Conventional wisdom holds that Trump won the 2016 election by appealing to voters left behind in Obama’s economy and may win re-election based on a stronger economy in 2020. But new research casts doubt on both stories. Sean Freeder finds that the effect of economic performance on the president’s re-election has been declining since the 1980s because citizens both misperceive the economy and selectively credit the president to match their partisan bias. Robert Griffin finds that 2016 Trump supporters were actually better off than Clinton supporters, especially racial minorities, once you minimize the effect of partisan bias. And real economic distress may still hurt Trump in 2020.

Studies: “It’s No Longer the Economy Stupid” and “In the Red

Interviews: Sean Freeder, University of California, Berkeley; Robert Griffin, Voter Study Group

You can subscribe to the Science of Politics on iTunes, Soundcloud, Spotify, or wherever you get your podcasts.


Matt Grossmann: This week on The Science of Politics: Will a good economy save Trump? For the Niskanen Center, I’m Matt Grossmann. President Trump gets much stronger approval on the economy than overall, and he’s hoping that bodes well for 2020.

Indeed, economic performance has long been key to the president’s reelection, and some say the 2016 economy might have helped Trump win those who felt left behind. But new research casts doubt on both that specific story on 2016, and even the role of the economy in presidential elections overall in an era of biased partisan voters.

Today, I talk to Sean Freeder of the University of California, Berkeley about his working paper, “It’s No Longer the Economy, Stupid”. He finds that the effect of actual economic performance on the president has been declining since the 1980s because citizens both misperceive the economy, and selective credit and blame the president to match their partisan bias.

I also talk to Robert Griffin of the Voter Study Group about his report with John Sides, “In the Red.” They find that it was actual Clinton’s supporters who were worse off financially, especially racial minorities. Real economic distress doesn’t follow biased economic perceptions, but based on who is feeling real distress now, it may actually hurt Trump. “It’s the economy, stupid” is the clear conventional wisdom on presidential elections, but Freeder finds there’s a weakening relationship because of partisan motivated reasoning.

Sean Freeder: This really solid finding over the past several decades that political scientists have talked about a lot, that the state of the economy is a major predictor of incumbent vote share, particularly presidential incumbent vote share, but it can also be shown to have effects on vote shares at all sorts of lower levels of people riding the president’s coattails.

And the paper that I wrote is focused on the weakening of this relationship over the past several decades. This is a factor that’s been built into a lot of economic models that predict how likely the incumbent president is to win reelection, and basically the argument I’m making is that scholars have kind of missed how much this has weakened over time.

Let’s not say that it has no power anymore, but it is quite a bit weaker as a predictor by a bunch of different measures than it was a few decades ago, and so that’s sort of the first half of what I talk about in the paper.

The second half is trying to provide something of a political psychology explanation for this, and so I see this as basically explainable by rising partisan polarization, that we used to live in an era where, for some subset of the population, it was more important to respond to economic indicators.

That’s just more naturally what you might do as a voter, and increasingly as partisanship has grown over time, people are much more interested in protecting their party’s value, protecting their partisan identity, and so in a choice between voting based on party, voting based on economic outcomes, increasingly, people are making the choice to do the former.

And people still really care about the economy. They indicate this in all sorts of polls. When you talk about what matters to them, people usually still talk about the economy. So then, kind of thinking like, “How is this possible?”

And the story that I really focus on is motivated reasoning, that people like to convince themselves that their side’s good, the other side’s bad, and they use some tricks to do that. And the two tricks I focus on is, people sort of looking at what I call selective perception, that people will assume that their side does a good job with the economy, regardless of the evidence in front of them, or that the other side just does a bad job with the economy, regardless of what they do.

But even in certain cases where the economy is really, really good, or the economy’s really bad, this might seem a bit silly. It’s hard to say during a depression, for instance, that my side’s doing a great job with the economy.

So in these extreme cases, and the second thing I argue is that people switch over to, instead of selectively perceiving the economy differently, they switch over to selective attribution of credit for the economy.

So they might say, “Yeah, the other side seems like it’s presiding over a good economy, but really it’s just luck.” Or, “My side seems to be doing a bad job with the economy, but it’s just bad luck.” So I focus on how people switch off between those two accounts.

Matt Grossmann: And Griffin and Sides find the conventional wisdom that Trump won last time based on economically distressed voters doesn’t really fit the evidence.

Robert Griffin: I think one of the things about the 2016 election that was really fascinating was how much this narrative around the economically distressed, and particularly white Americans who were having economic problems, sort of took hold in the media.

And again, it was this idea, this was a group who was suffering sort of uniquely, maybe, in the United States. And that this was they were flocking towards Trump in the 2016 election.

Now, what our work found, that there was a number of those pieces, that narrative, that just weren’t quite true. So one, we didn’t see that white Americans were uniquely economically distressed. If anything, they were doing a bit better than those belonging to other racial groups.

We found that Trump voters particularly were not uniquely economically distressed. That it was actually, if anyone, Clinton voters who were showing higher levels of economic concern. That a lot of these sort of features of economic distress were not correlated with support for conservative policies, but actually correlated when it came to economic policy or redistributive policy, to liberal policies.

And I think we found really interesting, and this is sort of like a post-election finding, is that part of the narrative is that Trump is still drawing potentially support from some of these distressed Americans. But that the real, what we were finding in the data was that people who were distressed were actually less likely to approve of the job that Trump was doing right now.

And this was true of both Republicans and independents, that there was a marked drop-off between people who were not experiencing that much distress, and those who were. So I think generally, if we were to sum up the findings of the paper, is it really runs counter to a lot of the narratives that people had in their head about what happened in 2016.

Matt Grossmann: Freeder came to his study with a focus on political psychology and how that leads to polarization.

Sean Freeder: My main level interests are American politics and political psychology, about equal levels. And within each of these, for American politics, I’ve really been focused and interested in affective polarization. You know, the extent to which we seem to all be increasingly disliking each other and having a difficult time trusting one another, and fighting based on partisanship.

And within political psychology, my focus has really been a lot on motivated reasoning. How we tell ourselves stories that allow us to maintain our priors and to maintain our attachments, and the esteem that we give to our parties.

And so, in one earlier paper in the dissertation, I kind of focus on what leads to affective polarization, so how our tendency towards motivated reasoning can sort of reinforce affective polarization.

But in doing that paper, I kind of realized, this is a feedback loop. Motivated reasoning can clearly cause people to be more polarized, but the reverse is also true. If you’re more polarized, you have more of a strong partisan attachment. That also is going to lead you to want to be more of a motivated reasoner.

So I was thinking of, what are things that people would start telling stories to themselves about, that they maybe didn’t used to tell those same stories about? And looking at all these different factors that seem to matter in political science, the economy seemed to be one of the few factors that, in previous studies, partisanship didn’t seem to subsume and absorb.

And that seemed really important to me, so the most cynical story I could possibly tell would basically be, even the economy really is no longer is this independent thing. And I wanted to see if there’s any evidence that that, too, as a result of partisanship and polarization, has declined over time. And so I started looking into the numbers, not necessarily expecting to find a huge effect, and kind of somewhat surprised by the degree of an effect I actually found.

Matt Grossmann: Griffin was trying to get around biases of partisanship to really measure economic distress.

Robert Griffin: So, what I think we were trying to do with this, I believe we did so successfully, is try to get people, to really pin them down and ask them about actual events in their life. So, did you have difficulty paying a credit card bill? Have you had any difficulty paying your mortgage? Did you lose your job? How do you feel about your savings?

So, getting people to start to assess these, as close as possible, some of these real life events that are occurring to them, to kind of help distant them from maybe the way in which partisanship is framing up some of these questions, and ultimately, we think, to some biased answers.

Matt Grossmann: They’re both running into strong conventional wisdom. Freeder says it’s not just that we have the wrong economic indicator, but that they’re all becoming less important over time.

Sean Freeder: The conventional wisdom seems to be, at the very least, let’s say, slipping. I would not go so bold as to say the economy doesn’t necessarily matter anymore, but just as a sort of primer for the audience, the findings are that the state of the economy is this major predictor of incumbent vote share, and in terms of all the different factors that go into these models, it sits right under partisan preference. People almost always just vote their party when they’re voting for president. It’s one of the effects that isn’t sort of naturally subsumed by partisanship.

So, people think about their party. They think about how well the economy’s doing in these traditional models. Maybe how long the incumbent party’s been in power, or is there a war, in some of the models. In a lot of cases, that’s really all you need to predict the economy.

And within that, what are we saying about the economy? What’s really predicting it? The measure that’s traditionally used is looking at real disposable income. At the national level, interestingly, at least in the history of these models, national data seems to have more of an impact on people than local.

And people are also myopic. They tend to not focus on the entire four-year term, how well the president’s done with the economy, but just kind of how things have been in the most recent year-to-year period.

So these are the indicators that I still focus on, and compared to other indicators that you might try a measure than already … You might try looking at the whole economy over four years. I try that a bunch in my paper, these different measures, and I still find the conventional wisdom seems to be right, in that national year-to-year real disposable income still seems to be, of all these different measures, the most consistently strong indicator and performer in predicting this stuff.

But when I track real disposable income as a predictor across time using either actual county-level vote share, or individual data from surveys like the American National Election Study, it shows that this relationship is, at the very least, starting to weaken compared to previous decades.

Matt Grossmann: And Griffin says economic influence is a slippery concept, but there’s no question people’s economic attitudes are now infected with their partisanship.

Robert Griffin: It is a little bit slippery when people start to talk about this topic, right? So, economic anxiety. What does that mean? Well, sometimes people mean, okay, they don’t think the US economy is doing that well. Sometimes they mean, okay, they don’t think their personal economics are necessarily doing that well.

And there’s another piece of it that’s even more than that, which is to say that they are experiencing actual distress. That their events in their life that are not going well at all. So I think some of what we were doing here was actually just starting to tease those apart, and the reasons we wanted to tease them apart is because, from a measurement perspective, they’re quite different.

And they’re quite different, and not just that, but that I think has become the case with a lot of things that we’re measuring in political science. It’s become increasingly correlated with … increasingly, you could say, poisoned by partisanship. That that has become an additional factor that’s made measuring some concepts within political science a lot more difficult, because people’s partisanship just starts to frame how they see everything.

So, on the one hand, you do have some of this economic anxiety stuff that is often measured as a function of how you think the national economy’s doing, how do you think your personal finances are doing? And what we find is that pre- and post-election, Republicans are switching around.

And especially with the data that we use, which is panel data, we find that this isn’t a sampling issue. It’s not people switching parties around. It’s people who were Democrats and Republicans post-election. As a result of who is now occupying the White House, they’re flipping their assessment, not just of what’s going on nationally, but what’s also going on in their own personal lives. Which is perhaps a new sort of feature of how partisanship is influencing our perceptions of these things.

Matt Grossmann: The declining influence that Freeder finds extends into the Trump administration.

Sean Freeder: It expands all the way from the beginning of the 1940s to just coming out of the Great Depression all the way up to the present. So I’m using change in real disposable income across this entire period, and then comparing it, as I mentioned before, to different, both aggregate and individual vote share rates.

And if you track this over time, basically, what is the correlation between RDI and vote share? And you track this over time, you find generally just a constant decline if you use a pretty simple OLS model or something. But if you use a model that’s more sensitive to picking up on changes within that period, you really kind of find two periods of decline.

The first one is a story that I think we’d all kind of expect to see, which is that you see a decline in the strength of that relationship in economic voting from 1940 to about 1960. And as far as I’m concerned, this is a story of the Great Depression made people really think about the economy. Once the economy starts doing better, they’re sort of freed up to think about other things other than just the economy and are we getting ourselves out of this mess?

The other major period you see the decline, and this is what I really focus on, is the period from about 1985 to 2005. You see this other kind of big dip during this period, and this is what I associate with polarization.

I wouldn’t go so far to as the effect of economic voting is now zero or anything, but as one indicator, if you just start your data at 1990 and you look at the correlation over time between these things, after 1990, correlation is just pretty much about zero. I’m skeptical that it’s really that low, but it’s a lot weaker than it used to be.

One thing that sort of confirms this, at least at the present, and we’ll see what happens with this, is that Trump is already very far off this trend line. He’s performing 20 to 30 net approval rating points below where he should be on this trend line just based on the strength of the economy.

So that’s something that sort of backs up this data in terms of what’s going on right now, and one thing that we see is that the economy, it still impacts true independents, so I still think that you’re still going to see people that have absolutely no partisan lean whatsoever. If this really is just a story of rising partisanship, those that truly have no partisan lean, they should still vote based on the economy, rewarding and blaming where appropriate.

But there’s potentially less people available to be in that train of thought than ever, and even independents can get distracted by some of the policy and partisan squabbles that we’re looking at now.

And the other time when this still seems like it matters is probably at the very extreme. If someone absolutely tanks the economy, if the economy’s really, really well, one would think that even in those cases, people are going to be able to squeeze some information out of the strength or extreme weakness of the economy.

But given that we are sitting in a very strong economy right now, at least by traditional indicators, what happens at the end of 2020 might be another test that we might be interested in looking at to see, just how extreme is this decline in actuality?

Matt Grossmann: And it’s clear, no matter what indicator you look at.

Sean Freeder: I was struck by the robustness of this finding. I really tried to throw the kitchen sink of alternative specifications at this finding and see if I could kill it, and pretty much nothing does.

I try a bunch of different controls for using the aggregate data, like district partisanship and other models of maybe indicating the importance of time, trying different levels of fixed effects, trying different lengths of time where you can consider windows where I’m looking at aggregated correlations, try a bunch of different economic indicators, so we move beyond RDI to maybe looking at the stock market, the unemployment rate. That doesn’t seem to make a difference.

I look at this across, as I said, both aggregate, individual data. Doesn’t seem to make a difference. Within the individual data, I play around with, is this sort of an artifact of the weighting schemes I’m using? Is this a change in the sample? Nothing changes with that. So what really just stands out to me is how strong this seems to be as a finding and robustness to alternative specifications.

One additional alternative specification that I tried that I think provides sort of its own finding, is breaking down, in the aggregate data, where is this decline happening? So, using the county vote share data I had can break it into four quartiles of counties, like going from very partisan counties that pretty much always go to one party, all the way down to counties that are much more competitive and might flip from election to election.

And you do find a story that confirms that this is likely about rising partisanship, where the decline is the strongest the more partisan the county is. So in very red, very blue counties, the economy has just kind of stopped mattering at all.

But that doesn’t mean that, in competitive counties, there has not been this decline. The effect in these counties is also highly significant. It’s just somewhat less than it is in partisan counties. So we see it everywhere, but it does seem to be contingent on partisanship.

Matt Grossmann: Griffin agrees that economics likely don’t matter as much as they used to, but at the individual level, he finds the most distressed are still less likely to support Trump.

Robert Griffin: Part of the continuing narrative that might even exist going into 2020 is that Trump still has special cachet with these communities, and they might even be part of his base and they’re not going to stop supporting him.

But again, when we look at actual economic distress, across the board, it’s those who are experiencing those highest levels of distress that are now the least likely to support Trump once you control for other factors.

So if you actually control for people being Democrats, independents, or Republicans, which does a large part of the explaining, frankly, in regards to whether or not people approve of Trump. But you break them out by people who are Republicans and experiencing very little distress, and Republicans who are experiencing a lot of distress, it’s actually those in the high distress that are least likely to approve of the job Trump is doing.

And that is doubly true among independents. The effect is even larger. So if part of the narrative is, this is still going to be a net positive for Trump, that he has a special connection to these communities, it’s not being borne out, I think, by some of the job approval numbers that we’re seeing, at least right now.

Now, going into the 2020 election, how much will this affect him? Well, I think as a global answer, we have to say maybe not as much as we’d imagine, I think, first off. Just because, again, partisanship is playing such a bigger role.

My own suspicion is that if we somehow rewound the clock 20, 30 years and ran this study back then, we’d probably be seeing a stronger relationship between some of these distress metrics and people’s approval of the president and whether they’re going to vote to reelect him.

So it’s not to say that it’s not having an effect now, but my own suspicion is it’s probably having less of an effect than it did in the past, just because once partisanship starts gobbling up some of the variation in all of our data, that’s got to come out of somewhere and part of it, I think, is coming out of these economic evaluations. And even accurate evaluations of the distress that re existing within people’s lives.

Matt Grossmann: Freeder finds two reasons that economic perceptions are now less important: Partisans can either deny economic reality, or they can attribute credit or blame to their side’s benefit.

Sean Freeder: Motivated reasoning is all about finding cognitive shortcuts. People don’t like to think about complicated issues. Those of us in political science, we might actually have a high need for cognition. We actually enjoy thinking through this stuff and evaluating the role of the parties in influencing the economy.

But most people don’t enjoy doing that. They don’t necessarily have the tools to do that, so you try and find cognitive shortcuts, and the first shortcut to just sort of maintaining your partisan identity is to do the simplest thing you could, which is you look at a reality that doesn’t really agree with what you’d expect, and you just deny the nature of that reality.

So, this might be sort of akin to what you’re saying about Elizabeth Warren, that you might just look at Trump’s economy right now and actually find reasons to say that, you know what? Like the unemployment rate, it looks good but people are working three or four jobs under that, and people are still struggling to make ends meet. And there’s many people that are not receiving the benefits of living in a booming economy.

And I think for the average person, you might just tell a simpler story, which is just to say that, “Nope. The economy’s just not good. I can’t tell you a more complicated one, but it’s not good,” if I’m a Democrat and I know that that would mean giving Trump some sort of credit for something he’s done potentially.

What I think is more likely in the current environment, and this is what I argue in the paper, is that in these extreme times, people will prefer a different method. So there’s some findings in motivated reasoning research, that people are good motivated reasoners, but overwhelming evidence can cause enough anxiety that people start to say, “You know what? This seems weird to maintain this perception of the economy when I know economic reality is pretty good.”

So in 2020, I imagine a lot more people are moving to the other method, which I call selective attribution. And that’s really all just about saying that … I think at least in the moment, this manifested or has previously manifested as people saying, you know, “Thanks, Obama. This is Obama’s economy. Trump just happens to be benefiting from it.”

Maybe at this point, because we’ve now gone a few years from the Obama administration, people could say, “The world economy is just really good and it’s not your fault. It’s not your credit that this going well, or maybe there’s other actors that are really responsible for it.”

They just move away from actually giving Trump credit, and I’ll mention quickly that in the paper, some evidence that I show of this is looking at the election of 1988 versus the election of 1992. In 1988, the economy is like, you could interpret it either way fairly. It’s decent. It’s actually pretty good, but if you want to tell a story in which it’s bad, you could.

And in that year, Democrats who like to tell somewhat of a negative story, most of them, given a choice between perception and attribution, they go with perception. They just say that the economy’s actually not doing so great. Very few people really make the effort to tell the attribution story.

In 1992, the economy is more arguably just negative. It’d be harder to tell a positive economy story, and so Republicans in this case, wanting to booster George H.W. Bush, choosing between perception or attribution, then it flips. Perception is very hard to maintain. Very few people actually try and argue in the survey evidence that, “Oh, the economy is pretty good.” Instead, they move to this other story which is that it’s not really Bush’s fault that the economy’s bad.

Matt Grossmann: Freeder’s experiments show that people switch attributions for success when they can’t deny the evidence.

Sean Freeder: I focus on the experiments within the attribution claim that I’m making, and the reason why I switch over to the experimental data is, I lose the ability to tell an over-time story with attribution.

Political scientists have been really good about asking questions about how people perceive the economy over time. They have been less good at consistently asking questions about people’s attributions of credit or blame for the economy.

So what I would love to be able to do is go from 1962 all the way through to the present with ANES data on attribution, but unfortunately, there’s only four election years in that period that really asked those questions. They’re all during the time of polarization, so I can’t show a clean over-time story in terms of an increase in selective attribution.

But what I can do is run some experiments to at least show that in the American context, this has in the British context been shown elsewhere, that people do selectively attribute the economy based on partisanship. I can at least show that people do this in the American context.

So, I run a couple pretty basic experiments, one in which I want to see if people are able to sort of generally do this with the parties. So I present different accounts telling them based off of different evidence, which depending on how you tell the story, you can make a case for either of these claims: that either the Democrats or the Republicans overall in the past several decades have been better at handling the economy, just looking at the numbers.

And then asking people to rate a series of explanations for why would that be? And they can, on a seven-point scale, say that, “Oh, this is all just a story about luck. They just keep getting lucky.” Versus, “No, it really indicates that this is a story about their skill in handling the economy.” And as it turns out, people will really, for their own party, prefer the skill-based explanation. For the other party, they prefer the luck-based explanation.

The other experiment I run is just using this sort of unique moment where Obama hands over a very strong economy to Trump and so, depending on whether the president is Obama or Trump, the answer should be, if you ask people how the economy is, it should be for either of them, it’s good. It’s mostly very strong evidence that it’s good, so I just vary which president people see, ask them, “How good is the economy and how responsible is that president for it?”

And what I find is that, for people that get a president from their own party, the relationship between perception and attribution is positive. So as they say the economy looks better, they’re more willing to give the president credit for a really good economy. And the relationships is the opposite if you get a president from the other party, that you would say, “Well, I think the economy is worse.” And as the economy gets worse over time, you’re more likely in that case to give blame to that party and then vice versa.

So you see different responses based on partisanship, and even though you can’t get that over-time story, you can at least see that attribution is something that people selectively process when they look at the economy.

Matt Grossmann: One big factor Sides and Griffin pinpoint is race. Even though working class white voters are more stressed than their educated white counterparts, they’re not nearly as economically stressed as racial minorities.

Robert Griffin: It’s not that, say, white non-college voters, which is a group we talk about a lot today, it’s not that they are not experiencing more distress relative, let’s say, to whites with a college degree, because it appears that they are.

But it is the case that they’re actually experiencing less distress, or just about equal amounts of distress, to even non-white populations, so even African-American and Hispanic populations, who have a college degree.

So the idea that they’re experiencing, I think, something that is unique in terms of the amount of distress, it doesn’t seem to be borne out in the data. Again, we look at non-college Hispanic and African-American populations, much higher levels of distress. And even these college-educated populations among African-Americans and Hispanics are pretty much experiencing equal amounts of distress.

Now, this gets played out even once you start to control for income. So this isn’t even just a function of, “Okay, let’s take a look at the education levels and even things out that way.” Even when you compare people who have very similar levels of income, it appears to be that there’s higher levels of distress among, particularly, this non-college, non-white population.

So Hispanics, African-Americans and Asian-Americans who don’t have college degrees, just showing much higher levels of distress even once we control for income. And part of what’s happening there, again, just by way of comparison, is that there’s probably not as much familial wealth that’s cushioning people during hard times.

So even at similar levels of income, there’s just more probably intergenerational wealth within the white community that’s cushioning the blow for some of these communities, so that even if their incomes are not very high, they’re just not experiencing some of these events at the same rate that other communities are.

Matt Grossmann: Freeder says his evidence fits with theirs. Economics may matter less when any quality is rising, and people know where each party stands.

Sean Freeder: There’s a few stories that you can tell, why race and class coalitions sort of hardened and sorted into the parties might impact economic evaluations and how important they are. I think one story to tell is that this is about people’s developed preferences over time, that we used to live in a world where maybe people didn’t have enough of a class identity, didn’t think enough about their racial identity in certain cases, that it rebounded automatically to the party, and now they think about these things.

And based on those things, they have actual policy preferences. They know the party’s have different things that they want to do, and they’ve seen those differences, and now they vote for a party that sort of does the things that they want them to do.

I think another story that sort of asks less of the voting public is really, perhaps especially with race, it’s not about, you can now perspectively vote and you have all these preferences, but it’s really just about simple partisan trust. You hear the rhetoric between the two parties.

If you’re a person of color voting, over time, it’s quite plausible that you’re going to develop a certain amount of trust for one party and a lack of trust for the other. And trust is something that you’re going to want to make accord with your natural beliefs about how the parties therefore do, in terms of helping you in the economy.

This is definitely a period where I find the decline, where you see rising challenges in terms of both class inequality, racial inequality, and you worry that possibly, people are getting left behind and they’re feeling like, “These indicators are no longer really relevant to me.” Or it seems like there’s a disconnect between, “Everyone’s saying the economy’s good. It doesn’t seem like it’s good for me.”

And one more possibility I’ll just throw out there is that the economy might just be getting crowded out by other stuff, as race, as class becomes more important, but particularly as race becomes more important, maybe people have other things on their mind to vote for than the economy. Maybe they want to vote based on civil rights. Maybe they really care about immigration policy.

There might be other issues that have become increasingly salient on people’s minds, and these really sort of combine with the existing, as you say, sorting of race and other factors into the party coalitions at this point.

Matt Grossmann: He does find that economic attitudes can be divorced from partisanship in some contexts.

Sean Freeder: I’m thinking maybe there’s something to be said about the context in which the question’s asked, rather than the actual question itself or the indicator that one’s using. So as an example, in addition to ANES data, I used data from the General Social Survey to see if I could replicate the results. And this is the one case in which I don’t find a perfectly robust replication of this finding, the economy doesn’t matter or that people are not changing in their perceptions of the economy.

Using the GSS, doesn’t seem like there’s much change over time in how people evaluate the economy based on partisanship and this is a survey, it’s asked at a different point in the year when campaigns aren’t as active. It’s asked alongside a host of non-political questions, so you might just be sitting there thinking, “I’m being asked this question as a consumer, rather than as a partisan.”

So, if you don’t activate people’s partisanship, it seems like they might still be able to see the economy for what it is, or not try and attribute things differently, but if this is a question of, “How do we get people to fairly judge the economy?” You could probably do it without that context.

If the question is, “How do we get people to vote if we want them to, based on the economy, the actual outcomes of the economy?” It’s hard to imagine a way to get people to not be thinking about politics when they’re about to make the biggest political decision they make every couple years.

So it seems like it’s difficult to find a way to necessarily make it matter in terms of their behavioral decisions. So at the moment that they walk into the voting booth, that refrigerator question might not come in handy for them so much anymore.

Matt Grossmann: Griffin says even with the decline of macroeconomics in importance, he’s still seeing individual effects with Trump now hurt among those facing economic stress.

Robert Griffin: If we do start to see a strong disconnect between some of these macro level measurements and the election results that we’re seeing, I think at least right now, what the measurements that we’re taking would indicate to us is that, if the storyline is that those macro level metrics are just zeroing out, that the predictive power is starting to sort of totally disappear.

So far what we’re seeing in our data is that once you really poke under the hood and find out whether people are actually distressed or not, we’re not seeing a huge relationship. It’s not maybe as big as we might expect, and again, just my own suspicions, it probably would have been bigger in the past, but that that relationship is still there at an individual level.

Matt Grossmann: That means overall economic performance might still be helping Trump a bit.

Robert Griffin: These effects are probably as big as we think they should be, because there are these other forces that are rolling around, and sometimes a lot of times overriding some of these economic evaluations, but a healthy economy has probably kept him afloat.

You know what I mean? That if we didn’t have the sort of economy that we have right now, which has been almost defined a increasingly small unemployment rate and increasing wages at most income margins, that he would probably be doing even worse than he is right now.

Matt Grossmann: Freeder says it could be good news that most people are not voting on the nature of the times alone, but it could also mean that they’ve just come to dislike the other party.

Sean Freeder: I think that there is potentially a rosy, or at least a rosier interpretation of this. I think that that account, that people are just actually starting to become higher processes, might be a bit more ideological. It’s certainly consistent with some of the polarization evidence that have Abramowitz and others have found over the past couple decades. Of course, it’s inconsistent with some of the other polarization evidence found by Fiorina and some other people.

So I do think there’s enough evidence to suggest maybe people are just becoming more close watchers of politics, and to a certain extent, that’s a good thing. I’ll cut that positive story with another potentially negative interpretation, which is, as sort of a sub-finding I look at in the paper, I look at which types of people is this decline really happening with?

Is it happening with people that are really strong partisan believers that really like their own parties, so a sort of positive partisanship? Or is it happening at a rate that’s more consistent with people that really increasingly dislike the other party, so negative partisanship.

And I don’t find much connection in the decline to positive partisanship, and I guess that’s kind of what I’d expect to see if this was a story about, “I really believe in what my party believes, and I’m going to support it based on that.”

It seems like the people for whom this economic voting effect really goes away are those that just absolutely hate the other side, and that seems like an emotional driven account than sort of a rational choice. “I know my preferences and I’m sticking with them” account.

Of course, you could really hate the other side because you know how much worse their policies are or something like that. So I wouldn’t say that’s a smoking gun that this is a story that’s not so rosy, but it definitely gives one pause that this is about rising good citizenship.

Matt Grossmann: Freeder’s next step is to look at race and class differences, and whether partisan cheerleading is the only motivation at work.

Sean Freeder: Turning this back into really looking at the potential race and class story that’s here. So one thing I didn’t get a chance to do yet with this data is really looking and see, among demographic characteristics, socioeconomic characteristics, which people, which regions perhaps, are specifically seeing these declines?

And so, in the next year or so, the next kind of project I want to work on with this is to look at those differences across race and class, and see how much of this really is a story of demographic sorting and people feeling sort of left behind by an unequal economy.

And then, another application that I think is sort of there in my data, is there’s been this sort of debate on partisan cheerleading versus actual economic beliefs. So when people say the economy is the way it is, and they’re wrong, do they really believe that? Or is it kind of just cheap talk, they’re just sort of cheering for their side?

The fact that this is coming out in changes in actual voting behavior suggests that it’s less about cheerleading, but there’s probably more that could be done to probe this, either through experiments or just asking people some open-ended questions and getting them to really reflect on the economy. Maybe getting people to make costly predictions, where they either get money or don’t based on whether or not they’re right about their perception. So, this is stuff I’d like to explore going forward, and hopefully I’ll get a chance to do that pretty soon.

Matt Grossmann: And Griffin will be working more on the Voter Study Group, an amazing public resource with open panel data going back to 2011.

Robert Griffin: One of the things that I’ve loved about it, it’s a very different type of tool, relative to a lot of survey research. Again, these snapshots that we get in time that are a lot of survey research, they don’t give us that ability to tell narratives over time.

So it is difficult for me to conjure up a time when people who voted for one candidate and then voted for another candidate has been a stronger narrative, and I’m referring here to Obama-Trump voters, but has been a stronger piece of the narrative about understanding our time than any other point in US history.

It’s just, this is a group that people are really interested in. And something like panel data is unique in that it allows us to have these contemporaneous measurements of how people voted, so that there’s not misremembering or potentially even lying several years down the road, in regards to how people actually voted and what their behavior looked like.

So these panel measurements, which, again, we’ve been going since 2011, we’re going to go at least through the 2020 election. We’re trying to be able to tell a story that spans that eight or nine year period, and tells us distinctly about how people change during that time.

And again, what it allows us to do, separate from a lot of these snapshots, which we can kind of aggregate over time, is it’s not just, how did people switch their vote? But what did those particular individuals look like who decided to switch their vote?

If you did switch a party, it’s not just that we can identify that certain demographic groups are maybe moving around, but we can sort of pinpoint the actual individuals who have changed their party identification, and start to talk about them. What they look like demographically, what they look like in terms of their ideology, to really I think tell a richer story about the types of evolutions that are occurring within the American public.

Robert Griffin: And what I like particularly about the Voter Study Group, it’s something we’re dedicated to, is releasing this data in a timely fashion and making sure it’s as open as possible to people. We’re very happy to see that there were, I think, at least six abstracts at the upcoming APSA conference that mentioned Voter Study Group right in the abstract. I happen to think there’s probably more who actually have the data floating under the hood.

And again, that’s really great. We like to sort of get the first bite of the apple at the Voter Study Group, in terms of writing up results. But our ultimate goal is to push this data out, have it serve a second and a third life in academia and helping us understand this kind of vital time period that we’re living through.

Matt Grossmann: Stay tuned for more. The panel is running through the next election.

Robert Griffin: We’re going to be continuing our panel through the 2020 election, so between now and the end of 2020, we’re going to be touching base with our respondents at least three more times.

So we’re going to be making sure that we track people before and after the Democratic primary, to be able to really key in on who voters were really supporting at any given time, and the extent that people switched around or didn’t get the candidate that they wanted during the Democratic primary or anything like that.

What happens to them through the 2020 election? And how we can understand their behaviors as that big narrative piece rather than sometimes getting these snapshots that tell an incomplete story?

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 Grossmann. Thanks to Sean Freeder and Robert Griffin for joining me. Please check out “It’s No Longer the Economy, Stupid” and “In the Red.” Both are freely available online. And then listen in next time.

Photo Credit: