Obama-Trump vs McCain-Harris Counties

The class signal

In 2024, Donald J. Trump won 480 counties that Obama had won in 2008. In contrast, Harris secured just 19 counties that McCain had secured back then. The gap between these two contains a much stronger signal of the class-partisan cleavage than that revealed by the covariates of vote change, which is the standard approach. The reason is that 2,557 or 83% of all counties are either deep red of deep blue. Swings that took place in them (and their underlying social cleavages) are not as directly implicated as these swing counties in the outcome. And conditioning on them allows us to isolate a cleaner signal.

Basically, the characteristics of McCain-Harris counties are much more certain to reflect the sorts of people who really bought the Democrat line of Trump being a threat to American democracy. Conversely, the characteristics of Obama-Trump counties sheds special light on the character of the swing voters who bolstered Trump’s coalition.

Here’s some aggregate statistics. A bit less than 10 million people live in McCain-Harris counties; some 52 million live in Obama-Trump counties. The median college graduation rate—which captures the diploma divide central to the contemporary class struggle—is 42.8% in the former and 21.4% in the latter; literally twice as high. The median household income in the median McCain-Harris county is about eighty thousand dollars; in Obama-Trump counties it is just fifty-eight thousand. McCain-Harris counties are largely metros or affluent suburbs at the upper end of the rural-urban continuum; whereas Obama-Trump counties are somewhat towards the rural end. These are all textbook signatures of the working class-professional class divide. Significantly, the population of McCain-Harris counties has increased by 5.2% since 2020; while Obama-Trump counties have decreased by 0.5%. This is another class signal: Obama-Trump counties are being literally left behind.

As we shall see, these are highly significant differences, allowing us to make reliable inferences about what just happened. In what follows, we report t-tests for the difference between the two, and display violin plots to get a sense of the socioeconomic gap between them and their variation.

We start with the college graduation rate. The difference is very large (21.4% or 2x) and highly significant (t = 9.54, P < 0.0001). Notably, McCain-Harris counties are socioeconomically even more of the blue type than the deep blue Obama-Harris counties. This is what I meant when I said our selected contrast contains more signal.

We find the same pattern in median household income. The difference we noted above (80k vs 58k) is also extremely significant (t = -7.79, P < 0.0001). By way of inference, the Never Trump Republicans who voted for Harris are not only on the right side of the diploma divide but considerably more affluent. Indeed, these are residents of counties that are even more affluent than deep blue counties. So, the signal from income and education has become completely congruent in terms of what it tells us about the class-partisan rotation.

The McCain-Harris counties are also doing well, in the sense that they are attracting people; while Obama-Trump counties are losing people. The difference is again highly significant (t = 6.33, P < 0.0001).

The difference in the rural-urban continuum is also significant (t = -5.41, P < 0.0001). And once again we find the same pattern of McCain-Harris counties being socioeconomically bluer than blue. But note that Obama-Trump counties are not just rural—they span the entire rural-urban spectrum, attesting to the broad national basis of Trump’s triumph.

Again, this difference estimator contains a cleaner class signal than that revealed by change in vote share. And we can make two quite unambiguous inferences from this more kosher analysis.

First, the educated, affluent, urban elites—even those who had traditionally been Republican enough to vote for McCain over Obama—turned against Trump presumably because they bought the argument made by Democrats that he posed an existential threat to the American republic.

Second, large numbers of Obama voters—who can hardly be accused of racism—voted for Trump this year. And these voters are disproportionately on the wrong side of the diploma divide, have more modest incomes, and live in more rural areas that are bleeding people. This the classic signature of the working class.

So what just happened can be best described as the revenge of what is condescending called “flyover country” against the hated “coastal elites.” Trump won on the backs of a powerful multiracial working class coalition.