Tyranny of the majority: when the larger number decides what the smaller number may do

In plain words

The tyranny of the majority is the risk that the larger part of a population uses voting, law, or social pressure to take away the rights or fair treatment of a smaller part. The worry is as old as modern democracy: James Madison, Alexis de Tocqueville, and John Stuart Mill all wrote about it, and courts in many countries exist partly to answer it.

Here is the simple version. If 60 people want something and 40 people do not, majority rule says the 60 win. That is usually fair. But some questions should not be decided by counting. Should a group of neighbors be allowed to vote that one family cannot buy a house? Should most of a town vote on whether a minority religion may build its place of worship? Most people agree some things are off the table. The hard part is deciding who draws that line and how.

The problem, stated precisely

The tyranny of the majority is not simply “losing a vote.” Losing is normal and acceptable in a democracy. The problem has three narrower parts:

1. Rights put to a vote. A majority decides a question that concerns the basic rights or equal legal standing of a smaller group, such as religion, housing, marriage, or political participation.
2. Persistent losers. The same group loses again and again on questions central to its life, so the outcome is not a fair rotation of winners and losers.
3. Social pressure. Tocqueville and Mill also worried about a softer form: the pressure of majority opinion that makes people afraid to say what they think, even with no law involved.

What this page does not cover: who gets to vote at all is covered in ballot access and access and exclusion; outcomes skewed by uneven turnout are covered in unequal participation; and failures of the referendum device itself are covered in referendum failures.

Defined terms

• Majority rule. A decision rule under which the option with more than half of the votes cast wins. For two alternatives, Kenneth May proved in 1952 that majority rule is the only rule that satisfies four conditions together: it works for every possible set of votes, treats all voters alike, treats both options alike, and responds in the right direction when a vote shifts.⁠¹
• Faction. Madison’s term in The Federalist No. 10 for a group of citizens, “whether amounting to a majority or a minority of the whole,” united by “some common impulse of passion, or of interest,” that is adverse to the rights of others or to the community’s long-term interests.⁠²
• Counter-majoritarian check. Any institution that can stop a majority decision: a written bill of rights enforced by courts, a supermajority requirement, a second chamber, a federal structure, or a review of a ballot measure before it goes to voters.
• Judicial review. The power of courts to strike down laws, including laws adopted by voters directly, that conflict with a constitution.
• Discrete and insular minorities. A phrase from footnote 4 of a 1938 U.S. Supreme Court opinion, suggesting that “prejudice against discrete and insular minorities” may weaken the ordinary political protections minorities rely on.⁠³

Evidence

The founding arguments

Madison’s argument in The Federalist No. 10 is the classic statement. He warned that “When a majority is included in a faction, the form of popular government” enables it to sacrifice “both the public good and the rights of other citizens”.⁠² He argued that “a pure democracy,” which he defined as a small society whose citizens govern in person, “can admit of no cure for the mischiefs of faction,” and he offered a large republic with elected representatives as the “republican remedy”.⁠² In No. 51 he put it in one line: “If a majority be united by a common interest, the rights of the minority will be insecure”.⁠⁴

Tocqueville, writing about the United States in the 1830s, gave the problem its name in a chapter titled “Unlimited Power Of Majority, And Its Consequences.” He called it “an impious and an execrable maxim” that a people may do whatever it pleases, and he wrote: “I know no country in which there is so little true independence of mind and freedom of discussion as in America”.⁠⁵ Mill, in On Liberty (1859), recorded that “’the tyranny of the majority’ is now generally included among the evils against which society requires to be on its guard,” and added that protection is needed “also against the tyranny of the prevailing opinion and feeling”.⁠⁶

These are historical arguments, not measurements. They tell us what thoughtful people feared, not how often it happened.

Documented cases where a majority vote restricted a minority

California, 1964. Voters adopted Proposition 14, which added a provision to the California Constitution barring the state from limiting a property owner’s right to refuse to sell or rent housing to anyone “in his absolute discretion.” According to the dissent in the later case, it passed 4,526,460 to 2,395,747. In Reitman v. Mulkey (May 29, 1967), the U.S. Supreme Court held that the provision unconstitutionally involved the state in private racial discrimination.⁠⁷

Colorado, 1992. Voters adopted Amendment 2 by statewide referendum. It repealed local ordinances protecting people from discrimination based on sexual orientation and barred governments from adopting such protections. In Romer v. Evans (May 20, 1996), the Court struck it down 6 to 3, holding that “A State cannot so deem a class of persons a stranger to its laws”.⁠⁸

Switzerland, 2009. On November 29, 2009, Swiss voters accepted a popular initiative banning the construction of minarets, with 57.50 percent voting yes, 19.5 of 23 cantonal votes in favor, and turnout of 53.76 percent.⁠⁹ The ban affects one religious minority’s buildings and was adopted by the people directly.

Switzerland, 1959 to 1990. Women’s suffrage shows the problem in its starkest form, because the excluded group could not vote on its own exclusion. A 1959 federal referendum on women’s suffrage was rejected by 66.9 percent of voters. The February 7, 1971 referendum was approved by 65.7 percent of male voters. In one canton, Appenzell Innerrhoden, women gained the cantonal vote only after the Federal Supreme Court ruled in 1990 that their exclusion violated the constitution’s equality clause.⁠¹⁰

How often does this happen?

Single cases do not show a rate. The systematic evidence is mixed. Barbara Gamble’s 1997 study, as summarized by Hajnal, Gerber, and Louch, found that nationwide, initiatives restricting civil rights passed more often than other types of initiatives.⁠¹¹ Hajnal, Gerber, and Louch then examined California ballot propositions from 1978 using exit poll data. They found that racial and ethnic minorities, “in particular Latinos,” lose regularly on a small number of racially targeted propositions, but that these measures were “less than 5% of all ballot propositions,” and that across all propositions, most Latino, Asian American, and African American voters were on the winning side.⁠¹¹

Both findings can be true at once. Majorities rarely target minorities, but when a measure does target a minority, it tends to pass. A system that counts outcomes across all measures will look fair; a system that looks only at targeted measures will look dangerous. The denominator decides the story.

Causes and competing explanations

Several explanations compete, and they lead to different remedies.

• Interest. Madison’s explanation: a majority with a shared material interest will use its numbers to pursue it at others’ expense.⁠² The remedy is to make durable majorities harder to form, through size, federalism, and separated powers.
• Prejudice. The Carolene Products footnote suggests that prejudice against isolated groups can block the ordinary process of coalition and bargaining that normally protects minorities.⁠³ The remedy is closer judicial scrutiny when that condition holds.
• Direct versus representative decision. Some scholars argue that representatives, who must face many groups and explain themselves, are less likely to enact targeted measures than voters deciding one question in private. Others reply that legislatures have passed plenty of discriminatory laws themselves. The evidence above does not settle which channel is worse overall.
• Agenda. A targeted measure can only pass if it reaches the ballot. Who writes the question and how it qualifies matters as much as the vote, a point developed in agenda capture.
• Social conformity. Tocqueville’s and Mill’s softer concern is not about law at all but about fear of disapproval.⁠^5,6 Its remedy is cultural and procedural: protected private expression and a secret ballot.

Who is harmed

The direct harm falls on groups that are numerically small, geographically scattered, or socially disfavored at a given time: religious minorities, racial and ethnic minorities, sexual minorities, noncitizen residents, and, historically, women. The harm is not only the lost legal right. A measure adopted by popular vote can carry a message that the community, not just a government, rejects a group, a point the Romer majority made when it said a state cannot make a class of persons “a stranger to its laws”.⁠⁸

A second, wider harm falls on everyone. If people believe their rights depend on winning every vote, they have reason to treat each election as existential, which feeds polarization. And if majority opinion silences dissent, the whole public loses information it would have gained from hearing minority views.

What has been tried elsewhere

Entrenched bills of rights with judicial review. The U.S. model withdraws certain rights from ordinary politics. The Supreme Court’s 1943 Barnette opinion states: “The very purpose of a Bill of Rights was to withdraw certain subjects from the vicissitudes of political controversy”.⁠¹² Results: courts have struck down majority-adopted measures, as in Reitman and Romer.⁠^7,8 Failures: courts have also upheld exclusion for long periods and struck down protective laws; Waldron notes the striking down of 170 labor statutes by state and federal courts in the Lochner era.⁠¹³

Pre-vote validity checks. Switzerland requires its Federal Assembly to declare a popular initiative invalid, in whole or in part, if it violates rules on form or subject matter or “mandatory international law”.⁠¹⁴ Results: the check exists and is automatic. Limits: it is narrow. The minaret initiative passed this check and was adopted,⁠⁹ which shows that a pre-vote screen limited to mandatory international law will not stop many measures that minority groups experience as targeted.

Federal structure and courts working together. The Swiss suffrage history shows a federal referendum extending rights in 1971 and a court finishing the job in one canton in 1990.⁠¹⁰ Neither institution acting alone closed the gap.

Deliberation before voting. Citizens’ assemblies, which put a random sample of residents through structured learning before a recommendation, are one response; the deliberation research and comparative cases pages cover their record and limits. Deliberation can change opinions, but it does not by itself bind a later vote.

Competing interpretations

The rights-first view. Many legal scholars, and the Barnette Court, argue that some subjects should never be put to a vote.⁠¹² On this view, judicial review and entrenched rights are the price of a democracy that does not turn on its own members. Supporters point to Reitman and Romer as evidence that the check works.

The democratic-process view. Jeremy Waldron, in “The Core of the Case Against Judicial Review” (Yale Law Journal, 2006), argues that in a society with working democratic institutions, functioning courts, a general commitment to rights, and good-faith disagreement about what rights require, “judicial review of legislation is inappropriate as a mode of final decisionmaking”.⁠¹³ His point is not that majorities are always right, but that judges are also a small group whose views on contested rights questions are not obviously better, and that letting them decide moves power from the many to the few. His essay addresses the tyranny of the majority directly in its own section.⁠¹³

The self-government view from the right. Justice Scalia, dissenting in Romer, argued that the Court “has mistaken a Kulturkampf for a fit of spite” and that the question should be settled by “normal democratic means,” including state constitutional amendments.⁠⁸ Many conservatives see courts overriding voters on contested moral questions as itself a form of minority rule.

The fairness-of-counting view. May’s theorem shows majority rule is the only two-option rule that treats all voters and both options equally while responding to changes in votes.⁠¹ Any supermajority or veto rule protects the status quo, and the status quo is not always the side of the vulnerable. Swiss women in 1959 were harmed by a rule that kept the existing exclusion.⁠¹⁰

The empirical-skeptic view. Hajnal, Gerber, and Louch argue that critics have “overstated the detrimental effects of direct democracy” because most measures do not target minorities.⁠¹¹ Others respond that the right test is how targeted measures fare, not the overall average.

These views share more than they admit. Nearly all accept that some rights should be protected; they disagree about who decides which ones, and by what process.

Application to this framework

My Vote Forward separates three things that the tyranny-of-the-majority debate often blurs. Priorities say which subjects matter most; they are public in aggregate but are not decisions about anyone’s rights. Opinions are specific, voluntary, and revisable. Private voting is the official decision, which in a real ballot remains governed by the constitution and law of the jurisdiction.

That separation matters here. A published priority showing that most participants rank a subject highly does not mean most participants agree on what to do, and it gives no one a mandate to restrict anyone. The representation chapter covers why an official who knows the majority’s priorities still has a duty to explain decisions and protect minority rights. The proposals chapter covers how authority, legality, and review shape what can be on a ballot at all.

Unresolved questions

• Should a participation tool ever publish a breakdown that shows one group’s priorities against another’s, or does that invite the “us against them” framing it hopes to avoid? See published releases.
• Is a pre-vote legal review of questions more protective, or does it simply move agenda power to reviewers? See governance.
• How should a system weigh intensity, so that a minority that cares deeply about one issue is not always outvoted by a majority that cares mildly? See single-issue mobilization.
• What is the right denominator for measuring harm: all measures, or only those that target a group?