mainstream media. - The New York Times, of course, recently released its latest midterm polling. And it dug into this question. And Republicans have a 14-point advantage over Democrats when it comes to the question of illegal immigration, and the parties are even on the issue of legal immigration. So that does suggest that if immigration in general becomes a more salient issue, potentially, it could help Republicans. In that same polling, 62% of Americans said abortion should always or mostly be legal. So the numbers on immigration are better for Republicans. We talked about one data point, basically-- the Google Trends search. But, Geoffrey, I know that we are conducting-- we have conducted another round of issue polling with Ipsos that we're going to be releasing later this week. Have we seen in that polling a change in terms of how important these two issues are to Americans, abortion and immigration? - Right. So in our polling just after the Dobbs decision, the share of people who said that they viewed abortion as a top issue facing the country definitely jumped. It jumped from 9% to 19% from our previous poll. Now, our polling is a little different because we allow people to pick up to three options in terms of issues that they think are important for the country. Abortion has now come down a bit since then. It's, among all respondents, down to 11%. But it's much higher among Democrats than, say, Republicans. And then to speak of another issue that was brought up here was, like, immigration. Well, immigration consistently in our polling has been very much a Republican issue. Republicans are the ones that are answering immigration as an important issue facing the country. But the number has hardly changed. It's been between 32% and 38% in every wave of Republicans saying immigration, and just under 20% overall in most of the polls. So it's been very consistent in that way. - So at least as far as this polling is concerned, immigration is a more salient issue to Americans than abortion. - Yeah. I think so. So there are a few things that come to mind with this that I think complicate our understanding of it a little bit. Because we've been citing that New York Times Siena College poll a lot because they do good work. But you even if a lot of people aren't answering abortion necessarily as, like, a top issue that they care about, you will notice in that poll, as just a comparison point, slightly more respondents said that they view the Republican Party as extreme than the Democratic Party. And if you look under independents, it was a strong plurality of independents said they view the Republican Party as more extreme. So to me, the importance of the abortion issue is not so much that people are naming it as the top issue or one of the top issues for them personally, or for the country. It's, is it playing into attitudes about the parties? And I think that's where this issue is different from a lot of midterms in that abortion-- the issue of abortion has created circumstances for people to view the Republican Party as more extreme. Now, there are other things that could go into that, obviously, in the aftermath of January 6. But for me, that's, like, an important part of this. And it's not so much just about what people are naming as an important issue, but how their views of the parties have been affected by things that are going on. - To wrap up here, Nate, how are you processing what this sort of change in priority or change in salience means heading into the midterms? Because I think it was-- like, back in June when we talked about the outcome of the Dobbs decision, we're like, oh, this seems like a big deal. It's also unclear whether this will be a big deal in five months or four months. Now, we're getting-- now we're much closer to the election. - First of all, I agree with Geoff's point that it kind of gets into attitudes about the parties. And also, abortion's a very rare instance where the party that's out of power had a huge influence on an issue despite not controlling congress or the presidency, right? The Supreme Court overturning Roe v Wade is a huge deal, one of the most important decisions of the past 50 years. And so it serves in a more substantive way to remind voters of what the stakes are than supposed future threats, right? Oh, the next election, well, you elect these election deniers, it'll be stolen. Well, that might be true, but it hasn't happened yet, right? And also, the people who make those claims, I think, are correctly worried about that, also often cry wolf about other things that may not be as legitimate. So to have an actual case in hand of this thing that most Americans think is a fundamental right for women, to have that taken away, when the other party is supposedly out of power, that's a very important and unusual change. And it kind of reinforces concerns about the GOP and the Supreme Court in a way that a future harm might not. - Yeah. I think that's a really excellent point, because there's this concept of, like, thermostatic public opinion. And it's like if one party's got control, and they pass-- full control, like Democrats do, obviously, very narrowly, but they have full control right now-- it's like they pass a bunch of things that their party wants, but voters begin to feel like, oh, the country is drifting too far in that direction. And then, you know, there's a reaction, which is often the midterm election-- I mean, there are other things that go into that. But in this case, you had this sort of shock to thermostatic public opinion coming from an unexpected direction, in this case, which is from the party that's not in power. And that's, as Nate says, very unusual. So has that created a different set of conditions than we would normally expect in a midterm environment? - All right, so to come back to the beginning question at hand-- a good or bad use of data. It sounded like we had two bad uses of data and one good use of data. Has anyone been convinced throughout the course of this conversation that this is, in fact, the opposite? And that's going to be a no. - No, but I do really-- oh. I was going to say no, but I do actually really like Google Trends data and I do think there are really good uses of it. I know Nate recently used it for sort of talking about how Trump has remained in the news, and I think that was, like, a really excellent use of that data when thinking about politics as a whole and what's going on. So, like, there are really good uses of Google Trends data. So I'm not trying to pooh pooh that generally speaking. - I think that's true. I also think it's a really good point that we don't know necessarily the significance, from a partisan perspective, of Google Trends. Data I did an interview last week with Max Fisher, who wrote a book about social media. And one of the points in the book that I took issue with was basically saying, there was so much more shared about Donald Trump, there was so much more, like, news about Donald Trump on social media than there was about Joe Biden, this shows that this social media, maybe even Google Trends, whatever, can sort of be biased, and promote Republican ideas, or promote misinformation X, or Y, or whatever. And it's like, a lot of people are searching about this stuff because they're mad about it. And one of Biden's strategies was to stay out of the limelight and make the election a referendum on Trump. And I can tell you with almost 100% certainty that Biden's campaign would not have preferred Biden being a higher search trend feature than Trump in the run-up to the 2020 election. So it's important when you see these trends to think about, OK, why might this be peaking? It may not be the most obvious reason at first. - Yeah, Galen, this is funny you bring that up-- the 2020 election. Because I actually remember in 2016 doing some research on Google Trends and Clinton and Trump. And it tended to be whoever was in the news more, their polling numbers tended to slide. And then when they reversed who was in the news more, you know, the candidate that wasn't in the news as much tended to go up. That was not like-- I don't know about statistical significance on that. I ran some tests later that sort of said, eh, maybe. But nonetheless, it was an interesting pattern that appeared. - So when both candidates are underwater with the public, you just want to stay out of the news and let the other unpopular candidate do the talking. And that'll maybe increase your poll numbers. - I mean, in theory. In theory, yes. - All right, well, speaking of potentially unpopular candidates, let's move on to Biden and the midterms. All right, let's talk about what President Biden has said about his intentions to run for re-election and how that ties into the 2022 midterms. So when first asked in March of 2021, here is what Biden said about his plans to run for re-election. - Yes, my plan is to run for reelection. That's my expectation. - A week ago, he was asked on 60 Minutes about his plans and here's what he said then. - Look, if I were to say to you I'm running again, all of a sudden, a whole range of things come into play that I have-- requirements I have to change, and move, and do. - In terms of election law. JOE BIDEN: In terms of election laws. And it's much too early to make that kind of decision. I'm a great respecter of fate. And so what I'm doing is I'm doing my job. I'm going to do that job. And within the time frame that makes sense after this next election cycle here, going into next year, make a judgment of what to do. - You say that it's much too early to make that decision. I take it the decision has not been made in your own head. - Look, my intention, as I said to begin with, is that I would run again. But it's just an intention. But is it a firm decision that I run again? That remains to be seen. - As you can hear, there has been some shift in how he's talking about a reelection campaign. So I want to talk about why, whether it has anything to do with the midterms, and how his popularity will affect the midterms, regardless of what he does in 2024. So, Alex, let's start with you. Why, less than two months before the midterms, is Biden saying that a re-election run is just an intention and, quote, "remains to be seen?" - It could be because he sees himself as a drag on members of his own party, which I think is evident from his low approval ratings compared with how Democrats are-- how Democrats are performing on a generic congressional ballot. And there's only so far that you can swim above water. And if Biden is 10 to 12 points behind a Senate or House candidate in a competitive race and Biden ends up dragging those people down, he might see that as a sign to step back and let someone else run. And I think Jeff, Nathaniel, Sarah Frostenson, and I kind of answered this question a few weeks ago of if Biden doesn't run, who is the obvious, like, Democrat to replace him? And I don't think any of us came up with a clear answer there. But I think the split that we're seeing on Biden's approval versus the generic congressional ballot, you know, it kind of speaks to the range of outcomes that could happen this November. You know, it could be a Republican sweep. It could be Democrats holding onto both chambers. Or it could be something in the middle. And I think the better the results are for congressional Democrats, I think the more likely it is, then, that Biden decides to run for re-election in 2024. - We should dig into whether that is the takeaway that Biden should take away. But let's put some numbers to what you said. So Biden's current approval rating is 43%. His disapproval rating is 53%. He's 10 points underwater. Meanwhile, on the generic ballot, the question that asks whether you plan to vote for a Republican or Democrat in the midterms, Democrats are leading Republicans by 1.3 points according to the FiveThirtyEight average today. So, Nate and Jeff, I'm curious-- do you agree that this is just a potential strategy on behalf of President Biden to background himself more during this midterm cycle because Democrats seem to be outperforming him? - Yeah, that seems pretty logical to me, right? You know, if it becomes a referendum on Biden-- approve or disapprove-- Democrats do worse with that question than the choice between their party and the Republican Party. And I guess, really, kind of, like, either way-- if Biden says something definitive either way, that creates a big story about Biden and distracts from abortion and other issues that Democrats think that they have a good shot of persuading people over. I'm not sure I agree whether the midterms themselves will affect Biden's decision. That's a separate point that we can maybe debate later on in this segment. - So I think there are a couple of things going on here. It's like on the one hand, presidents aren't usually-- or first term elected presidents aren't usually getting asked repeatedly if they plan to run again, because usually, the default expectation is, yes, they are running again. Biden is unique, to some extent, in this way. In fact, I would say he is unique. He's the oldest person ever elected president. And so there is real uncertainty about whether he's going to run again. And that-- you know, that, to me, is, like, why this question keeps coming up maybe more often than it has in the past. I mean, I know with Reagan, there were some questions. But I think just, like, Biden, given his age, is getting this question even more than past presidents. At the same time, there is-- so in that sense, it's like, well, let's not give a definitive answer, to Nate's point about, like, we don't need this to become a page one story-- page A1 story. The other thing going on is that presidents don't actually usually-- I mean, candidates of any stripe-- don't usually formally declare their campaign intentions until after the midterm election. So even incumbent presidents-- like, for example, everybody knew Obama was going to run again. But he formally announced he was running again in April of 2011. So for Biden, there's also just, like-- there's, like, no need to rush making a formal decision on that. It's just that you also have this extra wrinkle of, well, people are actually a lot more interested in that question than maybe they have been for other first term incumbents, because it's a real uncertainty. It's a real question. - Do you think Biden's being asked so frequently-- I know his age is probably a factor, but the fact that Trump is kind of teasing a 2024 run too, do you think that's part of the reason why Biden's being asked about this more as well? - I mean, campaigns are starting earlier and earlier. And Trump kind of already sort of running for 2024 as part of that. But yeah, I think-- I mean, the elephant in the room is Biden's age. He is going to turn 80 in November. So he would be 82 at the start of his second term, if re-elected. That's quite old, right? You don't find a lot of governors historically who have been that old, not to mention presidents. You don't find a lot of CEOs, people in executive positions, who are that old. It's completely appropriate for the media to focus on it. I think, if anything, the media doesn't focus enough on the effects of you appoint one guy who's commander-in-chief who has a huge amount of power in the American political system, you know, I think the media should spend more time on candidates' mental fitness. - So, yeah, I mean, yes, I think his age is obviously part of that. And we don't know, as Jeff said, as you all suggested, part of this could be strategy. Part of this could be a real answer in that he doesn't want to literally announce that he's running for president again. Maybe he also literally doesn't know whether he is. Who knows? But nonetheless we're, still left with this fact that for months, Democrats have been outperforming Biden. Historically, a party's performance is pretty closely tied to a president-- the president who's in the White House's approval rating. That has not been the case. It's not been the case for months. We questioned when this became a trend when we started getting a lot of generic ballot polling, whether the two numbers would converge over time. We're now a month and a half out from the election, they still have not converged. Why is that? So set aside strategy, like, why are we seeing the somewhat unique situation with the numbers? - First of all, I think the relationship between presidential approval and midterm results is not actually all that strong historically. But that's a kind of data question we can-- - Yeah, I feel like we've addressed this. - It's strong directionally, but maybe not-- - Directionally, it's strong. - Yeah, no, I get what Nate's saying. It's like if you're going by, like, seat counts, like, there is a lot of variation and you have to consider, like, where were the parties at going into the election? Where-- you know, et cetera, et cetera. So directionally, yes, but maybe not so much in terms of-- - And directionally in this case, Biden is underwater and Democrats are above water. - Look, I feel like election analysts, or whatever you want to call them, or journalists more broadly, like, don't give people enough credit. People are able to say, I'm not thrilled with Biden, but I don't like the alternative, right? Like, that's not-- - Of course. - That sophisticated. - Well, no, but I'm curious-- - This, like, cliche-ridden-- it's like, oh, it's all about Biden. - But it does seem unique. This election-- - I mean, people are able to think about more than one thing. - Do you disagree that this election seems more unique, or seems unique compared to others in terms of the split between the party and the incumbent? Because I think it does seem unique, and I'm trying to understand why that is. - Well, every election is unique in its own way, right? I don't think that's the most interesting thing about this election. - I was going to-- not that there's, like, teams in this, but I would lean toward team Galen if there were teams. - Yes. Yes. - Wow. - I don't know, Nate and I have been at odds this entire podcast. So what I will say is that-- but Republicans have elected a number of controversial or disliked candidates endorsed by Trump in a number of battleground states-- you know, Arizona, Georgia, Pennsylvania, Ohio. And in the House too, I think Democrats have had some success propping up candidates viewed as too extreme for a general election. So it is possible that voters don't want these more controversial Republicans in office. They'd rather see Democrats elected, but then are more mixed or split on whether they want Biden to seek a second term. I mean, 2024 so far out from now. So much could happen in between then. I think the immediate is that maybe they don't want these controversial or disliked Republicans in office, but are kind of split, again, on Biden's political future. - I feel like that's compatible with both-- with what both Nate and I are saying, in a way. That, yes, like, voters are smart enough, obviously, to distinguish between an individual president and two parties. But at the same time, this is a somewhat unique situation. So let me frame this another way, Nate, and see if I can get you to play ball with me. So we have seen, like, protest movements basically against the last, I think, four presidents. We had the resistance against Trump. We had the Tea Party against Obama. We had the anti-war movement against Bush. We had the Republican revolution in 1994 against Clinton. For some reason, we haven't really seen a grassroots protest movement spring up in opposition to Biden. And I am curious if there's something about Biden that just is not particularly getting people amped up, people aren't thinking about him very much, and so he takes up less oxygen in the political space, and maybe it's easier for voters to say, Biden is one thing, set him aside, I'm going to look at the two parties. That could be pure punditry, I don't know. But there seems to be some kind of unique situation going on here. Do you agree with that at least, Nate? - Yeah. Well, I mean, unique-- I don't like that word unique, Galen. I don't like that word, unique. There's a situation-- every election people think is unique, right? We're describing why this election is this election and not a different election. I mean, I think it's, like, certainly relevant to point out that, like, Biden is not as preoccupying to people as, yeah, as other recent presidents. And maybe that explains why there's some type of apparent disconnect. But I just don't think this disconnect is something that is all that strange, right? I mean, it's not a presidential election. It's a congressional election, right? It would be strange if in 2024, Biden was really unpopular, and the other candidate was really popular, but somehow Biden was winning anyway. That would be strange, right? But this is an election for congress. It's not that weird that like congressional preferences are not perfectly aligned with presidential preferences. And yeah, maybe it's easier when Biden doesn't suck up all the oxygen in the room in the way that Donald Trump, or even an Obama, or Bill Clinton did. Sure. - You know, so when I think about this midterm election, I see sort of both points-- like, Galen, you are right-- if Democrats were to do something like gain seats in the House, which, you know, at this point, still seems more unlikely than not-- but if they were to and Biden's approval rating were in the low to mid-40s, that would certainly be striking and unusual. Because usually, when the president's party has made very small gains, like in 1998, or 2002, or even very small losses like in 1962, like, the president's approval rating was very high. And so, like, it was easy to point to that as a part of the reason why the president's party did well. If Biden's approval rating is that low and Democrats are in that position where they lose five seats, like, that would-- six seats or something, like, that would be, for a mid-term election, a real win for the president's party. I mean, that would be really striking. And I think that gets-- what comes to me is that-- what we were talking about in the previous segment has to be part of this, which is the Dobbs decision and how that has altered the usual sort of out-party advantage. I don't-- at the end of the day, the election is still in November. We don't know exactly how things are going to play out. But, you know, it's not a coincidence that our generic ballot number flipped from Republicans having an edge to Democrats having a very slight edge after the Dobbs decision. I think, you know, it shook up the electoral environment in a way that is atypical for a midterm. And whereas you can point to, like, 9/11 and the impending Iraq war with Bush's popularity before the 2002 midterm and Republican overreach-- so maybe there actually is a similarity there, in a way, between 1998 and 2022 in terms of the concept of Republican overreach maybe costing Republicans. In this case, the Republican overreach came from the Supreme Court, not necessarily from one of the chambers in congress or congress as a whole. But in that case, Clinton was very popular, and they were impeaching him, and it came off as Republican overreach. So it would be very different, obviously, if Democrats gain this time or what have you. But at the same time, to talk about Republican overreach, I could see, you know, some of that. So to Nate's point, unique is maybe not the right word. - Fair enough. So not unique, but striking. - Sure. Yeah. - Or at least atypical. I don't know. Anyway, let's-- - Atypical, I think, is reasonable. - Let's put semantics aside. There's still six weeks left. You know, there are still new developments in the economy. Interest rates are rising. The economy may be slowing. There are all sorts of different things that could happen. We're not even in October yet. We haven't even had time for an October surprise. So obviously, no one is counting any chickens before they're hatched. It's just that so far, after months of speculating that those two numbers might converge, either Biden might become more popular, and he has become a little bit more popular, or that Democrats' position on the generic ballot would decrease-- we still see this gap. Anyway, I thought it was notable. But to wrap up this segment, who-- what are we seeing on the ground in terms of how candidates are positioning themselves vis a vis Biden in competitive races? Like, do they seem to see him as-- do they seem to see him as a liability or an asset? Neither? - I mean, it seems mixed based-- you'll see lots of campaign ads in swingy districts where there's a Democratic incumbent and they'll say, you know, such and such against the Republican agenda. I fought against this. But then they'll be like, and I broke with the Biden administration to duh duh duh duh duh. So, like, this is something you see. And it's not weird. And, like, a lot of midterm environments for the president's party's candidate to sort of try to separate themselves, try to be their own man, or woman, or what have you in their race. So in that sense, it's not that strange. I will say, though, that, you know, Biden recently campaigned in, like, Pennsylvania, for instance. And now, maybe that's because Scranton is there. And we know Scranton and Joe, they're real tight. So you know, maybe that played a role. But it was interesting to me that that Biden was out doing some campaign stuff. You know, but at the same time, like, I could point to examples like-- I lived in Charlottesville, Virginia for a long time, and President Obama actually came and campaigned for the then-sitting member of the House from there in 2010, right before the election, Tom Perriello, the Democratic incumbent there. And so it's like, even in that case, Obama, who wasn't terribly popular, you know, heading into the 2010 election, was on the ground doing stuff. So it's like-- it just kind of depends. It's like maybe you bring the president, if you're the president's party, to places where you know there are lots of Democrats, in this case, or Republicans, when Trump was president, to try to rally the base. But in your campaign ads, when you're trying to get in touch with, like, people watching the 11:00 news, you're trying to show yourself being a separate-- separate from the president of your party. - I feel like every time I read a story about Democrats dodging Biden, I feel like Tim Ryan, who's running in the Ohio Senate race-- he's always mentioned, and I think on the one hand, it makes sense, you know, Ryan's running a statewide race there and Biden lost Ohio to Trump by about 8 points in 2020. And I think Ryan's spokesperson went so far as to even tell ABC News that their campaign had not asked the White House nor Biden to campaign with Ryan and had no plans to do so. But one example that's been particularly striking to me, and I know we'll talk about Georgia a little bit later in the podcast, is Senator Raphael Warnock. You know, according to ABC, again, Warnock wouldn't say, when pressed, if he supports Biden coming to Georgia to campaign for them. And that, of course, stands in pretty stark contrast to last year, when Biden campaigned for both Warnock and Jon Ossoff in their respective runoff races. - So we will see what ends up happening. Maybe that is a good indicator to watch how Biden views himself as a figure in the Democratic Party-- whether he ends up in Georgia. Nate, final thoughts on this. - What's atypical is that Democrats might have a pretty good midterm despite being the party that controls the presidency, right? To me, that's the interesting story, if it happens. And we don't know what's going to happen. There's still lots of uncertainty. That, to me, is kind of the headline and the disconnect between Biden's-- I guess sometimes when it's happened, it's because you've had a very popular president, like Clinton or Bush, right? Biden would not be that. So maybe it is a little bit-- - Exactly. That's what I'm trying to say, Nate. Glad you could come around to my position. But it could also say something about America politics overall, which is that, like, you're probably not going to have presidential approval in the 60s. And in a world where like everything is squashed, both in terms of how much they can sour on a political candidate and how much they can really like a political candidate. - OK, so here is one more thought, which is I do wonder how much of this is kind of soft disapproval of Biden over his kind of age and kind of mannerisms, right? Because, like, the policies he's enacted are, for the most part, reasonably popular. We can debate individually. But when you're constrained by having only 50 senators, two of whom are Kyrsten Sinema and Joe Manchin, then you can't do sweeping progressive change, right, to the extent the left might want. And so, you know, people might-- I don't know, it's hard to know. But I think there are people who disapprove of Biden because he's old. And that might not translate into reason to vote Republican in the midterms. - All right, well, there we have it. As you previewed, Alex, let's move on and talk about Georgia. And our colleague Elena is going to be joining us. So, Geoffrey, we're going to say goodbye to you. Thank you so much for joining us today, Geoff. - Hey, thanks for having me. It was fun as always.

This transcript has been automatically generated and may not be 100% accurate.

{"duration":"42:54","description":"The team debates whether tracking Google search terms over time is a better barometer than traditional polling.","mediaType":"default","section":"ABCNews/fivethirtyeight","id":"90582892","title":"Can Google searches tell us something the polls can't?| FiveThirtyEight","url":"/fivethirtyeight/video/google-searches-us-polls-fivethirtyeight-90582892"}