Here’s the folk theory of how elections are supposed to work.
Citizens care about the country. They examine the issues facing it. They form views about which policies would best address those issues. They assess which candidates and parties are most likely to implement the policies they support. They evaluate how current office-holders have performed — have economic conditions improved, has crime gone up or down, have promises been kept. They weight this evidence and vote accordingly.
When something significant happens — a recession, a pandemic, a war, a major policy failure — citizens update their votes. They reward parties that handled difficulty well and punish parties that handled it poorly. The collective result is a system in which governments are held accountable for performance, and in which the public will on substantive issues gets translated, through elections, into policy.
This folk theory has problems. It’s roughly what most schools teach, and most election coverage assumes. Unfortunately, when political scientists have examined whether voters actually behave this way, the answer has consistently been not really. The gap between how people vote and how the folk theory says they vote is one of the most robustly documented findings in political science — and it has important implications for how you understand your own vote, and for what democracy is actually doing.
The book that forced the conversation
The most comprehensive modern articulation of this evidence came from two American political scientists, Christopher Achen and Larry Bartels, in their 2016 book Democracy for Realists. The book pulled together decades of research on voter behaviour and argued, bluntly, that the folk theory of democracy is empirically wrong in its central claims.
What the evidence shows, according to Achen and Bartels:
Voters mostly don’t hold coherent policy preferences. Surveys find that on most political questions, a large fraction of voters don’t have stable views. Asked the same question a few months apart, they give different answers. Asked to explain their views, they often can’t. Asked about the specific policies their preferred party favours, they often misidentify them. The image of the informed voter carefully considering policy options bears little resemblance to how most voters actually relate to policy.
Voters largely don’t hold elected officials accountable for performance. One would expect, under the folk theory, that voters would reward governments when things go well and punish them when things go badly. In fact, Achen and Bartels document, voters often punish governments for events that are clearly beyond the government’s control — natural disasters, shark attacks, droughts. And voters often fail to punish governments for events clearly within their control — policy failures, economic mismanagement, broken promises. The accountability signal is there, but it’s noisy and often aimed at the wrong things.
Voters largely vote on group identity, not on policy. The strongest predictor of how someone votes, across most democratic elections, is not their views on the issues — which are often vague and unstable — but their membership in particular social groups. Religion, race, region, class, occupation, family tradition, education level. Voters pattern-match their identities to parties, and vote accordingly, regardless of whether specific party policies align with specific voter interests.
Achen and Bartels were not the first to make these arguments. Their book draws heavily on earlier research, particularly the foundational work of the American political scientist Philip Converse, whose 1964 paper The Nature of Belief Systems in Mass Publics had already documented that most voters don’t hold coherent ideological views. What Achen and Bartels did was gather the accumulated evidence, organise it into a comprehensive case, and insist that political theory and democratic institutions take the evidence seriously rather than continuing to assume a voter who doesn’t really exist.
The symbolic-politics tradition
A complementary research programme, developed by the American political psychologists David Sears and Donald Kinder, has examined what voters do vote on, given that they don’t vote on policy in the way the folk theory describes.
Their framework — sometimes called symbolic politics — suggests that voters respond to political messages primarily through emotional and identity-based associations rather than through substantive analysis. A voter hears a candidate make a specific policy claim. They don’t primarily evaluate the claim on its merits. They evaluate it based on: does this candidate seem like one of us? Does this message resonate with values I hold as part of my identity? Does it trigger positive or negative feelings associated with my group’s heroes and enemies?
This isn’t irrationality exactly; it’s rationality operating under conditions of high complexity and low information. Most voters cannot possibly evaluate policy claims directly — they don’t have the time, information or expertise. So they use identity as a proxy. If someone like me trusts this candidate, I can probably trust them too. If this candidate’s rhetoric makes my tribe uncomfortable, something about their proposals is probably wrong for my interests. The heuristic isn’t always accurate, but it’s tractable in a way that genuine policy evaluation isn’t.
What’s become clearer in the last two decades, particularly through the work of researchers like Lilliana Mason at Johns Hopkins, is that identity-based voting has intensified in many democratic countries. The different parties have become more clearly aligned with distinct social groups — different religions, races, regions, educational levels, urban-rural divisions. This is sometimes called affective polarisation — not primarily a difference in what the parties would do (which is often smaller than the rhetoric suggests) but in how intensely members of each party feel about members of the other. You’re not voting for your party because of specific policies. You’re voting for it because the other party represents people you don’t want to be in power.
The affective-polarisation finding is one of the more disturbing results in recent political psychology. It suggests that many democratic conflicts aren’t really about substance — about policy, about the measurable effects of different approaches on real problems. They’re about identity. And identity conflicts are much harder to resolve than policy conflicts, because they can’t be compromised or split the difference in the way substantive disagreements sometimes can.
The counter-view from deliberative democracy
Before accepting the bleak Achen-Bartels picture wholesale, an important counter-tradition. Researchers working on deliberative democracy have argued that voter behaviour looks different when voters are given different conditions than the ones ordinary elections offer.
The most sustained experimental programme here has come from the American political scientist James Fishkin at Stanford, who has run what he calls deliberative polls for several decades. The design: select a representative sample of ordinary citizens, bring them together for a weekend, give them detailed balanced information about specific policy questions, have them discuss the issues in small groups with trained moderators, and measure their views before and after.
Fishkin’s finding, across dozens of deliberative polls conducted in many countries on many issues, is that ordinary voters change their views substantially when given the chance to actually deliberate. The views that emerge from deliberative polls are more nuanced, less polarised, more policy-focused, and closer to what traditional democratic theory would expect. Voters, given conditions conducive to thinking, do think. The conditions just aren’t present in ordinary political life.
This matters because it suggests that the Achen-Bartels critique isn’t about voters being fundamentally incapable of substantive political reasoning. It’s about the conditions under which they usually encounter politics. Ordinary political life gives voters fragmented information, emotionally charged messaging, limited time, and strong group-identity signals. Under those conditions, the voting patterns the critique describes are what you’d expect. Change the conditions — through deliberative poll design, or through other institutions that support serious deliberation — and you get different voting patterns.
Whether this is a response to the Achen-Bartels argument or an extension of it is a real question. One reading: voters can reason well when given the chance, so democratic reform should focus on creating more chances. Another reading: voters can reason well in artificial conditions that real political life can’t reproduce at scale, so democratic design has to work with the voter behaviour we actually get rather than the voter behaviour we might hope for.
What this suggests practically
For readers about to vote for the first time — or reflecting on how they’ve voted in the past — the research has some uncomfortable implications worth holding.
Your own vote is probably less substantively informed than you feel it is. This isn’t an insult; it’s almost universal. Most people, examined carefully, turn out to hold their political positions for reasons that are more about identity and group membership than about careful policy analysis. This isn’t a reason to stop voting. It’s a reason to be appropriately humble about what your vote is actually measuring.
The specific issues you care about probably matter less to your vote than you think. Research on vote-switching suggests that even voters who hold strong positions on specific issues rarely switch parties based on those issues — they stay with their usual party and adjust their positions to match it, rather than adjusting their party allegiance to match their positions. If you notice yourself doing this, you’re doing what most voters do, and it’s worth being aware of.
Affective polarisation is something to resist in yourself. The tendency to experience politics primarily through dislike of the other side, rather than through positive commitment to your own side’s program, is strongest in societies with weakening information environments and intensifying identity alignments. It produces voters who vote against rather than for, and it’s corrosive to democratic quality. Noticing it in yourself is the start of counteracting it.
Small amounts of deliberate deliberation beat large amounts of passive news consumption. The Fishkin research suggests that focused engagement with specific policy questions — reading balanced treatments, considering counter-arguments, thinking through implications — produces more meaningful political opinions than any quantity of ambient media consumption. Most voters don’t do this on most issues. Doing it on even a few issues each year puts you in a tiny minority of citizens who actually form their views through something like the process the folk theory assumes.
The question that remains
The deepest and most honest reading of this research isn’t that democracy is a failure or that voters are stupid. It’s that democracy does different work from what the folk theory claims, and that the work it does is still genuinely valuable.
What democracy actually does, reliably, is allow incumbent parties to be removed from power. This is smaller than the folk theory’s picture of informed rule by the people, but it’s larger than it sounds. Countries that have this feature — the capacity to non-violently change their rulers — have, across many centuries, avoided many of the worst outcomes that countries without it have suffered. Peaceful transfers of power. Protection against sustained tyranny. Some floor of accountability against incompetence, even if not much ceiling on competence. These are real goods, and they’re what democracy delivers.
What democracy doesn’t reliably deliver is informed rule by a rational public responding to substantive evidence. It delivers something messier and less inspiring: collective choice among options offered by political elites, shaped by group identities, heavily influenced by factors outside the control of those who supposedly benefit from voting. This isn’t nothing — it’s quite a lot — but it isn’t the noble picture most democratic rhetoric depicts.
The question worth carrying, especially as you think about your own voting:
When you cast a vote, what are you actually voting for — the specific policies of the party you’re supporting, or your sense of which kind of people should be in charge? And is the answer one you can live with honestly?
Key research referenced: Christopher Achen and Larry Bartels, Democracy for Realists (2016); Philip Converse, “The Nature of Belief Systems in Mass Publics” (1964); David Sears and Donald Kinder on symbolic politics; Lilliana Mason, Uncivil Agreement (2018) on affective polarisation; James Fishkin’s deliberative-polling research (When the People Speak, 2009).