In 1960, Americans were asked whether they would be pleased, displeased, or unmoved if their son or daughter married a member of the other political party.
Respondents reacted with a shrug. Only 5 percent of Republicans, and only 4 percent of Democrats, said they would be upset by the cross-party union. On the list of things you might care about in child's partner — are they kind, smart, successful, supportive? — which political party they voted for just didn't rate.
Fast forward to 2008. The polling firm YouGov asked Democrats and Republicans the same question — and got very different results. This time, 27 percent of Republicans, and 20 percent of Democrats, said they would be upset if their son or daughter married a member of the opposite party. In 2010, YouGov asked the question again; this time, 49 percent of Republicans, and 33 percent of Democrats, professed concern at interparty marriage.
For Shanto Iyengar, director of Stanford's political communications lab, the marriage polls were yet more evidence that something important was changing in American politics.
The big institutions and broad outlines of our political system have been so stable for so long that it makes it hard for people to see when the tectonic plates of American politics are actually shifting. There's been a Democratic Party and a Republican Party for most of the country's history, and they've always bickered, so it's easy to assume — particularly in a country with a short historical memory — that the partisanship we see now is simply how it's always been.
But Iyengar was coming to believe that today's political differences were fundamentally different from yesterday's political differences; the nature of American political partisanship, he worried, was mutating into something more fundamental, and more irreconcilable, than what it had been in the past.
Political scientists have mainly studied polarization as an ideological phenomenon — in this view, party polarization is really another term for political disagreement, and more polarization simply meant more severe disagreements. But it's hard to find the evidence that the disagreements among ordinary Americans have really become so much more intense.
"If you look at Americans' positions on the issues, they are much closer to the center than their elected representatives," Iyengar says. "The people who end up getting elected are super extreme, but the voters are not."
But even as American voters remained relatively centrist, they seemed to be getting angrier and more fearful of the other side.
After every election, researchers ask voters an almost endless series of questions, creating a rich record of why Americans vote the way they do. The result is called the American National Election Survey, and beginning in the 1980s it began to show something puzzling.
What caught Iyengar's eye was a section of the survey known as "the thermometer." The thermometer asks people to rate their feelings toward the two political parties on a scale of 1 to 100, where 1 is cold and negative and 100 is warm and positive. Iyengar noticed that since the 1980s, Republicans' feelings towards the Democratic Party, and Democrats' feelings towards the Republican Party, had dropped off a cliff.
In 1980, voters gave the opposite party a 45 on the thermometer — not as high as the 72 they gave their own party, but a pretty decent number all the same.
After 1980, though, the numbers began dropping. By 1992, the opposing party was down to 40; by 1998, it had fallen to 38; in 2012, it was down to 30. Meanwhile, partisans' views towards their own parties had remained pretty much unchanged — that 72 from 1980 had only fallen to 70 by 2012.
Iyengar's hypothesis was that rising political polarization was showing something more fundamental than political disagreement — it was tracking the transformation of party affiliation into a form of personal identity that reached into almost every aspect of our lives.
If he was right, then party affiliation wasn't simply an expression of our disagreements; it was also becoming the cause of them. If Democrats thought of other Democrats as their tribe and of Republicans as a hostile tribe, and vice versa, then the consequences would stretch far beyond politics — into things like, say, marriage.
And the data was everywhere. Polls looking at the difference between how Republicans viewed Democrats and how Democrats viewed Republicans now showed that partisans were less accepting of each other than white people were of black people or than black people were of white people.
But there was no way partisanship — an identity we choose, and that didn't matter much to us 50 years ago — could possibly have become a cleavage in American life as deep as race, right? That seemed crazy.
So Iyengar decided to test it.
The experiment was simple. Working with Dartmouth College political scientist Sean Westwood, Iyengar asked about 1,000 people to decide between the résumés of two high school seniors who were competing for a scholarship.
The resumes could differ in three ways: First, the senior could have either a 3.5 or 4.0 GPA; second, the senior could have been the president of the Young Democrats or Young Republicans club; third, the senior could have a stereotypically African-American name and have been president of the African-American Student Association or could have a stereotypically European-American name.
The point of the project was to see how political and cues affected a nonpolitical task — and to compare the effect with race. The results were startling.
When the résumé included a political identity cue, about 80 percent of Democrats and Republicans awarded the scholarship to their co-partisan. This held true whether or not the co-partisan had the highest GPA — when the Republican student was more qualified, Democrats only chose him 30 percent of the time, and when the Democrat was more qualified, Republicans only chose him 15 percent of the time.
Think about that for a moment: When awarding a college scholarship— a task that should be completely nonpolitical — Republicans and Democrats cared more about the political party of the student than the student's GPA. As Iyengar and Westwood wrote, "Partisanship simply trumped academic excellence."
It also trumped race. When the candidates were equally qualified, about 78 percent of African Americans chose the candidate of the same race, and 42 percent of European Americans did the same. When the candidate of the other race had a higher GPA, 45 percent of African Americans chose him, and 71 percent of European Americans chose him.
But Iyengar and Westwood wondered whether these results would really hold outside the laboratory setting. After all, the study's participants knew their answers were being judged by the researchers. Perhaps discriminating against members of the other party was socially acceptable in a way discriminating against people of the other race simply wasn't. In other words, perhaps people are willing to show their partisan bias whereas they hide their racial bias, and that was what was behind the results.
So Iyengar and Westwood came up with another test — a test that would be much harder to fool.
Taking an implicit-association test is a humbling experience. Your job, as the test taker, is to hit a letter on your keyboard when certain word and images flash together. The instructions are to go as fast as you can, but the fastest you can go is sluggish compared with the pace of the program. You get nervous, your finger stumbling over the keys, hitting the wrong ones. And the whole time you're doing it, you know you're being judged, and you realize, with a sickening certainty, that the verdict isn't going to be good.
The point of IATs is to measure the snap judgments your brain makes at speeds faster than conscious thought. Mountains of psychological research shows that these judgments are powerful — that much of what we consciously think is an after-the-fact rationalization for the instant judgment we made before we had time to think. IATs are meant to expose those judgments.
The test is grounded in studies of racism: Researchers will ask subjects to pair positive words with black and white faces, and see which they have more trouble doing. The underlying insight is that the task is easier to complete when it aligns with people's automatic, unconscious reaction than when it isn't — you're faster when you can go with instinct then when you have to suppress it. Study after study shows that IATs are at least somewhat predictive of real-world racial bias. They have since been extended to measure bias in gender, age, weight, and more.
Iyengar and Westwood's idea was simple: Why not use an IAT to measure partisan bias, too?
So they built one. And the results were fascinating. But before we run through them, Iyengar and Westwood kindly shared their code with us, and so you can take the test here, or at the top of this article, and see how strong your bias is.
The results showed, as you might expect, that Democrats exhibit an automatic bias against Republicans, and vice versa. What was surprising was that the bias partisans exhibited for their out-group exceeded the bias white participants showed for black people, or that black participants showed for white people. According to the test, Americans are more automatically partisan than they are automatically racist. (If you want to know your own results on the racism test, you can take a version meant to test racial bias here.)
This was, to Westwood, a bit of a shock. "To be honest, I didn't expect this to work at all," he says. "The common story is most Americans don’t care about politics, they don’t understand politics, they don’t understand policy. So you wouldn’t expect Americans to have strong preferences. That’s where I started."
Together, the two experiments suggest that partisanship now extends beyond politics — it's becoming a fundamental identity in American life, and may well lead to discrimination in completely apolitical contexts.
I asked two other political scientists — John Sides and Danny Hayes, both of George Washington University — whether they bought Iyengar and Westwood's data. Both said they did, though they noted that opportunities for partisan discrimination are less common than opportunities for racial or gender discrimination, if for no other reason than partisanship is less visible.
Still, the impulse to discriminate against the out-party is real. "The more partisanship becomes a social identity — and I think this is as true today as it's been in modern American politics — the more we should expect people to engage in in-group favoritism and out-group discrimination," Hayes said.
Iyengar's hypothesis is that partisan animosity is one of the few forms of discrimination that contemporary American society not only permits but actively encourages.
"Political identity is fair game for hatred," he says. "Racial identity is not. Gender identity is not. You cannot express negative sentiments about social groups in this day and age. But political identities are not protected by these constraints. A Republican is someone who chooses to be Republican, so I can say whatever I want about them."
You can see an example when you look at the media, Westwood observes. There are no major cable channels devoted to making people of other races look bad. But there are cable channels that seem devoted to making members of the other party look bad. "The media has become tribal leaders," he says. "They’re telling the tribe how to identify and behave, and we’re following along."
Iyengar and Westwood's research is a fundamental challenge to the way we like to believe American politics works. A world where we won't give an out-party high schooler with a better GPA a nonpolitical scholarship is not a world in which we're going to listen to politicians of the other side on emotional, controversial issues — even if they're making good arguments that are backed by the facts.
Iyengar's initial insight was that political polarization might be less about policy than it is about identity, and his research more than proves it. "The old theory was political parties came into existence to represent deep social cleavages," he says. "But now party politics has taken on a life of its own — now it is the cleavage."
That changes the playbook for cynical presidential candidates, policymakers, pundits, and so on.
"The major take-home here is it’s relatively easy to score points by attacking the opposition and touting the goodness of one’s own party," says Westwood. "If you’re trying to get the largest return from voters, it would make sense for politicians to try to activate social identity rather than focus on policy."
Winning an argument, at least when you're talking to co-partisans, is less about persuasion than about delegitimization — the savvy move isn't to try to build a better case than the other side, but to make clear that the other side is the other side.
Westwood is quick to note that the comparison to racism doesn't mean that partisanship is somehow worse than racism, more pervasive, or more damaging. It's easier to see — and thus discriminate — against people based on their skin color than their partisanship, for instance. And as Jenée Desmond-Harris has written, political beliefs are a choice with moral implications while race is not. Judging someone on whether they support gay marriage, universal healthcare, or gun laws is far different than judging someone on the color of their skin.
What Iyengar and Westwood's research shows, however, is that partisanship is no longer just a political phenomenon. Party and ideology have become powerful forms of personal identity, and the way they inform our lives — who we listen to, who we help, even who we love — now stretches far beyond the political realm.