How much Americans can trust the polls
RICK NEWMAN: From Yahoo Finance, this is "Electionomics." I'm Rick Newman.
ALEXIS CHRISTOFOROUS: And I'm Alexis Christoforous. Thanks so much for joining us. As Election Day approaches and President Trump continues to trail Joe Biden by high single digits both nationally and in some key states, their respective bases are buzzing with either hope or dread that the polls could be wrong again. We saw this play out in 2016, certainly.
Here to discuss polling as we near the election is Rob Griffin. He is Research Director for the Democracy Fund Voter Study Group. Rob, good to have you on the "Electionomics" podcast again. So let's talk about this idea of polls possibly being wrong again. Who might that actually benefit in this case-- Biden or Trump?
ROB GRIFFIN: And thanks for having me. You know, I think just to dive right in, one of the most important things to know about this is that polling error isn't necessarily correlated election to election. That is to say, if you see an error one year, it doesn't necessarily tell you that you're going to have the same type of error or even in the same direction as you do the next election cycle. So in this case, I think what people sort of need to be prepared for is an error that could go either way, right? We could either be in a territory where this is a much closer race than we think it is, or it could be one where it's kind of a Biden blowout, and you know, you're just not quite sure until you actually start seeing election results roll in.
RICK NEWMAN: So Rob, there are two different ways to look at the polls, which we've all learned since 2016. One is the national numbers, which tell you the trend of the country, and Biden looks pretty good there. As Alexis mentioned, he does seem to be up in large single digits over Trump, and that's been consistent basically for the last six months.
But then there are the swing state polls, which are the ones that are really going to determine the outcome, and Biden does look pretty good there. And I've seen a lot of polls and analysis of the old saying that in at least a few of the key states, Biden seems to be leading by more than the margin of error in those polls, and he also seems to be leading by more than the polls were mistaken in 2016. So do you buy that, and what is your best guess, you know, just a few days before the election where this is going to come out?
ROB GRIFFIN: Yeah, no, I think all that analysis sounds reasonable to me. You know, I think one of the questions we all have is, right, will we see another 2016 repeat? And I think part of the answer to that is have we responded to the errors of 2016, right? So we knew that-- or we know now, I should say, that we weren't surveying enough white non-college voters, right? So that's something that by and large pollsters have responded to.
Another one of the features, which wasn't necessarily an issue with the pollsters, but we had a high number of undecided voters, and we had a turn pretty late in the election. Now, this time around, we don't have as many undecided voters, and so far, things have kind of stable. So by and large, we might not expect to see the same types of error that we saw in 2016, although it's always possible.
I guess the thing that keeps me up at night, or you know, what people should keep in mind, is we might have corrected for the issues of 2016, the question is whether we've corrected for the issues of 2020, right? There could be a whole new slew of issues that we haven't really identified yet and that we don't know about, because they're just unique to this year. And again, what's unique about this year? Well, we're in the middle of a pandemic and a recession, and there was an impeachment earlier this year, right? There's all these other things that actually might possibly upset polling in a sort of a macro way that we're just not totally aware of, and we won't know until we start seeing results come in.
RICK NEWMAN: Where would we be seeing that show up? You know, for example, we've got early voting at a much larger scale than we've had before because of the pandemic, and those, you know, people seem to be turning out in very large numbers. You know, we could have the largest turnout in a century as a percentage of the population, but nobody, we don't have a like a baseline to compare that to, because we've never had early voting in a pandemic. Can you draw anything from this, or do you just have to say, I don't know until I see--
ROB GRIFFIN: I think you've kind of landed on one of the stories here, right? So every time around what you might hear from polls is that while this is a registered voter poll, and then as we get closer and closer to the election, it's a likely voter poll, right? So this is pollsters modeling their expectations about what the electorate is going to look like and who is actually going to show up to vote.
And this year, because of voting methodologies have shifted so much, because right now, at least what we're expecting is a historic level of turnout, probably the highest level of turnout that we've seen in the presidential election for over 100 years, in the same way that 2018 was a high turnout election, it just might be the case that maybe we're getting that likely voter model wrong. And again, that's not to say it's even wrong in a given direction. Again, this could kind of go either way when you start to think about how we might be biased this election cycle.
ALEXIS CHRISTOFOROUS: You know, one of the things that we saw in 2016 was that there were a lot of voters in key states who hadn't made up their minds very close to the election, and I'm wondering if that is a game changer this time around, because Biden's lead is a little bit larger than Clinton's was during this time in 2016 and a bit more stable, because you do have far fewer undecided voters as we move closer. So do you think that reduces the chances of some late break toward one side or the other?
ROB GRIFFIN: Yeah, I think it absolutely does. You know, one way of thinking about undecided voters, it's almost like cargo that's not tied down on a ship. Do you know what I mean? Big ships can kind of tip over the boat, it can tip over expectations and really change a race if there's enough of undecided voters floating around. This time around, we're just really not in the same territory when it comes to undecided voters. There are still a number of people who will tell you they're undecided, but it's not nearly as high as it was in 2020.
And at least at this point, again, as you've said, we're just not seeing the same type of instability that we saw in 2016. That this just appears to be a much more stable election year, and you know, at least some of the analysis I've seen, to the extent that people are undecided, this is not a group of people who are necessarily still super happy with the president. That is to say that, you know, this isn't a group that's all that favorable towards him, so it's unclear how even this last group of undecideds might break as we head into the final week of the election.
RICK NEWMAN: Third party candidates are less of a factor this time around too, right?
ROB GRIFFIN: Yeah, absolutely, right. So if you put third party candidates on a questionnaire and you ask people about, you'll get a percentage of people who will say yes to it, but the traditional wisdom is that it's going to be lower than what people traditionally say, that people kind of over report third party voting relative to what we actually see on Election Day. So if that's going to happen this time, this is already much lower than it was in 2016. Most people couldn't name the major third party candidates that are running and that, you know, depending on the state, they might not even be on the ballot. So all expectation is that third party vote is going to be much lower this time around.
RICK NEWMAN: So Rob, we want to ask you about how the demographics have changed in the last four years, and how that has changed the base of voters for each party? But before we do that, one last effort to pin you down on what your best guess is on what's going to happen. So FiveThirtyEight is doing their probability model again, and they're giving Biden, you know, somewhere between an 85 and an 87 or 88% likelihood of winning. Do you think that's in the ballpark?
ROB GRIFFIN: Yeah, I think that's right. I mean, like I always like to boil this down and just to say if you had to be somebody, who would you be? And it's pretty clear that you would be Joe Biden. There's no world in which you would want to be Donald Trump heading into Election Day. You know, again, that's not to say that something couldn't happen here, right, that we could have a tightening in the race. It's possible, although that would be a historic late shift.
It's possible that there could be polling error, but again, it would have to be historically large, would have to be a combination of those two things that could lead to some alternative outcome. But again, if you had to be somebody, be Joe Biden. And again, the numbers that we're seeing from both FiveThirtyEight and "The Economist," which is the other big forecasting model right now, those seem pretty well calibrated that they're both telling an answer that's pretty close to one another, so it seems like there's kind of a convergence of the common wisdom coming from the data.
RICK NEWMAN: And then the other scenario is, does Biden win with a Democrat controlled Congress, which means the Democrats would retake the Senate, or does he end up with a split Congress? And again, those probabilities are around 55% that there is a blue wave, Democrats take the Senate and the White House. You go along with that as well?
ROB GRIFFIN: Yeah, no, that seems right. That's much more of a toss up, right? You know, I think anybody who's making sharp predictions about the direction that the Senate's going to go in, you know, that's pretty close to a coin toss right now.
RICK NEWMAN: So at the Voter Study Group, you do look at the, you do sort of what I guess you could call longitudinal research looking at how voting groups change over time, and I know you've looked at the Trump base from 2016 and just how it is changing based on if only on demographics and aging and things like that. So what do you know about how the Trump base is changing in just this fairly short period of four years from 2016 to 2020?
ROB GRIFFIN: Sure, I mean, you know, so I think the biggest thing to know about demographic change is it's kind of like an iceberg. It moves slowly, but it keeps moving election cycle over election cycle. So you know, the big story almost for the last 40 years has just been the decline of white non-college voters. And particularly since 2016 showed an increase in the educational divide between white voters, that is to say white non-college voters and white college voters started voting in very different ways. You know what I mean? That has just become more and more important.
So this vital base for Donald Trump, these white non-college voters that were the key to his success in 2016, they're probably going to decline about three percentage points as a share of the electorate between 2016 and 2020, and that's, again, probably going to keep happening election cycle over election cycle going forward into the future. So it's not impossible that Trump could put together a different coalition that leads to success, but I think it's worth mentioning that just for the Republican Party, how the coalitions are currently aligned? You just create this headwind election cycle over election cycle as one of your key voting demographics just continues to shrink in size. That's happening pretty much across the country in almost every state.
ALEXIS CHRISTOFOROUS: You know, Rob, in 2016 and then again in the midterm elections in 2018, it seems like the state polls chronically underestimated Republican strength in the Midwest and in Florida and underestimated Democrat strength in the Southwest. Is that happening this time around, or is that true this time around?
ROB GRIFFIN: I mean, it's certainly possible, and we won't kind of what the errors look like until we actually get to Election Day this time around, but you know, if it were to follow the pattern of 2018, 2018 in terms of polling errors of looked like sort of looked like 2016 but sort of smaller. That is to say that the same types of errors were there. They were happening in the same places, but they were smaller than they were in the 2016 election.
So you know, we could be on course for something like that, where again, it just looks like a smaller version in 2016. But again, just noted here how much uncertainty there is, it could go in the opposite direction too where maybe all of our likely voter models are kind of off, and the fact that everyone has changed methodology in terms of how they're voting this year, voting by mail, voting early, in-person, just really sort of shifts the electorate in a way that we can't really predict so far.
RICK NEWMAN: So to go back to your point about the decline of white non-college educated voters, why is that group shrinking?
ROB GRIFFIN: So there's essentially two forces at play. The first is that we've got a rapidly sort of racially diversifying country, so right, so there's immigrants coming in from parts of the world where the immigrants are essentially not white, right? So Central and South America as well as parts of Asia, these are the big growth communities in the United States. So part of it is just a racial diversification story as well as people aging into the electorate, so the children of first and second generation immigrants that came here, you know what I mean, like 20, 30 years ago, right, they're having kids who are suddenly becoming of voting age. These forces are essentially causing less and less of the population to be white over time.
And the other piece of it is rising educational attainment. So the number of people who are getting college degrees over time is just going up, up, up. So those two forces combined has resulted in the share of voters who are white non-college declining a really substantial amount over time. It's pretty much close to like 36 points between 1980 and 2016. And again, that's just something that we expect to continue in about two to three points per election cycle, at least for the next like 16 to 20 years.
RICK NEWMAN: So if you were President Trump in 2017 taking office in the White House and you were aware of these demographic changes, which are not, as you point out, not new within the last four years. You could have seen this coming. Is there any obvious way you might have said to yourself, just based on the kind of data you look at, Rob, here's what I have to do to expand my base. These, you know, these are the groups. I have to start going after to counter set the people that we're going to lose just because of demographic forces.
ROB GRIFFIN: Yeah, I think there's two routes, and there's the one that I think he kind of went and the one he kind of did. So we'll start with the one he did go, which you know, I think one of the ways that he thought about expanding his base was just to double down on white non-college voters, right? So they were about 44% of voters in 2016. Even in 2020, you still expect them to be like 41, 42% of voters. That's a big chunk of the electorate. If you could gain another five points with that group of voters, right, that would be a huge windfall for the Republican Party.
RICK NEWMAN: I mean, but then you would have to get more of them who didn't go out to vote, right?
ROB GRIFFIN: Yeah, or either increase the turnout of the group or just increase your margins. Just do five points better, make the Democrats do 5 points worse. That would be a strategy that would pay tremendous dividends if you could pull it off, because white non-college voters are one of the best geographically distributed groups in the country, right?
So we can talk about the strength of white college voters or Latino voters or Black voters, but they're often sort of geographically located in places that won't necessarily be big tipping point states. They won't have a huge impact on the electoral college. White non-college voters are important just about everywhere. There's almost not a county in the country where they don't make a big difference and certainly not a state where they don't make a big difference. So that's the route I think he did go.
RICK NEWMAN: He did, so he did. And we don't know yet whether he will get a larger chunk of that voting group or whether he will increase turnout, but you can tell from his strategy that he does seem to have gone after them?
ROB GRIFFIN: It does seem like that was the intent on that. He took away, the lesson he took away from 2016 is immigration was a huge winning issue, you know, campaigning on cultural change and racial change in the United States was a winning issue for him, and I think he tried to double down on that throughout his presidency. Sure.
RICK NEWMAN: So you said there was another pathway that it looks like he missed. What was that pathway?
ROB GRIFFIN: Yeah, and I think the other pathway is actually, you know, trying to win more of these growing populations, right, expanding the base in a way that the people have been talking about for a while now with the Republican Party, to go after more white college who are still a pretty stable portion of the electorate, large portion of the electorate, go after growing populations like Hispanic and Asian-American voters, go after African-American voters, go after younger voters, right? So get better margins among Gen Z and millennials who are a growing portion of the electorate every single year. That's going to just keep happening, but that so far have been deeply, deeply Democratic in nature.
ALEXIS CHRISTOFOROUS: Rob, what happened with Trump and suburban females? Right on the stump recently, he actually said, why don't you like me as sort of tongue in cheek. But what went wrong there or is going wrong there with that demographic?
ROB GRIFFIN: So I think it's tough to disentangle, right? We're kind of in a period where so many things are happening at once that it's actually tough to sort of tease out precisely what's driving changes, right? So we've got, again, an impeachment, pandemic, a recession, one of the largest protest movements in American history as was the president's responses to all these things, that could be things that are moving people around.
I guess what we had seen so far is that those events, these sort of let's call them post-February events, seem to have had some impact on how people view the present and how they're planning to vote. But even before all of this, there was some level of dissatisfaction with his presidency, right, that this was not a person, who even if you go back to like July of 2019, was already trailing Biden in these match-up questions that we asked, right? So there's a little bit of this is already baked in, and then a lot of the events of 2020 probably made it worse for him. So I mean, you know, I think it's hard to disentangle.
One of the things that I've been focusing in on and I've noticed a lot is Trump lost the mantle of moderation, right? That there was this view in 2016 that he was actually the more moderate candidate relative to Hillary Clinton, and Joe Biden is actually seen just about as moderate as Hillary Clinton is. If you ask those people this question, they're actually seen just about equally as moderate.
The big change has been among Trump. People no longer see him as a moderate figure in American politics when you ask them about this. So he's lost that label that so many people might be using to vote, right? There's just this common wisdom that moderation is rewarded at the ballot box, and that holds up when you start to study this stuff that people, generally speaking, the, you know, moderates will get a higher vote share, because it's just something valued in American politics.
ALEXIS CHRISTOFOROUS: But when you say no longer necessarily the moderate candidate, talk to us a little bit more about that. Is he being viewed as the more conservative of the two?
ROB GRIFFIN: Yes.
ALEXIS CHRISTOFOROUS: I mean, you just get-- you have Amy Coney Barrett now going to be, you know, part of the Supreme Court. Is that obviously part of his legacy, but that has to play into voters' thinking, I think.
ROB GRIFFIN: Yeah, I think it's more people see him as conservative than they did back in 2016, and you know, I think some of that, you know, we could talk about recent events in 2020, but this was true even back, you know, in 2019 or the years before that some features of his actions over the course of his presidency has just caused people to reassess him and no longer see him as a moderate candidate, which he was seen as, again, back in 2016.
RICK NEWMAN: So Rob, another voting group that seems important here, older voters, seniors. I think Trump got them by a considerable margin across all income and education levels in 2016. You can correct me on that if I'm wrong, but he seems to be losing them in 2020, and it's not hard to imagine why because of the perception of his poor handling of the coronavirus.
ROB GRIFFIN: Yeah, and let me-- I'll back up and give a little historical context, and then we can talk about 2020. So historical context, I think the image that a lot of people have in their head is that younger voters and older voters always voted in really different ways. This actually wasn't true. You go back to even 2000, and there wasn't that much of a difference between how younger voters and older voters were voting, so this age divide, this age gap is something that's developed just in the last 20 years, really, and it's never ever been as big as it is now essentially, or at least as it was in 2016.
What we're seeing in 2020 is that age gap kind of collapse, that older voters, for whatever reason, we're not quite sure yet, have started leaning away from Trump and towards the Democratic candidate Joe Biden. So I think one of the stories that people tell about this is, again, this sort of post-February story of pandemic, recession, maybe a response to the protests that were happening around the country, as well as the police reaction to them, and I think there are elements of truth to that. But the truth is even before those things, it appears that seniors had already kind of abandoned Donald Trump, that they were already, a majority of them, were already voting for Joe Biden or leaning towards Joe Biden.
Or at the very least, if they didn't have a lead among them, that Trump had lost ground relative to 2016. So I think there's sort of a "yes, and" story, right? That COVID and maybe the concerns that are particular to seniors as well as, again, all these other events post-February, these are contributing to how seniors are evaluating him, but there was something happening even before that that, again, was already kind of cooked into the numbers.
RICK NEWMAN: You think it's the same thing you just mentioned? People view him as less moderate, and that would be true for older voters as well, or do you think it's something particular to older voters?
ROB GRIFFIN: I think, you know, I think there's a good story to be told about it being that moderation. You know what I mean? You know, again, it's a bit of a summary statistic for how people are assessing candidates, and you know, maybe Trump was walking in just being seen as sort of, you know, atypical, not your usual candidate. He was running as a Republican, but this is a guy who used to be a Democrat and isn't he really a business person first and foremost, and he's there to make great deals. And you know, all these things that might suggest some level of moderation because of the unusual nature of Trump. And again, that label, that impression that he had with the American public, it just appears that he's lost it over the course of his presidency.
RICK NEWMAN: So Rob, in just the few minutes we have left, we're talking here about Biden and Trump who represent the Democratic and the Republican Party, but they're not exact overlaps, and there's going to be a lot of analysis after the election based on the outcome of how each party is changing and perhaps how either one or both of them needs to do damage control to repair itself. We won't know a lot until after the election, but is there anything you can see that are obvious changes between who considers himself a Democrat and who considers themself a Republican between 2016 and 2020?
ROB GRIFFIN: So I think there's a question there about whether 2020 is going to be a good model for thinking about people's long term political identity, right? You might have people who are voting Democratic this time around, but it's really an evaluation of Trump, right? It's really some statement about his presidency rather than a long term behavior that they're adopting.
Now, I think where the exception might be is among younger voters, because what we know about younger voters is that the political environments that they grow up in, that they come of political age in, are formative, right? They start to inform how they think about themselves for the rest of their life. So if you have an entire political generation coming of age essentially during a Republican administration where things have gone pretty pear shaped, that's going to leave potentially a lasting impression on how they view themselves and who they view as a political party that can be effective in office, that can speak for their values, that can run the country efficiently.
RICK NEWMAN: So I'm a registered independent, and I always like to make the case, everybody should be a registered independent. Don't put your lot in with either party. You know, think for yourself and so-forth. Is there any chance-- I mean, independents are like 40% of the electorate at this point, and I know they tend to either lean D or lean R. But have you see anything that increases the likelihood of a meaningful third party emerging from all this?
ROB GRIFFIN: Not especially. I mean, I think one of the things, as you kind of even already mentioned, is that we do have this number where it's like 40% of the electorate identifies as independent. But also, as you mentioned, they lean towards a political party. What we know about these leaning independents is that if you take a look at their behavior, if you take a look at their attitudes, they look very much like partisan. That is to say that they vote consistently Democratic when they say they lean towards Democratic Party. They vote consistently Republican when they say they lean towards the Republican Party.
So they're kind of partisans in all but name in some way, as near as we can tell, and that is a group that if you want to think about that, the true independents underneath, the people whose behavior actually varies election to election or actually have more of these moderate both sides opinions, that's a group of people that has been shrinking over time. So at least at present, it's not clear that there is like a coherent third party coalition that somehow could leverage this 40% of voters who actually appear underneath the hood to be made up largely of partisans, right, people who are in all but name partisans and have very different views on things.
ALEXIS CHRISTOFOROUS: All right, we're going to leave it there. Pretty soon, we're going to know the answer to whether or not the polls are right. Rob Griffin, Research Director for the Democracy Fund Voter Study Group, thanks so much for joining us on this podcast, and be sure to rate and review what you just heard and saw. And as always, you can follow me @AlexisTVNews.
RICK NEWMAN: And me @RickJNewman, and boy, are we going to have a discussion on there on November 3. Rob, you want to put a social media handle out there?
ROB GRIFFIN: Sure, and you can follow me @RP_Griffin.
ALEXIS CHRISTOFOROUS: All right, everybody, thanks so much, and get out there and vote. Even if the lines are long, stick to it, make it happen, and make your voice heard. Thanks for joining us.