Sympathy for the Devil?

Refusing to engage with Trump's base will only guarantee the growth of the far right.

Flip Schulke / The US National Archives

Vox’s Dylan Matthews wants the media to stop making excuses for Trump supporters. The Trump phenomenon isn’t about “post-industrial decay,” he writes. It’s not happening because “neoliberal capitalism is failing.” Such depictions, by Beltway media types on one side and “leftists and social democrats” on the other, willfully ignore the obvious.

“The press has gotten extremely comfortable with describing a Trump electorate that simply doesn’t exist,” Matthews says. If you want to know the Trump supporters’ concerns, you just have to look at what they’re saying in polls: It’s about racial resentment. It’s about white nationalism. Matthews has the data to prove it.

For one thing, they’re not working class, or even economically struggling; “Trump’s supporters are not the wretched of the earth.” Matthews points out that at the national level the median income of Trump’s primary voters was higher than Hillary Clinton’s. Support for Trump in polls was “correlated with” higher income, even among whites. Yet, perversely, the media has been pumping out feature stories about the fervor for Trump in hardscrabble places like Leetonia, Ohio, or Creston, Iowa — places that, statistically, don’t even exist by Matthews’s calculations.

It’s not that Matthews doesn’t care about real economic suffering; he’s no conservative. He recognizes that globalization has been hard on many Americans. He knows that there are poor whites in this country as well as poor blacks and Latinos. “The government should help people who are materially struggling,” he writes. “And Hillary Clinton, to her great credit, has offered programs . . . that will leave millions of white Trump supporters much better off.”

But for Matthews, the key point is this: “This isn’t worth doing to win back their votes; it’s worth doing because it’s the right thing to do.”

That’s a curious idea. If Matthews supports the Democratic Party’s agenda, why wouldn’t he want it to win back as many Trump votes as it can? How can the Democrats gain the kinds of majorities they need to push through all the beneficent policies he cites if they fail to win votes away from the other side? Isn’t that one of the ways parties win elections — by taking votes from the other side? In fact, isn’t that why Hillary Clinton’s campaign is now wooing anti-Trump Republicans?

Obviously Matthews wouldn’t want Democrats to use racist appeals to win Trump votes, and neither would I. But neither does he claim that anyone’s calling for such a step. He certainly doesn’t cite anyone who is.

What he does complain about is the “unprecedented outpouring of sympathy” for Trump voters that he sees in the media: the earnest pleas, often written by conservatives or Beltway-pundit types, that we should listen to the concerns of “the millions of white voters living on the edges of the economy,” the “decent, sincere people who feel disregarded,” and so on.

Matthews finds these stories exasperating. And you do have to agree with him when he points out that there was never any comparable litany of hand-wringing about the “concerns of Mitt Romney voters” or the “interests of Hillary Clinton supporters.”

But then again, when I look at my Twitter feed, which is full of elite media types, the main outpouring I see is just the opposite: a flood of contempt and disgust for Trump supporters, not just for the irresponsible votes they cast, but for their defective character as a group. Come to think of it, that might help to explain the rash of sympathetic pieces.

Matthews’s admonition against trying to win back Trump voters reminds me of one my favorite quotes from the ever-colorful Grover Norquist. In a 2006 American Prospect roundtable, the conservative anti-tax impresario was asked about the gay rights issue — at the time, he was taking heat from fellow Republicans for his work with gay organizations:

I speak to the Log Cabin Republicans and work with them on a whole host of issues . . . The Human Rights Campaign on certain things . . . So I get trashed from time to time by some of my friends. I think it’s a mistake to write off any group.

I was in Romania, they’re having elections in four weeks, and I was organizing the non-communists. And I had them write on a blackboard: Who’s Voting for Us, Who’s Voting for Them. And they had to list [the voter groups and] understand why everybody was [voting that way].

They had the gypsies voting for the Communists. And I said, “OK, I get why the Communists are voting for the Communists, and the Army and the police and the guys with government jobs. But why the gypsies?” If I were a gypsy I’d want to live outside [even] touchy-feely US law, much less harsher Communist law.

And they said, “Well, the Communists buy them liquor and then they vote for them.”

And I said, “We can do this. George Washington did this, it’s OK.” And they said, “No, the gypsies are scum and we won’t talk to them.” And I said, “OK, I guess you’re not getting the gypsy vote then.”

In politics, you want to have as few gypsies as possible, as few groups of people who are not voting for you because you’re not talking to them.

Maybe it’s not surprising that a conservative like Norquist and a liberal outlet like Vox would have differing views of politics. What’s surprising is Vox’s preference for the outlook of the Romanian right.


But aren’t the Trump voters inseparable from the racist appeals of Trump himself? That’s what Matthews seems to argue, with the aid of a raft of studies and data.

In doing so, he sheds useful light on the standard liberal way of thinking about politics. Matthews seems to take individuals as the elementary particles of political life. The individual is apparently endowed with a more or less well-defined set of attitudes on all the major issues of the day.

If you sum up the aggregate of those opinions, what you get is “politics”: election returns, opinion polls, legislative roll calls, all the quantitative mass phenomena of national politics. These are just the sum of individual brains.

As for how those opinions got manufactured and sorted into all those brains, that’s a question for psychology, or history, or economics, not politics.

You can see this methodology at work in the studies Matthews cites to make his case. Almost all of them follow the same approach. First, they measure the attitudes expressed by Trump supporters in multiple-choice polling questions. Then they compare them to the answers of non-Trump supporters. (Usually they control for other factors as well.) Whichever issues most sharply distinguish the Trumpist group from non-Trumpists are assumed to reveal the Trump supporters’ innermost feelings, hopes, and fears — in short, their motivations. Individual motivations, it’s assumed, can be inferred from group differences.

In summing up all the correlations and cross-tabs, Matthews is very clear on this point: “Trump’s voters [are] motivated by genuine political disagreement about race”; “these people [are] motivated by racial resentment”; “Trump’s supporters are not, in fact, motivated by economic marginalization.”

I’ve seen most of the studies Matthews links to. As far as I can tell, none analyzed polling questions that actually asked people what was “motivating” them. Instead, they used standard polling questions like: What is your household income? Do you approve of Obama? Should taxes be cut? Should immigration be reduced? Is black poverty caused by a lack of effort? Actual motivations were never recorded: they were inferred by researchers, using math, and then imputed to Trump supporters en bloc.

For example, one of the studies Matthews cites analyzed questions from the 2016 ANES pilot survey. According to Matthews, the study concluded that while “support for Trump is correlated most strongly with party ID, the second biggest factor . . . was racial resentment.” The study’s author concluded that “Trump support isn’t about the economy.”

Meanwhile, the same ANES survey had actually included a battery of questions that did try to ask respondents about their motivations. It listed twenty-one different political issues and asked: “Which of the following issues are the most important to you in terms of choosing which political candidate you will support?”

Trump’s three signature racially coded themes — immigration, terrorism, and crime – were among the possible choices; a third of Trump supporters picked one of those as their top issue. Two-thirds did not. 51 percent chose traditional kitchen-table issues like the economy, health care, Social Security, taxes, or the national debt. Another 8 percent chose culture-war issues like abortion, gay rights, or “morality.” And the remaining 8 percent chose “military strength,” “foreign policy,” or gun control.

Measuring Hate

Let me be clear. All of the following are true: From the start, Trump has put naked appeals to racism at the center of his campaign. In the process, he has magnetized a congeries of alt-right eugenicists, Confederate flag-wavers, and paranoid Mexican-haters to his cause. And then he went on to win 52 percent of the Republican vote in the primaries; he’ll probably win at least 40 percent of the popular vote in November.

Those facts aren’t in dispute. The question is what to make of them. There’s no doubt that the nation’s “white nationalists” provide disproportionate support to Trump’s racist campaign – and given the campaign’s tone, it would be very strange if they didn’t. Presumably, the nation’s socialists also provided “disproportionate” support to the Bernie Sanders campaign. And that sort of effect seems to account for how all the studies Matthews cites arrived at their findings, mathematically speaking.

For example, Matthews points to a study by UCLA’s Michael Tesler who found that “support for Trump in the primaries strongly correlated with respondents’ racial resentment,” and did so more strongly than McCain’s support in 2008 or Romney’s in 2012.

What that means concretely, if you look at Tesler’s charts, is that on the one hand Trump did a lot better than Romney and McCain among the more racially resentful half of Republicans; but on the other hand, he did equally well as them among the less racially resentful half. From eyeballing Tesler’s charts, it appears that the more racially resentful half of Republicans contributed a bit under 50 percent of Romney and McCain’s primary support. For Trump, the number was about 60 percent.

If that difference doesn’t seem all that big, it’s because while Trump has been very effective at mobilizing the most obsessively racist fraction of Americans to his cause — and great at winning Republican votes overall — he hasn’t been manufacturing more racists.

Indeed, amid the flood of “explainer” articles reporting the findings of complicated regression studies on racist attitudes, it’s striking how rarely you see simple aggregate numbers. The heated polarization of the Obama and Tea Party era in particular gave rise to an outpouring of intricate studies on the political correlates of “racial resentment” – dozens of which have been reported in Vox.

Meanwhile, in a co-authored academic article published this year, Donald Kinder, the University of Michigan social scientist who first developed the concept of racial resentment, reported: “Racial resentment is essentially stationary over the last quarter century, as measured by the ANES or by the GSS. We detect no sign here that White Americans’ racial resentments hardened during the Obama Presidency.”

Likewise, Gallup regularly asks the question, “Should immigration be kept at its present level, increased, or decreased?” Anti-immigration sentiment has been in long-term decline among non-Hispanic whites. In 2002, those wanting less immigration exceeded those wanting more by 43 percentage points. This year that number was 22 percentage points.

And that seems to be the case all around the world. Take the example of France, where the level of racism in political discourse seems to reach new heights every week and the far-right has been on the ascendant for decades. Yet the percentage of the French who say there are “too many immigrants in France” fell from 75 percent in 1988 to 50 percent in 2012. The percentage who think immigration is a “source of cultural enrichment” rose from 44 percent in 1992 to 75 percent in 2009. The percentage who agree that immigrant workers “should be seen as being at home here, since they contribute to the French economy” rose from 66 percent in 1992 to 84.5 percent in 2009.

In one sense, these figures — taken from a comprehensive 2013 study of long-run French public opinion by a team of political sociologists led by Vincent Tiberj of the Center for European Studies at Sciences-Po — shouldn’t come as a surprise. As the authors note, it’s long been understood that tolerance rises with education levels, and education levels have been rising for decades. Older and less-educated groups, in turn, are affected by the liberalizing cultural climate driven by younger and more-educated cohorts, albeit with a lag.

Thus, over the long run, each generation tends to express more tolerant attitudes than the last, and each generation tends to get more tolerant as it ages. “In all Western countries,” Tiberj says, “electorates are, generally speaking, more open and more tolerant than they’ve ever been.”

Yet you’d never guess any of that from watching the reactionary spiral of French political discourse. Tiberj explains the “paradox” this way:

Historically, France has never been more tolerant. Yet polarization on cultural values has never been as strong, either. The explanation is simple: If there’s a rightward shift, it’s above all a shift in the political debate.

When Mitterrand was reelected in 1988, the electorate still voted according to its socio-economic values. To put it simply, white-collar professionals voted for the right and workers voted for the left.

Starting with Nicolas Sarkozy’s [hard-right] 2007 campaign, things changed: voters started to vote according to both their socio-economic and their cultural values . . .

[And] while Nicolas Sarkozy didn’t hesitate to assert a right-wing ideological narrative to push a wedge into the debate, François Hollande has a lot more trouble with that . . .

Generally speaking, political discourse re-shapes the logic of voting: the rightward shift isn’t a demand coming from the electorate, it’s a result of the political supply . . . Our study tends to demonstrate the primordial importance of ideological combat.

Sensational events like riots, scandals, or terrorist attacks do cause short-run declines in tolerant attitudes, Tiberj finds. Eventually they’re forgotten and tolerant attitudes resume their long-run rise. But in the meantime, they can have long-run effects on politics by altering the terms of political debate — the ideological formulations offered by visible representatives of the Left and the Right. These are crucial in determining the tenor of the political discourse. And that tenor, in turn, alters how individuals understand their own relationship to politics, their own interests — even their own “motivations.”

“The same individual can simultaneously present dispositions to tolerance and to prejudice,” write Tiberj and his co-authors, “with the prevalence of the one over the other being strongly dependent on the environment, the information received, and recent events that have made an impression on them.”

The Union Effect

In one of those journalistic forays to the struggling pro-Trump hinterland that Matthews finds so annoying, the Guardian’s Paul Lewis and Tom Silverstone recently traveled to West Virginia for a video reportage on the most pro-Trump county in the state’s Republican primaries: McDowell County. A former coal mining area that lost its mines, McDowell is about as destitute and decrepit as you might expect.

One man they spoke to, a poor and elderly former coal miner of twenty-six years, was a lifelong member of the United Mine Workers. “I voted for that black guy two times,” he says with a laugh. Asked how he’ll vote in November, he says he’ll vote for Trump. His reason is, “Donald’s going to put all the coal mines back.”

Historically, being a member of a union — especially a combative union with an active internal life — is the most important instance of the kind of external force that Tiberj points to as having the potential to transform how individuals translate their “dispositions to tolerance and to prejudice” into their political outlook and behavior.

West Virginia’s history is a fine example of this. In the 1920s, when the UMW was weak and declining, the state’s politics were reactionary, dominated by ruthless coal operators and the Ku Klux Klan. But the resurgence of the UMW in the 1930s on the back of militant mass strikes transformed West Virginia politics, integrating its otherwise insular working class tightly into a national labor-liberal New Deal coalition that depended for its survival on black workers and black voters.

That’s why in 1964, four months after Lyndon Johnson signed the Civil Rights Act, West Virginia voted for him against the anti-labor Goldwater by the sixth-largest margin of any state. It’s why in 1968, when the Democrat on the ticket was the labor-liberal Hubert Humphrey — a figure who for twenty years had been more visibly committed to the civil rights issue than any other national politician — West Virginia’s vote for him was the seventh-largest in the country. (And its vote for George Wallace was far lower than the other border states.)

And remarkably, that union effect continues today, on a smaller scale: in 2012 Barack Obama’s deficit among non-college whites in the top half of the racial-resentment scale was 26 percentage points smaller among those who belonged to unions than those in nonunion households, according to data from the large-sample Cooperative Congress Election Study.

What’s remarkable about this is that union membership today is so often a lamentably low-intensity, low-commitment affair. It’s almost surprising to find it having any aggregate effect at all. And yet by reshaping the individual’s understanding of the stakes of politics, being in a union still has a powerful effect on how “dispositions to tolerance and to prejudice” translate into political behavior.

The Trump Vote

Trump’s voters in November will come in different varieties. There will be McMansion-dwelling evangelicals in the Atlanta suburbs. There will be owners of prosperous construction businesses in rural California. And there will also be voters like that West Virginia ex-coal miner in the Guardian report.

A charitable reading of Matthews’s piece is that he merely wants us to keep those first two types in mind, lest we succumb to the illusion that the Trump phenomenon was all about downtrodden coal miners. A less charitable reading is that he wants us to forget that it was about that, too.

Given the long, slow slide in the Democrats’ performance in the House of Representatives, the governors’ mansions, and the state legislatures, many will ask what the party could do to strengthen its position. As analysts sift through the returns, Trump’s eye-popping margins among non-college whites will generate a great deal of commentary. (In 2012, Obama lost even whites with a college degree but without postgraduate training.)

The numbers will be clear: downscale whites are a big pool of untapped votes. Yet if a cordon sanitaire is placed around that demographic territory and hung with the notorious label, “Trump Vote,” the Democrats will be even more likely to let the party system drift down its current path: into the culture-war politics of the reactionary Tammany-versus-Klan 1920s, rather than the class-based politics that followed.