Political scientists will be analyzing the bases of Donald Trump’s dramatic victory for a very long time. So far, the evidence points toward a toxic brew of economic and cultural anxiety, populism, racial resentment, sexism and authoritarianism, among other factors. But from a broader perspective, the search for specific explanations for Trump’s support is probably misguided. An extraordinary campaign has produced a remarkably ordinary election outcome, primarily reflecting partisan patterns familiar from previous election cycles.
As John Sides has already noted, the national election outcome was consistent with forecasts based on “fundamental” factors like incumbency and the state of the economy — though he and I and many others imagined that it would not be, given the many remarkable features of this year’s race, and of Trump’s campaign in particular. By the conventional standards of high-minded democracy, many commentators believed that Trump’s manifest unsuitability for the presidency should have doomed him at the polls. But millions of ordinary voters, employing their own less rarefied standards, viewed Trump as a strong leader who would “say what he meant” and “get things done.”
Most state election outcomes were also surprisingly consistent with past voting patterns. Here is the relationship between Trump’s popular vote margin in each state on Tuesday (based on preliminary returns, with the heavily Democratic District of Columbia omitted in order to make the other results more legible) and Mitt Romney’s margin in 2012:
Utah is a dramatic but not very surprising outlier, with unusually strong support for Romney (a Mormon) in 2012 and for independent candidate Evan McMullin (also a Mormon) in 2016. No other state saw a popular vote margin shift by more than 16 points — a net shift of 8 percent of the electorate. The places in which Trump ran ahead of Romney included such famously white working-class states as West Virginia, Iowa and Ohio, but also Hawaii (Barack Obama’s home state), Maine and Rhode Island. The average net shift (weighted by state population) amounted to just 1.2 percent of the electorate, and the correlation between 2012 and 2016 results (likewise weighted by population), even with Utah in the mix, is 0.93.
It may be helpful to put these shifts in historical perspective by comparing them with similar shifts in previous election cycles. Elsewhere I have tracked continuity and change in presidential election outcomes from 1868 through 1996, noting that recent elections have been marked by “a period of partisan stability and predictability unmatched since the end of the 19th century.” That period of unusual stability has continued through the first two decades of the 21st century. Here is the trend in “partisan continuity,” a measure of the extent to which each set of state-level election outcomes has reflected the partisan forces evident in the three preceding elections:
The extent of partisan continuity evident in 2016 — at the far right of the graph — is slightly lower than in 2012, but well above the average value for the past century, and very much in line with typical values in the past several elections.
I have also tracked the magnitude of new national and local forces producing changing electoral patterns in each election year — a combination of the shift in the national partisan tide (the absolute value of the intercept in a simple regression analysis) and the variation in the average magnitude of state deviations from that national shift (the standard error of the same simple regression analysis). Prominent peaks in 1912, 1932 and 1972 represent major shifts in American electoral patterns, at least in the short term, with a typical state moving by as much as 15 percentage points. In contrast, the 2016 value (4 percentage points) is quite low by historical standards — and even lower than has been typical over the past 30 years:
The impression of partisan stability suggested by the state-level results for 2016 is generally echoed in national exit polls, which showed Clinton winning 89 percent of the vote among Democratic identifiers and Trump winning 90 percent among Republicans. These percentages precisely match those in 2008, the last time there was no incumbent president on the ballot. (The corresponding percentages in 2012 were only slightly higher, 92 percent and 93 percent.) To a good approximation, Trump won because the overwhelming majority of Republicans voted for the Republican candidate, undeterred by the qualms of party leaders and conservative intellectuals.
Nonetheless, as usual, pundits are trotting out visions of realignment. Even before the election, Michael Lind proclaimed that “the old political system is crumbling, and a new American political order is being born.” Jonathan Haidt called 2016 “a global turning point … the year that the battle between globalists and nationalists became the central axis of conflict” in the United States and Europe, adding that “the battle will grind on … long after Donald Trump leaves the national stage, as the United States becomes a ‘majority minority’ country sometime between 2050 and 2060.”
Perhaps. We won’t know anytime soon, since major realignments of the party system usually take years or even decades to unfold. In the meantime, however, it would be a mistake to read too much into Trump’s victory, remarkable as it was — and remarkable as his presidency may well be.
Christopher Achen and I have argued that in most democratic elections, “the choice between the candidates is essentially a coin toss.” In 2012, the “fundamentals” predicted a close election and the Democrats won narrowly. In 2016, the “fundamentals” predicted a close election and the Republicans won narrowly. That’s how coin tosses go. “It is a blunder to expect elections to deliver more.”
Larry M. Bartels is a political scientist at Vanderbilt University. His books include “Democracy for Realists” (with Christopher H. Achen) and a newly revised edition of “Unequal Democracy: The Political Economy of the New Gilded Age.”