Chapter 25 — Political Polarization: How America Sorted Itself and Whether It Can Unsort

"What we call polarization is not one thing. It is at least three things, possibly five, depending on how you slice the data — and the relationships among them are exactly as messy as the relationships among the political identities, geographic communities, media diets, and primary electorates that produced them." — Lilliana Mason, paraphrased from Uncivil Agreement (2018)

25.1 Why this chapter is hard to write

Most chapters in this book describe an institution: Congress, the courts, the bureaucracy, the parties, the media. You can name the actors, describe the rules, and trace the history. The institution sits there to be examined.

Polarization is not an institution. It is a property of the system — a description of how the actors are arranged, how they relate, how they feel about each other. It shows up in roll-call votes and Thanksgiving dinners. It is measured by political scientists with formulas and felt by ordinary citizens as the unhappy moment when an old friend posts something on Facebook that makes them stop being friends.

That makes polarization both extremely real and exceptionally easy to mis-describe. Each of us experiences it from inside our own political position. The other side looks polarized; our side looks like normal politics. Each side has a story about why the polarization is the other side's fault, and each story has some grain of truth and some grain of self-flattery. The political-science literature is robust, contested, and not always intuitive. Reform proposals abound, and none of them is obviously right.

This chapter does what it can. It will describe the phenomenon as carefully as possible, present the major causal accounts honestly, and steel-man both partisan diagnoses without endorsing either. It will treat the empirical claim of asymmetric polarization — that congressional Republicans have moved further right than congressional Democrats have moved left over the past forty years — as a claim with substantial empirical support, and will also present the most serious responses to that claim.

Read this chapter looking for the gap between what we can measure and what we cannot. Most of what fuels polarization is feeling — and feeling is not nothing, but it does not always fit the categories political scientists have built to describe it.

25.2 What polarization is — and isn't

The first move in studying polarization is to break the word into pieces. Loose talk treats it as a single thing. The literature distinguishes at least three things that travel together but are not the same.

25.2.1 Ideological polarization

Ideological polarization is the distance between the parties' policy positions. If you imagine a left-right line and place each party's median legislator on it, the distance between the two medians is one measure of ideological polarization. As that distance grows, the parties become more programmatically distinct.

The standard way to measure this in Congress is the DW-NOMINATE score, developed by Keith Poole and Howard Rosenthal in the 1980s and updated continuously since. DW-NOMINATE uses every roll-call vote in congressional history to place every member on a scale from roughly -1 (most liberal) to +1 (most conservative) on a primary economic-and-redistribution dimension. It is not a survey of what members say they believe; it is a structural model of how they actually vote.

What DW-NOMINATE shows is striking. From the New Deal era through the early 1970s, the parties overlapped extensively. There were liberal Republicans (Jacob Javits of New York, John Lindsay of New York) and conservative Democrats (the Southern bloc). The median Democrat and the median Republican were not far apart, and many votes crossed party lines.

Beginning in the late 1970s and accelerating through the 1990s and 2000s, the two parties pulled apart. By the 2020s, the most conservative Democrat in the House was to the left of the most liberal Republican — there was no overlap between the parties' ideological ranges. By 2024, the partisan distance on DW-NOMINATE was at the highest level recorded since Reconstruction.

In the electorate at large, ideological polarization has grown more modestly. The percentage of voters who hold consistently liberal or consistently conservative views (across multiple issues) has roughly doubled since the early 1990s, but most voters still hold mixed views — liberal on some issues, conservative on others. The shape of the change is real but smaller than what we see in Congress.

25.2.2 Affective polarization

Affective polarization is something different. It refers to how much partisans dislike, distrust, and feel emotionally distant from members of the other party. It is measured most cleanly by the feeling thermometer — a survey item used by the American National Election Studies (ANES) since 1964, in which respondents rate their feelings toward various groups on a scale from 0 (cold/negative) to 100 (warm/positive).

The ANES findings are dramatic. In 1980, the average partisan rated their own party at about 71 and the opposing party at about 47. By 2024, partisans still rated their own party around 71, but their rating of the opposing party had fallen to roughly 18 — a decline of nearly 30 points. The cross-party gap had grown from 24 points to 53 points.

Shanto Iyengar and colleagues, in a series of influential papers beginning around 2012, showed that affective polarization had grown to levels at which Americans now disliked the other party more than they disliked any other social group, including racial out-groups, religious out-groups, and immigrants. They argued that party affiliation had become the most powerful "social identity" in American life — more powerful than race, religion, or class — at least in terms of how strongly Americans express in-group favoritism and out-group hostility.

Crucially: affective polarization can grow even when ideological polarization is stable. The two are not the same. People can dislike each other without disagreeing about substantive policy. And the empirical evidence suggests that affective polarization in America has grown faster than ideological polarization — meaning Americans hate each other more than the substance of their disagreement justifies.

25.2.3 Sorting

Sorting is the alignment of party with other identities. In 1960, both parties contained roughly equal proportions of Catholic and Protestant voters, urban and rural voters, college-educated and non-college voters. Republicans had a strong following among Black voters in some Northern cities; Democrats included a substantial bloc of white evangelical Christians.

Today, those alignments have shifted. White evangelical Protestants are reliably Republican. Black voters are reliably Democratic. College-educated white voters lean Democratic; non-college white voters lean Republican. Urban voters lean Democratic; rural voters lean Republican. The parties have become more internally homogeneous and more externally distinct along multiple dimensions simultaneously.

Sorting is not the same as polarization. Sorting could in principle happen without anyone changing their views — voters with consistent worldviews simply moved to the party that matched. But sorting reinforces polarization, because when multiple identities align with party, partisanship becomes a "mega-identity" (Mason's term) that bundles together race, religion, geography, education, and lifestyle. Disagreeing politically becomes disagreeing about everything.

25.2.4 The three together

Imagine a three-dimensional object: one dimension is ideological distance between parties (positions), one is emotional distance between partisans (feelings), and one is identity overlap with party membership (sorting). All three have grown in the United States since 1980. They are correlated but not identical. Different reform proposals target different dimensions, and a reform that reduces one might leave the others untouched.

The single biggest mistake in popular writing on polarization is collapsing these three dimensions. When a commentator says "America is more polarized than ever," ask: which kind?

25.3 The empirical picture

Let us walk through the data.

25.3.1 Congressional polarization

Voteview.com, the public archive of DW-NOMINATE scores, makes the following picture available to anyone with a browser. The two charts that matter most are:

• Party medians over time. The median House Democrat and median House Republican were within about 0.25 of each other on the NOMINATE scale in the 1960s. By 2024, the gap was about 0.95 — nearly four times wider.
• Distribution overlap. In the 1960s House, the most conservative quarter of Democrats overlapped with the most liberal quarter of Republicans. By the late 1990s, this overlap had vanished. Since roughly 2010, no overlap exists; the most conservative Democrat is to the left of the most liberal Republican.

Senate polarization tracks House polarization closely, with a slight lag and slightly less extreme magnitudes. The disappearance of "moderates" — members near the center of the NOMINATE scale — has been continuous and is now nearly total.

25.3.2 Mass polarization

Pew Research Center's "political typology" surveys, run periodically since 1994, show a different but related pattern. Pew tracks the distribution of ideological consistency among voters — how many liberal or conservative answers they give across a battery of policy questions.

In 1994, the median Democrat held views slightly to the left of the median Republican, but the distributions overlapped substantially. About 35 percent of Republicans were more liberal than the median Democrat, and about 30 percent of Democrats were more conservative than the median Republican.

By 2024, those overlaps had shrunk to under 10 percent on either side. The distributions had pulled apart. Crucially, however, the shape of the distributions had also changed — they had become bimodal for partisans (clustering toward each end) but remained unimodal for the broader public, including the substantial minority who do not strongly identify with either party.

This is one of the most important and easy-to-miss findings. The American electorate as a whole is more moderate than partisans within either party. The polarization is concentrated in those who are most engaged.

25.3.3 Affective polarization data

The ANES feeling-thermometer time series shows three patterns:

1. Warmth toward one's own party has been roughly stable since 1980 (around 70).
2. Warmth toward the other party has fallen sharply (47 → 18).
3. The decline has accelerated in roughly four periods: the late 1990s (Gingrich/Clinton), the late 2000s (Obama/Tea Party), the late 2010s (Trump/Resistance), and the early 2020s (post-Jan-6).

Iyengar and Krupenkin's 2018 work showed that the cross-partisan gap in the feeling thermometer had become the single strongest predictor of vote choice — stronger than any specific issue position. People do not merely disagree with the other party; they affirmatively dislike it.

25.3.4 Sorting in the geographic data

Bishop's The Big Sort (2008) documented a striking pattern: in 1976, 27 percent of Americans lived in counties where the presidential race was decided by 20 points or more. By 2004, that share had nearly doubled. By 2024, it was roughly 60 percent.

This is partly cause and partly effect. Americans appear to have moved toward communities that match their politics, and once they are there, the politics of those communities reinforces and intensifies. Whether the migration is because of politics is contested — most movers cite jobs, climate, family, or housing costs, not ideology — but the political consequence is the same regardless of motivation.

25.3.5 The institutional data

If we step back from individual voters and look at institutional behavior, the polarization picture is unmistakable:

• The number of bipartisan votes (where majorities of both parties supported a measure) declined from about 70 percent of major legislation in the 1970s to under 25 percent in the 2020s.
• The use of the filibuster (or threats thereof) in the Senate rose from a handful per session in the 1970s to over 200 per session in the 2010s and 2020s.
• The number of confirmation battles for judicial nominees has risen dramatically; cloture votes on nominations were rare before 2003, routine after.
• Speaker turnover, leadership challenges, and internal-party rebellions have increased.

Polarization is not simply a feeling. It is a property of how the institutions function.

25.4 Causal accounts

Why is this happening? The literature offers many candidate explanations. Most of them have empirical support. None of them, alone, fully accounts for the pattern. They are best understood as overlapping causes that reinforce each other.

25.4.1 Geographic sorting

If like-minded people live near like-minded people, the local political environment becomes homogeneous. Local norms reinforce in-group views. Cross-cutting conversation becomes rare. Bishop's argument is that residential mobility — which has fallen overall in America since 1980, but which remains powerful for those who do move — has produced a country in which most Americans do not regularly encounter strong disagreement in their daily lives.

The data support a moderate version of this claim. Counties have grown more lopsided. Suburbs that were politically mixed in 1990 are politically uniform in 2024. The strongest urban cores are deeper blue; the most rural counties are deeper red. The suburban ring has been the contested zone in every recent presidential election.

The claim that individuals deliberately choose where to live based on politics is harder to substantiate. Most movers cite economics. But the cumulative effect of millions of separate, non-political decisions is a more politically segregated country.

25.4.2 Media

Chapter 18 covered media in detail. The relevant finding here: partisan media reinforce in-group views, and the breakdown of the broadcast-network monopoly that ended in the late 1980s produced a media environment in which most Americans choose their information sources, and many choose sources that match their politics.

The empirical evidence is mixed on whether media causes polarization or reflects it. Some studies find that exposure to partisan media does increase ideological extremism and affective polarization. Others find that media exposure mostly reinforces views people already held. The honest answer is probably both: media shapes a small slice of the politically engaged, and that slice has outsized effect on the rest of the conversation.

25.4.3 The primary system

Primary electorates are smaller, more ideological, and more attentive than general electorates. A typical congressional primary draws 10 to 20 percent turnout; a competitive general election draws 50 to 70 percent. The voters who show up for primaries are more likely to be highly engaged, ideologically consistent, and motivated by issues that energize the base.

The structural consequence: primary winners are systematically more extreme than the median voter in the district. A House member who wants to win a primary cannot afford to be too moderate, because doing so risks a primary challenge from the more ideological wing of the party. The result is a Congress filled with members whose proximate threat is from their own side, not the other side, and who therefore behave more accordingly.

This effect is amplified in safe districts — districts where the general election is uncompetitive, so the primary is the only meaningful contest. In a safe Republican district, the only competition is the Republican primary; in a safe Democratic district, only the Democratic primary. The outcome in such districts is determined by the most engaged primary electorate, which pulls representatives toward the wings.

25.4.4 Gerrymandering

Chapter 35 covers gerrymandering in depth. Briefly: when state legislatures draw congressional and state-legislative districts to favor one party, they typically pack opposition voters into a few overwhelmingly opposition districts and spread their own voters across more districts of moderate majority. This produces a small number of safe opposition districts and a larger number of safe own-party districts.

The polarization consequence: more safe districts means more districts where the primary is the decisive election, which means more representatives who answer to primary electorates, which means more polarization. Gerrymandering does not cause all of polarization — even non-gerrymandered districts have polarized — but it amplifies the underlying dynamic.

25.4.5 Money

Chapter 34 covers money in depth. The relevant finding: donors are more ideological than voters. People who give money to political campaigns hold more consistent and more extreme views than people who merely vote. As campaigns have become more donor-dependent — driven up by post-Citizens United spending and by the rise of small-dollar online fundraising — politicians have become more responsive to donor preferences.

Small-dollar online donors are particularly noteworthy. These donors give to candidates whose rhetoric and positioning excite them; they reward conflict, intensity, and ideological purity. ActBlue and WinRed, the major online fundraising platforms, raised a combined $11 billion in the 2024 cycle. The candidates who raise the most online tend to be the most rhetorically intense, not the most legislatively productive.

25.4.6 Negative partisanship

A theme running through the affective-polarization literature is the rise of negative partisanship — the phenomenon of voters being mobilized more by dislike of the other party than by affection for their own.

Alan Abramowitz and Steven Webster, in a 2018 paper, argued that since the late 1990s, the dominant driver of partisan voting has been negative affect toward the opposing party rather than positive affect toward one's own. Voters who give their own party a tepid 60 on the feeling thermometer but the other party a frigid 10 will turn out to vote against the other party even when they are uninspired by their own.

Negative partisanship is self-sustaining. Politicians who run on "the other side is dangerous" mobilize their base, which rewards the rhetoric, which rewards more politicians who run that way, which deepens the affective gap, which makes the other side seem more dangerous. The dynamic does not require either side to be wrong about anything specific; it requires only that intensity wins primaries.

25.4.7 Identity sorting

Lilliana Mason's Uncivil Agreement (2018) argued that the alignment of identities with party — the sorting described above — is the most powerful single driver of affective polarization. Her argument: when partisanship overlaps with race, religion, geography, and lifestyle, every political loss feels like a loss for "people like me" across multiple dimensions. The stakes feel higher because the in-group is more cohesive.

In a less-sorted era, when both parties contained Catholics and Protestants, urbanites and ruralites, Northerners and Southerners, college-educated and non-college, a political defeat was experienced as one of the many things that go up and down in life. In a sorted era, a political defeat is experienced as a defeat for the entire bundle.

25.4.8 Political entrepreneurship

Polarization rewards a particular kind of politician — one who speaks intensely to the base, draws sharp contrasts with the other side, and refuses to compromise. These politicians often outperform institutionalists in primaries, in fundraising, and in media attention. The system therefore selects for them, and they in turn deepen polarization by their behavior.

Newt Gingrich's tactics in the 1990s House — confrontational rhetoric, refusal to socialize across the aisle, framing politics as moral war — represented one early example. The Tea Party in 2009-12 represented another wave. Donald Trump's 2016 campaign and presidency intensified the dynamic and reshaped expectations. On the Democratic side, the rise of "the Squad" after 2018 represented a parallel intensification, although operating from a different ideological direction and without comparable institutional power.

The point is not that any particular political entrepreneur is uniquely responsible. It is that the system rewards intensity, and intense politicians, once rewarded, raise the equilibrium intensity for everyone else.

25.4.9 Social media

Engagement-optimized algorithms reward content that generates strong responses, and strong responses are most often produced by content that triggers tribal in-group/out-group reactions. The result: social media platforms systematically amplify the most polarizing content from each side, while algorithmically demoting calm, nuanced, or cross-cutting content.

The empirical evidence here is contested. Studies attempting to identify the causal effect of social media on polarization have produced mixed results. Some studies, including a series of 2023 papers by Meta-affiliated and independent researchers, have found smaller effects than popular commentary assumes. Others, looking at specific platforms or subpopulations, have found substantial effects. The likely truth is that social media is a polarization amplifier rather than a polarization creator — it makes existing divisions more salient, more intense, and more visible.

25.4.10 The rural-urban divide

The geographic story has a particular structure: density tracks party. Urban areas have grown more reliably Democratic; rural areas have grown more reliably Republican; suburbs are mixed and contested.

The 2024 presidential election showed this pattern in extreme form. Counties with population density above 1,000 per square mile gave roughly 70 percent of their votes to the Democratic candidate. Counties with density below 50 per square mile gave roughly 70 percent to the Republican. The same pattern held within almost every state, and it had grown sharper in every cycle since 2000.

This is more than a statistical curiosity. The rural-urban divide maps onto economic structure (manufacturing/agriculture vs. services/knowledge), educational attainment (lower vs. higher college rates), demographics (whiter and older vs. more diverse and younger), religion (more vs. less religiously observant), and lifestyle (more vs. less car-dependent, more vs. less public-life-oriented). Each of these correlates produces another dimension along which Americans sort into politically homogeneous groups.

25.4.11 The Trump effect

Donald Trump's emergence and presidency function in the polarization story as both cause and symptom. He is plainly a symptom — a phenomenon impossible without the prior decades of Republican Party realignment, the conservative media ecosystem, the small-dollar donor economy, and the rural-suburban demographic shifts that made his coalition possible. He is also plainly a cause — a politician whose rhetorical style, contempt for political norms, and willingness to question election outcomes intensified polarization in ways that affected both parties.

The reaction to Trump from Democrats has likewise functioned as both cause and symptom of polarization. The "Resistance" framing, the focus on Trump as existential threat, the willingness to expand rhetorical and procedural conflict (court-packing proposals, filibuster-elimination proposals, expanded executive action) — these were responses to Trump that themselves intensified polarization.

The honest framing: Trump is a major polarizing event in recent American politics, and the reactions to Trump are also major polarizing events. Neither phenomenon caused the underlying conditions, but both intensified them.

25.5 Asymmetric polarization: the debate

We come now to the most contested empirical claim in this literature: that the two major parties have not polarized symmetrically.

25.5.1 The Mann/Ornstein/McCarty position

In It's Even Worse Than It Looks (2012), Thomas Mann (Brookings) and Norman Ornstein (American Enterprise Institute) — a centrist Democrat and a centrist Republican respectively — argued that congressional Republicans had moved further from the center than congressional Democrats had. Their argument leaned heavily on DW-NOMINATE scores: the trajectory of the median House Republican from 1970 to 2010 was steeper than the trajectory of the median House Democrat over the same period.

Nolan McCarty's Polarization: What Everyone Needs to Know (2019) developed this further. McCarty showed that on the standard measure, the median House Republican moved roughly 0.45 NOMINATE units to the right from 1970 to 2018, while the median House Democrat moved roughly 0.20 NOMINATE units to the left. The Republican shift was, by this measure, more than twice the Democratic shift.

The Mann/Ornstein/McCarty position has had real influence in academic political science and in centrist commentary. It is sometimes summarized as: "Both parties have polarized, but the Republicans have polarized more."

25.5.2 The pushback

The asymmetric thesis has been challenged from several directions, and it would be intellectually dishonest to ignore the challenges.

The issue space changes. DW-NOMINATE measures legislators on a stable scale derived from roll-call votes. But the content of those votes changes over decades. Issues that did not exist in 1970 (climate, gender identity, expansive views of executive enforcement priorities, novel financial regulations, contemporary debates over racial classification) now drive votes. Frances Lee, Eric Schickler, and others have argued that on these newer issue dimensions, Democrats have moved substantially leftward of mid-century positions. NOMINATE captures left-right structure but does not perfectly capture which side is "moving."

Selection bias in roll-call votes. Congressional leadership controls what comes to the floor. If Democratic leadership brings progressive bills to a vote that would have been outside the mainstream in 1985 — and these are now the bills available for measurement — the resulting NOMINATE scores will record Democratic positions on those new issues. Whether that represents Democrats moving left or the issue space expanding is a definitional question.

Some Democratic policy positions have moved substantially. On immigration, the average elected Democratic position in 1995 (when Senator Barbara Jordan's commission recommended significant reductions in legal immigration and stricter enforcement) is well to the right of the average elected Democratic position in 2024. On gender and identity, the shift is dramatic. On policing, on affirmative action, on the role of the state in regulating speech, on certain economic policies — Democratic positions have moved leftward of mid-twentieth-century Democratic positions.

Institutional behavior may not capture what voters experience. A measure that focuses on roll-call votes captures legislative behavior. It does not necessarily capture the felt reality of, say, a small-business owner in 2024 who experiences the regulatory environment, the cultural environment, and the educational environment as having moved substantially in directions she did not choose.

25.5.3 The current state of the debate

The asymmetric-polarization claim has empirical support, particularly in legislative behavior data. The magnitudes are contested. The interpretation depends on whether you weight institutional roll-call behavior more or less than substantive policy and cultural change.

A reasonable summary of the current state of the literature: by the standard quantitative legislative measures (NOMINATE), Republicans have polarized more than Democrats over 1970-2025; by some other measures including policy substance on certain issue domains, Democrats have moved substantially as well. Both claims are defensible. A reader who concludes that one side's polarization is the only real polarization is probably reflecting their own priors more than the data.

Both partisan diagnoses ("the other side has gone crazy") have grain of truth and grain of self-flattery. Republicans look at Democratic positions on immigration, identity, education, and the administrative state and say: those positions are far from where Democrats stood in the Clinton era. Democrats look at Republican positions on election integrity, the 2020 election outcome, the status of the administrative state, and the use of constitutional hardball, and say: those positions are far from where Republicans stood in the Reagan era. Both groups are right about the other group. Neither is entirely right that their own side has not also moved.

25.6 Comparative perspective

Is the United States uniquely polarized? Comparative data complicates the picture.

The V-Dem Institute's Liberal Democracy Index tracks liberal-democratic governance worldwide. The U.S. score has slipped on this index since 2016, reflecting concerns about institutional norm erosion, election administration, and rule-of-law indicators. Through 2025, the U.S. remains classified as a liberal democracy, but with downgraded scores on several sub-indicators. By the most commonly cited cross-national polarization measures (the Polarization-Ideological Distance index and the Affective Polarization Index used in the American Political Science Review literature), the U.S. ranks among the more polarized advanced democracies — though not the single most polarized.

Other advanced democracies that score similarly or higher on cross-national polarization measures:

• Israel has experienced extreme polarization, particularly during the 2023-24 judicial-reform crisis and following the October 7, 2023 Hamas attacks and the subsequent war.
• Hungary and Poland (categorized as "electoral democracies" or "electoral autocracies" depending on the year and measure) saw substantial polarization through the 2010s, with Hungary's slide on democratic indices particularly steep.
• Brazil experienced intense polarization during and after the Bolsonaro presidency.
• France and Italy have substantial polarization, with electoral systems that produce frequent realignments.
• The United Kingdom experienced significant polarization around Brexit (2016-19) and continues to show partisan affective polarization at U.S.-comparable levels.

Some democracies have managed lower polarization, though most have seen rising polarization recently:

• Germany has historically had relatively low affective polarization, but the rise of the Alternative für Deutschland (AfD) since 2013, the continuing strength of the Linke and Bündnis Sahra Wagenknecht on the left, and the migration debates of the 2010s and 2020s have raised polarization measures.
• The Nordic countries (Sweden, Denmark, Norway, Finland) historically scored low on polarization but have seen rises driven by migration politics, the rise of Sweden Democrats and Finns Party, and the Russian threat following the 2022 Ukraine invasion.
• Switzerland and the Netherlands have multi-party systems that have historically diffused polarization across many parties, though both have seen rises in cross-party hostility.

The comparative lesson is mixed. The U.S. is high but not unique. Most advanced democracies have seen rising polarization since 2010. The U.S. pattern is distinctive in some ways — particularly the binary two-party structure that maps polarization onto a single cleavage rather than dispersing it across many — but not in being the only polarized country.

25.7 Affective consequences

How does polarization affect daily life?

25.7.1 The marriage-and-family question

Iyengar's classic finding: in 1960, fewer than 5 percent of partisans said they would be "displeased" if their child married a member of the other party. By 2024, that number had risen to roughly 40 percent for Republican parents and 50 percent for Democratic parents.

Compare this to other social distances. In 1960, a substantial share of Americans would have been displeased if their child married someone of a different race or religion. By 2024, those numbers had fallen to single digits — even as partisan-marriage displeasure rose. Partisanship has become the most acceptable form of out-group hostility in American social life.

25.7.2 Workplace, friendship, and neighborhood

Pew and More in Common surveys consistently find that Americans report:

• Avoiding political conversations at work (over 60 percent)
• Having lost friends over politics (about 30 percent)
• Knowing fewer people of the other party than they did a decade ago (about 40 percent)
• Choosing to live, attend church, or join organizations partly on the basis of political climate (smaller but rising shares)

These patterns are stronger among the politically engaged than among the disengaged. Most Americans are not political junkies, and most retain cross-cutting relationships. But for those who follow politics closely, the social and personal consequences are substantial.

25.7.3 Trust

Interpartisan trust is at historic lows. The percentage of Republicans who say they "trust" Democrats to act in the country's interest has fallen from about 30 percent in 1990 to under 10 percent in 2024. The Democratic figure for trust in Republicans tracks similarly. Both parties' perceptions of the other have moved from "wrong on the issues" to "actively dangerous to the country."

This is the most concerning data in the polarization literature, because it indicates that the disagreement has shifted from being about what to do to being about who is legitimate. Democracies require, at minimum, mutual recognition that the other side has the right to govern when it wins. When that recognition erodes, the system itself comes under stress.

25.8 Policy and governance consequences

25.8.1 Gridlock

Polarized parties produce legislative gridlock. The classic data point: the number of significant federal laws passed per Congress declined from a peak in the 1970s to roughly half that rate in the 2010s and 2020s. Major legislation — comprehensive immigration reform, climate legislation, infrastructure modernization, healthcare reform — repeatedly stalls.

Defenders of gridlock argue this is a feature, not a bug, of the constitutional system. The American government was designed to make change difficult; the absence of cross-party consensus means change should be slow. Opponents argue that the constitutional design assumed cross-party coalitions would form as needed, and that polarization has prevented the formation of those coalitions, leaving major policy challenges unaddressed.

Both arguments have force. The constitutional system was indeed designed for deliberation. It also assumed that compromise was possible. Polarization has made compromise rarer.

25.8.2 Decline of regular order in Congress

Chapter 8 covered congressional procedure. The relevant finding: the use of regular committee procedures, open debate, and conference committees has declined markedly since 1990. More legislation moves through extraordinary procedures — leadership-driven bills, omnibus packages, reconciliation, post-deadline continuing resolutions — that bypass the deliberative committee structure.

Polarization is not the only cause. Procedural change has many drivers. But polarization makes regular order harder, because regular order requires cross-party cooperation that polarization punishes.

25.8.3 Veto-proof majorities

The Senate filibuster requires 60 votes for most legislation. In a polarized environment, 60-vote coalitions are difficult to assemble. The result: most major legislation passes through the reconciliation process (which evades the filibuster but is limited to budget-relevant provisions), through executive action (which is reversible by the next administration), or not at all.

This produces a pendulum-swing pattern in major policy. The ACA passes via reconciliation under Democratic control; Republicans attempt to repeal it for a decade. DACA is implemented by executive action under Obama; Trump attempts to rescind; the Supreme Court blocks the rescission on procedural grounds; Biden re-implements; the next Republican administration may try again. Climate and energy policy swings between executive frameworks based on which party holds the White House.

25.8.4 Compromise as betrayal

In a polarized primary system, members of Congress who cooperate across party lines risk primary punishment. The Tea Party challenges to incumbent Republicans (Bob Bennett of Utah, Richard Lugar of Indiana) and the progressive primary challenges to incumbent Democrats (Joe Crowley of New York, Eliot Engel of New York) sent the same signal in different directions: cross-party cooperation can cost you your seat.

This produces the dynamic in which most members vote the party line on most votes, and the small number of moderates remaining are systematically targeted by the wings of their own parties.

25.8.5 Norms erosion

Chapter 37 will cover this in depth. Briefly: when politics becomes existential — when the other party is not just wrong but dangerous — the temptation to break norms in the service of preventing the other side from winning becomes powerful. Constitutional hardball, expansive interpretations of executive power, court-packing proposals, election-administration fights, and rhetorical questioning of opponents' legitimacy all become more frequent.

These norms-erosion dynamics happen in both parties. Each party's defenders argue their own norm-breaking is justified by the other side's behavior. Both arguments contain truth. The cumulative effect is degradation.

25.9 Can America "unsort"?

We turn now to the prospects for reducing polarization.

25.9.1 Optimistic accounts

There are real reasons for cautious optimism.

Cross-pressured voters still exist. Roughly 10-15 percent of the electorate holds policy positions that do not align cleanly with either party. These voters are open to persuasion in ways that highly partisan voters are not. They tend to be less ideological, less politically engaged, and more focused on immediate practical concerns. They remain a real constituency, and elections still hinge on them.

Populist alignments cut across traditional party lines. On certain issues — trade policy, foreign-policy restraint, regulation of tech companies, antitrust enforcement, agricultural and small-town concerns — coalitions form that include both populist Republicans and populist Democrats while excluding establishment members of both parties. These cross-cutting coalitions show that party-line voting is not absolute.

Younger voters are less rigidly partisan. Voters under 30 register as Independents at higher rates, and their attachment to either party is weaker than older voters'. Whether this represents a generational shift toward less polarization or merely a delayed partisan crystallization is unclear, but the data is real.

The "perception gap" research. More in Common's Hidden Tribes report and subsequent work showed that Americans systematically overestimate how extreme the other party is. Republicans overestimate how many Democrats hold particular extreme positions; Democrats overestimate the same about Republicans. The perception gap is larger than the actual position gap. Studies suggest that correcting this misperception — exposing partisans to accurate information about the other side's median views — can reduce affective polarization, at least in the short term.

25.9.2 Pessimistic accounts

There are real reasons for caution.

Sorting is structurally locked in. Residential, identity-based, and educational sorting do not unsort easily. They are reinforced by housing markets, marriage patterns, and college-educated migration. Even if individual minds change, the structural pattern persists.

The information environment does not reward depolarization. Engagement-optimized algorithms continue to amplify polarizing content. Partisan media remain economically successful. Cross-cutting media outlets struggle for audience.

Negative partisanship is self-sustaining. As long as voters are mobilized by dislike of the other side, politicians have incentives to deepen that dislike. The system does not naturally exit this pattern.

Generational change does not always reduce polarization. The "young voters are less partisan" claim is partially undermined by the observation that millennial and Gen Z voters who have crystallized their politics are more polarized than older crystallizers were at the same age. The disengaged are less partisan; the engaged are more so.

25.9.3 Reform proposals

Many reform proposals have been advanced. Each has supporters and serious critics.

Ranked-choice voting (RCV). Voters rank candidates by preference; if no candidate gets a majority, the lowest-ranked is eliminated and votes are reallocated. Defenders argue RCV reduces incentives for negative campaigning (because candidates need second-choice votes from supporters of opponents), encourages broader coalitions, and reduces the "spoiler" problem. Critics argue empirical effects are smaller than advocates claim, that it complicates ballot administration, and that it changes the strategic logic of campaigns in unpredictable ways. Maine and Alaska have implemented RCV statewide; results are mixed but not catastrophic.

Open primaries. Allowing all voters, including unaffiliated voters, to participate in primaries. Defenders argue this reduces the ideological extremity of primary winners. Critics argue it weakens parties as institutions and that the empirical effects on extremism are modest.

Anti-gerrymandering reforms. Independent redistricting commissions, court-imposed standards, multi-member districts. Critics argue commissions are themselves political, that "fairness" lacks a clean definition, and that the effects on polarization are smaller than effects on competitiveness.

Lobbying restrictions. Limits on who can lobby, what they can do, and the speed of the revolving door. Critics argue restrictions are easily evaded, that "lobbying" is a constitutionally protected activity, and that the connection to polarization is indirect.

Civic education. Curriculum mandates for civics in schools, programs to teach democratic deliberation. Critics argue the empirical evidence for educational interventions reducing polarization is thin.

Structured cross-partisan conversation efforts. Braver Angels (formerly Better Angels), Bridge USA, More in Common, Citizens' Assemblies. These programs convene small groups of partisans across the aisle for facilitated conversation. Defenders argue they reduce affective polarization at the individual level. Critics argue they reach only a self-selected, already-curious audience and cannot scale to the mass level. (We examine these in case study 02.)

Each of these reform proposals deserves serious analysis. None is a silver bullet. The best honest summary: polarization is multi-causal, and any reform that addresses only one cause will have at best a marginal effect.

25.9.4 Reform proposals: deeper consideration

Each of the reform categories in 25.9.3 deserves a closer look. The point is not to endorse any of them but to make clear what the evidence suggests, what the strongest critiques are, and what the trade-offs look like. A reader who comes away from the polarization literature thinking "well, then we should obviously just do RCV everywhere" has not yet encountered the strongest critiques. Likewise a reader who concludes "no reform works, so why bother" has skipped over real-world implementations whose results are not as dismal as cynicism suggests.

Ranked-choice voting in practice. Maine adopted RCV for federal and state-primary elections starting in 2018. Alaska adopted RCV plus a "top-four" open primary (the top four candidates from a single primary advance to a general-election RCV runoff) starting in 2022. New York City uses RCV for primaries. Several other cities (San Francisco, Minneapolis, Cambridge) have used it for decades. The empirical results across these implementations show: turnout effects are small and inconsistent; campaign tone in some races has moderated, in others has not; the share of "wasted" votes (cast for eliminated candidates whose preferences are never reallocated) is small but not zero; voter understanding of the system is not universal. Defenders point to the 2024 Alaska Senate race, where moderate Republican Lisa Murkowski won re-election in a context that previously would have plausibly produced a more ideological winner. Critics point to administrative complexity, occasional anomalies (the so-called "monotonicity" failures, where ranking a candidate higher could in theory hurt them), and the observation that the strongest claims for RCV's depolarizing power are not yet supported by definitive evidence. The honest conclusion: RCV is a real reform that produces modest, generally positive effects on competitiveness and tone. It is not a silver bullet. It is not catastrophic. Whether the trade-offs are worthwhile depends on what you are trying to optimize.

Open primaries in practice. California's "top-two" primary (since 2012) and Washington State's similar system (since 2008) have produced mixed evidence. Some studies find modest reductions in legislator extremism in California; others find effects concentrated in specific district types. Alaska's top-four-plus-RCV system is too new for definitive results. Critics — including some political scientists who study parties as institutions — argue that open primaries weaken the parties' ability to shape their own nominations, which weakens accountability and may push incoherence rather than moderation. Defenders argue the goal is voter responsiveness, not party institutional strength.

Anti-gerrymandering reforms. California, Arizona, Michigan, Colorado, Virginia, and several other states have implemented forms of independent redistricting commissions. The empirical results: commission-drawn districts are more competitive on average, but the depolarization effects are modest. The reason is structural: even non-gerrymandered districts contain populations that have themselves sorted geographically, so a "fair" map still produces many safe districts. Anti-gerrymandering reform is necessary to keep things from getting worse but is not sufficient to reverse the underlying sorting. Critics also argue that "fair maps" lack a clean definition and that commissions are subject to political manipulation in their composition.

Civic education and democratic-deliberation programs. The empirical evidence here is the weakest of any reform category. Studies of civics curricula find small effects on civic knowledge and somewhat smaller effects on civic engagement; effects on polarization specifically are difficult to identify. Programs like Generation Citizen and iCivics have produced positive evaluations but cannot be shown to address the polarization problem at scale. Civic education may still be intrinsically worthwhile, but it should not be sold as a polarization remedy.

The "small reforms" argument. Some scholars (notably Lee Drutman, Larry Diamond, and various contributors to the Liberal Patriot / "expanding-democracy" tradition) have argued for a bundle of reforms: RCV, open primaries, multi-member districts, anti-gerrymandering, expanded voting access, campaign-finance transparency, and structural reform of media incentives. The argument is that no single reform addresses the multi-causal nature of polarization, but a sufficiently large bundle might. Critics from the right argue this bundle is itself ideologically loaded — that the reforms favored by reform-minded scholars happen to be reforms that benefit certain political constituencies. Critics from the left argue that the reforms are insufficiently radical and miss the underlying problem of concentrated economic and corporate power. Both critiques have force.

The most honest summary of the reform conversation: We do not know how to depolarize a polarized democracy. We know some reforms produce modest improvements on certain margins. We do not have a confident playbook.

25.9.5 Cross-cutting research and the "perception gap"

A subset of the depolarization literature is more empirically encouraging than the structural-reform literature. The "perception gap" research, pioneered by More in Common and replicated by other research organizations, is worth understanding in detail.

The basic finding: ask Republicans what percentage of Democrats hold particular extreme views (e.g., "the U.S. should have completely open borders" or "police should be defunded entirely"); ask Democrats what percentage of Republicans hold particular extreme views (e.g., "no immigrant should be allowed to enter the U.S." or "America is a Christian country and Christianity should be its official religion"). Then compare the estimates with what Republicans and Democrats actually say in surveys.

The estimates are systematically wrong. Both sides massively overestimate how extreme the median member of the other party is. The estimation errors are largest among the most politically engaged — those who follow the news closely, who consume partisan media, and who have the most-developed political identities.

The implication: a substantial share of affective polarization is driven by inaccurate beliefs about the other side. Correcting those beliefs — exposing partisans to accurate information about the actual median positions of the other party — has been shown in experimental research to reduce affective polarization, at least temporarily.

This finding is hopeful, but its limits matter. The reduction in affective polarization is modest and tends to fade over time. Correcting one misperception does not correct the broader pattern of consuming information from sources that systematically misrepresent the other side. Scaling perception-gap interventions beyond experimental studies has proved difficult.

Still, the perception gap research suggests that the polarization we see is partly a perception problem rather than a substance problem. Americans dislike a caricature of the other side that is more extreme than the median member of the other side actually is. This creates room for hope, even if the path from "experimental evidence of perception correction" to "actual mass-level depolarization" is not yet clear.

25.9.6 The case for polarization (or, at least, against alarm)

A minority of analysts argue that polarization is, at minimum, less catastrophic than mainstream commentary suggests, and possibly even healthy in some respects. This view is associated with scholars including Frances Lee (whose Insecure Majorities (2016) is the most thorough academic statement), some commentators on the political right (Yuval Levin, Christopher Caldwell), and some commentators on the left (Matt Stoller, Ryan Grim).

The argument has several strands.

Polarization clarifies stakes. A non-polarized politics, in which the parties are nearly indistinguishable on policy, may be a politics in which voters cannot meaningfully choose. Polarized parties present voters with real alternatives. The mid-twentieth-century era of bipartisan consensus often celebrated by polarization-worried commentators was also an era in which significant policy disagreements (on civil rights, on Vietnam, on the role of women, on the regulation of business) were poorly represented in mainstream political competition.

Some "polarization" is just democratic responsiveness. When the median voter is far from the policies the political class would prefer, polarization may reflect mass-level demands that the establishment is reluctant to satisfy. The rise of Bernie Sanders, the rise of Donald Trump, and the rise of various populist figures internationally can be read as parties belatedly responding to electorate dissatisfaction with the prior consensus.

Affective polarization may be downstream of substantive disagreement. If the parties really do disagree about important things, intense feeling is a natural response. The question is not whether feeling is intense but whether the disagreement is real. If it is, intensity is appropriate; if it is not, intensity is misplaced. The scholars in this tradition argue that the substantive disagreements are real.

The institutional dysfunction critique conflates two things. Polarization makes legislation harder; the constitutional system was designed to make legislation harder. The two effects compound, but the constitutional system was not malfunctioning when it required cross-party coalitions. Whether the contemporary difficulty in forming such coalitions reflects a problem with polarization or a problem with the constitutional design (which assumed cross-cutting coalitions would form) is itself contested.

This minority view does not deny that polarization has costs. It argues that the costs are smaller than mainstream commentary suggests, the benefits are larger, and the assumed alternative (a less polarized politics) may not have been as healthy as nostalgia paints it.

A reader should not conclude from this section that polarization is fine. The mainstream-political-science consensus that polarization has serious costs is broadly correct. But it is intellectually honest to note that the consensus has dissenters, that the dissenters have arguments, and that pretending polarization is uniformly bad is a form of motivated reasoning.

25.10 Polarization in your district

Your District callout. Polarization shows up locally as well as nationally. Some questions for your own congressional district: What is the Cook Partisan Voting Index (PVI) — the measure of how much more Democratic or Republican your district is than the country as a whole? What was the margin in your district in the 2024 House race? Did your representative face a primary challenge? If so, was the challenger to their left or to their right? What share of your county was decided by 20 points or more in 2024? Did your local newspaper endorse a candidate? Do you know neighbors who vote differently from you, and do you talk politics with them? Document what you find. The polarization you read about in this chapter is not just an abstraction; it is the local political environment you live in.

25.10.1 Two conversations about polarization that miss each other

Before closing, it is worth describing two characteristic ways the polarization conversation goes wrong, both of which a careful reader should learn to spot.

The "they're crazy" conversation. A common pattern in elite political commentary: each side describes the other side's polarization in detailed, well-sourced terms while describing its own side's polarization, if at all, as a defensive response or a misperception. Right-leaning commentary documents the leftward shifts of the Democratic Party on identity, immigration, education, and the administrative state, and treats Republican shifts as either justified responses or media misrepresentations. Left-leaning commentary documents the rightward shifts of the Republican Party on election integrity, the administrative state, climate policy, and constitutional norms, and treats Democratic shifts as either justified responses or media misrepresentations.

Each set of analyses has empirical content. Each is also missing the other half of the picture. A reader who consumes only one side's polarization analysis will come away with a one-sided understanding. A reader who consumes both will come away noticing that the two analyses describe different parts of the same phenomenon.

The "both sides are equally bad" conversation. A different but equally common pattern in centrist commentary: collapse all asymmetries into a single "both sides are polarized" framing that does not engage with the empirical question of whether the magnitudes are actually equivalent. This framing has the advantage of avoiding partisan offense. It has the disadvantage of being potentially false. The asymmetric-polarization literature is contested, but it is not nothing; pretending the parties have moved equivalent distances is a different kind of motivated reasoning than the partisan-on-only-the-other-side framing, but it is motivated reasoning nonetheless.

The honest middle ground is harder. It involves engaging with the actual data, acknowledging that both parties have moved (in different directions, by different magnitudes, on different issues), and accepting that the precise magnitudes are themselves contested. The honest middle ground does not produce satisfying conclusions on either tribal axis. It is, however, what the evidence supports.

25.10.2 Polarization and the future of American democracy

Several of the most influential recent books on American democracy — Steven Levitsky and Daniel Ziblatt's How Democracies Die (2018) and Tyranny of the Minority (2023), Larry Bartels' Democracy Erodes from the Top (2023), Yascha Mounk's The People vs. Democracy (2018) — argue that polarization is not merely a procedural inconvenience but a precursor to democratic deterioration. The mechanism: when polarization becomes intense enough, the temptation to treat the other side as illegitimate rises; when illegitimacy framing rises, the willingness to use anti-democratic means to defeat the other side rises; and the combination of polarization with weakening institutional norms is the path by which democracies in other countries (Hungary, Venezuela, Turkey) have transitioned to less-democratic forms.

These analyses are taken seriously by mainstream political scientists. They are also contested. Critics argue that the U.S. has institutional protections (federalism, judicial independence, separation of powers, regular elections) that have not failed in countries that experienced democratic backsliding, and that comparing the U.S. to those countries overstates the parallels. The Levitsky/Ziblatt thesis has had real influence — it shaped much of the post-2016 commentary on Trump and democratic norms — but a reader should be aware that the analogy to other countries' backsliding is not universally accepted by specialists.

Where this chapter lands: polarization is a real concern with real costs. Whether it is an existential concern that threatens American democracy directly, or a serious-but-survivable challenge of the kind democracies have faced before and overcome, is genuinely contested. Both views are held by serious scholars. The honest answer is that we do not know, and the answer probably depends on choices that have not yet been made — by political leaders, by media institutions, by citizens.

The next several chapters look at adjacent and related phenomena: populism and political violence (Ch 26), civic engagement (Ch 27), money in politics (Ch 34), gerrymandering (Ch 35), voting rights (Ch 36), and democratic erosion (Ch 37). Each is in part a chapter about polarization seen from a different angle. Together they constitute a portrait of a democracy under stress but not yet broken — a system whose ability to adapt is being tested.

25.11 What the chapter has done — and where this leaves us

Polarization is real. It is multi-dimensional (ideological, affective, sorting). It has grown substantially since the 1970s in Congress and in the most engaged segments of the electorate, more modestly in the broader public. It has multiple causes — geographic, media, primary system, gerrymandering, money, identity sorting, political entrepreneurship, social media, the rural-urban divide — and any account that attributes it to a single factor is incomplete.

The asymmetric-polarization claim has empirical support, particularly in legislative-behavior measures, but is contested in important ways. Both partisan diagnoses contain truth and self-flattery. Comparative perspective shows that the U.S. is not uniquely polarized; many advanced democracies have polarized along different lines. The consequences are real: legislative gridlock, declining trust, social distance, norm erosion.

Whether America can unsort is genuinely uncertain. There are optimistic and pessimistic readings of the same data. Reform proposals are many; none is a silver bullet; each has serious defenders and serious critics.

What to take from this chapter: polarization is not someone else's problem. It is the political environment in which the rest of this book's institutions — Congress, the courts, the bureaucracy, the parties, the press — operate. The next chapter examines populism and political violence; the chapter after, civic engagement and what citizens actually do. Both are downstream of, and contribute to, the polarization examined here.

The disagreement is real. The disagreement also looks larger than it is. Both things can be true at once. That is roughly where the political-science literature stands, and it is roughly where this chapter stands.

You will form your own view. The discipline of this chapter has been to give you the tools to do so without either side's most caricatured version winning by default.

A final note on the texture of polarization. Most readers of this chapter are not professional political scientists. You are students, citizens, people trying to make sense of what you experience. What you experience is partly the structural phenomenon described above — the sorting, the affective gap, the institutional dysfunction — and partly the mundane reality of disagreeing with relatives, neighbors, coworkers, and the strangers you encounter in online life. The structural and the mundane are connected, but they are not identical. Some of the structural pattern is amenable to reform; some of the mundane experience is amenable to ordinary social effort. Neither is amenable to the kind of clean policy fix that political-science writing sometimes implies.

If this chapter has done its work, you should now be able to do three things you could not do before. First, you should be able to distinguish ideological polarization from affective polarization from sorting, and identify which is being discussed when commentary uses the loose word "polarization." Second, you should be able to evaluate causal claims about what is driving the pattern — to recognize when a single cause is being elevated above its actual weight, and when a multi-causal account is being sloppily reduced. Third, you should be able to engage with the asymmetric-polarization debate in both directions — to take seriously both the empirical case for asymmetry and the most serious responses to that case, and to form your own judgment without it being entirely a function of which side you root for.

Those three capacities are the practical product of this chapter. They will be useful in the rest of the book, in the rest of your political reading, and in the rest of your political life. Polarization is not going away soon. Understanding it more carefully than the loudest voices on each side allow is the most realistic form of resistance to its worst effects.

Continue to exercises.md, quiz.md, case-study-01.md, case-study-02.md, key-takeaways.md, and further-reading.md.