Where did that come from?

How were the pollsters so wrong? What accounts for the discrepancy in predictions and the final result?

Anyone who went to sleep before the shock exit polls were announced last night could be forgiven for thinking they were still dreaming when they awoke to news of a likely Conservative majority.  This election more than any in living memory would appear to show how wrong the pollsters and pundits can be, with almost all predicting a finely balanced, hung parliament.  So how did they get it so badly wrong?

At the start of the week pollsters such as ComRes were still showing the two main parties locked in a dead head – a picture that fuelled the ongoing expectation of a hung parliament.  Indeed, since Labour has been able to achieve a majority with a smaller share of the vote historically, if anything, Labour looked like they would win the most MPs.

However, at time of writing, The Conservatives are expected to have won a 37% share of the national vote, Labour 31%, UKIP 13%, the Lib Dems 8%, the SNP 5%, the Green Party 4% and Plaid Cymru 1%.

There has been some discussion already of the varying accuracy of different polling methods – with online polling deemed in particular to be problematic.  Further problems are that people answering a telephone poll may not turn out to vote, or worse – they may not even by registered to vote!  Recent changes to voter registration rules have prompted concerns among senior Labour figures that Labour would suffer particularly hard from low voter registration.

Certainly it was the exit polls that proved to be spot on.  Not only are they conducted face-to-face, more importantly, exit polls ask people on their way out of the polling station, who they actually voted for.   Only they can capture the voter who changes their mind in the last 24-hours, who may have told a canvasser days earlier they would probably support their candidate.

National polls are of some use, and certainly the current predictions of national vote share match with a Conservative majority.   But in the first-past-the-post system, the battle in the marginal seats is crucial.  Lord Ashcroft’s seat-level polling has ensured there are more seat-level polls in the public domain than ever before.  Still, these failed entirely to predict the ultimate outcome.

Disillusioned activists this morning in marginal seats are scratching their heads at the difference between their canvas data (based on what local voters had told them) and the face of their new MP. Previously ‘safe’ Lib Dem seats like Vince Cable’s in Twickenham and Ed Davey in Kingston and Surbiton fell to the Tories. Labour failed to win the seats it expected to, such as North Warwickshire, Nuneaton, Enfield Southgate and Croydon Central.  Meanwhile Scotland was a wipe-out for Labour, with senior party figures like Douglas Alexander losing from a position of having a 16,000 vote majority in 2010.

In accounting for this late national swing, the national campaign strategies of the parties, the performance of their leaders and their national messages, all come to the fore.  Calls for Miliband’s resignation, due to his personal poor ratings, already abound.  Paddy Ashdown cited the successful ‘fear factor’ in the Conservatives’ strategy of scaring voters with the prospect of a Labour government beholden to the Scottish nationalists. Over the next hours and days the political futures of both Ed Miliband and Nick Clegg’s look gloomy.  The weeks and months that follow will see these two parties embark on a full review to understand how they failed first to capture voters in the numbers they hoped, then lost many who may have voted for them in the critical final hours of the campaign.