Dylan Matthews, in a new piece for Vox, is back to rejecting “the idea that [Trump] voters are motivated by economic struggles.” This time around, he has a new argument:

Taking Trump voters’ concerns seriously means listening to what they’re actually saying . . . describing these people as motivated by racial resentment . . . [is] supported by extensive amounts of social scientific research and indeed by the statements of Trump’s supporters themselves.

This, of course, is supposed to cleverly turn on its head the notion that we’re being empathetic when we acknowledge that economic anxiety may play a role in motivating Trump voters. Which might seem like a nuance that avoids completely demonizing them — but in fact, Matthews argues, such explanations are “insulting” because they don’t take “the stated concerns of Trump voters . . . seriously in the slightest.” In fact, if we take Trump voters seriously, we should dismiss these personal theories of motivation and take them at their word when they say they’re motivated by racism.

Fair enough. But if Matthews wants to lean on deference to other perspectives, why does he then invoke “social scientific research” in addition to “the statements of Trump’s supporters themselves”?

These two things are not the same! Almost exclusively, however, Matthews relies on a collection of studies that statistically correlate support for Trump with various indicators of racism, and on his inference about the “message this research sends.” It’s true that those indicators are usually self-reported racism — people making statements like “If black people would only try harder they could be just as well off as white people,” for example. But from this evidence, researchers are concluding that one thing motivates the other, and pundits are concluding that one exclusively motivates the other. These may be defensible conclusions about the evidence, but they are not, as Matthews suggests, the conclusions of the respondents themselves.

That distinction is crucial, because Matthews is not making an argument about rigor, but about deference. He wants to cast people who see an economic role in support for Trump as ideologues who are even ignoring the explanations that Trump voters give for their own politics — that way, he gets to turn around the empathy argument so often wielded against liberals who are accused of ignoring economic hardship.

But to pull off that move, you would have to rely on what Trump voters are explicitly saying about their own motivations. And that point is fairly clear: when asked “why they support their candidate,” 76 percent of Trump voters credit his “views on the economy,” while only 28 percent credit his “views on race relations.” The single tweet Matthews offers as evidence that “it’s not about economics” doesn’t somehow overthrow these results.

Personally, I think it’s pretty ridiculous to defer to the theories Trump voters give about their own motivations. For example, when only 28 percent say they’re motivated by his views on race, I suspect that the long-understood problem of social desirability bias is coming into play and that a lot of them are being dishonest.

I also suspect that when 78 percent credit Trump’s “views on terrorism” for their support, what they actually mean are his views on Muslims and other minorities, meaning that the role of racism is actually much higher than 28 percent suggests. Additionally, I also suspect that Trump supporters are unlikely to even realize when they’re motivated by implicit racism, even though that almost certainly plays a huge role in their politics. These are all good reasons why we should not “take the stated concerns of Trump voters . . . seriously” in the sense of allowing them to discredit a more sophisticated analysis.

But of course, when we pursue a more sophisticated analysis, we often get more sophisticated conclusions. Thus for example, in the same Gallup study that Matthews thinks “confirms” his analysis, the author explicitly affirms the role of economic factors in Trumpism: the evidence, he writes, indicates “support for the idea that Trump supporters are less prosperous than others” and he goes on to speculate that “material circumstances caused by economic shocks . . . are the true underlying causes.”

Acknowledging a role for economics in the Trump candidacy does not, of course, preclude a role for racism as well — and as Jeff Spross points out, that was never the left argument to begin with. Regardless, Trump voters are obviously not telling us to rule out economic motivations, and even if they did, it would be ridiculous to defer to their analysis.