UK referendums are constitutionally advisory but politically binding in practice. Since 1997 they have built the devolution settlements, sealed the Good Friday Agreement, closed the AV question, kept Scotland in the Union and taken the UK out of the EU. This walk-through covers how the device works, the named referendums with their results, the case for and against referendums in a representative democracy - then finishes with a worked 30-mark essay. Three short quizzes break the tour up.
A referendum is a vote of the whole electorate on a single question - direct democracy grafted onto a system built around representative democracy. The UK does not use referendums as a default; it uses them for the biggest constitutional questions, and since 1997 they have decided the shape of the state itself. The spec (P1.3.2) asks two things: how referendums have been used since 1997 and their impact on UK political life, and the case for and against referendums in a representative democracy. The constitutional starting point matters for both: UK referendums are advisory under constitutional convention - Parliament could legally have ignored the 2016 result, and the Supreme Court in Miller (No 1) 2017 ruled that Article 50 had to be triggered by an Act of Parliament, not by the executive's prerogative. Advisory in law, binding in politics: no government has dared ignore one.
Scroll - each referendum lights with its result, its turnout and what it proves.
Spec point P1.3.2.a wants the record: how referendums have been used and their impact. The cases divide into two families - the settlement referendums that built institutions (1997, 1998, 2011 Wales), and the question referendums that settled or unsettled arguments (2004, 2011 AV, 2014, 2016). Every essay on this topic runs on these cases, so the results and turnouts need to be exact. Scroll through; the figure beside you holds the referendum cards.
For each referendum, carry four things: the question, the result, the turnout where it matters, and the argument it serves - because in the exam these are evidence for the case-for-and-against debate, not stories to retell.
The referendum that launched the devolution era. The 74% Yes vote gave the Scotland Act 1998 a foundation Westminster legislation alone could never have supplied, and Holyrood's 129 MSPs - elected by the proportional AMS system - have been politically irreversible ever since. The case-for evidence at its strongest: direct legitimacy on a constitutional question, delivered and accepted.
The same year, the opposite lesson. The 2023 Pearson mark scheme does the arithmetic itself: a margin of less than 1% on a turnout of just over 50% meant Welsh devolution had the active support of no more than around a quarter of the electorate - and the mark scheme draws the reform conclusion, that a turnout threshold and a wider margin should be required for constitutional change. The repair came in 2011, when 63% voted Yes to primary law-making powers - the settlement re-founded on a proper majority.
The strongest single case for the device. The 2023 mark scheme calls the referendum a pivotal landmark in the peace after the Troubles: Northern Ireland is vastly more secure and prosperous than before, power is shared across the political divide, and the mark scheme concludes that without a referendum carrying over 80% approval, such a change would not have been possible. When an essay needs one referendum that representative politics could not have delivered alone, this is it.
The forgotten case that does useful work: referendums can kill a policy as decisively as they can found one. The emphatic rejection ended the plan for elected regional assemblies in England - and it shows the device working as a check on government, not only as a rubber stamp for it.
The Alternative Vote referendum settled the FPTP question decisively - and it is the lead case in the 2024 Pearson mark scheme's argument that referendums serve party purposes: the AV vote was the price the Conservatives paid to the Liberal Democrats in coalition, and a safe choice for the Conservatives, since AV was a system few reformers actually wanted. One referendum, two arguments: a clear settled answer, and a device used to manage a coalition rather than to enhance democracy.
The participation showpiece: 84.6% turnout, with 16 and 17 year olds voting - the standard counter-evidence to the participation crisis, and the 2024 mark scheme's evidence that including younger voters broadened democracy. But the result did not stay settled: independence support has remained around 40-50% in polling, the question returned within years, and the 2024 mark scheme's sceptical column reads the referendum as a move to maintain popularity in Scotland rather than to enhance democracy. High engagement and an unsettled answer in the same case.
The case study that reshaped the whole debate. The mechanics first: no minimum-turnout threshold, a simple-majority rule, and a binary question covering what turned out to be a complex multi-option choice between forms of leaving. The origins next: the 2024 mark scheme states the referendum was held to thwart UKIP, and the 2023 mark scheme adds that Cameron never wanted it but called it to placate a section of his party. The aftermath: the result still divides opinion given how close the vote was, and Miller (No 1) 2017 forced the trigger back through Parliament. And the engagement point cuts the other way: 72.2% turnout beat the general elections either side of it - people participated when the question mattered.
The Brexit lessons were not legislated: the Elections Act 2022 introduced no minimum threshold and no supermajority rule for future referendums. The comparisons reformers cite - Switzerland's supermajority and cantonal-majority rules, Italy's 50% turnout requirement for binding referendums - remain things the UK does not do. The reform case is live precisely because nothing was reformed.
Scroll - each argument lights with its evidence.
P1.3.2.b is a pure debate point: the case for and against referendums in a representative democracy. The for side runs on four arguments, each anchored to a named case - and the 2024 Pearson mark scheme supplies the official versions. Scroll through.
Legitimacy, settlement, participation and education. Each argument needs its best case attached - an argument without a named referendum scores AO1 only.
Major constitutional decisions gain a public approval that no parliamentary vote can match - the 2024 mark scheme's conclusion is that such decisions are made by the people rather than politicians, which carries long-term weight. The cases: Scotland 1997 (74% Yes) made devolution politically irreversible, and the Good Friday Agreement 1998 - over 80% approval - delivered a peace settlement Westminster legislation alone could never have legitimised.
Some questions cut across party lines, leaving the representative system gridlocked - the 2023 mark scheme notes the Conservative Party was in exactly this stalemate over EU membership, and concludes a public referendum is sometimes the only option to bring clarity. The cleanest settlement case is 2011: the AV vote closed the electoral-reform question 67-33, and it has stayed closed.
Referendums move people: 84.6% turnout in 2014 - the highest in UK referendum history - and 72.2% in 2016, above the general elections either side. The 2023 mark scheme concludes that referendums produce more legitimate results than elections do; the 2024 mark scheme adds that including 16 and 17 year olds in Scotland broadened democracy. In any participation-crisis essay, these are the counter-evidence numbers.
The 2024 mark scheme's fourth column: referendums get the public involved between elections and enhance the political understanding and education of the electorate - renewing democratic legitimacy and addressing the participation crisis. The honest pairing: the same campaigns that educate can also mislead, which is where the against case begins.
Scroll - each argument lights with its evidence.
The against side has two layers: the design problems of the device itself, and the political-purposes charge - that governments call referendums to serve party interests, not democracy. The 2024 Pearson question put the second layer directly: referendums since 1997 have not supported democracy but have been used for other political purposes. Scroll through.
Oversimplification, weak mandates, party purposes and campaign quality. Then the reform agenda that follows from them - and the fact that none of it has been legislated.
A referendum reduces a complex choice to Yes or No. 2016 is the standard case: Leave/Remain was binary, but leaving turned out to contain multiple options - no-deal, soft, hard, a second vote - that the ballot never asked about. The deadlock of 2016-19 was, in part, Parliament trying to answer the question the referendum had not. Representatives elected to use their judgement found themselves bound by an instruction with no detail in it.
Two versions. The weak-mandate case: Wales 1997 - a sub-1% margin on a 50% turnout, support from around a quarter of the electorate, which the 2023 mark scheme reads as proof that some referendum results are not the true will of the majority. The unsettled case: the 2023 mark scheme notes that 2016 still divides opinion given how close the vote was, and 2014 was followed by years of continued argument - leading to its conclusion that real and lasting legitimacy needs more than a simple plurality. The same mark scheme adds the union point: referendum-driven devolution fuelled the case for Scottish independence and created uncertainty in Northern Ireland.
The 2024 mark scheme's agreement column, in its own terms: referendums have been manipulated and created to benefit the party in office. The 2011 AV vote was the price paid to the Liberal Democrats in coalition - and a safe choice for the Conservatives. The 2016 referendum was held to thwart UKIP, born of fear that anti-EU sentiment would cost Conservative seats - and Cameron, per the 2023 mark scheme, never wanted it. Devolution referendums, on this reading, were moves to maintain popularity in Scotland. The conclusion the mark scheme offers: greater democracy or participation is rarely the reason for calling referendums.
Referendum campaigns inform unevenly and can mislead - the Vote Leave campaign in 2016 was found to have broken electoral law - and a binary public vote carries a majoritarian risk: a bare majority binds everyone, including minorities whose rights are at stake, with no court or chamber to check it. These are the arguments that make constitutional purists say referendums simply do not belong in a representative democracy.
The against case points at reform more often than abolition: a supermajority requirement on constitutional questions, a minimum-turnout threshold, and making the advisory status explicit on the ballot. The comparisons: Switzerland runs frequent referendums under supermajority and cantonal-majority rules; Italy requires 50% turnout for a binding result. The UK fact to land: the Elections Act 2022 introduced none of this - the Brexit lessons were not legislated.
How the topic is tested, with approaches to the recurring questions.
Referendums appears in Paper 1 as a 30-mark question - at Q1 with a source, or at Q2 as an essay. Both split AO1/AO2/AO3 at 10/10/10. The rules: two views weighed in a balanced way with a sustained line of argument; a one-sided answer is capped at Level 2. Structure by theme - legitimacy, settlement, design, purposes - using the referendums as evidence inside each, not as a chronological tour. Conclusions justify rather than summarise.
Recurring 30-mark questions to practise.
Approach: The Part 3 party-purposes section against the Part 2 case for. Para 1 - the purposes charge: AV as the coalition price, 2016 held to thwart UKIP, devolution as popularity management. Para 2 - the democratic record anyway: legitimacy (Scotland 74%, GFA over 80%), participation (84.6% in 2014, with 16-17s voting), clear answers on EU membership and independence. Para 3 - the education and engagement column against the campaign-quality counter. Judgement: motives were often political, but outcomes still widened democracy - judge the device by what it did, not why it was called.
Approach: The worked essay below answers it in full - the design problems and unsettled results against the settlements, the engagement and the peace.
Approach: The constitutional-principle framing. Para 1 - the purist case: representatives are elected to judge, binary ballots oversimplify (2016), bare majorities carry majoritarian risk. Para 2 - the practice: advisory in law (Miller No 1), so Parliament remains sovereign - the device supplements representation rather than replacing it. Para 3 - the record: settlements (devolution, GFA) that representative politics could not have legitimised alone. Judgement: referendums fit a representative democracy when confined to constitutional questions - the problem is design, not principle.
Approach: Para 1 - thresholds: Wales 1997 (the mark scheme's own quarter-of-the-electorate arithmetic) and the 2016 simple majority against the risk that thresholds let abstainers veto change. Para 2 - the question and the campaign: binary ballots on complex choices and the Vote Leave finding against the difficulty of writing multi-option ballots. Para 3 - the comparisons: Swiss supermajorities and Italy's 50% rule against the UK's flexible constitution. Judgement: reform - the Elections Act 2022 changed nothing, and the case-against problems are all design problems.
Judgement. Referendums have brought more advantages than disadvantages to UK political life. The disadvantages are genuine - oversimplified questions, weak mandates, party motives, campaigns that broke the rules - but every one of them is an argument for the reform agenda the Elections Act 2022 ignored, not for abandoning the device. Against them stand devolution, the Good Friday peace, the closed AV question and turnout figures no general election can reach. A tool that built the modern constitutional settlement has earned tighter rules, not retirement.
Referendum. A vote of the whole electorate on a single question - the UK's main instrument of direct democracy, used for big constitutional questions.
Direct democracy. Citizens deciding questions themselves rather than through representatives.
Representative democracy. Citizens electing representatives to decide on their behalf - the UK's default system, into which referendums must fit.
Advisory status. UK referendums are advisory under constitutional convention - Parliament could legally ignore a result, though politically no government has.
Miller (No 1) 2017. The Supreme Court ruling that Article 50 had to be triggered by an Act of Parliament, not executive prerogative - proof of the advisory point.
The 1997 settlements. Scotland: 74% Yes - the strongest founding mandate. Wales: 50.3% Yes on a 50% turnout - the weakest, repaired by the 63% Yes of 2011.
The Good Friday referendum 1998. Approved on both sides of the border, with over 80% approval in Northern Ireland - the peace settlement case.
The 2004 North East vote. A regional assembly rejected by 78% - the device as a check on government.
The 2011 AV referendum. No 67-33 - the settled-answer case, and the 2024 mark scheme's coalition-price case.
The 2014 independence referendum. No 55-45 on an 84.6% turnout - the highest ever, with 16 and 17 year olds voting.
The 2016 EU referendum. Leave 51.9-48.1 on a 72.2% turnout - the largest direct-democracy vote in UK history, run with no threshold and a simple-majority rule.
Turnout threshold. A minimum participation rule for a binding result - Italy requires 50%; the UK requires nothing.
Supermajority. A requirement for more than a bare majority on constitutional questions - part of the Swiss model; never adopted in the UK.
Elections Act 2022. The post-Brexit elections statute - which introduced no threshold and no supermajority. The reform agenda remains unlegislated.
Majoritarian risk. The danger that a bare majority binds everyone, including minorities, with no check - a core argument of the constitutional purists.