As is often the case with things that become widely popular, articles examining the voting preferences of white Trump voters in the Midwest are now dismissed as necessarily useless and repetitive. I can tell why, in the reporter's wake Trip Gabriel's article about voters near Youngstown, Ohio, published in the New York Times on Monday, the comments were one piece: Here we go again.
Well yes. Here we go again, looking at the president's support as he approaches re-election in a district where he narrowly secured the votes needed to win the Electoral College three years ago. There was often an element of defensiveness in the early 2017 treks across the Midwest and, at times, a sense of overcompensation in constantly checking in with the same voters in 2018. But now, it's safe to say the dynamic has shifted. How voters in Ohio, Pennsylvania, Michigan and Wisconsin view the president is information important to understanding the upcoming election.
The challenge now, though — as it was in 2016 — is to understand both how important the white working-class vote is to politics in these states and how much of that vote is influenced by Trump himself.
We can answer this last question quite directly. Data released by Gallup in April shows that the politics of whites without college degrees — a reasonable proxy for the “working class” — has become steadily more favorable to the GOP over the past 20 years.
In other words, it's not really that Trump motivated white working-class voters to abandon the Democrats. If anything, it mobilized more of them to vote. At the same time, he benefited from declining turnout among voters who supported Barack Obama in 2012. As we've reported, an estimated 4.4 million Obama voters did not turn out in 2016, a third of them black.
As a general rule, there are far more white working-class residents in most counties than black residents. Nationally, there are about four times as many white Americans without college degrees as there are black Americans. In more than 2,400 of the more than 3,100 counties for which the Census Bureau has detailed data, the black population is less than 20 percent of a county's working-class white population.
While the correlation between the ratio of black to white working-class voters in a county is not as strong, counties with more white working-class residents than blacks supported Trump by an average margin of 35 points. Counties with the largest number of black residents supported Hillary Clinton by 28 points.
(Why isn't it very robust? Partly because there are a lot of very white, very Democratic areas, like the Northeast.)
More interesting is a broader look at how white and non-white populations compare. There are more than 400 counties where the population of non-white residents is greater than the population of working-class white residents, compared to only 150 where there are more black residents.
Note, however, the relative lightness of the area we were discussing. In Ohio, the average county has about 18 times more white working-class residents than black residents. It's fair to note, as does Jamelle Bouie of the Times He madethat framing a look at the residents of Youngstown, Ohio, through the lens of the white working class is inaccurate (the town is the majority of non-whites). But it's also true that “Youngstown” was shorthand for “manufacturing workers in Northeast Ohio” long ago. Bruce Springsteen's song named after the city.
A look at this region of the state shows that most counties with larger cities are not overwhelmingly non-white.
Mahoning, Summit and Cuyahoga counties all voted for Clinton in 2016 anyway, based on the strength of turnout in those towns.
In Mahoning County, however, the margin was only about three points, after Obama won it by 28 points in 2012.
So we return to the questions we had at the beginning. How much of that change was a function of Trump — either through boosting turnout or persuading Obama voters to support his candidacy? How critical are these white working-class voters of Trump's hopes of winning a second term?
You have been forewarned: These questions require more reporting.
