The Role of Identity in Political Alignment

People do not usually choose a party because of its policies. They adopt policies because of the group they already belong to. The evidence for that inversion, and what it means for political argument.

Escrito por Ottavio Malatacca · Publicado 2026-03-14 · Actualizado 2026-08-30

The intuitive model of political choice is that a citizen examines the available policies, compares them against their interests and values, and selects accordingly. Six decades of research in political psychology have found that this model has the causal arrow backwards for most people, most of the time. Group attachment usually comes first, and policy preferences are adjusted afterwards to fit it. This is one of the most robust findings in the field, and one of the least comfortable.

The finding, and how it was established

The classic demonstration comes from panel studies that follow the same people across years. When a party's position on an issue shifts, a substantial share of its supporters shift with it, and they report their new position with the same confidence as the old one. The effect appears on issues where individual material interest is minimal and on issues where it is considerable.

A related family of experiments varies only the party label attached to an identical policy proposal. Support for the same text moves by wide margins depending on which party is said to have proposed it. The effect is strongest among the most politically engaged respondents — the group the naive model would expect to be most independent, because they know the most. Greater knowledge turns out to supply better tools for reaching the conclusion the group already holds.

None of this shows that policy never matters. It shows that for most voters, on most issues, the identity signal arrives first and does more work. A minority of voters — issue publics with intense, specific commitments — behave much closer to the intuitive model, and they matter disproportionately in close elections.

Why groups do this: the psychology

Henri Tajfel's minimal group experiments in the 1970s established the baseline. Participants divided into groups on an explicitly arbitrary basis — a coin toss, a stated preference between two painters they had never heard of — immediately began allocating rewards to favour their own group, including in conditions where doing so reduced their group's absolute payoff. There was no history, no conflict, no material stake. The category alone was sufficient.

Social identity theory explains this as self-esteem maintenance. Part of how a person evaluates themselves comes from the standing of the groups they belong to, so raising the relative position of the group raises the self. The implication for politics is direct: a political identity is not primarily an information-processing device. It is a component of the self, and evidence that threatens it is processed as a personal attack rather than as a correction.

Jonathan Haidt's work adds the moral dimension. In The Righteous Mind (2012) he argues that moral judgement is largely intuitive, with conscious reasoning functioning as a press secretary that produces justifications after the fact. He identifies several moral foundations — care, fairness, loyalty, authority, sanctity, liberty — and finds that self-described liberals weight the first two heavily while conservatives distribute weight more evenly across all of them. Whether the foundations are as cleanly separable as the theory claims remains contested, but the core observation is well supported: political opponents frequently are not disagreeing about facts. They are running different moral operating systems and each finds the other's priorities unintelligible.

Cleavages: why some identities become political and others do not

Countless group differences exist in any society and only a handful organise its politics. The classic account is Lipset and Rokkan's cleavage theory (1967), which traced Western European party systems to four historical conflicts: centre against periphery, church against state, land against industry, and owner against worker. Party systems, on this account, froze around the cleavages that were salient when mass suffrage arrived, which is why they proved so stable for so long.

The freezing has clearly ended. Across most democracies the class cleavage has weakened sharply — parties of the left in many countries now draw more support from university graduates than from manual workers — while a new cleavage has hardened. It is variously labelled cosmopolitan against communitarian, open against closed, or anywhere against somewhere. Its markers are educational attainment, population density and geographic mobility rather than income.

This realignment explains a pattern that puzzles many observers: parties that historically represented workers now often hold positions on immigration and national identity that a substantial part of their traditional base rejects, while parties of the economic right have gained support in exactly those communities. The economic axis did not disappear. It was crossed by a second axis that cuts through it — which is precisely what a two-dimensional compass is built to show, and why the older single left-right line has become so misleading.

Sorting and the rise of affective polarisation

Lilliana Mason's Uncivil Agreement (2018) makes a distinction that clarifies a great deal of confusion about polarisation. Issue polarisation is disagreement about policy. Affective polarisation is dislike of the opposing group. In most democracies the second has risen far faster than the first.

The mechanism is alignment rather than extremity. Historically, identities were cross-cutting: a person's religion, region, occupation, union membership and party did not line up neatly, so almost everyone shared some important identity with almost everyone else. Those overlaps functioned as social shock absorbers. As identities have sorted — as party has become predictable from religion, geography, education and consumption — the overlaps have thinned. Opponents are now out-group on every dimension at once.

The consequence is that measures of hostility toward the other party have risen sharply even in periods when measured policy distance moved little. People who agree on more than they think can dislike each other more than ever, because the dislike is not tracking the disagreement.

What this means for anyone taking a political test

Two practical implications follow, and both are worth holding onto.

First, a result is a snapshot of expressed positions, not a reading of a fixed inner nature. If some of your answers were shaped by group attachment rather than by independent evaluation — and for nearly everyone some of them were — the result reflects that too. This is not a flaw in the instrument. It is an accurate measurement of how political opinion actually forms.

Second, the most useful thing a compass can do is decouple axes that identity has fused. Group membership tends to deliver a package: adopt the party and you adopt its full position set. Answering fifteen questions one at a time, without a party label attached, sometimes reveals that your economic and social positions do not sit in the quadrant your affiliation would predict. Users report this often, and it is arguably the strongest reason to take such a test at all.

Can the effect be resisted?

Partially, and the interventions that work are less inspiring than the ones usually recommended. Simply presenting corrective facts frequently fails, and can occasionally strengthen the original position when it is identity-relevant. What has better experimental support is more modest:

• Self-affirmation on an unrelated dimension before encountering threatening evidence, which measurably reduces defensive processing.
• Requiring a person to explain a policy mechanism in detail, which reliably moderates confidence — people discover they cannot actually explain how the policy would work.
• Sustained contact with individuals from the out-group under conditions of equal standing and shared purpose, which reduces hostility more effectively than any argument.
• Making a cross-cutting identity salient — a shared profession, city, or interest — which temporarily restores the overlaps that sorting removed.

What these have in common is that none of them works by winning an argument. They work by lowering the identity stakes of the argument first.

Conclusion

Identity is not a distortion layered on top of real political reasoning. For most people it is the substrate on which political reasoning runs. That is not a claim that voters are irrational — attaching to a group and adopting its heuristics is a defensible strategy for a person with limited time facing an enormous number of complex questions. But it means that political argument aimed purely at the level of evidence and policy will keep failing to explain why it is not working. The disagreement is usually not where the argument is being held.