Duverger's Law

In brief

Maurice Duverger (1917 to 2014), a French political scientist and constitutional lawyer, observed a regularity so consistent that it came to bear his name. In elections held in single-member districts where the candidate with the most votes wins outright (the system usually called first-past-the-post, or single-member plurality), politics tends to settle into a contest between two major parties. Where seats are instead allocated in proportion to vote share across multi-member districts (proportional representation), many parties survive and coalition government becomes the norm. Duverger stated the first tendency so firmly that he called it something close to a sociological law; the second he treated as a strong tendency rather than an iron rule. The mechanism is not about ideology or national character. It is about arithmetic and the anticipations it breeds.

The full treatment

The problem it answers

Why do some democracies, such as the United States and the United Kingdom, orbit around two dominant parties, while others, such as the Netherlands or Israel, routinely seat a dozen? The naive answer points to culture or history. Duverger's answer is structural: the rules for turning votes into seats do much of the work. Change the counting rule and, over time, you change the shape of the party system. This makes electoral design one of the few levers by which a constitution can shape political competition in advance, which is why the finding matters far beyond academic taxonomy.

How it works: the two effects

Duverger identified two distinct forces, and the distinction is the heart of the model.

The mechanical effect is pure arithmetic. Under first-past-the-post, only the plurality winner in each district takes the seat. Every vote for a third-place party translates into no representation at all. A party that wins a steady fifteen percent of the vote spread evenly across the country can end up with almost no seats, because it comes second or third everywhere and first nowhere. The system systematically over-rewards the top two and punishes the rest, converting votes into wasted votes for anyone outside the leading pair.

The psychological effect is the response of rational actors who see the mechanical effect coming. Voters who prefer a third party but expect it to lose face a choice: cast a sincere vote that will be wasted, or vote strategically for the more acceptable of the two contenders who might actually win, to keep out the one they like least. Donors and ambitious politicians reason the same way, starving hopeless candidacies of money and talent. Over repeated elections, this anticipation hollows out the third party from within, and competition consolidates around two poles. The mechanical effect wastes the votes; the psychological effect teaches people not to cast them.

Proportional representation reverses both pressures. Because seats track vote share, a party with fifteen percent support wins roughly fifteen percent of the seats, so few votes are wasted, and a voter who backs a small party is not throwing the vote away. Small parties survive, new ones enter, and governments are usually coalitions.

Duverger and the strength of the claim

Duverger stated the plurality tendency with unusual confidence. In Political Parties (the 1954 English translation of his 1951 Les Partis Politiques) he wrote that the single-ballot plurality system favors the two-party system with a force approaching that of a true sociological law. He was more cautious about the reverse: proportional representation and the two-round majority system both tend toward multipartism, but he framed these as tendencies rather than laws. Later scholars formalized the asymmetry. William Riker, in a much-cited 1982 essay, kept the plurality claim as the "law" and demoted the PR claim to a "hypothesis," precisely because the exceptions to the plurality side are rarer and more explicable.

Distinctions that matter: the exceptions

The law is a tendency about national aggregates, and its most famous failures are instructive rather than fatal. The key exception is the regional third party. Duverger's mechanism operates district by district, so a party whose support is geographically concentrated escapes the trap: it comes first in its own region even while it comes nowhere nationally. Canada sustains multiple parties under first-past-the-post because regionally concentrated forces, such as the Bloc Quebecois, dominate particular provinces. India runs first-past-the-post yet hosts a sprawling multiparty system, because strong state-level parties win pluralities within their states. The United Kingdom shows the same logic: the Scottish National Party wins many Westminster seats on a modest national vote share because its votes are packed into Scotland, while a party with the same national share spread evenly would win almost nothing. The exceptions confirm the mechanism even as they break the aggregate prediction.

Lineage

The intellectual roots run through the study of how institutions shape political behavior, a tradition Montesquieu opened when he argued that constitutional forms mold political life. Duverger's specific contribution belongs to mid-century comparative political science. His 1951 book was a landmark attempt to treat parties as objects of systematic empirical study rather than as incidental features of constitutions. The idea that electoral rules have predictable consequences was sharpened by Douglas Rae in The Political Consequences of Electoral Laws (1967), an early cross-national statistical test, and reframed by William Riker (1982), who located the law in the history of the discipline and clarified what it did and did not claim. The most rigorous modern statement is Gary Cox's Making Votes Count (1997), which recast the whole phenomenon as a coordination problem: the number of viable competitors in a district tends toward the district magnitude plus one, a result that subsumes Duverger's two-party finding as the special case of single-member districts. In that form the law connects directly to game theory and equilibrium analysis, since strategic voting is exactly the kind of behavior a Nash equilibrium describes.

The strongest case for it

The evidence for the plurality half of the law is among the most robust in political science. Across dozens of countries and many decades, single-member plurality systems cluster tightly toward two effective parties at the national or district level, and the deviations are almost always explained by geographic concentration, exactly as the mechanism predicts. The model is parsimonious and causal: it names a precise arithmetic pressure and a precise behavioral response, and both are independently observable. Strategic voting under first-past-the-post is documented directly in survey and voting data, and the wasted-vote logic is not a hunch but a countable fact of any plurality count. The law also has design value. It tells constitution-writers that plurality rules tend to deliver a stable two-party system with clear single-party government and accountability, while proportional rules tend to deliver representation for minority views and many parties at the table. Few generalizations in the social sciences travel so well or rest on so transparent a mechanism.

The strongest case against it

The objections are serious, and most are directed at the law's precision rather than its existence.

William Riker himself argued that the proportional-representation half was never strong enough to count as a law, only a loose tendency, and that even the plurality half needed the escape clause for regional parties before it held. Critics press that this makes the theory hard to falsify: any two-party country confirms it, and any exception is explained away as geographic concentration.

Empirically, the fit is imperfect even on the plurality side. India has run first-past-the-post since independence and remains stubbornly multiparty at the national level, and Canada has sustained more than two competitive parties for over a century. Defenders answer that these are regional-concentration cases, but skeptics note that the "exception" then covers a great deal of the world's plurality-using population.

On the theoretical side, Gary Cox's coordination framework, while broadly vindicating Duverger, shows that the two-party outcome depends on demanding assumptions: that voters share common expectations about who is viable, that information is good, and that people vote strategically rather than expressively. Where those conditions fail, more than two viable candidates can persist even in single-member districts.

Finally, from within democratic theory, advocates of proportional representation argue that the very mechanism Duverger describes is a defect, not a virtue: it manufactures majorities that do not exist in the electorate, wastes the votes of large minorities, and can hand power to a party that won fewer votes than its rival. On this view the law is accurate but describes a distortion, and the tradeoff between the accountability of two-party government and the fuller representation of proportional systems is precisely what designers should argue about rather than treat as settled.

Where it stands now

Duverger's Law remains a fixture of comparative politics and one of the discipline's most durable empirical claims, taught in every course on electoral systems and invoked whenever a country debates reforming how it votes. The consensus has shifted toward Cox's more general and better-specified version: the law is real, but it is best understood as a claim about strategic coordination in each district, with the two-party national result following only under further conditions about geography and voter expectations. Practically, the finding sits at the center of live reform debates, from referendums on proportional or alternative-vote systems in Canada, New Zealand, and the United Kingdom, to arguments in the United States over whether plurality rules entrench a two-party duopoly. New Zealand's switch from first-past-the-post to a mixed-member proportional system in the 1990s, which promptly produced a durable multiparty parliament, stands as something close to a natural experiment in favor of the law's core claim.

Test yourself

Think of the last time you voted, or would have, for a candidate you liked less than another because the one you truly preferred "couldn't win." That reasoning is the psychological effect in action. Now ask: was your favored candidate genuinely unviable, or unviable only because everyone, including you, expected them to be? Duverger's mechanism runs partly on a self-fulfilling prophecy, and noticing when you are inside one is the whole point of understanding the law.