An Insurance Policy for Democrats

Democrats lately have been basking in good news. The fourth Trump indictment! Continued success for abortion rights (the defeat of the Ohio referendum)! Good news on “Bidenomics”  (slowing inflation and strong job creation)!

The sentiment seems to be: we got this! How could we lose to a candidate (assuming it’s Trump) who’s under a blizzard of legal scrutiny for undermining democracy and represents a party that wants to take away women’s right to choose—especially when we, the good guys, are doing such a great job with the economy?

This “how can we lose?” attitude is uncomfortably reminiscent of Democrats’ attitude in 2016. Then too they thought they couldn’t lose. And yet they did.

Perhaps it’s time to take out an insurance policy. It may be the case that a multiply-indicted Trump is now toxic to enough voters and abortion rights such a strong motivator that even a candidate with Biden’s weaknesses will beat him easily. But it might not and that’s where the insurance policy comes in.

Consider that right now the race looks very, very close. The RealClearPolitics poll average has Biden ahead of Trump by a slender four-tenths of a percentage point. If that was Biden’s national lead on election day, he’d probably lose the presidency due to electoral college bias that favors Republicans.

In the latest Quinnipiac poll, Biden has a one-point lead over Trump consistent with the running average. Among white working-class (noncollege) voters, he’s behind by 34 points, considerably worse than he did in 2020. If Trump (or another Republican) does manage to prevail in 2024, we can be fairly sure that a pro-GOP surge among these voters will have something to do with it.

States of Change simulations show that, all else equal, a strong white working class surge in 2024 would deliver the election to the GOP. Even a small one could potentially do the trick. In an all-else-equal context, I estimate just a one-point increase in Republican support among the white working class and a concomitant one-point decrease in Democratic support (for a 2-point margin swing) would deliver Arizona, Georgia and Wisconsin (and the election) to the Republicans. Make it a 2-point increase in GOP support and you can throw in Pennsylvania too.

So an insurance policy to prevent such a swing is in order.

The problem: these are very unhappy voters. In the Quinnipiac poll, white working-class voters give Biden an overall 25 percent approval rating versus 70 percent disapproval and 72 percent have an unfavorable opinion of him. On handling the economy, Biden’s rating is even worse—24 percent approval and 73 percent disapproval. Just 20 percent say the economy is excellent or good, compared to 79 percent who say it is not so good or poor. By 63 to 16 percent, these voters believe the economy is getting worse not better. Evidently they haven’t yet heard the good news about Bidenomics.

The temptation among Democrats is to ascribe the stubborn resistance of these voters to Democratic appeals and openness to those of Trump and right populists to misinformation from Fox News and the like and, worse, to the fundamentally racist, reactionary nature of this voter group. The roots of this view go back to the aftermath of the 2016 election.

As analysts sifted through the wreckage of Democratic performance in 2016 trying to understand where all the Trump voting had come from, some themes began to emerge. One was geographical. Across county-level studies, it was clear that low educational levels among whites was a very robust predictor of shifts toward Trump. These studies also indicated that counties that swung toward Trump tended to be dependent on low-skill jobs, relatively poor performers on a range of economic measures and had local economies particularly vulnerable to automation and offshoring. Finally, there was strong evidence that Trump-swinging counties tended to be literally “sick” in the sense that their inhabitants had relatively poor physical health and high mortality due to alcoholism, drug abuse, and suicide.

The picture was more complicated when it came to individual level characteristics related to Trump voting, especially Obama-Trump voting. There were a number of correlates with Trump voting. They included some aspects of economic populism—opposition to cutting Social Security and Medicare, suspicion of free trade and trade agreements, taxing the rich—as well as traditional populist attitudes like anti-elitism and mistrust of experts. But the star of the show, so to speak, was a variable labelled “racial resentment” by political scientists, which many studies showed bore a strengthened relationship to Republican presidential voting in 2016.

This variable is a scale created from questions like: “Irish, Italian, Jewish, and many other minorities overcame prejudice and worked their way up. Blacks should do the same without any special favors.” The variable is widely and uncritically employed by political scientists to indicate racial animus despite the obvious problem that statements such as these correspond closely to a generic conservative view of avenues to social mobility. And indeed political scientists Riley Carney and Ryan Enos have shown that responses to questions like these change very little if you substitute “Nepalese” or “Lithuanians” for blacks. That implies the questions that make up the scale tap views that are not at all specific to blacks. Carney and Enos term these views “just world belief” which sounds quite a bit different from racial resentment.

But in the aftermath of the Trump election, researchers continued to use the same scale with the same name and the same interpretation with no caveats. The strong relationship of the scale to Trump voting was proof, they argued, that Trump support, including vote-switching from Obama to Trump, was simply a matter of activating underlying racism and xenophobia. Imagine though how these studies might have landed like if they had tied Trump support to activating just world belief, which is an eminently reasonable interpretation of their star variable, instead of racial resentment. The lack of even a hint of interest in exploring this alternative interpretation strongly suggests that the researchers’ own political beliefs were playing a strong role in how they chose to pursue and present their studies.

In short, they went looking for racism—and they found it.

Other studies played variations on this theme, adding variables around immigration and even trade to the mix, where negative views were presumed to show “status threat” or some other euphemism for racism and xenophobia. As sociologist Stephen Morgan has noted in a series of papers, this amounts to a labeling exercise where issues that have a clear economic component are stripped of that component and reduced to simple indicators of unenlightened social attitudes. Again, it seems clear that researchers’ priors and political beliefs were heavily influencing both their analytical approach and their interpretation of results.

And there is an even deeper problem with the conventional view. Start with a fact that was glossed over or ignored by most studies: trends in so-called racial resentment went in the “wrong” direction between the 2012 and 2016 election. That is, fewer whites had high levels of racial resentment in 2016 than 2012. This make racial resentment an odd candidate to explain the shift of white voters toward Donald Trump in the 2016 election.

Political scientists Justin Grimmer and William Marble investigated this conundrum intensively by looking directly at whether an indicator like racial resentment really could explain, or account for, the shift of millions of white votes toward Trump. The studies that gave pride of place to racial resentment as an explanation for Trump’s victory did no such accounting; they simply showed a stronger relationship between this variable and Republican voting in 2016 and thought they’d provided a complete explanation.

They had not. When you look at the actual population of voters and how racial resentment was distributed in 2016, as Grimmer and Marble did, it turns out that the racial resentment explanation simply does not fit what really happened in terms of voter shifts. A rigorous accounting of vote shifts toward Trump shows instead that they were primarily among whites, especially low education whites, with moderate views on race and immigration, not whites with high levels of racial resentment. In fact, Trump actually netted fewer votes among whites with high levels of racial resentment than Mitt Romney did in 2012.

Grimmer and Marble did a followup study with Cole Tanigawa-Lau that included data from the 2020 election. The study was covered in a New York Times article by Thomas Edsall. In the article, Grimmer described the significance of their findings:

Our findings provide an important correction to a popular narrative about how Trump won office. Hillary Clinton argued that Trump supporters could be placed in a “basket of deplorables.” And election-night pundits and even some academics have claimed that Trump’s victory was the result of appealing to white Americans’ racist and xenophobic attitudes. We show this conventional wisdom is (at best) incomplete. Trump’s supporters were less xenophobic than prior Republican candidates’ [supporters], less sexist, had lower animus to minority groups, and lower levels of racial resentment. Far from deplorables, Trump voters were, on average, more tolerant and understanding than voters for prior Republican candidates…

[The data] point to two important and undeniable facts. First, analyses focused on vote choice alone cannot tell us where candidates receive support. We must know the size of groups and who turns out to vote. And we cannot confuse candidates’ rhetoric with the voters who support them, because voters might support the candidate despite the rhetoric, not because of it.

So much for the racial resentment explanation of Trump’s victory. Not only is racial resentment a misnamed variable that does not mean what people think it means, it literally cannot account for the actual shifts that occurred in the 2016 election. Clearly a much more complex explanation for Trump’s victory was—or should have been—in order, integrating negative views on immigration, trade and liberal elites with a sense of unfairness rooted in just world belief. That would have helped Democrats understand why voters in Trump-shifting counties, whose ways of life were being torn asunder by economic and social change, were so attracted to Trump’s appeals.

Such understanding was nowhere to be found, however, in Democratic ranks. The racism-and-xenophobia interpretation quickly became dominant, partly because it was in many ways simply a continuation of the approach Clinton had taken during her campaign and that most Democrats accepted. Indeed, it became so dominant that simply to question the interpretation reliably opened the questioner to accusations that he or she did not take the problem of racism seriously enough.

We are still living in that world. Scratch a Democrat today and you will find lurking not far beneath the surface—if beneath the surface at all—a view of white working-class voters and their populist, pro-Trump leanings as reflecting these voters’ unyielding racism and xenophobia.

This is neither substantively justified nor politically productive. Democrats desperately need that insurance policy for 2024 and getting rid of these attitudes toward 40 percent of the electorate (much more in key states!) should be part of it. Think of it as a down payment on the “de-Brahminization” of the Democratic Party. This attitude adjustment might irritate some of their activist supporters, but considering the stakes, that seems like a small price to pay for a potentially vital insurance policy.