How we Polarized

(2200 words; 10 min read)

When Becca and I left our home town in central Missouri 10 years ago, I made my way to a liberal university in the city, while she made her way to a conservative college in the country.

As I’ve said before, part of what’s fascinating about stories like ours is that we could predict that we’d become polarized as a result: that I’d become more liberal; she, more conservative.

But what, exactly, does that mean?  In what sense have Americans polarized—and in what sense is the predictability of this polarization new?

That’s a huge question. Here’s the short of it.

Polarization has always been with us. A set of basic psychological and sociological mechanisms explain why human societies have always been characterized by local conformity and global diversity: there tends to be agreement within small social circles, but disagreement between them. As a result, when people go off on different life trajectories, it’s always been normal for their attitudes to drift apart.

What’s changed is that a series of factors have come together to align these various mechanisms—and kick them into overdrive. As a result, now when people like me and Becca go off on different life trajectories, their opinions diverge in predictable and consistent directions, and do so faster and farther than before.

In sum: normally, the polarization process is a random walk; recently, it’s been transformed into a feedback loop.

Of course, there are many moving parts in a full story of modern polarization.  My goal is not to challenge the standard empirical accounts of it; instead, it’s to challenge their normative interpretations—to argue that our polarized politics arises from reasonable people who care about the truth but face ambiguous evidence. Given that, it’ll suffice to have a simple empirical story on the table; subtle or contested details will be reserved for the Technical Appendix.

As a representative example, focus on me and Becca. There are three questions we want to answer: In what sense have we “polarized”? What mechanisms lead to such polarization? And what has changed, making our current polarization different from past polarizations?


In what sense have we “polarized”?
There are three distinct “polarizations” that Becca and I—and the United States in general—have gone through in recent decades.

The first is ideological sorting: my views have become more consistently liberal and aligned with those of the Democratic Party; hers have become more consistently conservative and aligned with those of the Republican Party.

For example: in 2010 I was pro-choice and Becca was pro-life, but we were both un-opinionated about gun rights. (In high school, I had the experience—which at the time I thought was cool—of firing a friend’s AK-47 at a shooting range. Non-US readers: yes, you can buy those legally.) Yet in the decade since, my views have become more consistently left-leaning—I am now both pro-choice and anti-gun. Meanwhile, Becca’s views have become more consistently conservative—she’s now both pro-life and pro-gun.

The second is affective polarization: our views of the opposing party have become increasingly negative.

For example: in 2010, most of my friends were conservative, and as a result I had quite a bit of respect for the Republican Party.  Today, I have to wrack my brains to think of people I know who might’ve voted for Trump; and—I must admit—I’ve come to dislike the Republicans more, and understand where they’re coming from less.  A similar story, no doubt, governs Becca’s opinions toward Democrats.

The third is attitude polarization: our disagreements over political questions have become much sharper.

For example: if in 2010, you’d asked us both whether it’d be good for the country for a Republican to be elected president in 2020, Becca would’ve been mildly inclined to agree and I’d have been mildly inclined to disagree.  What if you ask us today? Then we’ll both admit that this election feels like a matter of life or death for our country.  Becca thinks, “If Biden wins, the police will be abolished and we’ll be cast into socialism!”  I think, “If Trump wins, the norms that uphold our democracy will be under direct assault!” 

(If your skeptical that American's political opinions have undergone attitude polarization--as some are—see the technical appendix.)

Upshot: Becca and I “polarized” in the senses that (1) our attitudes became more consistently opposed, (2) our feelings toward the other side became more negative, and (3) our disagreements became increasingly sharp.

This process is familiar. The United States has become increasingly ideologically sorted by politics—for example, the proportion of people with consistently liberal or  consistently conservative positions more than doubled between 1994 and 2014. Negative feelings toward the other side have skyrocketed—for example, those with a “very unfavorable” opinion of the opposing party nearly tripled between 1994 and 2016. And disagreements on concrete political questions has become increasingly sharp––for example, the Republican/Democrat presidential approval ratings averaged 75% / 34% in the time of Nixon, 81% / 23% in the time of (W.) Bush, and currently sit at 92% / 4% in the time of Trump.

So Americans have always been polarized, but that polarization has kicked into overdrive in recent decades.  Given that, we need to know two things: Why, in general, do societies polarize? And what has changed to make this more severe in recent decades?


Why do societies polarize?
Psychologists and sociologists have long known of a set of mechanisms that drive people to have stronger (and often more conflicting) attitudes over time—especially when they are put in different social and informational environments.

First mechanism: most obviously, and most simply, people are persuaded by arguments. That means that if two people enter different environments in which they’ll tend to encounter arguments for different positions, their opinions will predictable diverge.

For example, since I was headed to a liberal university, I could expect to hear arguments in favor of progressive taxation and the existence of oppressive sexism; and since Becca was headed to a conservative college, she could expect to hear arguments in favor of the value of capitalism and the importance of being a good woman.  As a result of these opposing arguments, it’s in some sense obvious that we could predict that we’d come to disagree. (But that apparent obviousness is also a bit misleading—as we’ll see, the rationality of this predictable disagreement hinges on the subtleties of ambiguous evidence.)

Second mechanism: when groups of like-minded individuals share and discuss their opinions, they tend to become both more homogenous and more extreme in those opinions.  This is known as the group polarization effect, and is one of the most robust findings in social psychology.

For example, when I started taking about gun rights in groups of mostly-liberal university students (the majority of whom had never held a gun), we all grew more confident that interpretations of the second amendment have been bastardized, leading to the rise in mass shootings.  Meanwhile, when Becca did the same in groups of mostly-conservative friends (many of whom owned guns), she grew more confident that the right to bear arms was central to American identity, and that since we’ve always had guns, the rise in mass shootings has other causes.

These first two mechanisms—persuasion and group polarization—are what gets polarization started: simply by putting me and Becca in different (liberal vs. conservative) social environments, they started to pull our opinions apart.

Now jump to Spring 2011, when our opinions had already been pulled slightly further apart by our new environments. Once this happened, a new set of mechanisms kicked in: our opposing beliefs started to ratchet up on their own.

Much of this phenomenon goes under the rather disunited label of ``confirmation bias”: people’s tendency to gather and interpret evidence in a way that confirms their prior or favored beliefs. This tendency can be helpfully divided into three different mechanisms.

Third mechanism: selective exposure—when given a choice, people tend to prefer to see new information that they expect to confirm their prior beliefs, rather than information they expect to disconfirm them.

For example, in Spring 2011 I began to consistently check the New York Times and the Washington Post to get my news. Becca, meanwhile, became more consistent in checking Fox News and the National Review.

Fourth mechanism: biased assimilation of evidence—when confronted with conflicting or messy evidence, people tend to interpret it in a way that favors their prior beliefs.

For example, in 2012 when Republicans stonewalled Obama’s proposals to use government funds to boost the economy, I took this to show that Republicans tend to put party over country—but Becca took it to show that Democrats tend to resort to government over-reach.

Final mechanism: motivated reasoning, a.k.a. identity-protective cognition—people tend to gather and make use of evidence in a way that confirms the things that they want to believe, especially when the belief is tied to their identity.

For example, when Tara Reade accused Biden of sexual assault, I must admit that I was inclined to be a bit skeptical—to spend a bit longer looking at articles that questioned her credibility than at those that supported it. Yet when Blaise Ford made an accusation against Brett Kavanaugh, I made no such skeptical effort.  Becca, no doubt, reacted in exactly the opposite way.

These mechanisms should sound familiar—we all know that people tend to be persuaded by arguments, that their motivations tend to affect the way they reason, and so on.

This is as it should be. Polarization is a familiar fact of life—it’s always been with us. Thus any adequate explanations of it will be built upon other familiar facts of life; an explanation built solely upon new or surprising findings would be missing the bigger picture.

Nevertheless, there is something new in our polarized politics—and there is something surprising lying behind these familiar mechanisms.

What’s new is that these mechanisms have become collectively aligned and individually kicked into overdrive.

And what’s surprising is that all of them—including howlers like biased assimilation and motivated reasoning—are to be expected from rational people who care about the truth but face systematically ambiguous evidence.

Explaining this latter point will take weeks; today, let’s close with the former.


What has changed?
What has led to recent the increases in American polarization? The story, in outline, is that a variety of societal changes have led to increased social and informational sorting.

Both the southern realignment and the civil rights movement started the process of making Democrats the party of consistent progressives and the Republicans the party of consistent conservatives. In turn, this increase in ideological consistency may have combined with the fading influence of religion to make political party the new key to many people’s identity.

Meanwhile, an increasing urban-rural divide has made it so that the political views of one’s (future) friends has become more predictable than ever. Combined with a precipitous fall in civic engagement, this has led to a decrease in cross-party social pollination and fewer friendships across party lines.

At the same time, an increasingly fragmented and political media landscape, along with the rise of web personalization has allowed people greater freedom in choosing their sources of information and opinion.

In short, decades ago, our social circles were ideologically diverse, our news sources constantly confronted us with differing opinions, and our political beliefs were not terribly consistent or central to our identities. Today, all that has changed. As a result, the effects of persuasion, group polarization, selective exposure, biased assimilation, and motivated reasoning all point in increasingly the same direction––pulling Democrats to the left, and Republicans to the right.

In my case: the more time I spent talking with Democrats, the more persuaded I was of their arguments and the more opinionated we all became. The more informational choices I had, the more inclined I was to listen to and trust liberal sources. The more confident I became, the more inclined I was to interpret ambiguous stories about Republican ideas and politicians in negative ways. The more I came to identify with being a Democrat, the more motivated I was to see new evidence as confirming my Democratic beliefs.  And the stronger each of these processes became, the more it reinforced the others.

Similarly for Becca––and for Americans everywhere.

Our question: what does this story—of old polarizing mechanisms, recently kicked into overdrive—mean for the rationality of politics?

It means that we if understand the old mechanisms, we’ll understand both why we’ve always been polarized, and the route through which polarization has increased. And it means that if we come to see these old mechanisms as rational, then we’ll be able to see our polarized politics—including our political opponents—as the result of individuals doing the best they can with the information they have.

That’s where we’re heading. We’re going to take an in-depth look at the mechanisms of persuasion, group-polarization, selective exposure, biased assimilation, and motivated reasoning. We’ll see how each of these are driven by rational attention to ambiguous evidence—and that as societal changes have made our political evidence systematically more ambiguous, they have all been kicked into overdrive.

But before we can dive into those details, we need to establish some background. What do I mean when I say that these mechanisms are to be expected from “rational people who care about the truth” and yet face “ambiguous” evidence?  And how could it be that rational people—like me and Becca—can be predictably, persistently, and profoundly polarized?



What next?
If you have comments or suggestions about this empirical story, please send them!
​Next post: a philosophical crash course on what it means to be “rational”, what exactly “ambiguous evidence” is, and how it is needed to explain predictable polarization.