3.2 - Referendums in representative democracy

Direct decisions and representative institutions

A referendum gives the electorate a direct choice on a particular proposal, whereas an election chooses representatives to make decisions over time. Direct democracy involves citizens expressing their own decisions rather than delegating that decision to representatives. Representative democracy involves choosing people or parties to exercise political judgement on the electorate's behalf. A referendum is one direct-democratic mechanism within the UK's mainly representative system; it does not replace that whole system.

Representatives still decide whether to authorise many referendums, specify the question and franchise, and deal with implementation. A franchise is the group legally entitled to vote. The referendum offers a decision on its question, not necessarily consent to every later detail. A majority choosing an outcome cannot by itself reveal why each voter preferred it or which implementation every supporter would accept.

The legal effect depends on the authorising law. It is incorrect to claim that every UK referendum is legally binding, or that none can have binding consequences. The 2016 EU referendum's statute authorised the poll without making the result itself change the law. In Miller in January 2017, the Supreme Court held that parliamentary legislation was needed before ministers could give the relevant Article 50 notice. The result had major political significance but did not supply that legal authority by itself. Supreme Court press summary, explaining the majority's reasoning at paragraphs 116–124.

Devolution and agreement since 1997

Referendums have been used to seek consent for changes to how the UK is governed, rather than only for ordinary policy choices. The 1997 devolution votes helped legitimise the creation of new representative institutions. Approval of a legislature did not remove the need for legislation establishing its powers and elections.

Sources: Scottish Parliament path to devolution; Commons Library election statistics, referendum section; Northern Ireland Assembly historical paper; Parliament's 2005 record of referendum results. Figures are shares of votes on the relevant question, not shares of every resident or of the full electorate.

These cases illustrate different kinds of consent. The Northern Ireland vote formed part of a negotiated peace and constitutional settlement. Scotland and Wales endorsed institutional proposals that elected politicians had prepared. London's vote addressed regional government within England. They were not interchangeable votes on the same policy, nor did their electorates cover the whole UK.

The Welsh result also qualifies the argument that referendums always produce overwhelming agreement. A valid narrow majority can authorise a political direction under the rules, but leaves a large minority unconvinced. Legitimacy includes accepting an agreed decision procedure, as well as measuring enthusiasm for its outcome. A later election can choose a government's policies within the institution the referendum helped establish.

Reform approved and rejected

A referendum can block a government's proposed change as well as approve one. In the North East of England on 4 November 2004, 77.9% voted against an elected regional assembly. The proposed assembly was not created. This qualifies any claim that a government can always use a referendum to obtain a rubber stamp. Parliament's record of the result.

Two referendums in 2011 addressed different types of institutional reform. In Wales on 3 March, 63.5% supported wider primary law-making powers for the National Assembly. It could subsequently legislate within the specified devolved subject areas without obtaining a new transfer of competence from Westminster for each expansion. This deepened the earlier settlement; it was not a new vote on whether the Assembly should exist. Commons Library account of the Welsh referendum; result recorded in the Electoral Commission's referendum table, Appendix 2.

In the UK-wide vote on 5 May 2011, 67.9% opposed replacing Commons FPTP with Alternative Vote. FPTP remained. AV ranks candidates for a single-member contest; it was not a choice between FPTP and every possible proportional system. Therefore, the result supports the claim that this specific reform was rejected, rather than proving permanent popular opposition to all electoral reform. Commons Library AV result.

Referendums have also operated at local level, including decisions about directly elected mayors. Such votes concern the local institution and electorate, not a UK-wide mandate. Parliament's record of national, regional and local polls includes accepted and rejected local mayoral proposals. The political impact depends on what the question authorises and which area is deciding it.

Independence, the EU and political impact

The 2014 Scottish independence and 2016 EU referendums show that a clear decision procedure can coexist with disagreement over its wider meaning. Scotland voted on independence on 18 September 2014. The result was 55.3% No and 44.7% Yes, with turnout of 84.6%. Scotland remained in the UK. The participation level supports the argument that a referendum can engage voters strongly in a constitutional question; the substantial Yes minority shows that the result did not produce unanimity. Commons Library result and turnout. The campaign also prompted discussion of further devolution; the UK government's September 2014 statement records its commitment to that process, rather than proof that all participants were satisfied afterwards.

On 23 June 2016, 51.9% voted Leave and 48.1% Remain in the EU referendum; turnout was 72.2%. Electoral Commission official declaration; Commission report. The referendum changed the direction of UK policy and created pressure to implement withdrawal. David Cameron, who had advocated Remain, announced the need for a new prime minister on 24 June. His official statement is evidence of that decision and his explanation; its wider claims about his record are partisan argument.

The UK eventually left the EU on 31 January 2020. Government's 2020 national reform report, paragraph 1.11. The interval and the Miller ruling illustrate the distinction between choosing withdrawal and determining its legal route and detailed terms. “Leave” did not specify one complete future relationship, and elected institutions still had decisions to make.

Turnout and the winning share answer different questions. The winning share measures support among valid votes for the available options. Turnout measures participation by the electorate. Neither directly measures agreement with every implementation detail. A majority may confer a powerful political mandate while leaving scope for representative judgement over how to carry it out.

Consent, expertise and a fair process

The case for referendums is strongest when the question is fundamental, the choices are intelligible and the electorate can make an informed decision under accepted rules. Direct voting can give constitutional change a legitimacy beyond a party's election mandate, which bundles many policies together. It can oblige representatives to listen, settle a specific choice and encourage participation, as the 2014 Scottish turnout demonstrates. A referendum need not abandon representative government: it can authorise a broad constitutional direction while representatives organise the detail.

The case against is that elected representatives have time, resources and continuing responsibility to examine complex options. A binary vote may compress several possible arrangements into two labels, and a narrow result can leave a society divided. Campaign claims, unequal resources and selective information can affect informed consent. These are reasons to assess the question, campaign rules and information available; they do not justify assuming that all ordinary voters are incapable or that every disputed claim determined an outcome.

Consider this original teaching source: “Citizens should give direct consent to major changes in who governs them; the Scottish independence vote showed that a single issue can mobilise participation. Yet a vote may choose a direction without explaining its detailed implementation. Elected representatives must then reconcile that mandate with legal duties and competing interests.” The first argument values consent and engagement. The second values continuing deliberation and implementation. Both allow a referendum and a representative legislature to perform useful, different roles.

Evidence must be used proportionately. Scotland's 84.6% turnout supports a participation claim, but cannot prove that every referendum will attract equal engagement. The EU case supports the implementation concern, but does not prove that public votes are always illegitimate. A balanced judgement can favour referendums for major constitutional decisions while requiring a clear question, fair campaigning, adequate information, agreed rules and transparent arrangements for carrying out the outcome. Another may prefer representative decision making if those conditions cannot be met. Explain why the chosen conditions matter rather than simply listing advantages and disadvantages.