The shape of polarization in America

We're not as badly divided as we think, and it's all thanks to nonwhite voters

It is cliché by now to say that the country is polarized. Despite the predictable nature of the polarization discourse, there’s strong empirical evidence for the underlying trend. You may have seen this animation from the Pew Research Center, for example.

The tl;dr of it is that while rank-and-file Democrats and Republicans used to agree on a lot, they now hardly agree on anything. When we think of a time that politics was less ideologically polarized, we think of a period like midcentury America, when you had a lot of liberal Republicans and conservative Democrats. But the reality is that politics was a lot like this at the turn of the millennium. In 2000, liberals only barely outnumbered conservatives in the Democratic coalition — 27 to 24 percent. Today, the liberal-to-conservative ratio among Democrats is more than three-to-one. 

The polarization story is often told in terms of the twin extinctions of the conservative white Southern Democrats and the liberal Northeastern Republicans. This is mostly in a figurative sense, since most didn’t literally die off—they just switched parties. Both were opposite manifestations of the same trend: resorting themselves into the party with the people who thought mostly like they did on social issues. 

The polarization story is well-known and serves as an all-purpose explanation for everything that’s gone wrong in politics today. But it hasn’t touched every voter in the same way. A large number of voters are not polarized in the way highlighted by the Pew chart earlier, showing how far apart partisans now are in their policy beliefs. Only a specific subset of the population has become hyper-polarized in this way: white voters—and especially those with a college degree. 

This chart prepared by Victor Lue at Echelon Insights using data from the 2020 Cooperative Election Study paints a stark picture:

Different groups are polarized in different ways, and by this I mean that while white college graduates are extremely polarized in their views, Black and Hispanic voters are hardly polarized at all, holding largely moderate positions on policy that don’t shift dramatically as a result of their partisanship. Whites with college degrees are outliers in the levels of polarization, with partisans in this group tending to cluster at polarized extremes on policy. 

Here, we’re measuring policy views on a granular 0-100 scale based on how people answered more than 50 different policy questions on the 2020 CES. A score of 100 means that one gave the most conservative possible answer on all questions, and a score of zero would mean the same thing for the most liberal possible answer. The index itself was calibrated to reflect a rough 50-50 balance between right- and left-leaning voters, reflecting the reality of a closely divided politics. The original questions selected by the academic team behind the CES tended to yield more liberal-leaning responses. In spite of that, the CES is a very good survey for this kind of analysis, with 60,000 interviews that allow for robust subgroup analysis. 

Another way to boil this down is to bucket people into different ideological camps based on a 75-percent cutoff for ideological consistency. So, giving the conservative or liberal answer more than 75 percent of the time places you in each of those camps. Otherwise, you’re in a non-ideological middle ground. The 75 percent cutoff is an important one. Above we find Assad-like margins for Donald Trump or Joe Biden in 2020 of more than 98 percent. If you’re above this threshold, you’re not persuadable in the slightest. In the middle, your vote is basically up-for-grabs, progressing from one candidate to other in sliding scale fashion according to your policy views.

In each group but for one, solid majorities are in the non-ideological middle: 83 percent of Black voters, 77 percent of Hispanic voters, 69 percent of Asian-American voters, 58 percent of white non-college voters, and 56 percent of Native and other voters. And here again, one of these groups is not like the other: just 38 percent of white college graduates are in the middle, with large groups of extremely polarized liberals and conservatives. White college graduates also stand out in their representation of polarized liberals, with a concentration of them that’s nearly double that of any other group. And in the ranks of polarized liberals, there’s a notable absence of voters of color. Within a group of voters who agree with liberal positions more than 90 percent of the time, white voters with college degrees outnumber Black voters by 20-to-1, 60 to 3 percent.  

You might ask if these differences are merely a function of education, since I’m dividing whites by whether they have a college diploma or not, but not other groups. After all, highly educated voters have likely spent more time thinking arriving at a position on various issues, and are highly aware of where the parties stand. And white voters have more degree-holders in their ranks.

It turns out there’s good reason to single out white college graduates. Nonwhites are mostly similar in their views across education levels, with only a slightly more liberal peak among those with graduate degrees compared with those who only graduated high school. For whites, it’s a different story. These two groups are on different planets ideologically. The more highly educated a white person is, the more consistent and extreme their policy positions. Meanwhile, nonwhite voters at all education levels tend to cluster more in the middle. 

It’s not that whites with college degrees are the only group that’s polarized. White Republicans without college degrees are highly polarized on the right. White Democrats without degrees are pulled a bit rightward based on the group norm, so they appear as a relatively moderate group. Hispanic Republicans are showing signs of becoming more ideologically polarized. 

But the situation for whites with college degrees is unique, with twin ideological peaks on the left and the right. Citing the work of the Hidden Tribes project, David Brooks has called this the “Rich White Civil War.” The CES data bears this out: ideological polarization is mostly a function of being white and having set foot on a college campus. Whites in the Some College category are much more ideological on the right than those with a high school diploma only.

And this matters because this group is vastly overrepresented, basically everywhere. They’re solid majorities of the elites in both parties, of the media, of political Twitter. Though unscientific, when I’ve run polls of my Twitter following, which is roughly evenly balanced politically, more than 80 percent are whites with college degrees. This group is less than 30 percent of the American electorate. If everything seems polarized these days, it’s probably because of the circles you run in. Not everyone is like this. And the people that aren’t—the multiracial working class—are wildly underrepresented in political media.

This subject came up again in the discussion of a recent paper by Hakeem Jefferson about the “Black conservative paradox.” In surveys, around 30 percent of Black voters identify themselves as conservatives, but their rates of voting for the party of the right are obviously a lot lower than that. Jefferson argues that traditional conservative-liberal ideology scales aren’t a valid construct; many simply don’t see the connection to partisan politics. But an ideological scale built on the positions voters take on issues lets us see this through a different lens. Here, there are a microscopic number of Black ideological conservatives—2 percent. That’s far less than the 10 percent or so Republicans get in elections. But Black voters aren’t that ideological on the left either: just 15 percent qualify as consistently liberal. The vast majority are in the center. And among these less ideological Black voters, Joe Biden won 91 percent. 

Nonwhite voters in the middle are the future battleground of American politics. There are a dwindling number of white swing voters, but a large number of nonwhite voters with policy views in the middle or even leaning to the right, and these are groups where Republicans currently underperform. Among Black voters, they underperform for specific historical reasons. But Trump’s gains with nonwhites in 2020 showed that a day of more ideologically polarized nonwhite politics might be coming, with shifts of about 40 percent towards him among self-identified conservatives in every nonwhite group. 

For voters classified in the middle according to our issue index, the data does show an underlying logic to how people vote. For all races, the more conservative on issues you are, the more Republican you vote. But the curve is shifted dramatically for African Americans, with anywhere between an 80 to 100 percent net difference from whites in Presidential voting margins among voters classified in the middle. So, for example, a white voter who takes conservative positions 60 to 70 percent of the time was Trump +64 in the 2020 election. A similarly positioned Black voter was Biden +34, a gap of 98 points. 

The counter-argument is made that you can’t apply a single ideological spectrum to people of different racial groups. Different issues matter to different groups of voters, and racial issues where Black voters may be more left-wing could explain a big part of their partisanship. There is probably something to this, but I’m dubious that on its own it’s enough to explain a 98-point margin gap among otherwise slightly conservative voters. Much of the gap is explainable only by voter race itself and the social environment different groups of voters inhabit. Steadfast Democrats: How Social Forces Shape Black Political Behavior is a great book on this subject.

This graph from Xenocrypt in his thread reviewing the Jefferson paper underscores the point. Also using CES data, it compares the liberalism of Black respondents on each issue question to that of white respondents. And racial issues do pull certainly Black voters left — particularly on the question of whether whites are advantaged and it being harder for Black Americans because of slavery. Economically liberal positions, like a $15 minimum wage or Medicare for All, also push Black voters left. But there are some issues that push them right, specifically the lately salient one of abortion or aggressive action on climate and the environment, a point of division between Black voters and Democratic elites that largely goes unnoticed. 

I make the point in my upcoming book that the moderate politics of nonwhite voters are a sign of hope for the future, not necessarily just for the Republican Party’s electoral chances. If you dislike polarization, nonwhite voters provide a needed antidote. That’s true not only in the way they serve as a moderating force in the Democratic Party, knocking down progressive candidates time after time after time. Nonwhite and working class voters are also poised to have a similar effect on the GOP, moving them off a high-class polarization that did the party no favors in 2022.