Referendum

A referendum is a direct vote by an electorate on a specific question, such as a law, a constitutional amendment, a treaty or the political status of a territory, rather than on candidates for office. It is one of the main instruments of direct democracy, alongside citizen initiatives and recall votes, and it may be binding or advisory depending on the legal system. Because it places a single issue before all eligible voters, it is often distinguished from legislative votes and from general elections, and its status varies considerably across constitutional traditions.

Referendums matter because they shape how sovereignty, federalism and representative government work in practice. They have been used since the nineteenth century to ratify constitutions, approve laws, settle border questions and decide membership in international organizations, and they remain central to debates over party competition and the role of citizens between elections. Their use has also raised persistent questions about voter information, turnout, minority rights and the limits of majority decision-making.

This entry first clarifies the terminology and the main types of referendum, then describes how they are initiated, how ballots are decided and how their historical origins and early spread unfolded. It next examines the Swiss system as the most extensively studied case and surveys practice in other countries and regions. The entry then turns to participation and campaigning, the relationship between referendums and representative institutions, the use of plebiscites on self-determination and territory, and the legal status of referendums together with the main criticisms they attract.

Definition and terminology

In its most general sense, a referendum is a direct vote in which the electorate decides on a specific question, usually a law, a constitutional text or a policy measure, rather than choosing a party or individual candidate. The defining feature is the object of the vote: a proposal is put to citizens as a question in itself, so the result answers that question and does not allocate offices. This distinguishes referendums from elections, even though both are forms of direct or representative democratic participation and both can take place on the same day. Definitions beyond this core vary, because each jurisdiction defines the instrument in its own constitution or statutes. Some systems, for example, reserve the term for votes on constitutional amendments, while others apply it to any popular vote on legislation.

The word "plebiscite" has a separate history. It derives from the Latin plebiscita, the resolutions of the Roman plebs, and it was applied to popular consultations in revolutionary France from 1793. Its modern meaning was shaped by Napoleon I and Napoleon III, who used votes to confirm their authority, and by the peace settlements after world-war-i, which used plebiscites to settle disputed borders. In these usages the plebiscite tends to be a yes-or-no endorsement of a ruler or a territorial arrangement, often one that is ad hoc and that the government controls. Because of this association, some English-speaking usage treats "plebiscite" as a pejorative term for votes held under undemocratic conditions. Other usage treats it as a synonym for a referendum. In Australia, for example, the word has been used both for a simple national vote and, in the constitutional context, for a distinct procedure. Scholars generally agree that no clear line separates the two terms, and the difference is mostly one of connotation and context.

Terms in German law

German usage shows how terminology follows statute rather than universal theory. A Volksbefragung is a non-binding consultation of the people, while a Volksentscheid is a usually binding decision on a bill or a policy. The preliminary stage, in which signatures are collected to force a question onto the agenda, is the Volksbegehren, or Volksinitiative. At the municipal level the same triad is known as Bürgerbegehren (citizens' initiative), Bürgerbefragung (non-binding local poll) and Bürgerentscheid (binding local decision). Because the Land (state) and municipal instruments are regulated by separate laws, the terms do not map neatly onto one another. Some city-states use different names for their state-level initiatives, and the official designation in Bremen is Volksentscheid even at municipal level. In everyday German, Volksentscheid, Volksabstimmung and Referendum are often used interchangeably, although in Baden-Württemberg and Switzerland Volksabstimmung is more common. The national level is distinctive. The Basic Law provides for a federal popular vote only in the case of reorganising the territory of the federal states, so the general term "referendum" in German politics mostly refers to state or local procedures.

Related instruments

Referendums are usually distinguished from several neighbouring devices. A citizen initiative lets voters place a proposal on the agenda by petition, and in many systems it leads to a vote only if the legislature does not adopt it; in this case the initiative is the trigger and the referendum is the outcome. A referendum may be held on a law already passed by the legislature, in which case it is sometimes described as a veto or repeal mechanism. A legislative referral is a vote that the legislature itself decides to send to the electorate, which is the more common form in some American states. A recall election, by contrast, asks voters whether to remove an officeholder before the end of a term, so it concerns persons and not policies, which places it closer to an election than to a referendum. Taken together, these instruments show that referendums are defined less by a single feature than by their function as direct votes on issues.

Types by authority and legal effect

Referendums are usually classified along two axes: whether a vote must be held or may be called (mandatory or optional), and whether its result has legal force (binding or consultative). Most national and subnational systems combine these in different ways, so a single term can describe very different procedures.

Mandatory referendums

A mandatory referendum, also called an obligatory referendum, must be held when a legal rule requires it, and it normally needs no petition or request. The most common example is a vote on constitutional change. Constitutions in many countries cannot be amended without popular approval. In Switzerland, the federal constitution requires a nationwide vote on every revision approved by parliament. It also requires votes on joining collective security organisations or supranational communities, and on emergency federal laws that depart from the constitution and remain in force for more than a year. Ireland and Denmark also require a popular vote for constitutional amendments, and a single encyclopedic summary reports that nearly every U.S. state does so as well, with Delaware as the exception.

Mandatory votes can also cover major public spending. Some Swiss cantons hold a "financial" referendum for expenditure above a set amount, and each canton's constitution decides which matters require a vote. Because the trigger is written into law, these votes are generally binding and do not depend on signature drives.

Optional referendums

An optional referendum is held only when a qualifying actor asks for it. Its usual purpose is to review a law the legislature has already passed. The Venice Commission's comparative report distinguishes this from a popular initiative: an optional referendum challenges a text already approved by a state body, whereas an initiative lets citizens propose a text that no authority has yet approved.

Switzerland is the standard case. Federal acts and other enactments of the Federal Assembly can be put to a nationwide vote if 50,000 valid signatures are collected within 100 days of publication. A request from eight cantons also triggers a vote. Optional referendums can exist at cantonal and municipal level as well. In other systems the trigger varies. In Hungary, according to the same comparative report, a referendum on a law launched by a popular initiative of 200,000 citizens is always binding, while in other cases parliament decides whether a referendum will be held. Italy's abrogative referendum, which repeals an existing law, is another optional instrument. Its result is valid only if turnout passes a threshold, a point covered in the section on practice across countries.

Binding and consultative votes

A binding referendum creates a legal obligation: the government or legislature must implement the outcome. A consultative (advisory) referendum serves only as guidance to decision-makers. The distinction is set by each jurisdiction's law, not by a common international standard. The Venice Commission found that several states provide only decision-making referendums, among them France, Ireland, Italy and Switzerland. Denmark makes decision-making votes the rule but does not exclude consultative ones. Belgium treats legally binding referendums as unconstitutional, and its 1950 consultative vote was held on the basis of a specific parliamentary decision, a step whose constitutionality has been strongly disputed. The Netherlands held a consultative vote on the European Constitutional Treaty in 2005 under an ad hoc law.

A non-binding result has no direct legal effect, but it can still carry strong political weight. In the United Kingdom, the statute for the 2014 Scottish independence referendum and the 2016 Brexit referendum was silent on the consequences of the result. Commentators note that a consultative vote can be treated as binding in practice, even though it does not bind in law. Where a statute does attach a legal consequence, the result takes effect automatically. Under the planned 2011 UK Alternative Vote referendum, for example, a statutory duty would have required ministers to bring the new system into effect had voters approved it.

Terminology reflects these differences. In Germany, a Volksbefragung is a non-binding ballot question, while a Volksentscheid is a binding plebiscite. At municipal level the parallel terms are Bürgerbefragung and Bürgerentscheid.

Classification by subject and level of government

Referendums can also be sorted by subject matter and by level of government. Constitutional referendums are often mandatory and binding, as in Switzerland and Ireland. Legislative referendums are more often optional, because they respond to a particular law. Territorial referendums, such as those on secession or on the status of a region, are usually ad hoc votes created for a specific question. Local referendums appear at municipal or district level and frequently follow the same mandatory and optional patterns as national votes. In Swiss cantons and communes, for example, both mandatory votes and optional referendums on local decisions are available. Federal systems such as Switzerland and the United States therefore have referendums at several levels at once, each governed by its own rules.

Initiation, ballot design and decision rules

Who can trigger a vote

Whether a referendum takes place often depends on who holds the power to call it. In some systems the constitution or the legislature decides, while in others a sufficiently large group of citizens can force the question onto the ballot. Switzerland shows the signature model most clearly. Citizens who oppose a law already passed by parliament can launch an optional referendum by gathering 50,000 valid signatures within 100 days of the legislation's publication. A popular initiative to change the constitution has a higher bar, requiring 100,000 signatures within 18 months. A mandatory referendum needs no signatures at all, because all constitutional amendments approved by Parliament are subject to a mandatory referendum, i.e. they must be put to a nationwide popular vote. Organisers usually aim well above the minimum, since not all collected signatures are found to be valid.

Italy relies on a different gatekeeping model for its abrogative referendum, which can be requested by 500,000 electors or five Regional Councils. Since 2021 signatures may also be gathered through a digital identity system, which has allowed some campaigns to reach the threshold very quickly. Petitions do not automatically reach the ballot. The Court of Cassation first checks the petition under ordinary law, and the Constitutional Court then decides admissibility, so that many fully valid petitions with the necessary 500,000 signatures have never been accepted as referendums. In 2017 the Constitutional Court ruled one CGIL question inadmissible despite about 3.3 million signatures, showing how judicial review can narrow the agenda.

Ballot design and official information

The wording of a question is one of the most contested stages, because framing can steer answers. In many systems the question is set by the statute or resolution that calls the vote, and the election authority then prints and administers it. In Australia, for example, voters write either 'Yes' or 'No' in the box opposite each question on the ballot paper, and the rules are contained in the Referendum (Machinery Provisions) Act 1984. Many countries also provide an official explanatory or information booklet to voters, and some assign the drafting of arguments or the review of misleading wording to an independent commission or to the courts. The degree of official neutrality in these documents is a frequent point of dispute.

Decision rules: simple, double and qualified majorities

The simplest rule is a majority of votes cast. Swiss optional referendums on federal laws work this way: they pass with a simple majority of voters. For constitutional questions, many systems require more than a headcount. Switzerland applies a double majority, under which a majority of voters across the country vote in favour of the referendum and it achieves a majority in a majority of cantons. Australia uses a similar test, where a proposal must win a national majority of electors in the states and territories and a majority of electors in a majority of the states (i.e. at least four out of six states). The design protects smaller units: one reason a double majority is needed is to prevent voters in the bigger states from changing the Constitution against the wishes of voters in the smaller states. The result is a strong bias toward the status quo, as only eight out of 44 proposals to amend the Constitution have been approved since Federation.

Turnout quorums and strategic abstention

A quorum sets a minimum turnout, so a result is valid only if enough electors participate. Italy's abrogative referendum requires a 50% participation quorum, calculated for each referendum question individually. The outcomes show how decisive this rule is. In 2011 the quorum was reached, with a turnout of 56.9%, and all four questions passed. In 2022, by contrast, all five referendums were declared invalid due to turnout below the 50% threshold, and the 2025 vote reached only about 30 percent turnout, leaving the referendum void.

Quorums create incentives for strategic abstention. Opponents of a proposal who expect it to fail on turnout can gain by staying home rather than voting no, and this makes the threshold a political instrument in its own right. Italian commentary notes that Council President Matteo Renzi's call to abstain was one such case of party leaders advocating non-participation. Validity thresholds therefore affect campaign strategy as much as the formal decision rule.

Campaign regulation

Campaign rules are generally designed to ensure that both sides can communicate and that public resources do not tilt the result. Common tools include caps on campaign spending, registration or designation of official campaign groups, and disclosure of donations. In the United Kingdom, the Political Parties, Elections and Referendums Act 2000 introduced spending limits for referendum campaigns and a system of designated lead campaigners. Broadcasting rules may require balanced airtime, and many jurisdictions restrict government publicity during a referendum period, so that officials do not use public funds to advocate one side. Such restrictions vary considerably, and their enforcement, including the role of courts and election commissions, is a recurring source of controversy.

Origins and early history

Direct votes on public questions predate the modern referendum by many centuries, but the earlier forms differ from it in important ways. In the Roman Republic, a plebiscitum was a resolution passed by the plebeian assembly, which voted by tribe. It was distinct from a senatus consultum, a decree of the Senate. Roman tradition holds that the Lex Hortensia of 287 BC made such resolutions binding on the whole citizen body, though historians debate how that law worked in practice. The word "plebiscite" comes from this institution, but a Roman plebiscite was a product of class conflict between patricians and plebeians and was not a popular check on a text drafted by a separate authority. The Athenian ekklesia was a general assembly of adult male citizens that decided on laws, decrees, war and treaties, and it used secret ballots in procedures such as ostracism. Its participants were a small minority of residents, since women, slaves and foreigners were excluded. Treating these bodies as ancestors of the modern referendum therefore requires caution, because they combined legislation, deliberation and executive decisions in one assembly, and their continuity with later practice is more a matter of analogy than of institutional lineage.

The modern use of popular votes on texts emerged with the late eighteenth-century idea that a constitution derives its authority from the people. In Massachusetts, a draft constitution was put to the towns in 1778 and rejected, and a revised text approved in 1780 was ratified by popular vote. This procedure became a reference point for American state constitutions. In revolutionary France, the constitution of 1793 was submitted to the electorate and approved, though it was never put into effect, and the constitutions of 1795 and 1800 were likewise ratified by popular vote. The French votes were framed as acts of the nation exercising its sovereignty, but they were conducted under conditions that gave the government considerable control over the question and the result.

Switzerland provides the clearest early development of the instruments that became the referendum. After the Regeneration movement of the early 1830s, several cantons adopted a "veto" that let citizens contest a law already passed by the cantonal parliament. St. Gallen introduced this device in 1831, in the wake of mass assemblies that forced a new constitution. Between 1831 and 1852, six cantons introduced various forms of veto, many of them inspired by the Montagnard constitution of 1793. The veto required a set number of signatures within a fixed period, and non-voters were counted as either approving or rejecting depending on the canton. These rules made success rare. Graubünden (1854) and Basel-Landschaft (1863) adopted the first lasting obligatory referendums on legislation, and several cantons, including Solothurn (1856) and St. Gallen (1861), moved from the veto to the optional referendum. Basel-Landschaft had earlier adopted an obligatory referendum on constitutional amendments in 1832.

The nineteenth-century plebiscite is closely associated with Napoleon III. The coup of 2 December 1851 was followed within weeks by a plebiscite on 20–21 December asking voters whether they wished to maintain the authority of Louis-Napoleon and delegate to him the powers needed to write a constitution. The yes vote was about 92 percent. A second plebiscite in November 1852 restored the empire, and the final one on 8 May 1870 approved liberal reforms and a new constitution, with about 7.35 million yes votes against about 1.54 million no votes. Napoleon III presented these votes as a direct bond between the sovereign and the people, which allowed him to bypass parliamentary and party intermediaries. Later authoritarian regimes, including Nazi Germany with its plebiscites in 1934 and 1938, used the same device to display popular assent. As a result, the plebiscite acquired strong associations with acclamation rather than genuine choice, and in many constitutional traditions the term "referendum" became preferred for votes held under established legal rules.

Switzerland as the central case

Federal and cantonal structure

Switzerland runs direct-democratic votes at federal, cantonal and communal levels, and this layering accounts for how often and how varied its ballots are. All the Swiss cantons provide for legislative referendums on legislation passed by the cantonal parliament, although in different cantons these may be mandatory or optional. Some cantons also subject all cantonal laws to a compulsory vote, as well as any expenditure above a set amount, which is known as a "financial" referendum. Local optional referendums can also take place at cantonal and municipal level with fewer signatures than federal ones. The instrument predates the federal state: the popular referendum was first introduced in the cantons, for example in the Canton of Zurich in 1869. At federal level, the optional referendum was introduced by the Federal Constitution of 1874.

Mandatory votes on constitutional amendments

Mandatory referendums are held without any citizen trigger. Signatures do not have to be collected, because the vote is compulsory. This always happens when Parliament proposes to amend the federal constitution. Article 140 of the federal constitution also covers membership in collective security organisations such as NATO, membership in supranational communities such as the EU, and federal emergency laws that lack a constitutional basis and remain in force for more than a year. Those items must therefore go to a popular vote. Such measures need what is called a double majority: a majority of the electorate and a majority of the cantons. Cantonal half-cantons each carry half a cantonal vote.

Optional votes on laws and popular initiatives

Optional referendums work in the opposite direction, as a challenge from below. Any member of the electorate opposed to a new or amended federal law can launch one, including citizens living abroad. In practice the threshold is 50,000 signatures collected within 100 days, or a request from eight cantons. The threshold was not always this high: the initial requirement was 30,000 signatures, and after population growth and the introduction of women's suffrage in 1971, it was raised to 50,000 valid signatures in 1977. The first optional federal referendum was held in 1875, when all previous federal referendums had been mandatory. Cantons did not use their own route until 2004, when the first referendum initiated by the cantons was held.

Optional votes need only a popular majority, while federal popular initiatives, which always propose constitutional amendments, require both a majority of the popular vote and a majority of the cantons. The initiative route requires 100,000 signatures from supporters. When Parliament answers an initiative with its own version, that counter-proposal can appear on the same ballot, so voters may face several options at once.

Turnout variation

Turnout differs widely from one vote to the next. Turnout in Swiss referendums fluctuates considerably, ranging from 30% to 80%. Issues with high salience draw more voters: the 1992 vote on possible EEA membership reached about 78 percent turnout. At the low end, the 2010 vote on unemployment insurance revision recorded a turnout of 35.84 percent of registered voters. Historical averages also shifted over time: turnout was approximately 62 percent just after World War II, but fell to between 35 and 45 percent in the decades leading up to the 1990s. Research on the period up to 2022 has found only a small gap between passed and failed votes in average turnout.

Consensus politics and institutional design

The main scholarly argument about Swiss referendums concerns their indirect effects. Their chief influence lies less in the votes themselves than in the bargaining that precedes them. Researchers argue that the referendum is a strong incentive, even a constraint, to cooperate in the form of an oversized coalition, because the risk of defeat in a popular vote would otherwise be too high. The same logic underlies the concordance system in the federal executive, since a sizeable opposition could in principle "paralyse" the government by submitting too many referendum proposals.

The mechanism also worked through co-optation. After conservatives used the referendum, introduced in 1874, to block decisions by the liberal majority, they were eventually admitted to the government, with the first non-liberal minister elected in 1892. Parties admitted in this way were expected in return to support legislative compromises strong enough to survive a referendum. In more recent cases, the threat persists: the Federal Council needs a broad consensus on international agreements to avoid a negative vote in an optional referendum or the launch of a popular initiative. Scholars also note that a referendum allows the government to be pressured into consensus with groups outside Parliament, so as to prevent those groups from overturning new legislation. Because the threat is hard to measure directly, the evidence is largely drawn from case studies and institutional analysis rather than from controlled comparison.

Practice across countries and regions

Germany

Germany is unusual among EU members because it has never held a nationwide vote. Germany belongs to one of the few countries in the EU, which have so far no experience of national-wide referenda. The Basic Law contains only one explicit federal provision: a federal statute that redraws the boundaries of the Länder must be confirmed by a vote in the affected territories (Article 29). A petition backed by one-tenth of eligible voters in a contiguous area of at least one million residents can force Parliament to choose between a boundary change and an advisory vote in the affected Länder. Practice under this provision has been thin. A 1951 merger of Baden and Württemberg was approved, while a 1996 proposal to merge Berlin and Brandenburg was rejected.

Most direct democracy operates below the federal level, under the terms Volksbegehren (initiative), Volksentscheid (binding decision) and Volksbefragung (non-binding question) for the Länder. On the municipal level the three types are paralleled with Bürgerbegehren, Bürgerbefragung and Bürgerentscheid. Some Länder originally offered general binding votes on popular initiatives, and Hesse and Bavaria also having a mandatory binding referendum on changes to the state constitution. The Berlin campaign to expropriate large residential landlords illustrates the system's reach. Its initiative gathered a record number of signatures, and the 2021 ballot passed with 59.1% of the total vote in favour. A thirteen-member expert commission then examined its constitutionality under Article 15 and reported in 2023 that expropriation with compensation below market value could be permitted.

United States

In the United States, direct democracy is organised by the states rather than the federal government. Roughly two dozen states allow citizen-initiated statutes or constitutional amendments, and most of them also provide for popular referendums on legislation passed by the legislature. The procedures vary considerably in signature thresholds, subject restrictions and the approval rules used. By contrast, legislatures frequently place constitutional amendments on the ballot themselves, but purely advisory referrals that ask voters for an opinion without legal effect are uncommon.

United Kingdom

The United Kingdom has held only two nationwide referendums. The 1975 vote asked whether to remain in the European Community, and it produced a clear majority for staying in, with roughly two-thirds voting yes. The vote was consultative in legal terms, and it did not displace the doctrine of parliamentary sovereignty. The 2016 vote on leaving the European Union was also advisory under the European Union Referendum Act 2015, but its result, 51.9 percent for leaving on a turnout of about 72 percent, set off a constitutional dispute. In the UK vote was postponed indefinitely is not the relevant precedent here; the more important ruling was the 2017 decision of the Supreme Court in R (Miller), which held that the government could not trigger Article 50 of the Treaty on European Union without an Act of Parliament. The subsequent Brexit process therefore rested on legislation rather than on the referendum result alone.

Italy

Italy's abrogative referendums, provided for by Article 75 of the Constitution, ask voters whether to repeal an existing statute. As abrogative referendums, they require the quorum to be met in order to be valid: at least 50% + 1 of eligible voters must go to the polls. The rule has produced repeated failures. In 1990, a referendum failed to reach the quorum for the first time since the 1948 Constitution, even though both referendums had a "yes" victory with more than 90% of votes. Later votes followed the same pattern: turnout was below 50% in 1999, reached only 23.31 percent in 2009, and was declared invalid due to turnout below the 50% threshold in 2022. The 2011 vote on water, nuclear power and ministerial immunity met the quorum with a turnout of 56.9%. That outcome owed much to the campaign, since Silvio Berlusconi, the then-Prime Minister, implicitly invited to boycott the vote in the hope that the required quorum would not be met. The structure encourages strategic abstention. Those favouring the status quo can win either by preventing the quorum being reached or by emerging ahead among voters.

Post-Soviet states

Several post-Soviet states have used referendums to approve constitutions and to alter the terms of incumbent leaders. Observers have repeatedly criticised such votes for weak competition, limited campaign freedoms and reported turnout and approval figures that are difficult to verify independently. Validity thresholds and quorum rules in these states have also been a recurring point of contention.

The 2005 EU constitutional treaty votes

The 2005 votes on the European Constitutional Treaty showed how referendums on international agreements could produce unexpected results. In France, 55% of voters rejected the treaty on a turnout of 69%. Three days later, the Netherlands held a consultative vote in which 61.54% voted against on a turnout of 63.3%. That Dutch vote was the first national referendum in more than two hundred years, and the government had said it would abide by a decisive result provided turnout exceeded 30 percent. Following the two rejections, the Danish, Irish, Polish and Portuguese votes were cancelled, and the planned British vote was postponed indefinitely. Luxembourg, by contrast, approved the treaty, by 57% to 43%.

Participation, information and campaigns

Determinants of turnout

Turnout in direct votes depends on several layers of factors. A study of 43 referendums in the Swiss canton of Geneva, which matched official individual participation records with ballot and community data, found that context-related effects were significant, stemming mainly from the referendum rather than from the community level. Even so, the main driver of direct democratic participation is individual determinants, in particular citizens' past participation record. Community variables such as wealth and urbanization were included, but the referendum-level factors, such as campaign intensity and the importance of the issue, carried more weight.

Issue salience is expected to matter, but the evidence is uneven. The question of Switzerland's relationship to the European Union should be more important to citizens than a minor change in a tax regulation for firms. Yet while the importance of ballot measures should intuitively influence turnout, the empirical evidence is mixed. Campaign intensity is the more consistently supported factor, since there is widespread agreement that intensive campaigns foster turnout.

Turnout also bears on legitimacy. In a survey-based study of European referendums, the likelihood that citizens think a result should be followed drops on average by 19 percentage points when the turnout level is 35% compared to when the turnout level is not known.

Complexity and the 2005 constitutional treaty votes

Complexity is another proposed depressant of turnout. Studies in Switzerland show that turnout decreases with the complexity of the ballot proposal. The argument rests on information costs: citizens may decide that acquiring "encyclopedic" information is not a worthwhile activity, and while one option is to rely on information shortcuts, an even easier solution is not to turn out at all. Complexity also rises when several proposals share a ballot day, because voters need to inform themselves about all respective topics and not just one. A long-run Swiss study of 1848–2009 found that when more referenda are held on the same day, constituents are more likely to refer to parliamentary recommendations when making their decisions.

The 2005 referendums on the Treaty establishing a Constitution for Europe are often cited in this context. Many citizens in France and the Netherlands reported not voting because the issue was too complex, according to survey evidence. Turnout, however, was far from low in these cases. In France, 55% of voters rejected the treaty on a turnout of 69%. In the Netherlands, the consultative vote produced a 61.5% "No" on a turnout of 63.3%, and the government had pledged to be guided by the referendum if turnout was over 30%. Across the wider set of 2005 votes, turnout ranged widely. Spain's approval, for example, came with 81.8% in favour and turnout of 41.8%, while Luxembourg approved the Constitution by 57% to 43%. The variation suggests that national campaigns and political context shaped participation as much as the text's complexity.

Information shortcuts and elite cues

When full information is costly, voters rely on heuristics. Political scientists often describe these as information shortcuts, and the Swiss evidence suggests one common shortcut is the recommendation of parliament. When constituents face a higher level of complexity, they follow the parliamentary recommendations rather than those of interest groups. Researchers treat the strength of party division as a measure of campaign intensity. One measure in Geneva studies compares the vote share of parties that recommend accepting a ballot and those that recommend rejecting it. Where all parties line up on one side, cues are unified and competition is low; where they split, voters receive conflicting signals and campaigns tend to be more intense.

Differential effects of campaigns

Campaign effects are not uniform across voters. Research on Geneva's 43 votes found that while intensive political campaigns overall foster citizens to turn out to vote, they do so especially for "selective" (or "intermittent") voters, who need to decide anew at each ballot whether to turn out or not. By contrast, frequent voters are largely unaffected by campaign effort. Frequent abstainers are not immune: they get almost as strongly mobilized as selective voters in highly intensive campaigns. Occasional voters appear especially sensitive, since the higher the campaign intensity, the higher the likelihood that occasional voters participate. Comparative work from the United States and Switzerland further indicates that campaign efforts have differential effects for citizens who are in doubt about whether or not to participate, in comparison to definite voters or abstainers.

Relationship to representative government and parties

A referendum can either supplement or challenge parliamentary government. As a supplement, it adds a check on measures that an elected legislature has already passed, or it settles a question the legislature has found hard to decide. As a challenge, it takes a question out of the chamber's hands and gives the outcome a claim to authority that rivals that of the legislature. The tension is sharpest when the result is binding and the government must implement a vote that most members of parliament opposed, which raises questions about parliamentary sovereignty.

Governments have also used referendums to manage their own parties and coalitions, and the strategy can backfire. The clearest recent example is the United Kingdom. By 2013, David Cameron judged that renegotiating membership of the EU and holding a referendum on continued membership was the best way to resolve the divisions in his own party. Under pressure before the 2015 general election, he committed to an in-out referendum by the end of 2017. The vote did not end the dispute. The intra-party divide was laid bare in the 2016 EU referendum, when 186 Conservative MPs voted Remain and 141 voted Leave. Cameron resigned after losing the national referendum to leave the European Union. A referendum called to enhance a leader's legitimacy can likewise become a verdict on the leader. In Italy, Matteo Renzi had staked his political future on the outcome by framing the 2016 constitutional vote as a plebiscite on himself and his government. The result was a decisive defeat, with nearly 60% voting no on a turnout of nearly 70%, and Renzi resigned.

Referendums also alter the relationship between voters and parties. Scholars link partisan dealignment to a decline in "party identification" with major parties, and research on Britain argues that dealignment can encourage secondary attachments to other parties. Single-issue votes may reinforce this trend. Referendums separate one question from the wider bundle of positions a party offers, which can weaken parties by unbundling platforms, enabling intense single-issue voters and interest groups to exert disproportionate influence on outcomes. Identities formed around a single question can also compete with partisan ones. Analysts of the Brexit vote observe that "Leaver" and "Remainer" identities played a role that may be as important as traditional party identities. Within parties, the cost of defying a referendum result can be concrete. A study of Conservative MPs found that those who resisted switching to a pro-Leave position faced significantly higher electoral losses.

Supporters and critics read these effects differently. Proponents see referendums as democratic enhancements that give voters a say on issues that might not be presented in the election choices between parties. Party leaders often see them as a threat because they unbundle platforms in ways that can undermine their policy goals and what they judge to be their party's electability over time. Research on single-issue politics has also argued that such environments can contribute to broader populist movements in modern democracies. In the Italian case, the No campaign was spearheaded by the anti-establishment Five Star Movement, which shows how a yes-or-no question can be turned into an opportunity for populist mobilisation.

The plebiscitary use of referendums raises a further concern. A leader can present a vote as a personal mandate, and a binary question can concentrate attention on the leader rather than on the policy's details. Critics argue that such votes bypass the deliberation that representative institutions, including committee hearings, amendment processes and party debate, are designed to provide. Defenders reply that deliberation is not absent from referendum campaigns, but that it takes place in public debate rather than in the legislature. The democratic legitimacy of a referendum therefore depends partly on how the question was framed, how the campaign was conducted, and how the result relates to the institutions that remain responsible for implementing it.

Self-determination, secession and territorial questions

Referendums have often been used to settle where a border should run or which state a population should belong to. After the First World War, the Treaty of Versailles arranged two votes in the former Duchy of Schleswig to fix the frontier between Germany and Denmark. The 1866 Peace of Prague had already promised a plebiscite within six years, which would let northern Schleswigers choose whether they wanted Danish rule restored. The vote was held in two stages, on 10 February and 14 March 1920. The northern zone (Zone I) voted to join Denmark, while the smaller southern portion (Zone II) voted to remain part of Germany. A third, predominantly German area was not polled at all. Zone II, in the middle of the region, consisted of about 80% Germans, and the southern area, which was primarily populated by ethnic Germans, saw no vote.

The Saar followed a different path. After the Treaty of Versailles, the Saar Basin was run by the League of Nations for 15 years. Its status was tied to the German loss of control over the coal-producing territory, which had been wrested forcibly from German sovereignty to provide war reparations to France. The 1935 vote offered three options: reunion with Germany, continued separate administration by the League of Nations, or annexation to France. Under Hitler, the campaign was heavily orchestrated, with Joseph Goebbels designing a concerted campaign to sway voters, and the support of the local Catholic authorities for a return also helping. The large majority for reunion with Germany was later cited as evidence of the regime's propaganda success, and the episode is often treated as a precursor to German expansion in the late 1930s.

International law offers no settled answer to whether a population may decide its own political status unilaterally. The debate turns on the right of peoples to self-determination, which is recognized in the United Nations Charter but is generally read as supporting internal arrangements rather than automatic secession. The most influential modern statement came from the Supreme Court of Canada in its 1998 opinion on Quebec. The Court concluded that external self-determination, meaning the right to become independent, arises in only three sets of circumstances: for colonies, for oppressed peoples such as those under foreign occupation, and for people denied meaningful access to government. Outside those cases, the Canadian government asked the court whether Quebec could secede unilaterally under domestic or international law. The court found it could not, but held that a clear referendum vote for secession would oblige the federal government to negotiate in good faith.

Quebec's two referendums illustrate this framework. The 1980 vote asked whether the provincial government should receive a mandate to negotiate sovereignty-association with Canada, and the proposal was defeated by a 59.56 per cent to 40.44 per cent margin. Prime Minister Pierre Trudeau then pursued constitutional change, which led to the patriation of the Canadian Constitution. The 1995 referendum was far closer. Turnout was 93.52%, and the proposal was rejected by 50.58% to 49.42%. Parliament responded with the Clarity Act of 2000, which established the conditions under which the Government of Canada would enter into negotiations that might lead to secession following such a vote by one of the provinces. The Act requires the House of Commons to say in advance whether a referendum question is clear, yet it does not define what constitutes a "clear majority." Quebec's own counter-legislation, Bill 99, was partly struck down by the Quebec Court of Appeal in 2007. The Supreme Court also rejected the implication that a qualified majority would be required for secession negotiations to begin.

Scotland's 2014 independence referendum was negotiated through the 2012 Edinburgh Agreement, and the question was put to voters with a majority choosing to remain in the United Kingdom. Two referendums on possible secession have been held within the UK, in Northern Ireland in 1973 and in Scotland in 2014, and in both cases a majority of electors voted to remain part of the UK. Later attempts to hold a second vote without Westminster's consent ran into legal limits. In the UK Supreme Court's view, organizing such a referendum concerns matters reserved to the British Parliament and therefore falls outside Scotland's competences. The court also applied the Quebec reasoning, which means the only legal route to a future independence referendum is via Westminster approval.

Several questions remain open. The first concerns the relevant electorate: whether a secession vote should be decided by the province or territory as a whole, by a local majority, or by a regional subgroup, and whether residents who are not members of the majority nation, including Indigenous and anglophone residents in Quebec, should count differently. The second concerns minority protection after a border change. The Schleswig votes left a German minority in Denmark and a Danish minority in Germany, and the two governments later set out their respective protections in bilateral declarations. Such guarantees remain a central issue in any plebiscite that redraws a boundary or changes sovereignty.

Legal status, constitutional limits and criticisms

Constitutional entrenchment and the role of courts

A referendum's legal force depends first on the constitutional text. Some constitutions spell out the device, its triggers and its procedures in detail. Others leave it to ordinary statute, and in some systems a vote held without a clear legal basis is advisory by default. Courts usually act as gatekeepers at two points. Before a vote, they may test whether the question falls within the powers of the body that set it and whether it is clear enough to put to voters. After a vote, they may decide whether the result can take legal effect.

Italy shows the pre-vote gate most clearly. Abrogative referendum petitions are first checked by the Court of Cassation for conformity with ordinary law. The Constitutional Court then rules on admissibility under the constitution and can reject a question outright. Many petitions carrying the required 500,000 signatures have never reached the ballot for this reason. In the United States, state courts routinely review ballot titles and summaries. They can order a ballot reworded, strike a measure, or in rare cases overturn a result already counted. The standards vary considerably. The Wisconsin Supreme Court has held that an amendment question need not explain the proposal and is defective only when it is fundamentally counterfactual. Courts in other states, by contrast, require questions to be easily understandable and to describe the measure fairly.

Limits on what may be decided

Most legal orders place some matters beyond the reach of a vote. A referendum generally cannot amend a constitution's protected core, and subnational bodies often cannot legislate on matters reserved to a higher authority. In Maryland, a court held that a charter amendment in the form of a detailed rezoning scheme was legislative in character and therefore not proper charter material. Fundamental rights are a further limit. Scholars of American direct democracy have argued that the political dynamics of initiatives weaken normal safeguards for minorities, so courts should apply closer scrutiny to direct legislation. Campaign finance is also constrained. US Supreme Court decisions from 1978 and 1981 restricted limits on spending and contributions in ballot-measure campaigns.

Canada offers the clearest case of a limit on secession. In its 1998 reference on Quebec, the Supreme Court held that a province cannot secede unilaterally under the constitution or international law. It also held that a clear majority voting clearly for secession would oblige the other parties to negotiate, and that secession would require a constitutional amendment. The Clarity Act of 2000 implemented this by letting the House of Commons decide whether a question is clear, but it does not define a "clear majority". Critics in Quebec regard this as giving the federal legislature a veto over provincial self-determination.

Procedural critiques

Critics of referendums focus on how the process is run rather than on the principle. Ballot wording is the most common target, since a question's framing can predetermine the answer. The Canadian experience illustrates this. The 1995 Quebec vote was narrowly lost, and the federal government's alarm centred partly on a question seen as muddled. Timing is another concern. Whoever controls when a vote is called can choose a moment favourable to their side, and governments can use referendums to set the agenda. Electoral rolls raise questions about who is entitled to vote and whether temporary residents or diaspora voters are counted. Procedural rules on spending, broadcasting and official communication, discussed elsewhere in this entry, are also contested, because they determine whether one side can outspend or outpublicise the other.

Unresolved debates

Two debates remain open. The first concerns whether complex or technical questions suit a direct vote. Defenders argue that voters can inform themselves and that the purpose of a ballot question is not primarily to educate. Critics respond that information costs and the difficulty of technical trade-offs make accurate judgement unlikely, and that questions can be simplified in ways that mislead.

The second concerns whether binding referendums undermine parliamentary sovereignty. In the United Kingdom, the 2016 vote on leaving the European Union was not legally binding. In R (Miller) v Secretary of State for Exiting the European Union (2017), the Supreme Court ruled 8–3 that ministers could not trigger Article 50 without an Act of Parliament, because the 1972 Act had created domestic rights that only legislation could remove. The Court stressed that the Referendum Act 2015 contained no express language making the result binding. Supporters of direct democracy see such outcomes as a failure of democratic responsiveness, while defenders of the doctrine regard them as essential to constitutional coherence. In Scotland, the question of whether the devolved parliament can legislate for another independence vote without Westminster's consent has been settled by the UK Supreme Court in favour of Westminster's consent requirement, illustrating how territorial referendums depend on the division of powers.

References

1. Referendums in Germanyen.wikipedia.org
2. Volksentscheidde.wikipedia.org
3. Bürgerentscheidde.wikipedia.org
4. Volksentscheid / Volksbegehrenbpb.de
5. Volksbegehren, Volksentscheid, Volksabstimmung: Begrifflichkeiten und Modellebundestag.de
6. Referendum or plebiscite: what’s the difference? - University of Strathclydepureportal.strath.ac.uk
7. Plebiscite or Referendum - What's the Difference - ABC Newsabc.net.au
8. Referendums and Plebiscitesencyclopedia.com
9. Referendumstudy.com
10. Differences between a plebiscite and a referendumdifferencebetween.net
11. Referendumen.wikipedia.org
12. Mandatory referendumen.wikipedia.org
13. Optional referendumen.wikipedia.org
14. Referendumen.wikipedia.org
15. Mandatory referendums in Switzerlanden.wikipedia.org
16. The referendumch.ch
17. Direct Democracyaboutswitzerland.eda.admin.ch
18. This document will not be distributed at the meeting. Please bring this copy.venice.coe.int
19. This document will not be distributed at the meeting. Please bring this copy.rm.coe.int
20. 1 A COMPARATIVE LOOK AT REFERENDUM LAWS February 2009iilhr.org
21. Pravar Petkar: Consultative Referendums and Constituent Power in the UKukconstitutionallaw.org
22. Swiss referendums explainedweforum.org
23. Terse rules, loquacious judges: a critical assessment of forty years of abrogative referendums in Italyblue-europe.eu
24. Referendums in Italyen.wikipedia.org
25. Referendums Overview - Australian Electoral Commissionaec.gov.au
26. Switzerland Federal Referendums 14 June 2026 — Resultselectiomap.com
27. Swiss votes, initiatives and referendalenews.ch
28. Referendums what they are and why they matter - Explore MoADmoadoph.gov.au
29. Italians' vote on citizenship and workers’ rightsdemocracy.community
30. 2011 Italian referendumsen.wikipedia.org
31. 2022 Italian referendumen.wikipedia.org
32. 2025 Italian referendumen.wikipedia.org
33. Double majorityen.wikipedia.org
34. Referendums - Australian Electoral Commissionaec.gov.au
35. Referendums and plebiscites - Parliamentary Education Officepeo.gov.au
36. Referendums in Australiaen.wikipedia.org
37. Referendumhls-dhs-dss.ch
38. Abstimmungenhls-dhs-dss.ch
39. Basel-Landschafthls-dhs-dss.ch
40. Napoleon Iiiencyclopedia.com
41. Plébiscites sous Napoléon IIIfr.wikipedia.org
42. Plébiscite du 8 mai 1870fr.wikipedia.org
43. French Constitution of 1852en.wikipedia.org
44. Swiss Direct Democracyaceproject.org
45. List of Swiss federal referendumsen.wikipedia.org
46. 1875 Swiss referendumen.wikipedia.org
47. Voting in Switzerlanden.wikipedia.org
48. www.ssoar.info How representative are referendums? Evidenced-nb.info
49. A political-economic analysis of Swiss referendums 1848 to 2022: turnout, acceptance rates and the double-majority thresholdlink.springer.com
50. 2010 Swiss referendumsen.wikipedia.org
51. Consensus Democracy: The Swiss System of Power-Sharing - PMCpmc.ncbi.nlm.nih.gov
52. Concordance systemen.wikipedia.org
53. Citizen participation through direct legislation: a road to success? A perspective from Switzerlandlink.springer.com
54. The domestic politics of the Swiss‐EU negotiations on the Institutional Framework Agreement (and beyond) - Gago - 2024 - Swiss Political Science Review - Wiley Online Libraryonlinelibrary.wiley.com
55. 1905 Swiss copyright referendumen.wikipedia.org
56. Constitutional Referendum in Germany — Country Reportlink.springer.com
57. Deutsche Wohnen %26 Co. enteignenen.wikipedia.org
58. United Kingdom European Constitution referendumen.wikipedia.org
59. Abrogative Referendums: Voting on Labour and Citizenship Set for June 8 and 9 - Centro di Ateneo per i Diritti Umaniunipd-centrodirittiumani.it
60. 1990 Italian referendumsen.wikipedia.org
61. 1999 Italian referendumen.wikipedia.org
62. 2011 Italian referendumsen.wikipedia.org
63. Full article: Italy at the beginning of 2026: the spring referendum and the year aheadtandfonline.com
64. 2005 French European Constitution referendumen.wikipedia.org
65. 2005 Dutch European Constitution referendumen.wikipedia.org
66. Treaty establishing a Constitution for Europeen.wikipedia.org
67. Basic Law for the Federal Republic of Germanygesetze-im-internet.de
68. 2009 Italian electoral law referendumen.wikipedia.org
69. (PDF) Voter turnout in direct democracy: A joint analysis of individual, referendum and community factorsresearchgate.net
70. Voter turnout in direct democracy: A joint analysis of individual, referendum and community factorscambridge.org
71. Beyond the Ballot: The Impact of Voting Margin and Turnout on the Legitimacy of Referendum Outcomes in Europe - Sveinung Arnesen, Troy Saghaug Broderstad, Mikael P Johannesson, Jonas Linde, 2025journals.sagepub.com
72. Voter turnout in direct democracy: A joint analysis of individual, referendum and community factors - GOLDBERG - 2023 - European Journal of Political Research - Wiley Online Libraryonlinelibrary.wiley.com
73. Bounded Rationality and Voting Decisions over 160 Years: Voter Behavior and Increasing Complexity in Decision-Makingjournals.plos.org
74. RESEARCH PAPER 05/45 13 JUNE 2005 The Future of the European Constitutionresearchbriefings.files.parliament.uk
75. Referendums related to the European Unionen.wikipedia.org
76. Archive ouverte UNIGE https://archive-ouverte.unige.ch Article scientifiqueaccess.archive-ouverte.unige.ch
77. Mobilizing different types of voters: The influence of campaign intensity on turnout in direct democratic votes - ScienceDirectsciencedirect.com
78. Voter Turnout Research Papers - Academia.eduacademia.edu
79. Mobilizing different types of voters: The influence of campaign intensity on turnout in direct democratic votesresearchgate.net
80. Gambling on Europe: David Cameron and the 2016 referendumlink.springer.com
81. David Cameronglobalsecurity.org
82. Conservative divisions on Brexit: the general election and beyond - UK in a changing Europeukandeu.ac.uk
83. 2016 Conservative Party leadership electionen.wikipedia.org
84. Reaction: Italian referendum and Matteo Renzi's resignationblogs.lse.ac.uk
85. Italy’s Prime Minister Matteo Renzi to resign after referendum defeatedition.cnn.com
86. “The centre must hold” Partisan dealignment and the rise of ...lse.ac.uk
87. (PDF) Partisan Dealignment, Party Attachments and Leader Effectsresearchgate.net
88. Party realignment and single-issue votersjournals.plos.org
89. The effects of party labels on vote choice with realistic candidate differentiationcambridge.org
90. Party realignment and single-issue votersncbi.nlm.nih.gov
91. Italian PM Matteo Renzi resigns as referendum lostirishtimes.com
92. 1920 Schleswig plebiscitesen.wikipedia.org
93. Territory of the Saar Basinen.wikipedia.org
94. Schleswig, plebiscite (1920) - Dead Country Stamps and BanknotesDead Country Stamps and Banknotesdcstamps.com
95. Plebiscites After the First World Wargcsehistory.com
96. The Saar plebiscite, 1935 : Hitler's triumphuh-ir.tdl.org
97. Allied Plebiscite Activity in the Saar Territory, 1935militaryphs.org
98. CEEAC/S6/25/29/1 Constitution, Europe, External Affairs and Culture Committeeparliament.scot
99. The Case for a Scottish Clarity Act - Coulter - 2024 - The Political Quarterly - Wiley Online Libraryonlinelibrary.wiley.com
100. 1980 Quebec referendumen.wikipedia.org
101. 1995 Quebec referendumen.wikipedia.org
102. Clarity Acten.wikipedia.org
103. Reference re: Secession of Quebec, in Context - Centre for Constitutional Studiesconstitutionalstudies.ca
104. Reference re: Secession of Quebec, in Contextconstitutionalstudies.ca
105. Scottish independence referendum: legal issuesresearchbriefings.files.parliament.uk
106. New Independence Referendum in Scotland? Is the United Kingdom moving closer to continental Europe?eurac.edu
107. Quebec referendum of 1995britannica.com
108. Quebec referendum of 1995ebsco.com
109. Clarity Actebsco.com
110. Reference Re Secession of Quebecen.wikipedia.org
111. High court takes up limits on referendumslmtribune.com
112. Constitutional Constraints on Initiative and Referendumscholarship.law.vanderbilt.edu
113. Maryland State Board of Elections, et al. v. Anthony J. ...courts.state.md.us
114. Voters beware: What to know about constitutional amendment ballot questions before voting in the November electionwjiinc.org
115. Brexit & Miller: what next for Parliament? - House of Commons Librarycommonslibrary.parliament.uk
116. R (on the application of Miller and another) v Secretary of State for Exiting the European Unionblackstonechambers.com
117. R (Miller) v Secretary of State for Exiting the European Union [2017] UKSC 5lawcases.net
118. The Brexit Legal Challenge- A Summarycjel.law.columbia.edu
119. repository.law.umich.edurepository.law.umich.edu
120. Provinces Leaving Canada Part II: The Clarity Act - LawNow Magazinelawnow.org
121. 1998 CanLII 793 (SCC)canlii.org
122. Background to the introduction of Bill C-20, the Clarity Bill (PRB 99-42E)publications.gc.ca
123. What Is Canada’s Clarity Act and How Does It Work? - LegalClaritylegalclarity.org
124. 885 ARTICLE RETHINKING THE UNCONSTITUTIONALITY OF CONTRIBUTION AND EXPENDITUREsoutherncalifornialawreview.com