Donald Trump speaks during a campaign rally in Charlotte on Oct. 14. (Erik S. Lesser/European Pressphoto Agency)

 

It is by now clear that Donald Trump strikes a chord among many Americans who feel marginalized by free trade and left behind in the post-Great Recession recovery. Our research has uncovered that Trump supporters who are struggling economically perceive U.S. economic performance overall to be much worse than it actually is. And their perceptions are more dire than those who, though also struggling economically, support Hillary Clinton.

Here’s how we studied this.

As part of a study of political participation, on Oct. 19-22 we interviewed a group of 1,000 American adults who are unemployed. We asked them what the country’s current unemployment rate is and, in a separate question, how high the unemployment rate is on a five-point scale, from “very low” to “very high.”

Trump supporters were significantly more likely to say the rate was high. They also grossly overestimated the unemployment rate. When asked what the current unemployment rate is, by far the largest number of them chose the highest answer we offered — “15 percent or higher.” Fifteen percent is triple the rate reported by the Bureau of Labor Statistics for September of this year.

Of course, many people would be unable to cite the current unemployment rate with precision. In our full sample, most people cited a number above the actual rate of 5 percent. About one in three people in our full sample of unemployed Americans believed that the rate was 15 percent or higher. Still, Trump supporters had an especially gloomy view of the labor market. Nearly 40 percent said the rate was 15 percent or higher.

It is possible that Trump supporters come from communities and regions where unemployment is especially widespread; their views of the national jobs situation might be colored by these local conditions. Though it is noteworthy that the impact of additional factual information about the national rate did not move their perceptions much.

Republican presidential nominee Donald Trump made a campaign stop at at Macomb Community College in Warren, Mich., on Monday. There, rally-goers said they believe some voters have been quiet about their support for Trump but they will make their voices heard on election day. (Zoeann Murphy/The Washington Post)

We saw this when we exposed some of our respondents to informative statements about the unemployment rate. One randomly selected subset of our sample read the following text: “The most widely reported [unemployment] rate is around 5 percent, but this doesn’t take into account people who have given up on finding a job. When these people are included, the rate nearly doubles, to around 10 percent.” They then were asked the questions about the unemployment rate.

Trump supporters were not particularly moved by this discouraged-worker framing. The number of Trump supporters choosing the highest rate was approximately the same in the discouraged-worker treatment group as in the control group, where respondents were simply asked what the current unemployment rate is, with no framing.

The highest rate of gross overestimation was among Trump supporters exposed to a different experimental framing. People in this blame treatment were told that “rather than being a purely economic issue, unemployment is mainly a political issue — so if the level of unemployment is high and there are lots of people who lost their jobs, politicians and the government share much of the blame.”

Of this group, a whopping 45 percent of Trump supporters cited the top figure. This means that close to half of the Trump supporters who were primed to blame politicians for bad labor-market outcomes saw the unemployment rate as three times what it currently is, or higher. The corresponding figure for unemployed Clinton supporters was 26 percent.



Source: Authors’ Survey

Since there is much evidence that anger motivates people to act, we asked our sample of unemployed adults to report their level of anger when thinking about their employment status after reading the framing texts for their respective treatment groups. (We asked the same question of our control group.) The blame treatment was the only one that elicited significantly higher levels of anger.

Interestingly, the Trump supporters’ emotions were less sensitive to this treatment. This was probably because of ceiling effects: They entered our surveys already angrier than the rest of the sample. In the control group, which received no framing text and was simply asked about the unemployment rate and their levels of anger, unemployed Trump supporters reported higher levels of anger than unemployed Clinton supporters; the difference is statistically significant.

Are struggling Americans drawn to Trump, or is it more complicated?

How should one interpret these findings? Are struggling Americans who overestimate the nation’s economic woes drawn to Trump, whereas those who are also struggling but who have more accurate assessments are drawn to Clinton?

It’s probably not that simple. Candidate attachments are likely to drive perceptions as much as perceptions drive candidate attachments. There is much evidence that Americans’ perceptions of the economy are refracted through the lens of partisanship; our findings are in line with this evidence.

But there is also evidence that the perceptions voters report are a mix of their real views of the economy and of declarations of partisanship and loyalty to candidates, what John Bullock and his co-authors call “partisan cheerleading.” These investigators offered experimental subjects monetary incentives to come up with accurate answers to questions about changes in the inflation and unemployment rates, among other questions. In response to monetary incentives, “partisan cheerleading” diminished; the accuracy of answers and admissions of ignorance increased.

Something similar may be going on with the unemployed Trump supporters we interviewed. If their declared estimates of the unemployment rate simply reflected their factual beliefs, we would have expected their responses to be influenced by the discouraged-worker framing, which exposed them to accurate information. But instead they were sensitive to the blame framing, the same one that elicited the biggest jumps — in the full sample — in anger. Thus their “perceptions” of national unemployment rates moved little in response to information, but spiked upward when respondents were told that politicians were to blame. Factual appraisals appear to be driven by emotional reactions and candidate attachments, rather than by pure information.

Whatever the mix of facts, partisanship, candidate loyalty, and anger, many down-on-their-luck Trump supporters who heed his call to “make America great again” are people who overstate the distance between here and greatness.

Erdem Aytaç is assistant professor of political science at Koç University, Istanbul.

Eli Rau is a doctoral student in political science at Yale University.

Susan Stokes is John S. Saden Professor of Political Science at Yale University.