Hung parliament? Perpetual coalitions? Ed Miliband in Downing Street with Alex Salmond by his side? So the pollsters got it a teensy, tiny bit wrong. With a few percentage points this way or that, all the major opinion polls put the two main parties neck and neck as the vote loomed.
On the night, though, it became clear the Tories had swept to an unexpected victory. it appears they won 37% of the vote. That may only be 3% more than the final BBC poll of polls before the election, but 3% can clearly make a big difference in today’s politics.
To say this is embarrassing for an industry with its fingers supposedly on the British pulse is an understatement. It’s not as though market research firms all use the same approach either.
The fifteen companies that are members of the British Polling Council use different methodologies, which they are required to publish alongside the detailed results. YouGov for instance has an online-only panel of over 400,000 people in Britain, while Ipsos MORI employs 16,000 staff in 84 countries for random telephone and face-to-face questioning, as well as online questioning.
The art is in the weighting, attempting to make the sample as representative of the whole population as possible. The success of these businesses, which also poll for private and public sector clients on social issues, brand loyalty and the effectiveness of ad campaigns, is clearly built on trust that they’re getting it right.
Will their obviously getting this one wrong hurt them? Probably not. It doesn’t matter whether they failed to adequately account for the ‘shy Tory’ voter as in the infamous 1992 election, or whether there was a sudden last minute swing that they can’t be blamed for missing. The fact remains that their polls are still better than nothing.
If a fashion firm wanting to know how its advertising campaign is doing has a choice between intelligence that’s largely accurate but might occasionally be a little off, or no intelligence at all, which is it going to choose? The same applies for media businesses wanting to speculate about an election.
Of course accuracy matters, but crucially the opinion polls were almost universally off this time round. No one firm looks especially bad, and the industry as a whole is too big to be hit by one shaky result.
Take YouGov, for instance. It appears to have underestimated the Conservative vote by 3% and overestimated Labour’s by the same amount. Its shares didn’t slide as a result – they’re down 0.1% to 121.4p. Hardly a sign of commercial disaster. It may have been an embarrassing night for the pollsters, but it could have been worse. Just ask Vince Cable or Ed Balls.