The demographic detail of Donald Trump’s victory

or democrats, the worst thing about Kamala Harris’s loss is that Donald Trump will return to power. The second-worst thing, however, may be their inability to explain away his victory. In 2016 the left attributed his triumph to racism and the electoral college. This time, he won the popular vote—thanks not to “deplorables” but, the exit polls implied, to surging support from the non-white working class.

Did Mr Trump really win Latino men by 12 percentage points, as exit polls showed? Such surveys tend to overestimate how many voters belong to Democratic-leaning groups, and thus underestimate the share of voters within each group who back Democrats. This bias can be extreme. In 2016 the exit poll found that Mr Trump had won white college graduates by three percentage points. More accurate post-election surveys had him losing them by 15.

The best way to verify the exit polls’ findings is to test them against election results. In theory, if Mr Trump made larger gains among Hispanic voters than among white ones, he should also show the biggest improvements in heavily Hispanic areas. And returns from the 3,000 counties that have counted at least 90% of votes show that “racial depolarisation” was both greater than many Democrats thought possible, and insufficient to account for the magnitude of America’s red shift.

The most striking aspect of Mr Trump’s gains is their uniformity. Around 90% of American voters live in counties where his vote share increased from 2020. Half live in those where it rose by at least 1.9 percentage points. But some counties swung more than others, and these differences confirm that the shift was concentrated among Latinos. Among the 10% of voters living in the most Hispanic counties, the president-elect’s vote share rose by 5.1 points. For the 10% in counties with the fewest Latinos, the swing was just 1.6 points.

However, this predictor leaves big regional patterns unexplained. In Florida and especially the New York City area, the swing was greater than the Latino population implies. Elsewhere, it was smaller.

A more comprehensive variable eliminates such discrepancies. The foreign-born share of a county’s population explains half of geographic variation in the electoral swing from 2020 to 2024, making it an even stronger predictor than the white-working-class share was of shifts from 2012 to 2016. Places like New Mexico, with lots of Hispanics but relatively few immigrants, swung in line with the national average. Meanwhile, areas with many foreign-born residents and modest Latino populations, like Loudoun County in Virginia, lurched right. Crucially, what aligns with voting patterns is the size of the immigrant population, not the growth rate. This implies that the people changing their minds were immigrants themselves, rather than their native-born neighbours.

Naturalised Republicans

Pre-election surveys did not foresee a big shift among immigrants. In polls by YouGov since August Mr Trump gained more ground (relative to how respondents said they voted in 2020) among native-born likely voters than among foreign-born ones. And since January 2023, the share of YouGov’s respondents whose top issue was “jobs and the economy” or “inflation/prices”—subjects on which Mr Trump had a big lead—was the same for immigrants and native-born participants. Where foreign-born voters did stand out was their emphasis on public services like health care, education and public safety. They may have judged Democratic-run local governments poorly on these issues.

Exceptions to the overall trend of immigrant-rich regions shifting the farthest right offer further clues about the nature of Mr Trump’s increased appeal. He did unusually well in counties where lots of people trace their ancestry to three Catholic European countries—Ireland, Italy and Poland. The Democrats’ unceremonious jettisoning of Joe Biden, a proud, practising Irish Catholic, may have hurt them in such areas. Meanwhile, Mr Trump’s gains were relatively muted in counties with lots of college-educated whites or same-sex couples. In culturally liberal pockets, no amount of nostalgia for the pre-pandemic price of eggs could overcome cosmopolitan voters’ revulsion for Mr Trump.

Regional economic conditions and foreign policy also appeared to shape voters’ choices. The red shift was greater than other data implied in places with high poverty rates (like Native American reservations or eastern Kentucky), and where housing is expensive. Discontent over the conflict in Gaza appears to have hurt Ms Harris in counties with lots of Arab-Americans. And Mr Trump, who often praises Vladimir Putin, did unusually well in counties with a high prevalence of Russian ancestry, and relatively poorly in those where Ukrainian ancestry is most common.

Ms Harris has faced harsh criticism since her loss. But the data tentatively suggest that the vanquished vice-president’s campaign was better not only than Mr Biden’s, but also than Mr Trump’s.

Democratic Senate candidates did outperform Ms Harris, enabling the party to hold seats in four states carried by Mr Trump. But such ticket-splitting is no indictment of Ms Harris: it was roughly what was expected given Democratic nominees’ advantages in incumbency, fundraising and experience. Moreover, Ms Harris appears to have eliminated Mr Trump’s edge in the electoral college. In 2020 Mr Biden’s margin of victory was four percentage points greater in the national popular vote than in Wisconsin, the decisive state. This year, Mr Trump is likely to win nationwide by 1.5 points and in Pennsylvania, the “tipping point”, by 1.9—a trivially small gap.

Ms Harris cannot take full credit for such outperformance: the Rust Belt swing states should have shifted less than other places, since they have relatively few immigrants and Latinos. But on average, Ms Harris’s vote share in the seven battleground states was a full percentage point higher than their demography suggests. The number of ads seen in each county accounts for some but not all of this effect; the rest is probably attributable to superior events or get-out-the-vote operations.

A superior politician would presumably have fared even better. But outside the battlegrounds, Mr Trump gained an average of 3.3 points of vote share, which by modern standards counts as a tidal wave. The hole that Mr Biden dug was probably too deep for any successor to escape.