LSE - Small Logo
LSE - Small Logo


December 23rd, 2019

The Exit Poll, BBC Election Night and systemic media bias

54 comments | 5 shares

Estimated reading time: 5 minutes


December 23rd, 2019

The Exit Poll, BBC Election Night and systemic media bias

54 comments | 5 shares

Estimated reading time: 5 minutes

Accusations of BBC media bias have flowed thick and fast from right and left, but the real scandal of the 2019 Election Night was that seats projections were announced at 10pm, the exit poll vote shares were only made public briefly (in ‘footnote to history’ fashion) 30 minutes later, while meaningful discussion on the parties’ national vote shares came only seven hours later, when almost all viewers had gone to bed. Pippa Norris and Patrick Dunleavy argue that focusing solely on seats (and power) formed the centrepiece of a thoroughly over-legitimizing representation of the UK’s election process, exaggerating the Conservative and SNP victories, overstating Labour’s poor performance, and ignoring the injustices meted out to the Liberal Democrats (who greatly increased their votes), Greens and others. A simple re-framing could easily combat the BBC’s and other broadcasters’ now firmly enrooted ‘bias against understanding’, entailing something of a move back to older and more accurate election night formats.

The main lines of British political culture over the next four and a half years were constituted by the election of 12 December, and by how the broadcasters, especially the BBC, represented the results overnight. Election night coverage remains one of the few ‘water cooler’ events in public service television. BBC One’s election night program peaked at 6.1 million viewers as the results of the exit poll were announced at 10pm. It drew in around 4.36 million on average from 9:55pm to 2am, with many others watching the BBC online. Millions more people were tuned to ITV, Sky and Chanel 4.

The centrepiece of election night programming across all broadcast channels was the single Exit Poll, conducted in 144 polling stations, with voters recasting their ballots anonymously for Ipsos MORI. From the change in votes since last time (at the self-same sites) an army of skilled analysts then dissects the new results to predict the overall seat outcomes for the BBC, ITV and Sky. ‘The principal aim of the exit poll’, said John Curtice the BBC guru in overall charge, ‘is to help viewers and listeners to navigate the initial hours of election night as the first results come in. By comparing the actual results with the forecast of the exit poll, we will be able to point to the political direction in which Britain is now apparently headed’.

In the event, the 2019 poll correctly predicted 368 Tory MPs (actual number 365), 191 for Labour (actual 203), 55 for the SNP (in fact they only won 48 of Scotland’s 59 seats, a major gaffe for the analysis here), and the Liberal Democrats 13 (actual 11). This precise prediction more or less eliminated all other perceptions, and incessantly dominated all further analysis and discussion for the first many hours of programming. A dominant narrative was established, with no effective counter-notes of any kind, proclaiming a Tory triumph, Labour wiped out in a historic defeat (widely represented as paralled only by Michael Foot in 1983), and an (as it turns out, overstated) SNP hegemony north of the border

The national vote share percentages from the Exit Poll were not quoted onscreen for half an hour, during which the dominant narrative in terms of seats and power was firmly established. It then emerged very briefly at 10:25pm that the Exit Poll rated Tory support at 46% and Labour at 32%, while a page on Sky News website gave other details. However, these vote share estimates did not form part of the running coverage. Instead, the BBC shifted over to showing a banner based on actual votes from the results announced so far. For the rest of the night, this alternated with the forecast seat numbers. Here is the display at 1 hour and 35 minutes into the show after the Blyth result was announced.

Despite the “GB votes” label, these are not vote shares but only vote change numbers (slightly misleadingly called % instead of a percentage point increases). As this example shows, with no start point provided to viewers these numbers were useless early on – notice the Labour share down 13 percentage points here and the Brexit vote share up 11 percentage points.

Of course, health warnings were given orally about this representation, but this way of giving information was not altered, although it did not help voters understand what national vote shares were. Even around four hours into Election Night, the alternating ‘GB votes’ display was not a lot of use to anyone – note the Brexit vote share rated up 5% and Labour’s decline still over-stated.

Constituency vote swings were occasionally highlighted, but without any background template and only then mainly in seats which experienced a particularly dramatic (and usually untypical) change. The narrative became a dramatic and exciting landslide of seats for Johnson – and the historic defeat of Labour MPs.

Only after 5am did the BBC’s Jeremy Vine at last prominently announce an estimated three-party national vote share for Britain, to a residual audience of insomniacs and election geeks.

And what a different story this told. Despite the Brexit Party standing down in their favour, the Conservative UK vote share increased by just 1.2% on their 2017 performance. And Labour’s 32.1% share of the UK vote under Corbyn was not historically poor, exceeding as it did Ed Miliband’s in 2015 (30.4%); Gordon Brown’s performance in 2010 (29.0%), or Neil Kinnock’s vote share in 1987 (30.8%). Indeed the 2019 Labour vote was just a couple of points behind their average performance since February 1974, when multiparty competition started to reduce the average two-party share of the vote. Labour’s vote share was down sharply on 2017 (-7.8%), driven by supply-side patterns of party competition which split the Remain camp. The Liberal Democrats under Jo Swinson had actually achieved a near 50% increase in their vote share, despite winning only two handfuls of seats. The divisions amongst the UK’s clear majority of the Remain voters were exacerbated by the UK’s electorally disproportional First-Past-the-Post electoral system. It returned to its typical form in 2019, vesting Boris Johnson with 13% more seats than his national vote share, and awarding four fifths of the Scottish seats to the SNP for 45% of votes there. There was no vast blue tsunami in the grassroots British electorate. Different choices on the ballot simply altered party fortunes, which the electoral system then reshaped and exaggerated.

Why did the broadcasters vest all their national analysis in the Exit Poll, an exercise which since methods were changed in recent years has not been primarily designed to generate an accurate estimate of the national vote share? Why was the exit poll vote share not compared with the plenty of reputable national polls conducted very close to the election day itself – which gave a vote share that later turned out to be pretty much spot on, as Figure 1 shows? Any political scientist could have told the BBC that the median result here was highly likely to be accurate on national vote share. And while the BBC had a self-denying ordinance of not really covering polls during the campaign, that all ended at 10pm on election night.

Figure 1: Vote shares (%) in the actual 2019 result and the last seven opinion polls

So, accurate vote share information was there for journalists and commentators to consider – it just could not be managed within the dominance of the Exit Poll ‘frame’ – with its implied claim that only seats outcomes count, that only what determines the immediate contours of power in Westminster matters, and that the UK’s biased electoral system in fact accurately captures the ‘will of the people’.

For the BBC especially, the 2019 election night was a conspicuous failure of the Reithian mission to educate and inform citizens at a critical juncture in political life in an open and multi-variant way. It ‘help[ed] viewers and listeners to navigate the initial hours of election night’ only in a one-sided, “only power matters” kind of way. Ironically this was a complete denial of the BBC’s valuable Election Night heritage. In the old days of David Butler and Robert McKenzie’s ‘swingometer’, changes in the national share of the votes provided a key focus of discussion and debate across the first hours of every election night. It filled the ‘empty hours’ while the seats results trickled in, and accurately located viewers in an overall view of votes and seats within the electoral process. A careful analysis of how voters actually cast their ballots nationally also celebrated citizens’ collective judgment, and then allowed citizens to better appreciate how vote national shares became converted to seats. By 2019, all this was long gone. The fancy graphics were all about seats, and all discussion focused on what seats alone implied about political power and future policy.

Journalistic framing conventions largely defined the seats-only Exit Poll perspective of the BBC and of other channels. Perceptions like these largely determine what is covered as newsworthy in public affairs. Frames reflect organized structural conventions in newsrooms, not individual choices or biases by reporters. Independent international media watchdogs have rated BBC news highly for their factual reporting, although perhaps slightly favouring the left in their news story selection. The Loughborough University content analysis of 2019 campaign news coverage, found a rough parity in coverage of the two major parties, but around two-thirds of TV news focusing on the Conservative and Labour campaigns.  Nevertheless, the Liberal Democrats and SNP were given more attention on TV news than in the printed press – although that is hardly a level playing field comparison.

Appreciating framing bias

A different frame would have been possible if the interpretative frames used by broadcast journalists were only just a little bit more prepared to ‘speak truth to power’. Here is how national vote shares could have been introduced, from the outset of the Election Night broadcast, based on a better-rounded Exit Poll (taken together with well-conducted national opinion polls) – at the same time as the mature seats projections were announced:

The Conservatives came out top convincingly, gaining their largest parliamentary majority (80 seats) since 1987. Yet their share of all votes under Boris Johnson was 43.6%, up by one percentage point from two years ago under Theresa May.

Under Jeremy Corbyn, Labour’s vote fell back from 40% in 2017 to 32%, a steep decline which saw the party’s MPs fall back towards 200. Yet its vote share was just 3 percentage points below the party’s average performance under successive leaders for the last half century. And Corbyn won a higher vote share than Gordon Brown in 2010, or Ed Miliband in 2015.

The biggest vote gain of the night went to the Scottish National Party, up over 8 percentage points to 45% of all votes in Scotland. Against a fragmented opposition, the party gained 48 (80%) of the 59 seats in Scotland under the Westminster election system.

Elsewhere the Liberal Democrats also grew their support by 4 percentage points, reflecting a surge of support from their clear Remain stance. Under Jo Swinson the party achieved their best share of the vote since 2010, but won only 11 seats.

The two parties advocating a hard Leave position towards Brexit were marginalized. The Brexit Party under Farage gained 2% of the overall UK vote, and UKIP just 0.1% support – a dramatic change since the Brexit Party won a 33% vote share in the May 2019 European elections.

Overall, reflecting the public’s position towards Brexit shown in recent opinion polls, the Leave parties won a combined share of the GB vote of 47%, compared with 53% for the Remain/2nd Referendum camp.

By contrast, here is the BBC’s actual final overall summary of the night’s outcome (focusing only on seats, and still ending with a salient exit poll mis-prediction of historically unprecedented poor Labour performance):

Boris Johnson will return to Downing Street with a big majority after the Conservatives swept aside Labour in its traditional heartlands.

With just a handful of seats left to declare in the general election, the BBC forecasts a Tory majority of 78.

The prime minister said it would give him a mandate to “get Brexit done” and take the UK out of the EU next month.

Jeremy Corbyn said Labour had a “very disappointing night” and he would not fight a future election.

The BBC forecast suggests the Tories will get 364 MPs, Labour 203, the SNP 48, the Lib Dems 12, Plaid Cymru four, the Greens one, and the Brexit Party none.

That means the Conservatives will have their biggest majority at Westminster since Margaret Thatcher’s 1987 election victory.

Labour, which has lost seats across the North, Midlands and Wales in places which backed Brexit in 2016, is facing its worst defeat since 1935.


Updated version: We thank all those who commented on an earlier version, and whose inputs are incorporated here.

About the Authors

Pippa Norris (@PippaN15) is the Maguire Lecturer in Comparative Politics at Harvard University and the author of numerous books on British and comparative politics, media politics, and (with Ronald Inglehart), Cultural Backlash (Cambridge University Press 2019).



Patrick Dunleavy (@PJDunleavy) is Professor of Political Science and Public Policy in the Government Department at LSE, and Centenary Professor at the Institute for Governance and Policy Analysis at the University of Canberra. His most recent books are The UK’s Changing Democracy (LSE Press, 2018) [co-edited], which is free to download; The Impact of the Social Sciences (Sage, 2014) [co-authored]; and Growing the Productivity of Government Services (Elgar, 2013) [co-authored].


All articles posted on this blog give the views of the author(s), and not the position of LSE British Politics and Policy, nor of the London School of Economics and Political Science.

Print Friendly, PDF & Email

About the author


Posted In: British and Irish Politics and Policy | Featured | Governance | LSE Comment


Creative Commons Attribution-NonCommercial-NoDerivs 3.0 Unported
This work by British Politics and Policy at LSE is licensed under a Creative Commons Attribution-NonCommercial-NoDerivs 3.0 Unported.