Polarization and the Political Elite
For years, the scholarly debate over the polarization of American politics focused on a simple question without a simple answer: Is the American public polarized, or is polarization limited to political elites? Though some dissenters remain, the issue is largely settled: Political polarization is a mass phenomenon. Americans today are in an angry, uncompromising mood.
An accurate diagnosis is always helpful in solving a problem, but this one came at a cost: It diverted attention away from the problem of elite polarization. Even if the masses are polarized, our political elites — the millions of Americans who follow politics closely, think about politics regularly, and care about politics intensely — still deserve special attention. After all, they're the ones who attend political meetings, contact their political representatives, donate to political causes, and volunteer for political campaigns. And, perhaps most important, they vote in elections — including low-turnout elections like primaries.
Elites will always exert a disproportionate influence on any political system. In recent decades, however, we have made our system more susceptible to their pressure, at a moment when their numbers have grown and their attitudes have hardened. While politically engaged Americans are not the sole source of polarization, they do intensify it.
WHAT POLARIZATION IS AND ISN'T
In hindsight, the confusion about whether our ongoing bout of polarization is a mass or elite phenomenon was understandable. For years, scholars measured polarization by surveying Americans' policy preferences, and the evidence they gathered was mixed. When they asked the public whether the government should guarantee all Americans a decent standard of living, whether the United States should increase military spending, or how government should strike the right balance between services and spending, they found that people's answers remained fairly stable for decades.
But this approach didn't account for the fact that while Americans' positions on policy issues haven't changed much since the mid-20th century, their views have become more ideological over time. In the 1960s and '70s, many Americans held a mix of conservative and liberal opinions on different policy topics. Today, Americans are much more likely to be consistently liberal or consistently conservative in their policy preferences, even across social and economic issues. If you knew a person's position on, say, gun control in 1970, it would have told you very little about their position on government spending. Today, it tells you much more. The fact that policy positions are more ideologically aligned than they were 50 years ago means that even if aggregate measures of opinion on political issues haven't changed appreciably, political polarization has risen.
This increase in ideological thinking means that more and more Americans either agree with their fellow citizens on everything or nothing. As common sense would suggest (and social-psychology evidence confirms), repeated disagreement intensifies antipathy; we humans are much more likely to be understanding toward someone who disagrees with us on gun control but agrees with us on taxes than toward someone who disagrees with us on both counts.
The focus on policy positions also didn't account for the fact that members of the increasingly ideological public were sorting themselves into opposing parties after the 1960s, with liberals tending to move into the Democratic Party and conservatives into the Republican Party. A Democratic Party overwhelmingly composed of liberals is by definition more liberal than a party composed of both conservatives and liberals. The same goes for a Republican Party composed primarily of conservatives. So even as the public's policy preferences have remained roughly consistent over time, ideologically driven sorting has expanded the divide between the two major parties.
What's more, forensic analysis of the data on Americans' policy positions overlooked a key methodological fact: Polling organizations did not start asking Americans about their policy positions on specific issues with any regularity until the early 1970s. As political scientist James Campbell concludes in his 2016 book Polarized, the public was likely already polarized on these issues when the polling started. It stands to reason that polarization has only intensified since then.
Eventually the analysis of partisan politics moved beyond political science and into the realms of political psychology and sociology. And as scholars began to discover, the parties of the early 21st century were split not only by political opinion, but by race and religious observance as well. Starting in the early 1980s, non-white voters came to identify more closely with the Democratic Party: By 2012, they comprised almost half of Barack Obama's voters but just 11% of Mitt Romney's. Meanwhile, religiously observant white voters began to identify almost exclusively with the GOP.
To the surprise of many, Donald Trump bucked this trend in 2020 by gaining support among blacks and Hispanics while losing support among college-educated whites. It's too early to tell whether Trump's performance presages a trend, but even so, the racial and religious sorting of the preceding decades remains a plain fact. That alignment of racial, religious, and partisan identities is a recipe for political and social conflict.
One final ingredient would make our partisan divisions more poisonous still. Toward the turn of the 21st century, economic and technological changes were allowing Americans to sort themselves into both geographic and online communities of people with similar values and lifestyle choices. And like race and religion, these identities were increasingly aligning with partisan identity. Put more simply, scholars were discovering what most Americans at the time knew from their own experience: Democrats and Republicans were no longer living in the same neighborhoods, driving the same kinds of cars, working in the same industries, or even watching the same television shows. In 1970, an observer would struggle to predict a person's position on government services and spending, or even his party affiliation, by knowing his position on abortion. Today, we can predict both — not infallibly, but tolerably well — by knowing where he shops for groceries.
The problem, in other words, is not only that Democrats and Republicans disagree with one another on more political issues; it's that they have increasingly less in common with one another outside of politics. This trend holds the key to one of the more curious findings of the literature on polarization: People's hostility toward their political opponents has grown much faster than their policy disagreements. James Madison foretold this phenomenon as far back as 1787, when he wrote that the human tendency to divide into mutually hostile groups is so strong that "where no substantial occasion presents itself, the most frivolous and fanciful distinctions have been sufficient to kindle [people's] unfriendly passions and excite their most violent conflicts." Social psychologists have since found substantial evidence to confirm Madison's observation.
Indeed, human beings need almost no excuse to form groups. And once allied with a group, they consistently malign out-groups and venerate members of their in-group, even if there is no conflict of interest between the groups. They react to group threats with strong emotions like anger (toward the out-group) and enthusiasm (in defense of the in-group). And the more time they spend with their in-group, the more extreme their views grow. They also begin to ascribe more extreme views to the out-group than members of that group actually hold.
All of this is quite instinctive. In fact, psychologists have discovered that humans use the same neural pathway to process information about themselves as they do to process information about members of their in-group; they use a separate neural pathway to process information about members of out-groups. Polarization in America today operates on the cerebral plane of politics, but it is so deep and worrisome because it also operates on the more visceral plane of group identity.
ELITE POLARIZATION
If the tendency to polarize into mutually hostile groups is a distinctly human phenomenon, why do we need to pay special attention to elites?
The first reason is that the polarization trends we observe among the public are present in a concentrated form among our elites, making them more susceptible to political tribalism than the average American. Second, the proportion of elites vis-à-vis the general population has been rising. Data from the American National Election Studies (ANES) showed a spike in several measures of political engagement — including the number of Americans who donated to a campaign, displayed a political button or sticker, or tried to influence another person's vote — in 2004, and these measures have since settled at a new, higher level. Meanwhile, the percentage of Americans who reported that they were "very interested" in the presidential election reached 40% for the first time ever that same year, and reached its highest level yet — 49% — in both 2016 and 2020.
Like the public as a whole, politically active Americans are increasingly likely to identify as either liberal or conservative and to align themselves with the "correct" political party — liberals with the Democratic Party and conservatives with the GOP. In the ANES surveys of the early 1970s, barely a quarter of white Democrats called themselves "liberal" while less than half of white Republicans identified as "conservative." By 2012, the numbers were 51% and 76%, respectively. Gallup polls that include all races and extend to 2020 show similar results.
Americans today are also far more likely than Americans of the late 20th century to assert their ideological and political identities. In 2020, 40% of Americans placed themselves at one of the far ends of the ANES's liberal-conservative scale — firmly identifying themselves as either liberal or conservative — compared to 19% in 1972 and 29% in 2000. Similarly, 44% identified themselves as a "strong" Democrat or Republican in 2020 — the highest percentage recorded in 70 years.
Politically engaged Americans tend to be more invested in their identity as a liberal/Democrat or conservative/Republican than the average citizen. And since they often hold consistent ideological views on a range of issues, they are also more likely to disagree with their political opponents on almost everything. This disagreement, combined with strong party identification, intensifies partisan antipathy. According to a Pew Research Center study published in 2019, of the partisans surveyed who follow politics most closely, 91% of both Democrats and Republicans gave a "cold" rating to members of the opposing party.
As political scientist Lilliana Mason has shown, group identity, far more than political beliefs, drives political activism today. In fact, Americans with consistently liberal or consistently conservative views on multiple issues are also significantly more likely than their less ideological peers to be politically active — to contact elected officials, to volunteer for a campaign, to donate to a candidate or cause, to vote in elections, and the like. These individuals spend much of their time and energy thinking about, discussing, and participating in politics, which means they exert a disproportionate influence over our democratic institutions.
Of course, in any democracy, the most engaged citizens will always have more say in political outcomes than the less engaged. What makes this development especially troubling today, however, is that we have spent the last several decades making our electoral system more open and more permeable — often at the behest of elites themselves. Elites may have framed these reforms as efforts to make the system more democratic, but as noted above, elites by definition participate in politics more often than other citizens. Thus, by "democratizing" our political institutions, we've made those institutions more susceptible to elite influence — and we've done so at a moment when elites are more numerous, more ideologically polarized, and more antagonistic toward the other side than they have been since at least the dawn of public-opinion polling.
POLITICS FOR PARTISANS
The day after the March 1968 New Hampshire primary — in which Senator Eugene McCarthy shocked Lyndon Johnson and the Democratic Party establishment by winning 42% of the vote — a young Johnson aide named Ben Wattenberg sent a memo to the president in which he marveled at "the young people who worked for McCarthy." Drawing on field reports from political scientist Richard Scammon, he noted that McCarthy's supporters "ran a European-style campaign, canvassing door-to-door throughout the state" in a way that Johnson's supporters could never have matched even if they'd thought to try. Having made opposition to the Vietnam War and Johnson himself the centerpiece of his campaign, McCarthy had precisely the type of supporters who would turn out for a primary election or volunteer for a political campaign: passionate grassroots activists. Wattenberg also made an insightful observation, almost as an afterthought: McCarthy might perform just as well in all the remaining primaries as he had done in New Hampshire — or even better — and he would still fail to win the nomination.
Wattenberg's memo is a window into two key facts about presidential primaries back then. First, candidates needed the roused and the committed on their side to win a primary. Second, a candidate could not win a nomination through the primaries alone. The former remains true today; the latter was about to change.
In the tumultuous months that followed the New Hampshire primary, Johnson would decline to stand for re-election, Robert Kennedy would enter the race for the Democratic nomination and trade primary victories with McCarthy before being assassinated on June 5, and during a party convention marked by violent clashes between police and protestors outside the convention hall and tense fights over the war and civil rights within, the nomination would go to Johnson's vice president, Hubert Humphrey — a Democratic Party icon who hadn't entered a single primary race.
Reformers were outraged. Many had been fighting the Democratic Party's vague and informal nominating procedures for years, as Colgate political scientist Sam Rosenfeld documents in his 2017 book The Polarizers. They saw this as the moment to demand "more direct democracy" and call for "an end to 'boss control' of the nominating machinery."
Humphrey and his supporters were sympathetic to the thrust of the reformers' arguments — many of the party's procedures were in fact archaic, arbitrary, and plainly undemocratic. And in any case, Humphrey needed reformers' support if he was going to have a shot at winning the general election. So when a group of delegates pledged to implement major reforms to the party's nomination process in time for the 1972 election, he lent them his support. The most significant of these reforms was to make primary elections the critical means of selecting the Democratic Party's candidate. The GOP would soon follow the Democrats' lead.
In adopting this form of direct democracy for party nominations, the reformers successfully wrested control from the party bosses and made the process more open to rank-and-file members. What they had not done was make it more representative.
To see why, imagine two equally educated Americans. The first follows politics closely and feels passionately about a range of issues, is a strong supporter of one of the two major parties, has occasionally volunteered for campaign work and issue-based organizing, and considers the other party misguided, if not an outright danger to the nation. The second has a mix of agreements and disagreements with each party on various issues and has voted for members of both parties in the past, but hasn't volunteered for a campaign and doesn't view either party as uniquely threatening to the republic.
Which of these two individuals is more likely to vote in general? And which is more likely to vote in a low-information, low-turnout election? All the empirical evidence affirms what common sense tells us: Strong opinions motivate people to vote, and identifying with a party or political group motivates people to vote even more so. The first individual is thus more likely than the second to vote in both scenarios.
Even in high-turnout elections like presidential races, the electorate is not representative of the total adult American population. But as we move down the ladder of interest and turnout — from a presidential election to a presidential primary to a midterm Senate race to a midterm House primary race — the concentration of strong partisans in each electorate only increases. In other words, voters in low-information, low-turnout elections — of which primaries are a common type — are more partisan, more ideological, and more polarized than the public as a whole.
Their influence is greater still thanks to gerrymandering and to the geographical sorting of the two parties. In many House districts, the primary election — with its highly partisan and deeply polarized electorate — is the only election that matters. In 1975, there were 26 "landslide" House districts (those in which the presidential race was decided by more than 20 points). In 2020, there were 250. Senate "districts" cannot be gerrymandered, of course, but as the parties have sorted geographically, the number of non-competitive Senate races (those in which the primary election is the only one that really matters) has also soared. In 1979, 27 states had senators from two different parties, meaning it was possible for candidates of both parties to win a statewide race in more than half the states. Today, only six states have senators from each party — the lowest number since the direct election of senators began in 1914.
These changes have produced an incentive structure with enormous selection effects. It's not just that moderates have a hard time winning congressional races today; it's that they no longer run in the first place.
Primary elections are by far the most visible way we have turned power over to our most partisan, polarized citizens — but they are not the only way. Candidates need money to finance their campaigns, and in the middle of the 20th century, they received funds almost exclusively from the party itself. This is no longer the case today: Though rules vary from state to state, the overall trend has tilted heavily toward rules that curtail party fundraising, forcing candidates to raise money through donations from individual donors and organized interest groups. Perhaps unsurprisingly, the most passionate, most partisan, and most polarized voters are also the ones most likely to donate to political campaigns. And how do candidates get noticed by these potential donors? Ezra Klein sums it up well in his 2020 book Why We're Polarized: "[L]oud gets noticed. Extreme gets noticed. Confrontational gets noticed. Moderate, conciliatory, judicious — not so much."
Again, data back up the intuitive logic of Klein's point. Political scientists Raymond La Raja and Brian Schaffner analyzed the varying campaign-finance laws of each state to determine whether differing laws had an effect on political outcomes. The results were clear: Legislatures in states where parties had a larger role in fundraising were less polarized than those in states where candidates had to rely on donors to finance their campaigns. The pattern is the same at the national level.
By now it's a cliché to observe that partisan media and the internet allow Americans to isolate themselves into information bubbles that are impervious to any perspective that doesn't reinforce their prior opinions. Less remarked upon are two equally important facts, the first of which is that modern media and technology provide those interested in politics with an endless diet of political news and commentary. Even in 1990, these individuals would have had to rely on a single newspaper, one or two weekly magazines, and 23 minutes of nightly network news — only a portion of which would have focused on national politics — for their political fix. Today, social-media platforms, news websites, streaming services, and 24-hour cable-news networks offer political junkies a quick hit of dopamine or epinephrine just about anytime, anywhere.
Second, as Klein points out, the structure of today's information environment means that the news and entertainment industries compete with one another, enabling those uninterested in politics to ignore the subject entirely. This was much harder to do back when television networks aired both news and entertainment, and when the sports section in the daily paper came bundled with the latest political updates. In short, the politically engaged today can consume a constant diet of political information — much of it designed to intensify pre-existing beliefs or spark outrage at opponents — while those less interested in the subject can tune it out. This structure further widens the gap between the engaged elite and the less-engaged masses.
Third, well-intended open-meetings laws have made legislative negotiation more difficult than ever — especially when coupled with modern technology. In making almost all legislative meetings open to the public, they have come to be dominated by ideologues and special-interest groups. And thanks to electronic communication, these activists can instantaneously alert thousands of fellow travelers whenever a legislator strays from the party line or proposes an unwelcome compromise. They can then leverage the ensuing outcry to threaten to back a lawmaker's primary challenger if he fails to back down. Our elected representatives take these threats seriously and adjust their behavior accordingly — which is to say, they align their words and actions as closely as possible with the wishes of the highly agitated, hyper-partisan primary electorate in order to avoid losing re-election.
Primary elections, candidate-centered fundraising, open-meetings laws, and the enormous volume and choice of political information available in today's media landscape make our political system more open to the public, and therefore more democratic — at least in theory. Unfortunately, the people who tend to take advantage of this openness are the passionate and the partisan, which makes for more hostile politics and less effective governing.
RESTORING BALANCE
If polarization is accelerated by the exaggerated power we've given to our most polarized citizens, how can we empower the less partisan among us?
One way we might do this, as Jonathan Rauch has suggested in these pages, is to re-empower the parties as institutions. To imagine what this might look like and what the implications might be, we can begin by examining a hypothetical case in which the Democratic Party's nominees for president are chosen solely by Democratic members of the House and Senate.
Would this group be representative, demographically speaking? It would be just over 40% black, Hispanic, or Asian, and just under 40% female, making it a bit more racially diverse than the country as a whole, though quite a bit less gender diverse. We can tweak things by adding elected Democrats from each state legislature, or possibly the chairs and vice chairs of each state's Democratic Party committee. Both changes would give us more racial and gender diversity, while the latter would include people elected by Democrats' most intense and loyal partisans. Adding sitting governors and former presidents would make the group slightly whiter and more male, but their numbers are small and come with the benefit that these individuals have successfully assembled the broad coalitions necessary to win statewide or national elections.
The precise mix in this scenario can be adjusted, but the point is that this hypothetical assemblage of power brokers is a pretty close proxy for the Democratic National Committee's (DNC) "superdelegates," who are free to support the candidate of their choice at the Democratic convention. Since 2016, Bernie Sanders's supporters have assailed the party's superdelegates as unscrupulous backroom deal-makers who are beholden to shadowy powers bent on maintaining an intolerable status quo against the will of ordinary voters. In response, the DNC has reduced superdelegates' role in helping the party pick a candidate for president.
And yet, as we've determined above, this group could be quite representative of the party as a whole in terms of its demographics. Since a candidate who can't win can't govern, the group would also have an incentive to select a nominee who has a legitimate shot at the general election — which typically translates to a candidate who can unite a party's diverse constituencies while appealing to as broad a swath of the full electorate as possible. And to ensure that their preferred policies are enacted, the group would be motivated to nominate a person who has the skills and relationships necessary to push legislation through the lawmaking process once in office. Of course, members of this group would be fools to ignore their party's most intense partisans in selecting a nominee for all the reasons that parties give special attention to their strongest partisans. The difference, however, would be that they are not beholden only to those partisans.
By comparison, we can ask what today's primary voters seek in a candidate. In 2018, the Brookings Institution sought to answer this question through a survey that listed five traits of a congressional nominee and asked primary voters to rank them from most to least important. Both Democratic and Republican primary voters overwhelmingly chose someone who "shares my values" as their top criterion in selecting a nominee. In other words, they wanted their party to nominate someone who thinks and behaves as much like them as possible — someone who reinforces their identity and reflects it to the rest of the world. Where do they place "ability to win," which usually requires assembling a coalition of people who don't agree on everything but are committed to working together? Republicans placed it fourth; Democrats placed it dead last.
The hypothetical system outlined above would not be a purely democratic one, but it would offer the benefit of being transparent. And regardless of the combination of officials involved, every individual (with the exception of the former presidents) would be a sitting elected official, and thus answerable to the voting public for their choice of presidential nominee.
Such a change would not be a panacea. Reformers of the late 1960s had plenty of legitimate gripes against the parties' nominating procedures at the time, and today's reformers would be wise to keep these concerns in mind. On a more fundamental level, political professionals are human beings, not soothsayers; they are thus subject to the same foibles as the rest of us. Party officials in the past selected candidates like Alf Landon, who won the electoral votes of just two states; and James Cox, who suffered the worst popular-vote defeat in a century. And yet they also selected Abraham Lincoln and both of the Roosevelts while denying power to authoritarians and demagogues like Henry Ford and Huey Long — something officials were helpless to do when Trump entered the Republican primary in 2015.
PARTIES AS INSTITUTIONS
As the founders well knew, the goal of effective political processes and institutions is not to eradicate human fallibility; it is to accept human fallibility as a given while structuring incentives in such a way as to make positive outcomes more likely and negative ones less so. Today, all the incentives embedded in our nomination processes do the reverse: They amplify the power of the country's most polarized and partisan voters, who in turn nominate ideological, polarizing candidates. Returning some power to the institutional parties is one example of how we might restructure those incentives to better balance the power of partisan activists and less-ideological voters.
The denouement to the Democratic Party's 1968 nomination fiasco is suggestive in this regard. In 1970, Wattenberg and Scammon joined forces to write The Real Majority, a best-selling analysis of the 1968 election. In it, they noted that for all the rancor Humphrey's nomination provoked, his policy positions — including his views on Vietnam — reflected the wishes of the Democratic Party's rank-and-file majority and those of the median American voter about as well as those of anyone the party could have chosen at the time. The results of the election proved them right. Democrats faced prodigious headwinds that year, with the party's highly publicized internal fractures being just the beginning: The war was unpopular, inflation and crime were skyrocketing, protests and high-profile assassinations created a national zeitgeist of menace and disorder, and after eight years of unified Democratic control in Washington, the public was ready for change. Against this bleak backdrop, Humphrey lost the popular vote by less than a point.
Four years later, party activists wrested control of the nominating process from the professionals. They nominated the candidate of their choice in a process unsullied by backroom dealmaking. But political activists are unusually susceptible to a cognitive fallacy that afflicts all humans: the belief that our views are more popular than they actually are. A quick survey of one's own family members should be sufficient to dispel this notion; extrapolate the results of that survey to a nation of hundreds of millions, and its preposterousness should become plain. But confirmation bias is a powerful force. So, against the wishes of the party establishment, Democratic Party activists nominated George McGovern in 1972. He lost to Richard Nixon by 23 points — which remains the largest popular-vote margin loss in any presidential election since World War II.
The power of the politically engaged has grown because, in Rauch's formulation, the parties as institutions have declined, only to be replaced by the parties as brands. Brands are empty vessels for expressing identity, and identity is uncompromisable — making it a powerful source of human conflict. By contrast, institutions — from the family, to sports teams, to drama clubs, to civic organizations and the military — mold, discipline, and elevate members, providing them with rules, norms, and structures that allow them to showcase their individual talents while making the compromises and sacrifices necessary to work with others.
Today, the two major parties as institutions play almost no role in a political party's foundational task: vetting and selecting candidates for office. They play a declining role in financing campaigns of candidates who run under their banner. And they are increasingly helpless to protect members who seek compromise — even if those compromises are broadly supported — against the wishes of highly partisan party activists. In the name of reforming our parties to make them more democratic, more transparent, and more responsive to the will of the broader public, we have instead handed their reins over to the political elite.
Social scientists have known for decades that democracy's platonic ideal — a polity of the uniformly engaged and the well-informed — is neither a realistic nor a healthy goal. A functional democracy needs voters who are unsure, voters who can see different sides of an argument, and voters who experience what political scientists call "cross-pressures" — tension stemming from holding a mix of liberal and conservative opinions, or a conflict between a person's political beliefs and some other facet of their identity (think an evangelical Democrat or an atheist Republican). It even needs the apathetic. These individuals leaven our politics, preventing all-out war between passionate partisans and allowing legislators to foster the compromises that are essential to democratic governance. To be sure, they are fewer in number than they were 50 years ago. But even those who acknowledge that polarization is a mass phenomenon recognize that a more polarized public produces a more polarized elite, which in turn generates further polarization among the public.
Breaking this cycle will involve many different approaches. In pursuing them, we should keep in mind that polarization is a mass phenomenon, and thus not limited to the few. But we should also remember that political elites play an especially powerful role in our politics, for good and for ill, and that we have ceded even more power than usual to them in recent decades. If polarization is among our most pressing political problems, bringing the power of the most passionate and polarized Americans back into balance with the power of the less partisan is critical.
Returning some power to the parties as institutions would be a good first step. Doing so would not exclude the passionate from our politics — parties will always give these individuals special attention as a reward for their loyalty and labor. But a party whose elected members are involved in selecting, funding, and protecting candidates is a party that has not handed over decision-making power to its most committed ideologues. Instead, it is a party incentivized to work with the ideological and the cross-pressured alike, since assembling a large coalition is the surest way to translate policy goals into law. It is a party that strives for inclusivity rather than ideologically purity. And it is a party that forges the compromises necessary for governance in a pluralistic democracy.
It is a party, in other words, whose incentives are structured precisely as a healthy democracy should want them to be.