At dawn on 9 June last year, people woke up, bleary-eyed to a shock. In the days leading up to the election, the Conservatives were sure-fire winners. Despite May’s crumbling campaign, the polls put the Tories well ahead, until the final exit poll on the evening of the 8th. The polls, which had buoyed the right and strengthened the resolve of the left, had been wrong.
This marked the third major political event that the polling industry had failed to call in as many years. They had got it wrong in both the 2015 election and the EU referendum, and now the 2017 election. In the aftermath of the snap election, countless frustrated headlines emerged questioning why the polls were so wrong, and whether they could still be trusted.
Now, in the run up to this year’s local elections, the House of Lords has published its investigation into the state of British polling – and the peers aren’t happy. David Lipsey, the chair of the committee, declared: “The polling industry needs to get its house in order.”
The report began scathingly: “To misquote Oscar Wilde, to get one election wrong may be regarded as a misfortune, to get two wrong looks like carelessness, and to get three wrong suggests something somewhere has gone horribly amiss.” Ouch.
So why is it so hard for pollster to get it right nowadays and are they actually getting less accurate? Partly to blame is the country’s shifting demographics – there’s been a huge decline in class-based voting. What once was a key predictor for polling organisations is now of little value, and in the EU referendum, education mattered much more than social class.
Turnout is also an issue. In 2017, the polling companies wrongly supposed that turnout patterns would be broadly the same as two years previously – but more under-50s voted than expected, and they were considerably more likely to vote Labour. To top it off, voters are much more volatile nowadays, and switch between parties more often. All of which makes the results harder to predict.
Also in the sights of the Lords is the media. We have a problem. Journalists have a tendency to obsessively report on polls, regardless of whether they show any significant change. They also suggested a culture of wrongly implying such surveys predicted the outcome of elections, rather than giving a snapshot of public opinion at the time.
The Lords called on the pollsters’ trade body, the British Polling Council, to “name and shame” poor examples of polling reporting in the media. (But thankfully for politicos, the Lords stopped short of recommending a ban on voting intention polls in the run up to elections.)
But is all this fuss over nothing? Analysis of political polls since the 1940s shows no evidence that the industry is becoming worse at its job. Political science professor Will Jennings told the committee “there is no evidence of a global crisis in polling” and that the historical accuracy of polling in the UK was typical of similar advanced democracies.
So the real problem is not a crisis itself, but the perception of one.