How Twitter got a big new study on Trump's supporters mostly wrong

This is an archived article that was published on sltrib.com in 2017, and information in the article may be outdated. It is provided only for personal research purposes and may not be reprinted.

It was racial and cultural resentment, not economic woes, that drove the white working class to vote for President Trump.

That was the conclusion prominent Democrats, conservative writers and political journalists drew from a major new survey published last week on the views of the white working class. Some even said the argument that Trump's voters had economic reasons to back him was just a cover story for uglier motivations.

The debate about Trump supporters' motivations has practical implications for battered Democrats, who — out of power in all three branches of the federal government and marginalized in state houses and governors' mansions nationwide — are desperately trying to figure out how to rebuild their party.

If white, working-class Trump supporters were motivated by economic anxiety, then the party could make inroads with an economic agenda that appeals to them. On the other hand, if it were a question of race and culture, those voters would be extremely unlikely to join Democrats' progressive, multiracial coalition.

The new data is "useful for debunking myths and narratives — particularly the ubiquitous idea that economic anxiety drove white working-class voters to support Trump," wrote Emma Green in the Atlantic, which published the study with the Public Religion Research Institute (PRRI). "When these voters hear messages from their president, they're listening with ears attuned to cultural change and anxiety about America's multicultural future."

Even critics of that conclusion agreed that the study attributed Trump's success to racial anxiety. "The ⅛PRRI⅜-Atlantic survey is just one of the latest attempts to assure liberals and leftists that Trump supporters are unsympathetic," wrote the National Review's Michael Brendan Dougherty.

But according to one of the study's authors, to claim that economic anxiety does not matter to Trump's white, working-class supporters is, at best, an oversimplification.

"We aren't saying that," said the author, PRRI's Dan Cox. "Obviously, there's a lot more to understanding this group, to appreciating the challenges that they're going through."

Why did so many read the study differently?

Some of it, undoubtedly, is a product of the high stakes for the Democratic Party. Democrats see the question as crucial for understanding both what went wrong in 2016 and the unresolved debate between primary rivals Hillary Clinton and Bernie Sanders about the direction of the party in the future.

And some of it is the nature of writing on the internet. The election had nearly 130 million voters. Twitter limits users to 140 characters. Survey toplines make for good headlines. Methodology sections rarely do. And often takes with unambiguous conclusions race across the Internet faster than anything else.

Here's a step-by-step look at what PRRI's study said.

The researchers surveyed more than 3,000 American adults, defining the white working class as white participants without a college degree who did not hold a salaried position.

To test levels of anxiety about cultural change, the researchers asked participants whether agreed that they sometimes feel like strangers in their own country, or that American culture needs to be protected from foreign influence. About half of the white working class (48 percent) agreed with the first statement. Just over two thirds (68 percent) agreed with second, and nearly as many (62 percent) described immigrants as a threat to American culture.

Despite these negative attitudes toward immigrants, an overwhelming majority of the white working class — 69 percent — favors some kind of legal status for illegal immigrants, and 59 percent even said they should be granted citizenship.

To measure resentment against progress by racial minorities, PRRI asked participants whether they felt that discrimination against white Americans was as serious of a problem as discrimination against African Americans, Latinos and other minorities. A majority of the white working class — 52 percent — said that discrimination against whites and against other groups was equally serious.

The pollsters also attempted to gauge how the survey participants felt about their economic well-being.

In short, they were not confident. About a third of the white working class said it would be "very difficult" or "nearly impossible" for them to come up with $400 in an emergency. A similar share said they had cut back on food and meals in the past year to save money.

Participants in the research are prominently quoted on making ends meet. "I buy the dented cans at the grocery store because they're cheaper," one woman said. "I don't think anybody in Washington today would be able to do what we do as far as living a life."

Asked how their financial circumstances today compared to how they lived as children, white, working-class participants were about as likely to say that they were no better off as to say that things had improved for them.

Also, 54 percent of the white working class described getting a college education as a risky gamble that may not pay off. That figures suggests many feel excluded from the country's economic progress and believe that they do not have options for improving their situation, Cox said.

Overall, 60 percent of the white working class reported being in poor or fair financial condition.

"The middle class can't survive in today's economy because there really isn't a middle class anymore," said one participant, a man. "You've got poverty level, and you've got your one, two percent."

A separate section of the report treats problems of mental health and addiction among the white working class, which Cox suggested could be in part a consequence of financial distress. Thirty-eight percent said that they themselves or a member of the household had suffered from depression, compared to 26 percent of white college graduates.

"You can't completely divorce it from the economic experience of these folks — the fears of economic security," Cox said. "That's certainly in the mix."

After learning more about their survey participants, the researchers looked to understand which attributes, attitudes and economic circumstances were the best ways of predicting who would support Trump.

To figure out which traits were the ones that really distinguished Trump voters, they used a statistical technique known as a regression. Regressions are essentially a mathematical way of asking, "If we hold all else equal, how useful is one thing for predicting something else?"

In this case, the regression was intended to show how much more likely respondents are to have supported Trump based on their answers to each question, even when compared to other respondents who answered the other questions on the survey in the same way.

Some were obvious. For example, survey respondents who identified as Republicans were about 11 times as likely to support Trump as those outside the party, even when other factors (their demographic traits, political views, experiences of the economy and so on) were held equal.

That figure is a useful reminder of just how little Trump, despite his unconventional campaign and unorthodox policies, shifted voters' established coalitions — a testament to the power of party loyalty.

Readers interested in the debate over the motivations of Trump supporters concentrated more, however, on the indicators of cultural versus economic anxiety.

On some indicators, high levels of cultural anxiety did suggest a greater likelihood of supporting Trump. Those who supported identifying and deporting undocumented immigrants were over three times as likely to support Trump as those who did not, again comparing respondents who were similar in other respects. The same was true of those who believed that American culture needed protection from foreign influences, or who said they felt like strangers in their own country.

For other indicators, however, the connection was less clear. For example, those who said that discrimination against white Americans and racial minorities is equally serious were no more likely to support Trump than those who said discrimination against people of color was more of a problem.

There was even less clarity in the connection between economic turmoil and Trump support. Those who described college as a risky gamble were about twice as likely to favor the president. On the other hand, those who described being in poor financial shape were somewhat less likely to support him than those who said they were better off.

With those results in mind, the authors wrote: "Besides partisanship, fears about immigrants and cultural displacement were more powerful factors than economic concerns in predicting support for Trump."

And for many readers, that sentence apparently closed the case — proof that racial resentment and cultural anxiety were the real drivers behind Trump's success.

That open-and-shut analysis is not the kind of conclusion, however, that regressions are designed to support.

For Cox, the regression shows that Trump's rhetoric about immigrants and American culture was what most powerfully appealed to his voters. It also shows that those who supported him were not among the worst off financially, but rather those in the middle and lower middle classes.

When it comes to drawing broader solutions, though, regressions have limitations as a statistical technique.

Regressions show which participants in a survey were most likely to support a given candidate, but they do not describe a candidate's typical supporter. For example, while those respondents who favored deportation are much more likely to support Trump, there are not that many who do. Even among Trump's supporters in the white working class, only 39 percent favor deportation, according to PRRI.

Regressions also show correlation, not causation. In this case, predicting whether participants support a candidate is not the same as showing why they do.

For instance, PRRI's regression reveals a connection between antagonism toward immigrants and support for Trump, but it does not provide evidence about the source of that hostility. It could be that the white working class is more likely to make undocumented immigrants into scapegoats for the country's problems because of years of economic frustration. That distinction does not reduce the burden of discrimination, but for everyone looking to understand where that animus comes from — or how it might be lessened over time — the distinction is important.

"This is typical of what has happened in the past in the United States, but in other Western democracies as well, when people have gone through a period of sustained stagnation in living standards," said Harvard University economist Benjamin Friedman, who was not involved in the study. "The Trump phenomenon fits the pattern."

Friedman has shown that extended period of poor economic performance are typically followed by reactionary politics. In the United States, for instance, stagnation at the end of the 19th century was accompanied by the Chinese Exclusion Act of 1882, which barred immigrants from China, and a spate of Jim Crow legislation in the South.

In the years after World War I, economists count four recessions excluding the Great Depression, and income per capita was increasing by only about 1 percent annually. At that time, Americans witnessed the resurgence of the Ku Klux Klan and more exclusionary legislation making it difficult for immigrants from southern and Eastern Europe to come to the United States.

President Reagan's tough-on-crime, anti-welfare politics followed a spell of inflation, stagnation and turmoil in the 1970s. And today, the American economy has been lackluster for almost two decades, and typical household incomes remain below their level in 1999.

"The overwhelming tendency I think is very, very clear," Friedman said.

Other researchers have reached similar conclusions. German economists who compiled data on over 800 elections held in 20 countries over the past century found that the share of ballots won by extreme-right parties increased by about a third in the five years following a financial crisis.

Like the PRRI study, research on populist movements has found that they typically appeal to the lower class, but not to the poorest of the poor. Their experience of the economy may have less to do with their own circumstances, and more to do with their perceptions of how their friends, family and neighbors are doing, and of whether they are getting their fair share.

"In periods of economic stagnation, when people are really feeling pinched, then that can exacerbate cultural fears," PRRI's Cox said.

Although regressions can be misleading, PRRI's results are informative when taken in the context of all the reports' other findings. The study offers a detailed examination of the lives and beliefs of the white working class — a group afflicted by material hardship and mental illness, frustrated with the existing economic system, hostile toward people of color and uncomfortable with the country's increasingly cosmopolitan culture. The white working class, the report shows, is becoming alienated from the dominant trends and institutions in American society today.

PRRI's regression showed that attitudes toward immigrants and anxiety about the changing culture were the most reliable indicators of who supported Trump, but that fact does not rule out the possibility that other factors were connected with his victory.

Cox said that if Trump does not deliver a real improvement in living standards, he might find the white working class giving up on him — potentially giving Democrats an opening to campaign on economic issues.

"It's possible in 2020 we could see a resurgence of economic issues," he said.

The strangest fact about the data on the white working class might be that the dry and technical debate over how to interpret the statistics continues to incite so much passion and outrage — even six months after the election. It has become less a mathematical dispute than a moral one.

It seems as though Trump's opponents are only willing to extend their sympathy to those who had some kind of legitimate reason for voting for him. Presumably, though, few Democrats would argue that because of how many white, working-class Americans voted, this group does not deserve help — that policymakers should not act to address stagnant wages and widespread addiction. On the other hand, prejudice is not more defensible just because it is a response to economic conditions, nor is it less of a burden on those who are excluded.

Cox argues that human experience is about more than just politics, and that people are more than their votes. For the white working class, the election is becoming "the solely defining moment or event for us in understanding who they are, or what they care about," he warned.

"We reduce people in some ways," he said. "I would worry about defining any group of people by a single election."

And until the next one, the white working class, like the rest of the country, will go on with their complicated lives, going to work, buying groceries, raising kids, getting by. Occasionally, making decisions with real moral consequences is a part of life, too — but very few of those choices will have anything to do with Trump.