TL;DR:

In reviewing academic research, efforts to fix political polarization by making partisans dislike each other less are failing, because the problem is not about individual feelings.  The real issue is a political structure that incentivizes conflict and extremism.

A common belief suggests that if Democrats and Republicans could just understand each other better, the nation’s political hostility would subside. This idea has launched countless civility initiatives, workshops, and media campaigns designed to bridge the partisan divide. A new wave of political science research, however, indicates this entire approach is fundamentally misguided.

It argues that targeting feelings of dislike, what researchers call affective polarization, does little to curb anti-democratic behavior.

The real problem is not the animosity in our hearts, but the incentive structure of our political system.

This analysis synthesizes findings from several recent, large-scale studies to build a case for a structural turn in polarization research. It suggests we have been trying to solve the wrong problem.

NOTE:  I typically don’t highlight research that has not been peer reviewed, but this research isn’t heavy on math, but rather a good review of the research on polarization.  Worthy of sharing, but please note, the underlying paper is not peer-reviewed.

How We Know

Methodology

• Rachel Kleinfeld’s 2023 review of a decade of polarization studies
• Boxell, Gentzkow, and Shapiro’s cross-national measurement of affective polarization across 12 OECD democracies (149 surveys over four decades)
• Voelkel et al.’s 2023 experimental study testing depolarization interventions on 8,385 participants
• Stewart, McCarty, and Bryson’s 2020 formal model of polarization as an evolutionarily stable response to economic decline

Finding 1: Global Data Refutes Common Scapegoats

Many explanations for America’s rising polarization exist. The internet, social media, economic inequality, and immigration are frequent culprits. When researchers compared polarization trends across twelve developed democracies, however, these explanations collapsed.

Inequality, internet use, and trade openness rose in nearly all countries studied, but partisan animosity did not. It rose sharply in the United States, rose modestly in five other countries, and actually *fell* in six nations, including Britain, Germany, and Japan.

The only factor that consistently tracked with rising affective polarization was elite polarization: the ideological distance between the political parties themselves. When political parties and their leaders move further apart, mass animosity follows.

I used a great graphic in the classroom that shows polarization over the years of the elite and then the public.  The elites polarized first and the public trailed.  

This points away from universal causes like technology and toward specific features of a nation’s political system.

Finding 2: Changing Feelings Does Not Change Behavior

The second pillar of the old polarization argument is also crumbling. In laboratory settings, it is possible to reduce partisan dislike. Experiments show that correcting misperceptions about the other party or showing videos of cross party friendships can make people feel warmer toward their political opponents.

But these warmer feelings do not translate into stronger support for democracy. One study tested twenty-five different interventions on over 30,000 people. While several successfully reduced partisan animosity, they did not reduce support for anti-democratic practices or partisan violence.  One intervention actually increased support for undemocratic actions.

The variable that *does* predict anti-democratic behavior is the belief that the other side presents a fundamental threat to democracy and intends to break its rules.

Dislike may correlate with this belief, but the belief itself is the primary driver of action.

Finding 3: Polarization, Once Triggered, Becomes Locked In

Why is polarization so persistent? A cultural evolution model offers a powerful explanation. The model shows that under conditions of economic decline or rising inequality, people shift toward risk-averse, in-group favoring strategies.  Interacting with one’s own group becomes safer, even if interacting with outsiders offers a higher potential reward.  This is a rational response to economic stress.

Crucially, the model shows that once a society shifts into this high polarization state, it gets stuck there.  The system exhibits hysteresis, meaning it does not revert even when the original economic pressures disappear.  Returning to a low polarization state requires a massive shock or a coordinated, structural change. Small interventions that slightly reduce individual animosity are not enough to break the system out of its polarized equilibrium.  

The Cross-National Test

The most useful constraint on theories of polarization comes from comparing countries. If the same thing caused polarization everywhere, candidate causes should rise alongside it everywhere.

Boxell, Gentzkow, and Shapiro measured affective polarization across 12 OECD democracies. The trends do not move together.

Affective polarization rose substantially in the United States (5.6 points per decade), rose modestly in five other countries, and fell in six. The U.S. showed the largest increase.

The authors correlated polarization trends with candidate explanatory variables. The results eliminate most popular theories:

• Inequality: negative correlation, not significant
• Trade openness: essentially zero, not significant
• Internet penetration: negative, not significant
• Online news consumption: negative, not significant
• Immigration (foreign born share): negative, not significant
• Elite polarization (ideological distance between parties): 0.782 correlation, p=0.011

The only variable tracking affective polarization across democracies is elite polarization itself: a structural feature of national politics, not a feature of individuals or media technology.

This matters because it rules out the most discussed explanations of American polarization. Social media, rising inequality, and immigration do not track polarization once you look beyond the United States.  What does track it is how far apart the political parties have moved.

A “Wicked Problem”

Following Horst Rittel and Melvin Webber (1973), polarization is a wicked problem.  It has no clear stop rule, and every “solution” reveals a deeper layer of the crisis.

We thought it was Ideology (Elites).

Then we thought it was Affect (Voters).

Now we maybe realizing it is Incentives (Structure).

Why It Matters

This research has profound implications.

The popular understanding of polarization frames it as a problem about how Americans feel about each other. The research reviewed here suggests it is not primarily that.

It suggests that vast resources are being spent on interventions that, while well intentioned, are aimed at a symptom, not the cause. Focusing on civility and interpersonal dislike distracts from the institutional arrangements that make extremism a winning political strategy.

The real danger is not that Americans dislike one another. The danger is that the political system incentivizes leaders to exploit that dislike for power, which leads voters to believe the other side is an existential threat, which in turn justifies breaking democratic norms to stop them.

Polarization is a problem about political structural conditions under which polarized feeling can be instrumentalized to win elections.  It is about institutional incentives that make extremist candidate recruitment a rational strategy.

It is the structure, stupid!

Critiques and Future Research

This argument is an interpretation of existing research, not a single new empirical finding. The evidence for any specific structural reform, such as ranked choice voting or nonpartisan primaries, remains uneven and requires more study.

Future research must shift from short term lab experiments on individual attitudes to long term studies of real world political systems.

The key question is not how to make an individual feel less hostile for twenty minutes, but how to change the rules of the game so that moderation and coalition building become the most effective paths to political power.

Deep canvassing has shown some promise. This research shows that while a mind may move a bit, behavior often remains unchanged.

What remains missing: longitudinal studies tracking structural interventions over years rather than attitudinal interventions over minutes.  We are also reminded by the research that cross national experiments testing whether the same interventions produce different effects in different institutional contexts are needed.

Practical Implications for Policymakers

• Re-evaluate funding for programs focused solely on reducing partisan animosity.
• Prioritize the study and implementation of structural reforms that alter political incentives.
• Examine changes to electoral systems, primary rules, and campaign finance laws that may reward moderation over extremism.
• Consider whether interventions emphasizing threats to democracy may backfire by increasing perceived threat, which is the variable that actually predicts antidemocratic behavior.

Practical Implications for Public Affairs Professionals

• Campaigns built on reducing partisan animosity should not expect downstream gains in democratic norms or reduced political violence.
• Messages that reduce perceived threat from the other side (credible commitments to honor norms) outperform messages that reduce dislike.
• Framing that emphasizes shared threat or shared identity can reduce polarization but may increase support for antidemocratic practices. Test carefully.
• Cross partisan coalition building faces coordination problems inherent to bistable systems. Small efforts will not shift the equilibrium.
• Elite signals matter more than mass attitudes. Political leaders committing publicly to democratic norms move the variables that predict behavior.

The Bottom Line

Polarization research spent a decade assuming that if you made Americans like each other more, democratic behavior would follow. The evidence now shows that assumption was likely wrong. You can change feelings without changing actions!

Elite cues really matter, and I have over the years come to fully appreciate just how much.   And as long as elites are “single-minded seekers of re-election” in a political structure that incentivizes and rewards extreme candidates, they will behave in a polarized, outrageous manner.

The conversation about polarization needs a fundamental reset.

The unit that needs to change is not the individual but the political structure. The intervention that needs testing is not the psychology lesson but the institutional reform.

The challenge is not to make opposing partisans hold hands.

The challenge is to create a political system where demonizing the other side is no longer a viable strategy for winning.

The extreme challenge is reforming a system that benefits the very people needed to make changes.