Voting behaviour and the media

Explanations of voting are often divided into long-term factors, which change slowly (social class, age, ethnicity, region and party loyalty), and short-term factors, which vary from one election to the next (the issues, the government’s record, manifestos, campaigns, leaders and the media). In the 1950s and 1960s class was the dominant long-term factor: the political scientist Peter Pulzer described class as ‘the basis of British party politics; all else is embellishment and detail’.^[3]

Class dealignment is the weakening of this link. MORI/Ipsos estimates show that skilled manual (C2) voters went from Labour 49%, Conservative 26% in October 1974 to a 41%–41% tie in 1979. By 2024 Ipsos found only small class differences in Labour and Conservative support: Labour won 36% of the professional and managerial AB group and 32% of the DE group, while Reform UK took 25% of C2 voters but 10% of ABs.^[4][5]

Partisan dealignment is the decline in voters’ long-term attachment to a party. The British Social Attitudes survey found that 46% identified ‘very’ or ‘fairly’ strongly with a party in 1987, compared with 37% in 2014, while the share identifying with no party rose from 8% to 19%. Weaker loyalties mean more switching: the Electoral Reform Society (a campaign group) calculated that aggregate volatility between 2019 and 2024 beat the previous record, set in 2015, and was the highest for nearly a century (only 1931 was higher). In 2024 Labour and the Conservatives together won only about 58% of the vote.^[6][7][8]

As loyalties weaken, short-term judgements matter more. Rational choice theory sees voters weighing up what each party offers them, looking back at the government’s record (retrospective voting) and forward at the parties’ promises (prospective voting). Issue voting means choosing the party with the best policies on the issues a voter cares about, as many did on Brexit in 2019. Valence voting focuses on which party is most competent to deliver goals almost everyone shares, such as a strong economy and a working NHS – judgements of governing competency.^[9][10]

Issue 4 of the specification names leadership alongside manifestos, campaigns and the wider political context. Leaders can matter a great deal, but leader ratings do not decide every result: in 1979 voters preferred Labour’s James Callaghan to Margaret Thatcher as Prime Minister by 50% to 31%, yet the Conservatives won. Manifestos are published about three weeks before polling day and their pledges can shape coverage; in 2017 a backlash against the Conservatives’ social care policy led the party to change its commitment during the campaign.^[11][12][13]

Because the UK uses first-past-the-post, how votes are spread also matters. Tactical voting – backing the candidate best placed to defeat the one you most dislike – helped the Liberal Democrats in 1997 and again in 2024, when post-election YouGov polling for the Electoral Reform Society suggested 28% of voters voted tactically.^[14][7]

Result. The Conservatives under Margaret Thatcher won 339 of 635 seats on 43.9% of the UK vote, an overall majority of 43. Labour won 269 seats (36.9%) and the Liberals 11 (13.8%). Turnout was 76.0%. Thatcher became Britain’s first woman Prime Minister, beginning 18 years of Conservative government. The election was triggered when James Callaghan’s minority Labour government lost a Commons vote of no confidence on 28 March 1979. See Voter Profile for the full breakdown.^[19][20][21]

Governing competency and context. The strikes of the 1978–79 ‘Winter of Discontent’ badly damaged confidence in the government. MORI found satisfaction with the government fell from 39% in November 1978 to 17% at the end of January 1979, when the Conservatives led Labour by 55% to 36%. In March 1979, 43% expected the economy to get worse over the next year, against 23% who expected it to improve.^[22][23]

Issues, leadership and campaign. In MORI’s first campaign poll, 40% thought the Conservatives had the best policies on strikes and industrial disputes, against 22% for Labour, and 54% said prices and inflation were what the election was about. Leadership was Labour’s strongest card: a week before polling day, 50% said Callaghan would make the better Prime Minister against 31% for Thatcher. The Conservatives ran a professional advertising campaign; Saatchi & Saatchi’s ‘Labour Isn’t Working’ dole-queue poster, first designed for an expected autumn 1978 election, was reused in the spring 1979 campaign.^[24][12][25]

Media. Hostile press coverage reinforced an image of a Prime Minister out of touch: after Callaghan returned from the Guadeloupe summit in January 1979, The Sun ran a headline suggesting he had denied there was a crisis, a report historians regard as misrepresenting what he said.^[26]

Social patterns. Class still structured the vote – the ABC1 middle classes voted Conservative 59% to Labour 24%, and DE voters Labour 49% to Conservative 34% – but the Conservatives won by drawing level among C2 skilled manual workers (41% each). Women were more likely than men to vote Conservative (47% against 43%), and age differences were small: the Conservatives won 42% of 18–24-year-olds and 47% of those aged 55 and over. First-past-the-post turned a clear but not overwhelming vote lead into a working majority, while the Liberals won 11 seats on 13.8%.^[4][19]

Significance. 1979 is often used to illustrate class dealignment beginning and the power of retrospective, valence-based judgements: voters punished a government whose competence had collapsed, even though they preferred its leader.^[4][22]

Result. Labour under Tony Blair won 418 of 659 seats on 43.2% of the vote, an overall majority of 177, ending 18 years of Conservative government. The Conservatives fell to 165 seats on 30.7%, which the House of Commons Library calls their worst performance in seats and vote share since 1918. The Liberal Democrats won 46 seats on 16.8%, and the Referendum Party took 2.6% but no seats. Turnout was 71.4%, then the lowest at any post-war general election. See Voter Profile for the full breakdown.^[31][32][20]

Governing competency. The Conservatives lost their reputation for economic management on ‘Black Wednesday’, 16 September 1992, when Britain left the Exchange Rate Mechanism after interest rates were briefly raised to 15%; Conservative poll support fell from 43% to 29% by October 1992 and did not recover despite later economic growth. Allegations of ‘sleaze’ also damaged trust: in Tatton the independent anti-corruption candidate Martin Bell defeated the sitting Conservative MP Neil Hamilton by 11,077 votes.^[33][34]

Leadership, unity and issues. In MORI’s poll of 1 April 1997, 46% named Blair as the most capable Prime Minister against 22% for John Major and 14% for Paddy Ashdown, and 44% described the Conservatives as divided compared with 12% for Labour. The NHS (49%) and education (43%) were the issues most often named as important, and 40% said Labour had the best policies against 18% for the Conservatives. Blair’s ‘New Labour’ repositioning towards the centre is part of the wider context. Not everything Labour did in government was in its manifesto: its decision to give the Bank of England operational independence is an example often used in debates about the mandate.^[35][36][13]

Media and tactical voting. On 18 March 1997, the first full day of the campaign, The Sun switched to Labour with a front-page editorial headlined ‘The Sun Backs Blair’. Ipsos estimates the Conservative-to-Labour swing among Sun readers at 15.5 points, against 10.5 among all voters. Anti-Conservative tactical voting helped the Liberal Democrats more than double their seats even though their vote share fell slightly: John Curtice found the Liberal Democrat vote rose where they were second to the Conservatives but fell in safer Conservative seats, where Labour’s rose most.^[37][38][14]

Social patterns. Labour won across classes: it led among C2 voters by 50% to 27% and DE voters by 59% to 21%, drew level among C1 voters (37% each) and won 31% of ABs. The swing to Labour was actually larger among AB and C1 voters (13.5 points) than among DE voters (10.0), a sign of class dealignment. Labour won 70% of non-white voters, and led in every age group, although estimated turnout was only 51% among 18–24-year-olds compared with 79–80% among those aged 45 and over. The swing was largest in Greater London (13.0) and the South East (12.0) and smallest in Scotland (7.5) and Wales (7.0).^[38]

Significance. 1997 shows a valence election: voters judged the parties on competence, unity, leadership and trust rather than simply on class. First-past-the-post magnified Labour’s lead into about 63% of seats, giving Blair a large majority with which to carry out his manifesto.^[35][32]

Result. Labour under Keir Starmer won 411 seats on 33.7% of the vote, an overall majority of 172, ending 14 years of Conservative government. The Conservatives won 121 seats (23.7%), their lowest total since the modern party was formed; the Liberal Democrats 72 (12.2%); Reform UK 5 (14.3%); the Greens 4 (6.7%); and the SNP 9, down from 48. Turnout was 59.7%, the lowest since 2001. A record 263 women MPs (40.5%) were elected. See Voter Profile for the full breakdown.^{[43][44][20][8]}

Governing competency and context. In Ipsos’s poll of 21–24 June 2024, 83% were dissatisfied with the way the government was running the country, the worst figure at that stage of a campaign since 1979, and 78% said it was time for a change. Labour was far ahead as the best party on healthcare (41% against 9%) and the cost of living (36% against 15%).^[10]

Leadership and campaign. Rishi Sunak’s net satisfaction rating was -55 in late June 2024, the worst for any Prime Minister at that stage of a campaign since 1979. Starmer was not personally popular (net -19) but 46% saw him as the more capable Prime Minister against 21% for Sunak. The Conservative campaign suffered damaging moments: Sunak left the D-Day 80th anniversary commemorations in Normandy early on 6 June to record a TV interview and apologised the next day. Labour’s manifesto promised that it ‘will not increase taxes on working people’; when the October 2024 Budget raised employer National Insurance, Full Fact described whether this broke the pledge as disputed.^[10][45][46]

Party competition and the electoral system. The Conservative vote split: Ipsos estimated that only 53% of 2019 Conservative voters stayed loyal, while 24% switched to Reform UK and 12% to Labour. First-past-the-post gave Labour 63.2% of seats on 33.7% of the vote, while Reform UK’s 14.3% won 5 seats; the Electoral Reform Society calls it the most disproportional result on record. Tactical voting helped the Liberal Democrats win 72 seats on 12.2%.^[5][7]

Social patterns. Age was the sharpest divide: Ipsos estimated the Conservatives won just 5% of 18–24-year-olds but 43% of those aged 65 and over, while the Greens took 19% of 18–24s. Class differences in Labour and Conservative support were small, though Reform UK did best among C2 voters (25%). Labour won 46% of ethnic minority voters. Labour won 37 of Scotland’s 57 seats, up from 1, as the SNP fell to 9. Estimated turnout ranged from 37% of 18–24s and social renters to 73% of those aged 65 and over.^[5][47]

Significance. 2024 combines a strong valence and retrospective verdict against the government with a dealigned, fragmented electorate: Labour’s landslide rested on a lower share of the vote than Labour won when it lost in 2017, according to the Electoral Reform Society, which matters for debates about its mandate.^[7][8]

2010 produced the first hung parliament since February 1974. The Conservatives won 306 seats (36.1%), Labour 258 (29.0%) and the Liberal Democrats 57 (23.0%), and David Cameron became Prime Minister of a Conservative–Liberal Democrat coalition on 11 May 2010. The economy dominated after the financial crisis (named by 55% as a key issue in April 2010), but neither main party had a decisive lead on it (29% Conservative, 26% Labour in March 2010). Labour lost 11 points among C2 voters, falling to 29%.^{[48][49][50][51]}

2019 was dominated by Brexit. The Conservatives under Boris Johnson won 365 seats on 43.6%, a majority of 80, while Labour fell to 202 seats (32.1%). Ipsos estimated that 73% of 2016 Leave voters backed the Conservatives. The Brexit Party did not stand in the 317 seats the Conservatives had won in 2017. Jeremy Corbyn’s net satisfaction rating of -60 in September 2019 was the lowest for any opposition leader since Ipsos records began in 1977. The British Election Study found the Conservative lead over Labour among working-class voters rose to 21 points, while their lead among middle-class voters stayed at 10, as Labour lost ‘red wall’ seats in the Midlands and the North.^{[52][9][53][54][3]}

Taken together, the case studies show the main trends. Class predicted the vote well in 1979 and Labour still led clearly among working-class voters in 1997, but by 2019 the Conservatives led among them. Age has replaced class as the clearest social divide. Party loyalty has weakened, so votes swing more between elections and smaller parties – the Liberal Democrats, SNP, Reform UK and Greens – take a larger share. Turnout has fallen from 76.0% in 1979 to 59.7% in 2024.^{[4][3][5][20]}

Why have patterns changed? Explanations include the decline of heavy industry and trade unions, the growth of higher education, the parties’ own moves to the centre (as with New Labour), new cross-cutting issues such as Brexit and immigration, and a more fragmented media. The British Election Study team argues that Brexit accelerated a realignment by raising the importance of social and cultural issues.^[55]

Age. In 1979 the age gap was small (the Conservatives won 42% of 18–24s and 47% of over-55s). By 2019 Labour won an estimated 62% of 18–24-year-olds and the Conservatives 64% of those aged 65 and over. In 2024 YouGov found the median age of Labour voters was 46 and of Conservative voters 63.^[4][9][60]

Gender. In 1979 women were more Conservative than men (47% against 43%). The gap has since narrowed and changed shape: in 2024 Reform UK won more support from men (17%) than women (13%), and YouGov found 23% of women aged 18–24 voted Green compared with 12% of young men.^[4][5][60]

Ethnicity. Ethnic minority voters have consistently leaned towards Labour: 70% of non-white voters backed Labour in 1997 and 64% in 2019. In 2024 Labour’s share among ethnic minority voters fell to 46%, with 11% voting Green; Ipsos warns that results for smaller ethnic groups are only indicative because of small samples.^[38][9][5]

Region. Where people live shapes the vote because regions differ in their economies, identities and party systems. In 1997 the swing to Labour was largest in London and the South East. In 2019 Labour lost working-class ‘red wall’ seats in the Midlands and the North. In 2024 Labour recovered in Scotland, winning 37 of 57 seats, and at the 2026 Senedd election Labour fell to third behind Plaid Cymru and Reform UK. See the Power Map for current seats.^{[38][3][47][59]}

Turnout. Turnout varies by group, which affects results. Ipsos estimated 2024 turnout at 67% among AB voters but 45% among DE voters, and 71% among outright homeowners but 37% among renters. Older, better-off groups therefore have more electoral weight. Falling turnout is often linked to disillusion and apathy: the British Social Attitudes survey found the share trusting governments to put the nation first ‘just about always’ or ‘most of the time’ fell from 38% in 1986 to 17% in 2013.^[5][6]

National data sources. Evidence comes from pollsters’ aggregated surveys (Ipsos, YouGov) and the academic British Election Study. Their estimates differ because of methods: Ipsos’s 2024 figures came from an online random-probability panel weighted to the result in Great Britain, so Northern Ireland is excluded, and its turnout figures are a share of all adults, not of registered electors. For example, YouGov put Labour on 33% among C2DE voters in 2024, where Ipsos’s separate C2 and DE figures were 31% and 32%.^[5][60]

Unlike broadcasters, newspapers are free to take sides. On polling day in 1992, 9 April, The Sun’s front page attacked Neil Kinnock, and after the Conservatives’ victory it claimed the credit with the headline ‘It’s The Sun Wot Won It’ (11 April 1992). Giving evidence to the Leveson Inquiry, Rupert Murdoch said he had thought the headline tasteless and wrong, and that the paper did not have that sort of power.^[61][62]

In 1997 The Sun switched to Labour (‘The Sun Backs Blair’, 18 March 1997). Ipsos data show Sun readers moved from Conservative 45%, Labour 36% in 1992 to Conservative 30%, Labour 52% in 1997, and Daily Mail readers’ Conservative vote fell from 65% to 49%. In 2010 The Sun backed the Conservatives (from 30 September 2009) and its regular readers swung to them by 13.5 points against 5 points nationally – but Ipsos found much of this movement happened before the endorsement.^[37][63][64]

This illustrates the reflect or shape debate. Readers tend to choose papers that share their views, and papers may follow their readers or back the likely winner, so it is hard to separate cause from effect. Newton and Brynin tackled this with panel data for 1992 and 1997 and found that newspapers had a statistically significant effect on voting, larger in 1992 than in 1997. In 2024 The Sun backed Labour only on the eve of polling day, saying ‘It is time for a change’; the Cardiff academic John Jewell argued that its late, lukewarm endorsement showed the paper had never had as much power as it claimed.^[65][66][67]

Newspapers also matter between elections by setting the agenda that broadcasters, MPs and party leaders respond to. Their direct reach is shrinking, however: Ofcom found that 22% of adults used print newspapers for news in 2024, and that newspaper brands including their websites reached 34% of adults, down from 51% in 2018.^[68]

Media bias is therefore most relevant for the press, which is openly partisan, and for online sources, which are largely unregulated for impartiality. Broadcasters’ impartiality is also disputed, as the GB News and BBC cases below show. Surveys suggest voters notice bias: after the May 2026 Scottish Parliament election, the Electoral Commission reported that 54% of people it surveyed saw perceived bias in the media as a problem in Scotland, one of the issues they were most concerned about.^[69][70][71]

Television and radio are regulated by Ofcom. Section Five of the Broadcasting Code requires due impartiality on matters of political controversy, and Section Six adds rules for election periods: due weight must be given to the coverage of parties and independent candidates, taking account of past and current electoral support; constituency reports must offer the chance to take part to candidates with significant support; discussion of election issues must end when polling stations open; and opinion poll results may not be broadcast on polling day until the polls close.^[72]

Broadcasting remains the most trusted and, at elections, the most used source. Ofcom’s 2026 research found that after the May 2026 elections television reached 62% of people aware of the elections as a source of election news, against 33% for online sources, with the BBC’s main bulletin the single biggest source. About seven in ten users rate TV, print and radio well for accuracy and trust, against four in ten users of social media.^[73][74]

Televised leaders’ debates began in 2010. After the first 2010 debate an average of four polls found 48% thought Nick Clegg had won; the Liberal Democrats reached 29% in one poll tracker but won 23.0% of the vote. In 2024 the first ITV debate between Sunak and Starmer on 4 June peaked at 5.5 million viewers, and YouGov’s snap poll found 51% thought Sunak performed best and 49% Starmer – effectively a tie.^[57][75][76]

Impartiality has also been contested between elections. In February 2025 the High Court quashed two Ofcom decisions that GB News had breached due impartiality rules when Jacob Rees-Mogg, then a Conservative MP, presented news items in his programme; the judge found Ofcom had made an error of law. In October 2025 Ofcom decided not to change its rule on politicians presenting news, issuing updated guidance instead. In November 2025 the BBC’s Director-General Tim Davie and News chief executive Deborah Turness resigned after criticism of a Panorama edit of a speech by Donald Trump; Turness said that allegations that BBC News was institutionally biased were wrong.^{[70][77][78][71]}

The way people get news has shifted online. Ofcom found that in 2024 online sources (71%) overtook television (70%) as the most used source of news for the first time, and that 11% of adults used TikTok for news, up from 1% in 2020; 88% of 16–24-year-olds used online sources. In 2025, 51% of adults used social media for news. Ofcom’s 2026 survey, which measures use in the past month rather than the past year, found 60% used social media or video platforms for news, including 78% of 16–24s, and 21% had used an AI app or service for news.^[68][74][73]

Parties now spend heavily on targeted online advertising, aiming different messages at different groups of voters. Researchers citing Who Targets Me data report that parties spent over £11.2 million on Facebook and Google advertising at the 2024 election. In the first week of that campaign Labour spent about £684,000 on online ads and the Conservatives about £280,000, according to Tortoise Media.^[79][80]

Online campaigning raises questions of transparency. In 2019 the Conservative press office renamed its verified Twitter account ‘factcheckUK’ during the ITV leaders’ debate; the Electoral Commission said voters were entitled to transparency and integrity from campaigners, and Twitter warned of action against further attempts to mislead. Since 1 November 2023, under the Elections Act 2022, most digital campaign material must carry an imprint showing who paid for and produced it. A study of the 2024 election found that only 68.9% of material that needed an imprint had one, falling to 59.9% on TikTok.^[81][82][79]

Claims about social media’s direct effect on votes need care. After 2017 many commentators credited online campaigning with a ‘youthquake’ of young voters, but the British Election Study found no surge in youth turnout. Social media may matter more by reinforcing existing views, spreading misinformation or setting the agenda for traditional media than by changing votes directly.^[83]

New risks emerged at the May 2026 elections. An Electoral Commission pilot scanned about 3 million posts and found 219 political deepfakes from 60 accounts, viewed 9.43 million times, but no evidence of deliberate use of highly convincing deepfakes to interfere with the election; the Commission noted that political deepfakes largely fall within a regulatory ‘grey zone’. The Reuters Institute’s 2026 report found trust in news in the UK at 30%, with 18% consuming content from a news-focused creator or influencer each week.^[69][84]

Opinion polls estimate public support from samples of voters. Their accuracy matters because parties, journalists and voters use them. In 1992 the polls underestimated the Conservative lead by nearly 9 percentage points, described by the House of Commons Library as the worst performance in UK polling history; the Conservatives won more votes than any party in British history.^[56][85]

In 2015 the polls put the Conservatives and Labour roughly level, but the Conservatives won by about 6.5 points and a majority. An inquiry for the British Polling Council and the Market Research Society, chaired by Professor Patrick Sturgis, reported on 31 March 2016 that the main cause was unrepresentative samples, which contained too many Labour and too few Conservative supporters; a late swing added at most about one point to the error, and there was evidence of ‘herding’ among pollsters.^[56][58][86]

Polls also shape campaigns. With polls pointing to a hung parliament in 2015, the Conservatives’ March 2015 poster showed Ed Miliband in the pocket of the SNP’s Alex Salmond, warning of a Labour government propped up by the SNP. Polls help voters decide whether to vote tactically, and they can affect party morale and how electable leaders are seen to be. Some argue that polls create bandwagon effects or discourage turnout when a result looks certain.^[56][87][7]

In 2024 the polls correctly showed a large Labour lead but overstated Labour’s vote share; the British Polling Council noted that YouGov’s model was the most accurate on seats. Polls also differ from one another: in October 2026 Ipsos (15–22 October) put Labour on 32%, the Conservatives on 24% and Reform UK on 20%, while YouGov (25–26 October) had Labour 27%, Conservatives 22% and Reform UK 20%. Ipsos notes a 9 in 10 chance that each party’s true figure lies within 4 points of its estimate.^[88][89][90]

Some countries restrict polls before elections, but in the UK only broadcasters face a limit (on polling day itself). YouGov’s Peter Kellner argued against bans, saying that they cannot stop polls being taken, only published, and that France’s former ban pushed polling into private surveys for wealthy clients.^[72][85]