Despite relatively strong economic data, the public is sour, judging President Biden poorly and putting him even or below former President Trump in early polling. How will polls and economic assessments evolve as we approach the 2024 election? Robert Erikson finds that early polls are not predictive but that presidents will eventually be judged by their economic standing. He’s co-produced some of the most important scholarship on using polls to predict elections and understanding how citizens judge the president based on economic expectations.
Guest: Robert Erikson, Columbia University
Study: The Timeline of Presidential Elections
Transcript
Matt Grossmann: Previewing 2024: How Voters Judge Presidents, this week on The Science of Politics. For the Niskanen Center, I’m Matt Grossmann. Despite relatively strong economic data, the public is sour, judging President Biden poorly and putting him even or below former President Trump in early polling. How will polls and economic assessments evolve as we approach the 2024 election? Is it all different this time, or are we likely to repeat historical patterns?
This week, I talked to Robert Erikson of Columbia University in a special edition previewing 2024. Erikson’s data is what tells political observers that early polls are not predictive, but also what presidents are judged by their economic standing. He’s produced some of the most important scholarship on using polls to predict presidential elections, and understanding how citizens judge the president based on economic expectations. We also talk about the fate of Republicans’ electoral college advantage, the divergence between the real and perceived economies, how voters balance the president and Congress, and whether we should still rely on polls. I think you’ll enjoy our conversation.
So where are we in the timeline of presidential elections? How good are polls at predicting the outcome now, and why aren’t they better?
Robert Erikson: Better is an interesting frame of reference. Polls right now, more often than not, are now showing Trump beating Biden, and the trend has been pro-Trump ever since 2021 began in the match-ups between the two. But it’s still close if you take any average of the polls. So let’s say it’s 50-50. Or we say, “Yeah, Biden’s ahead, but…” or it should be “Trump is ahead, but Biden’s probably slightly favored,” something like that. Then you say, “Well, what about the fact that Trump’s under indictment? Why isn’t Biden getting supported for good times or good performance if you see that?” and so on, and all those indictments. You might say, if you’re not looking forward to a Trump return to the presidency, it might seem very disheartening.
You look at history, and basically, this time of year, the [inaudible 00:02:24] polls, to the extent we have polls in December before the election, they don’t matter. If you take polls in January of the election year, with the two competing candidates, once we eventually get the nominations, that the correlation is about zero. And so basically, early polls in January have no predictive power. And so that could be reassuring to some people. On the other hand, we ask the reason why that is the case is because people generally aren’t tuning into a presidential race early in the election year. A month from now, it’ll still be way too early. But what might be different here is that people might be more attentive this round than normal. And of course, people know Trump, he’s not a stranger. So if it was Biden versus Nikki Haley, let’s say, I would say, “Well, we don’t know. People don’t know Nikki Haley yet, so we have no idea what’s going to happen.”
But people know Trump. So basically, people have already made up their minds. Of course, you have the age of polarization, Howard. Look at earlier examples electors would support. So at least maybe we saying we’re going to have a close election. But then again, you can have some arguments why it won’t be the case, because we have two elderly candidates who don’t know how well they’re going to perform on the stage. Obviously we don’t know for sure either one is going to be the candidate. They probably are, but there’s a chance each one would not be. I think Biden could still retire, drop out, if he thought somebody else had a better chance of beating Trump and he was really objective about it, or if he had worrying health issues.
And presidents had done that before in election year when they thought they better run. Most recently, Lyndon Johnson surprised everybody in 1968. He could run for another term, and in March he surprised the nation by not running. Truman did the same thing way back in 1952. So presidents deciding not to run when they could, that’s happened before. And of course, Trump could be caught, because it’s happened before. Frontrunners have been caught in the primaries. People say it’s not going to happen, but it could be. DeSantis might be flailing a little bit, but it’s Nikki Haley who’s got some increased financial support, maybe a little wind at her back.
Matt Grossmann: So the usual pattern is that voters learn more about the candidates as we get closer to the election, they start paying more attention, and so the polls get better. But we’re really, looking at a rerun of last time here, it seems like voters should have good opinions of the candidates so far. Can we really compare any historical cycle to this one in terms of the likelihood that voters have good information to make this choice already?
Robert Erikson: Yeah, so let me just add one thing, that… Think of all the reasons for uncertainty. Obviously, I do not believe… I should mention the trials that may be occurring before the election. And you’d think that that would have an impact on the race. Think of the reasons why we really can’t predict, so many reasons we really can’t predict what’s going to happen, and that would be one of them.
Matt Grossmann: So you don’t think that this time is different in that voters have pretty firm opinions of the not just [inaudible 00:06:08]?
Robert Erikson: Oh yes, oh yes, oh yes, they do, but they could change. I suppose I could produce this hypothetical, okay? Here’s a hypothetical. It’s obviously hypothetical. Suppose it turns out that Joe Biden and Hunter Biden really are the worst crooks ever. Suppose it turns out that the worst claims about Biden that people like Rudy Giuliani say turned out to be true. You got to face the facts. Wouldn’t Democrats turn on Biden in that situation? It’s a hypothetical now, but what if we’re all wrong? This nice old gentleman is really this crooked pol. If that were true, and similarly, if people don’t want to believe the worst about Trump and they have to face it, that could be a different result. So we don’t know that’s going to happen. I’m going to say that’s in the realm of possibility.
Matt Grossmann: So we’re usually in the business of predicting the popular vote, but of course, there’s been a divergence between the electoral college results in the popular vote in the last couple of elections favoring Republicans, meaning they haven’t had to win a majority in the popular vote to win the electoral college. There are some initial signs, comparisons between state and national polls this time, and Biden’s relative strength with older white voters that are leading some folks like Nate Cohn at the New York Times to suggest that the electoral college advantage for Republicans may go down this time. So how should we think about that advantage as it’s evolved so far and its likely future?
Robert Erikson: Oh, okay. I think we might want to concentrate on just a few states. We think of, let’s say, we’ll start with Michigan, Wisconsin, and Pennsylvania. Traditionally, generally in the Democratic corner turned into Trump in 2016, but back to the Democrats barely in 2020. Okay. And of course, Biden also barely won Arizona and Georgia, which may have been slightly unexpected, but with 52% of the vote nationally, two-party votes, 52-48, Biden barely won the electoral college. Probably if Trump got a half a point more across the board, it would’ve been a tie in electoral college. In other words, the advantage was about 51.5%. That’s a benchmark for the Democrats of 2020 anyway. You have to win %51.5 of the two-party vote, not 50%. So that’s a bit of a hurdle. Republicans only have to win 48.5. So we don’t know what it’s going to be exactly in 2024, but we could guess.
As recently as 2012, the Democrats had the advantage. You might recall that Obama got virtually the same national vote that Biden did, in 2020 for Biden, 2012 for Obama. And Obama had no problem winning the electoral college. He won the three pivotal states in Wisconsin, Pennsylvania, and Michigan, plus he won Ohio, he won Florida, Iowa, I think. So basically he won a whole bunch of states that Biden could only dream about, and yet with the same amount of votes. So there, if the votes had been shifted a little bit, so that was a tie, let’s say in a popular vote, Obama would’ve won. Basically, Democrats under Obama had the electoral college advantage. So it can shift. Obviously, it depends on a few states what happens. So the thing is that Michigan and Wisconsin and Pennsylvania and recent elections in both 19… excuse me, both 2020 and 2016, in both elections, those three states have been slightly more Republican than average instead of slightly more democratic than average, and that made all the difference.
Other states that were been close in the past, like Florida and obviously Ohio for example, used to be battleground states just went more in the Republican direction and may be unattainable for the Democrats. So how the states move now Florida, or let’s say Texas, one of those large states, Ohio, if they start moving back toward being more democratic than average, that would tilt the electoral college advantage again. So what’s going on in Michigan, Wisconsin and Pennsylvania? That’s the question, I think. And of course, they tend to have older voters. I’m not sure that that enough is going to make a difference, but we keep an eye on it.
Matt Grossmann: So you normally predict election outcomes, not only with polling directly on the election, but with economic statistics and particularly these leading economic indicators. And I noticed that they’re not quite as positive as some of the other indicators people are using, like employment and disinflation right now. Why are they traditionally predictive and what do you make of their predictions for 2024?
Robert Erikson: Well, if you try to predict the election from the economic conditions, unemployment rate, and inflation rate, but actually not the best. Probably the best predictor would be, let’s say, per capita income growth as you go from quarter to quarter, something like that. That’d probably be the best single predictor. Maybe inflation, secondarily. But obviously inflation, we’re talking about inflation, obviously inflation is slowing, but what might be important is the level of prices that people still feel from past inflation. But per capita income growth, I haven’t looked at that carefully, but I don’t think income has grown as much as you might think given the full employment. And that might be, that indicator is not as prosperous, let’s say, as we might think it might be from what we might look at. I think in one recent quarter it actually went down from looking at the preliminary data. So that could be why. Maybe more people have jobs, but they’re not great jobs, they’re not happy. That might be part of it.
But in any case, the other way to look at it, of course, would be the subjective indicator, so what people think the economy is doing. And it’s not great, people don’t think the economy is great right now. Normally the subjective economy of people think how it’s been going the past year reflects how it goes in the past year, and it hasn’t done that so much.
And of course, when people are asked how Biden’s doing in the economy, that’s one of its worst areas. 30-some percent maybe give up high marks for the economy when we have record unemployment, record employment, I should say. And inflation is now seemingly under control. So it adds a little bit of mystery to it to why Biden doesn’t get credit. Maybe it’s the same reason he doesn’t get credit for foreign policy prowess, because maybe it’s all the polarization that people can’t change their minds. Maybe it’s that people see it through the filter, that it’s just an old man struggling. I don’t know what it is. Or it could be a simple lack of charisma, he doesn’t basically excite people and people might fall asleep during his speeches.
Matt Grossmann: So there is a bit of a debate about income growth, and part of the reason is because of the stimulus programs that expired in 2022. So the people who say that this is important say that, well, no wonder people aren’t very happy because they got temporary gains in 2021 and then they lost them in 2022. Of course, they haven’t continued to lose them in 2023. But, same kind of question people have about inflation. Is it just the latest inflation or are people comparing to a previous era before Biden was elected? So what do we know about that? How long are voters memories? Is it really just a year of the economy or is it-
Robert Erikson: That’s a good question. People are coming to that conclusion, I suppose, that basically that’s how inflation works. Let’s say, I myself might try to make a purchase of something I haven’t purchased in a long time and find, that seems high. It still goes on, right? We still think prices are high compared to the way it used to be, they’re just not growing even higher.
Of course, you can say, well, the President doesn’t have that much to do with it, so he’s just a victim of events. But part of it is too, that’s not basically events that because how you play the cards you dealt and basically, even if he explain the cards well, people don’t necessarily see Biden doing that.
Matt Grossmann: So you have looked at this divergence between perceived and real or as measured economic performance before. And of course they’re usually more associated than they are now. But there were a couple of elections, I think that you found, 2008 and 2012, where there was a divergence. And in those, I think you found that perceptions weren’t all that important by the time we got to the election outcome, is that-
Robert Erikson: Well, the perceptions do become important as [inaudible 00:16:21]. So basically early on they’re not that important. Okay, so let’s say this is one of the most stunning facts, I think that Chris Elijah, and I found in our research that if you look at perceptions of the economy, let’s say April of the election year, it doesn’t even predict the poll results in April. If you’re looking at perceptions of the economy in October and November, they of course predict the election pretty well, obviously not perfectly. But somehow between April and November that the perceptions crystallize in a way, or more importantly, the vote moves in the direction of perceptions. So, that would be the way it goes.
So if let’s say Biden could get the perceptions to match what some people say is a good economy, that would have a bit of a delayed effect, but basically between April and November, that’d be the trajectory in favor of the good economy. Or even better yet, if you had a good economy, and let’s say you want to take an objective measure of what’s going to happen, be the leading indicators. But as you know, the leading economic indicators are looking at goods right now, but that doesn’t mean they’re not going to look good percentage-wise in a couple of months. So it’s possible. So that’s a bit of uncertainty that could work either way. In a past election, sometimes we see a president zooming to a landslide, often coasting on a good economy that’s growing during the election year, but it’s not necessarily foreseen statistically the year before. Like right now, the economy right now doesn’t have much predictive power, no matter how you want to measure it. So much is in the future, I want to see the objective economy in the future. We start looking toward the leading indicators.
If they start looking better, that’s good for Biden. If they don’t, that’s bad because it suggests the objective economy is going to not be good. It usually follows the reading indicators a few months later. If they are good, does the perceptions begin to follow? And if they do, and Biden’s not … So if they do begin to follow and perceptions are, hey, the economy’s good, but the polls aren’t showing it yet, wait a while, and then maybe Biden would be happy. So in other words, the events are sort of layered. You have to watch the different signs of leading indicators, the objective economy, the perceived economy, and obviously watch the polls along the way.
Matt Grossmann: So as you mentioned, polarization is a threat to some of these predictive relationships of the past. In particular, economic perceptions changed real fast around election results.
Robert Erikson: That’s right, that’s right,
Matt Grossmann: In 2008, suddenly Republicans hated the economy more after the election and Democrats liked it more. And then in 2016 you saw the reverse, and in 2020, the reverse again, and we’re talking big moves, like 40 point moves in how people feel about the economy just based on who won. So it sure seems like partisanship is coming first here, not that people are updating their political attitudes based on their economic attitudes.
Robert Erikson: Yeah, absolutely. Obviously, there’s a signal in the perceptions that the economy should not ignore it, but particularly if you look at individual perceptions, they’re obviously tinged by partisanship. And so for example, let’s say let’s, the example I would use, let’s say it’s 2008, it’s fall of 2008, you were interviewed about the election. You wouldn’t vote for McCain even though you have a Republican president, Bush, McCain trying to succeed him. And you’re asked, how’s the economy going? You might say it’s going great, because that’d be the rationalization, and if you did, we would probably dismiss it. Nobody would think the economy is going great and vote based on that perception.
You might think the economy is going great because you’re totally ignorant, but that wouldn’t be the basis of your vote. So basically you have to take the average and basically, I suppose we should look for the independent voters. Now, from the data I’ve seen that these partisan divides on the economy, which are great these days that you point out weren’t that bad not that many years ago, basically Democrats and Republicans saw the same economy, not so much anymore. But obviously there’s lots of independent voters, and they’re the ones who are going to decide the election anyway, and we should figure out what they have in mind.
Matt Grossmann: So you’ve also historically argued that citizens make prospective assessments rather than just responding to the current or past state of the economy.
Robert Erikson: Yeah.
Matt Grossmann: How well do these findings hold in the current era, and what do they imply about Biden’s approval?
Robert Erikson: Well, first of all, assessments of the future economy prediction, how people see the future economy matter the most, really matter whether the people approve the president’s performance or not. So people think the economy is gloomy. Now, that’s going to stay bad or get bad. That hurts Biden more than what they think has happened recently. But if you’re trying to predict the election perceptions of the future economy aren’t that good, don’t work that well. The reason is, if you think, if people generally think the economy is going poorly, they might think they have a reason for optimism later because they think the economy’s going poorly now, they think that the president might be replaced, which is a reason for optimism. So let’s say if you were back in 2008, you were a voter, thought the economy has been going bad, you might say it’s going to get better because we’re going to elect Obama.
Matt Grossmann: So there’s also a divergence between how people today describe their own personal finances and the economy as a whole. I think traditionally the perceptions of the economy as a whole have been more important. Is that true? And if so, why?
Robert Erikson: Yeah, that would be important because if people try to take into account how the government is affecting them, and let’s say they have a windfall or bad economic luck, they win a lottery or they don’t get fired, they don’t blame the president or praise the president. But if they’re aware of the economy around them to realize that could affect them, the economy around them can affect them, and maybe they think the president is responsible for that. And so that’s why the overall perception of the overall economy seems to be more important than your personal economy when it comes to presidential approval and also election results.
Matt Grossmann: So we’ve been talking about it as kind of an evaluation of performance of the current presidents, but there might be some ideological judgment going on as well. And I know that you’ve found that people do judge president’s ideological positions relative to their own, and that we might have an update as to if the president’s gone too far in one direction from some of this.
So right now we have a liberal but not left moving public, if you believe the latest Stimson public mood measures. And Biden’s position, maybe he’s perceived as less moderate than he was when he was elected the first time relative to the Democratic Party.
How do those fit in?
Robert Erikson: Yeah. So I think that the ideological positioning matters more than we might think. How to apply that in this case is difficult because it’s not all ideology, of course. Not claiming that at all. But Biden’s policy is relatively moderate for an electorate, which is actually a little more liberal than sometimes in the past right now. So that should help Biden.
On the other hand, a couple things; the Democrats don’t get such a good rating as a party these days for performance and some of the readings of party identification. The Republicans are really getting close to par with the Democrats, whereas historically, that’s not the case. So that would be a worry for the Democrats. It’s obviously … Ideological proximity, people want a president who is ideologically compatible, but they also want a party that can run the country. And somehow the Democrats seem to be failing that test to some extent, according to public opinion polls. That can be frustrating if you don’t believe that, but that’s what a lot of polls show.
The other thing is, if you’re judging Trump, what’s his ideology? He’s not an ideological person. He’s not about to disband social security or Medicare; at least he says he’s not. There might be a dog whistle in that direction, but that’s no more than that. And so basically, by traditional kinds of conservatism, maybe he’s not so much a conservative. More like a nationalist, let’s say, and hard to classify by the traditional conservative liberal dimension.
Biden, on the other hand, is a classical, I guess, moderate Democrat, a moderate left Democrat.
Matt Grossmann: So we’ve been talking about this as a US pattern, but obviously there’s evidence in democratic systems everywhere that ruling parties are judged partially on how the economy is doing, and there does seem to be a pattern now which is not good for ruling parties; that is, Biden actually has the highest approval in the G8 vote share. The post-COVID leadership seems to be declining, whether the ruling party is on the left or the right.
How should we interpret that? Does that mean that we are kind of interpreting this economy wrong; maybe inflation has more negative effects than we know? Or can we extrapolate from that to what’s likely to happen here?
Robert Erikson: Yeah. I suppose I can only speculate, but over the past decade, at least starting with Brexit, maybe before, we’ve had this rise of populism across democracies. And so the glory days of democracy after the fall of the Soviet Union and the satellites has sort of ended, and so former Soviet republics have often backslid, and other democracies have. And then of course, that’s in terms of moving toward a more authoritarian right.
And then in Western Europe, we’ve had a sort of populous movement, which is toward the right. Certainly the Brexit movement in Britain was an example. I mean, for example, I think the Prime Minister at the time, Cameron, made a big blunder by calling for a referendum on belonging to the European Union, and he lost big time. Should have been aware there’s a big undercurrent of opposition to the EU in this country.
And we see it in other European countries now; perhaps in the Netherlands and other countries. It’s hard to find a country in Europe where you can say that old-fashioned liberal democracy is totally healthy. Often it’s the right wing that’s basically more dominant than usual.
And of course, Argentina is another example where … And in a lot of countries, things like immigration support, obviously inflation, as in Argentina.
So it’s a variety of things that are basically a worldwide phenomenon, I think.
Matt Grossmann: So we saw the pattern also after the Great Recession, that ruling parties didn’t do well if they were in power at the time of the Great Recession.
Robert Erikson: Yeah.
Matt Grossmann: Seems to be true very recently this time that Biden’s not that popular, but neither are the ruling party’s not popular in Canada or the UK or Australia, New Zealand.
Robert Erikson: That’s right. That’s right.
Matt Grossmann: So it seems like maybe there’s some general unhappiness beyond just right/left trends that-
Robert Erikson: Yep. Achen and Bartels, in their book, make one interesting point that during the Great Depression of the thirties, whoever was in power, left or right, became unpopular. Whoever it is. It’s not ideological, like left or right, but basically, “You’re in charge. We like the other team better.” And that would be the direction where elections went at that time.
Maybe the same thing is going on right now, and maybe in some countries, people on the left were unfortunate enough to be in power at the wrong time.
Matt Grossmann: So we’ve been talking about the presidential election so far, but obviously we’ll have elections for the US House and the US Senate next year, and certainly they’re going to be strongly associated. But the Republicans in the house did outperform Trump in both of the previous elections by a few points. And I know that you’ve found in the past that you kind of have two phenomenon; one where the congressional vote goes with the president’s party, and another where maybe people might anticipate the president would get elected and vote for the other party.
So explain how those dynamics work and how it might matter next time.
Robert Erikson: Okay. So there’s two phenomena going on. Presidents have coattails. We’re familiar with that, that if people are going to vote for the president, they’re more likely than not to vote for the president’s party for other offices like Congress. So basically, say the Republicans win in a landslide for president, they’re going to better an average for Congress because of coattails.
On the other hand, if they think the president is for certain going to be, let’s say, a Republican, they might want to say, “Well, it’s more reasonable [inaudible 00:31:46] democratic to balance.” So we have two processes going on.
So let’s say if a president wins in a real landslide, his coattails would have to give him a majority, let’s say, in Congress, but not so much than if people didn’t see the landslide coming.
So for example, presidents often have a better Congress when they first get elected because people don’t anticipate it, they think it’s going to be close. They don’t vote to expectations, they just vote for coattails. Then in the second election, when the president wins by a bigger margin, there are coattails, but they’re offset by the … The coattails are still there, but they’re offset by the balancing that people might be cautious about supporting that party which they’re certain is going to get reelected. So if the election’s really close and people see it’s going to be close, that [inaudible 00:32:44] gets offset. It doesn’t matter.
But the interesting circumstance would be if the election is seen as one-sided, but it’s an upset. The party expected to win loses. Like let’s say 2016 when Trump won in an upset, that didn’t get him a special bonus because the winning upset president gets extra congressional seats from voters who were voting for his party to balance the loser who did not win.
In other words, a lot of people voted Democratic in 2016 for Congress, not a lot, but enough did, to … Voted for Hillary Clinton … No, excuse me. They voted for a Republican to balance the fact that Hillary Clinton was going to be elected president, so they thought. Obviously she was not.
The biggest instance of that case was back in 1948 when the famous upset where Dewey was supposed to beat Truman and the Republicans were supposed to win the presidency, and the Democrats won Congress, won the House of Representatives by gaining something like 80 seats or something like that. Some tremendous gains in the house. They were voting democratic to block President Dewey, and yet instead they elected Truman with a big democratic Congress.
And so if the perceptions are a factor, but they’re wrong, you sort of get the unintended consequence. In that case, the Democratic president upset, what apparently the elected wanted, but with the big Democratic majority, which they did not intend, assuming that Truman would’ve won.
So how does that affect 2024? I don’t know, except that if we’re sure that Biden’s going to beat Trump and he doesn’t, Trump will have his majority in Congress.
Matt Grossmann: So there used to be a debate about whether we should be using, for forecast, things like poll results, which are directly asking people about the election, but aren’t necessarily what we used to call the fundamentals; things that might matter for the election outcome independent of just preferences over the candidates.
What’s the state of that? Have we learned anything about how much these fundamentals matter? Or what is a fundamental?
Robert Erikson: That’s a great question.
I think that early in the campaign, and maybe at this stage pre-campaign, we’d say, “Well, the polls show such and such”, maybe something like a close election, “But the fundamentals favor party X.” And so we’d predict with some confidence that the poll trend between now and election day would move in favor of party X because the fundamentals favor them.
Well, what are the fundamentals now with the … Yeah, the economy’s pretty good, [inaudible 00:35:52] Biden’s got to get better, but the people don’t think so. We might take into account … Oh, people are afraid that democracy’s in trouble that Trump wants to be a dictator, or something, and people take that into account. All these trials coming up, these are part of the fundamentals, I guess, now, with Trump. Or not. We don’t know until after the fact, I guess, what these are going to be. But normally, we would say the economy is an obvious one, and all the signs go in one direction. But we’ve been talking about how conflicting the various signals are. So we don’t know what those fundamentals are, and I think that’s why we have a certain degree of uncertainty.
Things can happen. Like in 2020, you couldn’t predict that election because of COVID, and the lockdowns, and the government response. You couldn’t use any normal indicators, at least I wouldn’t.
Matt Grossmann: So one alternative is to go with prediction markets, which can incorporate some of this additional information. But in previous work, you’ve shown aren’t necessarily better than polling, if you just account for how the polls are likely to move over time. So is it the best that we’re going to do even though it’s not very predictive at this point? Or…
Robert Erikson: Oh, I’d look at the prediction markets even though they typically reflect the polls. So there’s a question if they offer anything beyond the polls. Like the polls, do the polls predict the prediction markets? But basically, with a little tweak, which might be wrong rather than right. And so I may pay attention to them, particularly at an early stage. If you can anticipate certain things that might happen and their effect. Like right now I’d say, what’s the effect of all these trials going to be? Or what’s the effect of Biden’s age, or whatever you want it to say, it’s all reflected there in the prediction market, you’d think, and so you might want to take that into account.
As the campaign progressives, we’re talking about next year or we’re really trying to get serious, that maybe the prediction markets don’t have that much more to offer. Because they have the polls, and just total guesswork. Everything else is pretty much laid out. That might be the case, let’s say, in summer or fall of 2024. And one thing about prediction markets, that the prices often reflect too much anticipation of volatility at that stage. In other words, as the election gets closer, the cake is baked, not much is going to happen. If somebody’s ahead by a lot on election day, they’re at least going to win the popular vote. Okay?
But we might still think that all kinds of things can happen magically over the next two months, but rarely does, or at least enough to change it. So prediction markets might reflect, I’m thinking volatility, but in other words, you over predict the volatility. And therefore, they might overestimate the chance of long shots.
Matt Grossmann: So what about your subjective personal forecast for 2024, Bob? When do you allow your own view to differ from the quantitative models, and how often is your personal prediction the same as your models?
Robert Erikson:First of all, if I start plugging in some kind of quantitative model, now I’m floating anything I might come up with, based on the past, it would not be very good. Look at that low R square. It doesn’t predict much at all. Okay, so that’s the first thing.
And secondly, so normally I’d say it’s early. Okay. Obviously, there’s reasons why things might be stuck in place because for example, we know the candidates, et cetera, we talked about that. But obviously, things where they might be… you don’t know where they’re going, how are the candidates going to evolve through the geriatric campaigns, what about the trials, et cetera. These are obvious unknowns. More than usual. So I would say that the Republicans started out with a better position, we might think, given what we otherwise know right now.
Matt Grossmann: So the polls, although not predictive now will get predictive as we get closer to the election according to our historical data. But is a poll conducted in previous elections really the same as a poll conducted now, when only 2% of people that we try to reach actually want to respond to the poll? And a lot of what is released as a public opinion poll now is more like from an online panel of people who’ve opted in to take surveys regularly, and a lot of modeling.
So of course, we always used to weight data, but what comes out as poll results now is more of a mix of modeling and polling than we used to do. So when we’re looking at historical data on the relationship between polls and election outcomes, is there a reason we should doubt that that’s going to work, keep working the same if the quality of polling is deteriorating?
Robert Erikson: That’s a good question. If we could talk about how sausage is made and how polling is done these days, as you point out, so maybe 2% response rate, et cetera, and opt-in polls, so obviously that modeling is necessary, particularly… I think particularly if we consider post stratification. We try to weigh voters so that… weigh respondents so that you get the same proportion, different demographic categories. But then again, is that always enough? It’s that kind of question.
So one answer is, pollsters have to get it right or to be out of business. So let them do their work, I think. Because the pressure’s on them, and they know that. So they do the best they can. And if Trump is not on a ballot, they do very well these days, to my surprise. I mean, now they do it. The last midterm elections, where quality polls did very well. Same… I know there’s some junk polls out there, but the quality polls did very well predicting 2022, 2018, Trump got on the ballot off-year elections. By 2023, I think, they did fine. They didn’t have any big upsets, right? They got the state elections. We got these things right.
With Trump on the ballot, maybe it’s a little different, it’s a little harder, and Trump will be on the ballot probably. And so we have to have another caution. What that means is that people who support Trump don’t want to reveal it. They don’t want to be polled. At the same time, we don’t want to push that too hard because basically even the 2020 election wasn’t… or even 2016 or 2020 weren’t that far off, both just mis-predicted 2012 in the other direction, and they under-predicted Obama support a little bit. So they’re never going to be perfect. There’s always something out there to miss.
Matt Grossmann: Now we’ve been talking about it mostly as the rationale for voters making decisions between the two candidates over ideology, over the economic performance, et cetera. But there’s a whole another category of people who believe that elections are relative turnout contests between one side or the other. And certainly, some politicians who believe that they have majorities, if only they could get them to participate. So should we be taking… How much should we be taking that into consideration, or is that just likely to move in the same direction?
Robert Erikson: Yeah. Turnout. Turnout. It’s turnout. For example, does high turnout help Democrats? It would be one… typically a way we think about it. Like somehow Democrats may be some kind of low achiever, lazy voter that only comes out under certain circumstances, where Republicans always turn out, something like that. And a bit of evidence in that direction is that Republicans seem to do better likely voter polls than registered voter polls. So people who are registered as Republican, as a large subset of them who are likely to vote as supposed to unlikely.
That gap is lowering, I think, in large part because Democrats represent the kind of people who love to vote. And so I don’t think turnout is becoming that big a deal. Now, of course, we could get surprised. Most people who register do vote in the presidential elections. I’m surprised how high it is. Maybe if you believe the polls on this, it might be over 90%. If you’re registered, you’re almost certain to somehow make it to cast the ballot. So you’re worrying about an 8% that doesn’t show up. That’s trivial almost.
So maybe the registered and the registered and the likely or actual voters are almost the same people now. So it’s not as big a worry as before. I think there is a danger, particularly early in the campaign, of overestimating likely voters. Because likelihood… Certainly at this stage, if we’re looking at likely voters, it’s a year from the election. Who knows who’s going to vote? If we base likely voters on a very instrument in terms of… about interest in the election, those people express interest now. Or even the people who are willing to talk to pollsters now might be different from those who will do the same thing a year from now.
And so I think that’s meaningless. So I would tend to look at registered voters, if there’s… Look at the gap between registered and likely voters, if there is one. If there is, maybe you have… particularly if the likely voter vote goes slightly more in Republican direction, maybe add a little edge to the Republicans beyond what registered voters show, and I’d take that into account.
Some polls show… You have a very heavy screen on likely voters, which means that the pollster… their polling bounces around a lot because it depends on who’s excited, which party’s excited. And I don’t think that’s very useful because that kind of excitement is short-lived.
Matt Grossmann: There’s a lot more to learn. The Science of Politics is available biweekly from the Niskanen Center. I’m your host, Matt Grossman. If you like this discussion, here are the episodes you should check out next, all linked on our website: How Republicans lost 2018 by being too close to Trump; Interpreting the early results of the 2020 election; When information about candidates persuades voters; Does the 2022 election show how Democratic campaigns win; and, Will a good economy save Trump?
Thanks to Robert Erikson for joining me, and please listen in next time.