What drives polarization (by Timothy Ryan)

Guest post by Timothy Ryan, graduate student in political science at U. Michigan

Social scientists delight in few things more than when we can puncture a bit of prevailing wisdom, and the idea of a new political schism among American citizens presents a very tempting target. As political candidates sling invective, CNN frets about broken government, and Charles Murray bemoans a country “coming apart at the seams,” the idea that a sea change has occurred in citizens’ orientation toward politics seems ripe for scrutiny. Politics is inherently confrontational. Are things really different than in the past? How much of this talk of a grand cultural divide in America is just the parochialism of the present?
Most of it, according to political scientist Morris Fiorina. In a widely acclaimed book and numerous scholarly articles (such as this one), he and his collaborators examine several decades of survey data, looking for evidence that attitudes have become more polarized over time. They turn up a parade of flat trend lines that suggest the ideological character of the American public looks quite moderate and, well, more or less like it did forty years ago. The proportion of the electorate that calls itself “middle of the road” in terms of ideology? It’s down just one percentage point from 1972 to 2004. The proportions that identify as liberal or conservative? They’ve hardly changed. A seismic shift in the distribution of attitudes about spend/save issues? Not really. Social issues like abortion? Nope.
The lack of a “smoking gun” with respect to mass polarization has directed scholarly attention elsewhere. It is one reason researchers have hunted for institutional changes that might explain what seems to be more division and confrontation among our leaders. The idea here is that, if citizens haven’t changed in any pivotal ways, perhaps the highly cantankerous political climate stems from changes in how the “game” of politics is structured. In this vein, Sean Theriault directs our attention to the gerrymandering of congressional districts and how Congress uses procedural rules differently now than in the past. Alan Abramowitz and Markus Prior have written about declining voter turnout rates among moderates.
A few weeks ago, though, Marc Hetherington interjected a helpful reminder that there may be something to the idea that citizens have changed after all – if we think about it in a different way. He calls our attention not to ideological positioning or issue stances, but changes in how partisans feel about each other. Both Democrats and Republicans express much colder sentiments toward the other party than in the past, a trend called affective polarization. (“Affect” here taking its jargony psychological meaning: a rough synonym for feelings or emotions.)
The trend Hetherington notes gets even starker if we break it down by strength of partisanship, as I do below. It holds for strong partisans and weak ones alike. Journalists and scholars often think about polarization as a phenomenon where individuals become more extreme in their partisanship or ideology, yet the graphs below show how the country can become more polarized even if everyone’s political identity and views on the issues stay the same: we could hate each other more. Democrats or “extreme” Democrats, for instance, need not be more numerous for there to be more hostility toward Republicans. Partisans of the same stripe can express more animosity without being greater in number. And apparently they do.

The trend also speaks to one concern raised by polarization skeptics. Where wider schisms in attitudes toward individual people (e.g. approval of Presidents Bush and Obama) have been referenced as evidence of polarization, Fiorina and colleagues rejoin that what looks like polarization could really come from more extreme choices, rather than shifts in voters’ underlying character. (As Fiorina and Abrams write, “If President Bush had never invaded Iraq, we daresay that his approval ratings would look different [better] today even if Americans’ attitudes on the issues had not changed in the slightest.”) But because the affective polarization trend focuses on attitudes about groups, rather than an individual, and because it highlights a trend that has unfolded at a more or less steady pace for thirty years, it seems to capture something larger than the behavior of any particular person.
Another possibility that Fiorina (and, in a different way, Matthew Levendusky) note remains. They draw an important distinction between polarization – an increasing schism in attitudes – and “sorting.” In the past, the argument goes, there were a fair number of liberal Republicans and conservative Democrats. Nowadays, party labels and policy positions are much more in sync. Thus, Republicans and Democrats have not become more distant. It is just that these labels better coincide with ideological cleavages that were always there.
I don’t think sorting explains affective polarization – at least not fully. For attitudes toward ideologies to do the heavy lifting – for Democrats’ dislike of Republicans to increase because the word “Republican” is a better proxy for “conservative” than in the past – we would expect Democrats’ feelings toward conservatives to be even more negative than their feelings toward Republicans. (Similarly, we would expect Republicans’ feelings toward liberals to be more negative than their feelings toward Democrats.)
But they aren’t. The figure below shows how partisans evaluate the more distant ideology (that is, how Democrats evaluate conservatives and how Republicans evaluate liberals) over time. What is noteworthy in comparing this figure to the one above is that, in recent history, the ideologies tend to “run ahead” of the parties associated with them. For instance, strong Democrats exhibit disdain for Republicans, rating them only about 25 degrees in 2008. But they were much warmer toward conservatives, placing them above the 50-degree midpoint. Similarly, strong Republicans place Democrats at the 30-degree mark, but liberals higher, at nearly 40 degrees. The evidence here suggests that Democrats hate Republicans despite feeling somewhat warmer toward their ideology, not because the Republican label predicts conservatism better now than in the past – a pattern that seems to cast partisanship as the driving force. Moreover, the trend lines are relatively flat. If affective polarization emerged because (as “sorting” hypotheses might suggest), the people who really hate conservatives are gravitating toward the “Strong Democrat” label, we would expect, for instance, strong Democrats to exhibit less warmth toward conservatives over time. But they don’t.

The “sorting” idea suggests that partisans like each other less because they are more ideologically distant. Is it possible that the relationship is actually reversed – that citizens have become more ideologically distant because they like the other team less? Counterintuitive though that might seem, there is a lot of evidence that partisan ties can guide thinking about policy in this way. (For one example of many, see Geoffrey L. Cohen’s article in the Journal of Personality and Social Psychology, titled “Party Over Policy: The Dominating Impact of Group Influence on Political Beliefs.”) In this perspective, increased affinity for one’s own party or dislike for the opposition could cause ideological homogeneity, rather than the other way around.
The idea is tentative, but leads to some interesting questions: What, if not policy opinions, undergirds affinity for one’s own party and dislike of the opposition? Have there been systemic changes that might have widened the gulf in recent years? And perhaps most constructive, how might things change? Are there ways to talk and think about politics that might lead citizens to see the opposition in a friendlier light?
Timothy J. Ryan is a Ph.D. Candidate in political science at the University of Michigan. His dissertation investigates how public opinion influences the likelihood of political compromise.