Why are swing states so white?

America’s racial geography gave Trump a boost in 2016, and will again in 2020.

Hillary Clinton won the popular vote. Donald Trump won the Electoral College. What happened?

The Electoral College gives a disproportionate amount of power to non-Hispanic white voters, who were far more likely to support Trump than any other racial or ethnic group.

This is not a result of small states like Wyoming and the Dakotas having more electors per person than large states like California and Texas. Rural, white states actually reduce the election-tipping power of white voters,¹ since they are solidly red and thus unlikely to be electorally important.

Rather, white voters have disproportionate power because they are overrepresented in swing states:

*Data on 2016 voter turnout broken down by race is not yet available from a non-partisan source such as the U.S. Census Bureau. These values are estimated based on 2012 turnout and changes in turnout 2012–2016 in states where good data is available.

This isn’t a coincidence, according to academics I spoke with who study race and electoral politics. Barring significant party realignment on racial issues, the states that decide elections will continue to skew white.

To understand why, we need to break it down into two pieces:

  1. Why are some states much whiter than the national average?
  2. Why are these “very white states” more likely to be swing states?

Why Appalachia, the Midwest, and the Rockies are almost exclusively white

Why is a vast swath of the country — from Western Pennsylvania to Idaho — inhabited almost entirely by white people, outside of a few urban centers like Chicago and Detroit?

Natalie Masuoka, a professor of political science at Tufts, said white Americans have a large presence throughout the country owing in part to a history of racially exclusionary immigration policy, public land allotment, and general freedom of movement. Regional differences are better understood by looking at migrations of communities of color, which Masuoka broke down into three core elements.

First, there’s the legacy of slavery and the Great Migration. “The largest black population was transported from Africa and the Caribbean to Southern states,” Masuoka said. “While a good share of black families have historically remained in the South,” the 20th century saw “significant internal migration of blacks” to the industrialized urban areas of the Northeast, Midwest and California. Today, about half of all black Americans live in the former Confederacy; another quarter live in the metro areas of New York City, Chicago, Washington, Philadelphia, Detroit, Los Angeles and Baltimore.

Second, there are Latin@ communities in the Southwest that have existed since before the areas around them became part of the United States, Masuoka said. About 30 percent of the continental U.S. was once owned by Mexico; these Spanish-speaking regions were brought into the Union by the annexation of Texas and the Mexican-American War. Much additional Latin@ immigration has gone to places like California, Arizona and Texas, where existing communities “could help new immigrants adapt to the U.S.,” Masuoka said. Today, these three states are home to half of all Hispanic and Latin@ Americans.

Finally, the easing of immigration quotas has permitted a boom in Asian immigration since 1965. While specifics vary by national origin group, “Asian immigrants tend to be highly educated professionals because our current immigration policy prioritizes that type of person for a visa,” Masuoka said. As a result, many “settle in areas related to their occupations” — generally, large metropolitan areas. Today, 33 percent of Asian-Americans live in the Los Angeles, New York or San Francisco metro areas.

By process of elimination, then, the non-urban portions of the Midwest, Appalachia and the Rockies are still almost entirely white. Recently, though, these areas too have begun to integrate as “new Latino and Asian-American immigrants have been moving into Midwestern, Southern and some more rural areas,” Masuoka said. “So for whites in these areas, the demographics of their areas could be seen as in flux.”

Why these very white areas are more likely to decide presidential elections

So why are whiter states more likely to be swing states? There’s a simple answer and a not-so-simple answer.

The simple answer: “Swing voters are overwhelmingly white,” said Dave Hopkins, a political scientist at Boston College. “Nonwhite voters are pretty loyal Democrats, in good years and in bad. So the more nonwhite voters you have in the state electorate, the less swing there’s going to be, all else being equal.”

We can measure this. Hopkins is describing elasticity, a measure of how sensitive a state is to changes in political conditions. A more elastic state has, more or less, a wider bell curve around its current poll numbers — it’s more likely to be pushed and pulled with the tides of the campaign.

Not all “very white states” (80%+) are swing states, of course; in 2016, they ranged from voting for Clinton by a 26 percentage point margin (Vermont) to favoring Trump by 46 points (Wyoming). But wherever they are on the ideological spectrum, if two states were producing similar poll numbers, the whiter state would be more likely to flip parties if national conditions changed.

Here’s where it gets a little complicated. It’s not just voters of color that make diverse states inelastic. Rather, the white voters who live in states with large nonwhite populations also tend to be more consistently partisan, Hopkins said.

“Southern conservative whites in places like Mississippi and Louisiana are pretty loyal to Republicans,” Hopkins said. “So they’re not going to be very elastic either. And the same can be said for white liberals in California and New York — those voters are reliable Democrats and they just happen to live adjacent to significant non-white populations.”

Is this just historical coincidence, or is there some kind of causal relationship here? I spoke to Ryan Enos, a Harvard political scientist who studies the politics of intergroup contact.

“When you have proximate nonwhite populations that are segregated, that’s going to tend to move [white] people to the right,” Enos told me. “If you have enough intergroup contact over time, that’s going to be correlated with voting on the left.”

To illustrate this pattern, Enos ran a clever field experiment on a Boston commuter rail line. He hired pairs of Latino actors to have conversations in Spanish while waiting on platforms in predominantly white suburbs. The mere presence of these actors had a significant causal effect on the political views of white commuters, according to Enos’s experiment: The treatment group was more likely than the control group to say that the number of immigrants from Mexico who are permitted to come to the United States should be decreased.

Crucially, though, “this exclusionary impulse started to soften over time,” Enos said. Some of the commuters were surveyed after three days of consistently seeing the Latino men on the platform, and others were surveyed after 10 days. While both groups had more exclusionary views than the control group, the treatment effect was smaller after 10 days.

Enos warned against over-extrapolating from a short-term study, but Enos’s other research and historical evidence confirm the pattern. “When African-Americans started to move into the Northeast and the Midwest in the mid-century, there was this really strong backlash — it turned to violence in some cases,” Enos said. “But over time, those attitudes seemed to soften somewhat.”

Enos also cited the anti-immigration Brexit referendum in the United Kingdom — where areas with large increases in immigrant populations voted to leave while areas like London that had a longer history of diversity voted to remain — as evidence of a temporary hump in exclusionary sentiment following migration of communities of color. Donald Trump, who made exclusionary immigration policy the signature issue of his campaign, may owe his Midwestern success in part to the recent demographic shifts in this region.

Of course, there are other factors that may lead white voters in diverse states to be dedicated to one party, many of which are historical reasons that vary from region to region. For instance, Enos pointed out that “in industrialized places like the Northeast and the West Coast,” left-leaning whites “have been attracted to these areas for the same reasons as nonwhite voters: working-class jobs and available housing.” Democratic strength among whites in these regions may owe more to a tradition of labor politics and the New Deal coalition than to racial integration.

Whatever the reason, though, the consequence is the same: Campaigns are more likely to find persuadable voters among the white population in predominantly white states.

As a result, very white states have disproportionate power in the Electoral College, and white voters have disproportionate sway in our presidential elections. This allows candidates to ignore issues voters of color care about, particularly when addressing those issues might alienate white voters. More consequentially, this allows presidents to make governing decisions — such as entrusting the Justice Department to a man who opposes the Voting Rights Act of 1965 — that might not be politically viable if voters of color had more electoral power.

What’s noteworthy here is not that whites hold a preponderance of the electoral power in the U.S. — which is, after all, still a majority-white country. It’s that this population advantage is reinforced by an assortment of institutional advantages like the Electoral College held over from a history of overt, legal white supremacy.


This feature was written for FiveThirtyEight in January, but was cut due to scheduling conflicts.

¹ FiveThirtyEight calculates a Voter Power Index for each state, measuring the relative likelihood that a voter in that state tips the election. A non-Hispanic white voter chosen at random has a higher expected VPI (0.95) than a Hispanic/Latino voter (0.70), a black voter (0.87), or an Asian voter (0.56) chosen at random.