Something I’m fascinated with right now — expressive responding in polls. Consider this chart which recently made the rounds:
That’s right, after the election, Republican confidence in the economy skyrocketed, from -46 to +27. Meanwhile, Democratic confidence moderately declined. But what does that mean, really? Does that mean that partisan identity is so strong that people literally change their opinion overnight?
The thing is it can’t mean that. Look at that steady level of economic despair among Republicans about the direction of the economy. That -47 score Republicans gave the economy in 2016 is equivalent to the level of confidence in the economy during the Great Recession. That’s really low. That’s world-is-going-to-end-tomorrow low.
A naive reading of this says that for the eight years of the Obama administration Republicans were just as suspicious of the economy’s stability as when the entire economy was on the brink of collapse in 2008/09. They were scared every day that they were going to lose their jobs and the economy was going to collapse around them.
But here’s U. S. car and light truck purchases over the past fifteen years.
As the chart shows, when you’re nervous about the economy, you don’t run out and buy a new car. Even with the Cash for Clunkers program in 2008/09, demand for new cars in those years collapsed. No one buys a car when they think they may be out of a job tomorrow.
As the economy came back, however, demand returned to normal. This was a direct expression of people’s confidence in the economy — their willingness to take out a five year loan to get a new car. We can see the purchase of cars as a proxy for not what people want to express, but what they actually believe.
Fine, that’s just U.S. autos. But you can see proxies of confidence wherever you look. They bottom out during the crash and slowly come back.
So was it just Democrats, with their confidence, buying all these cars and engaging in new mortgages? No, of course not. The truth is the Republicans in the poll are not truly pessimistic, at least not at the level of economic intuition. Rather, they are using the poll to express a feeling about (or prejudice against) the current administration.
How far do people take this? Well consider this: when Trump voters were shown these two pictures of the Trump (photo A) and Obama (photo B) inauguration and asked which picture showed a more people, 15% said photo A had more people.
As the researchers note:
If there were no political controversy, any respondent who took the time to look at the photographs would see more people in the image on the right than the one on the left.
Clearly, some Trump supporters in our sample decided to use this question to express their support for Trump rather than to answer the survey question factually.
Polls, in other words, might slowly be becoming junk. We don’t actually know how many people believe Obama was a Muslim, or how many people believe the DNC rigged the election. The answer is surely some people believe each thing, but they are lumped together with people who use the question as a general proxy for expressing something else, as a way of advancing what is seen as a team agenda.
One thing to note, this seems to be more prominent on the Republican side (asymmetric, as they say). Aside from the consumer confidence pattern above there’s the Obama bombs Syria/Trump bombs Syria question. Thirty-eight percent of Democrats supported Obama’s plan to bomb Syria in 2013, and a nearly identical number of Democrats supported Trump’s action. Republicans, however, were lopsided: 22 percent supported Obama’s plan, but over 86 percent supported Donald Trump.
And while race has something to do with that asymmetry, it’s not just race: Republican trust in government, for example, has been mostly dependent on which party controls the presidency, whereas Democratic trust in government has been stable since Carter came into office, with a dip during the years of the Iraq war and Hurricane Katrina:
But again, I’m not quite sure how to read this. Is this really the level of trust? Or is it expressive responding? Or are the two things so intertwined it’s hard to tell?
Interestingly, you see that pattern emerge in the Carter administration. and of course the Carter administration is where partisanship begins to take off in the DW-NOMINATE scores as well, which are about voting patterns of legislators (and not generally considered expressive):
I’m aware of the many explanations of the DW-NOMINATE score increases, and also aware this is a big dig into a small signal, but it’s something I’ve been thinking about.
Also, I’m thinking that these poll responses — going back to the 1970s — show a lightweight, non-digital form of participatory propaganda. You get the call from the pollster, and you see this as an opportunity to further the interests of your party by ever-so-slightly affecting the poll results, bringing them in line with the party talking points (the economy is strong, the government is trustworthy, the crowd was huge). It’s like an early form of online reposting, trying to amplify the signal of the message you want to advocate, even if you may have some reservations about the truthfulness of the claims.