What do past elections tell us about the accuracy of the polls?

Labour has been gaining ground in the polls. The degree of its success, however, and whether or not the term “momentum” can be applied to anything other than its grassroots organisation, is still up for debate. That said, clearly Jeremy Corbyn’s party has improved on its position on 18^th April, when Theresa May called a general election hoping for a landslide.

According to YouGov, Labour has cut a gap of 24 points to 3, with the prime minister projected to be short of a majority. Other polls such as Opinium maintain that the Tories have a firm lead, though that is still down from 20 points to 11. Although the numbers vary, they have all swung from a massive Conservative lead in the direction of a minor-moderate Conservative advantage. Irrespective of their methodology, and whether they are predicting regional swings or an influx of young voters, Corbyn has jumped over a very low bar.

Of course, the polls can get it wrong, including exit polls (such as in the 1992 election), in which people say who they have voted for rather than merely who they intend to vote for. Polling is not an exact science. Yet the overall record of predicting the performance of the two biggest UK parties, Conservatives and Labour, has been historically solid.

The 1979 election, which saw Margaret Thatcher secure a 43-seat majority at the expense of James Callaghan’s ailing Labour party, was in keeping with the polls. A rogue Daily Mail/NOP poll placed Labour neck-and-neck with the Tories two days before the election, but otherwise the polls had predicted a Conservative victory since January of that year. In the month before the election, the number-crunchers had pegged Thatcher’s lead at seven points; she duly won 44 per cent of the vote to Labour’s 37 per cent.

A similar story can be found in the 1983 election, where the polls were unanimously predicting a double-digit lead for the Conservatives as early as June 1982. The caveat here would be that Thatcher’s second term was chiefly secured after winning the Falklands War, though the Labour manifesto, dubbed the “longest suicide note in history,” may well have contributed. In the end, the Conservatives won by 14 points with the polls correctly tracking the shift from early 1982, when Labour was over 10 points ahead.

The 1987 election was very much in line with polls, a 12-point Conservative victory. The outlier was the 1992 election, where the majority of polls showed a slight Labour lead, with the exit polls predicting a hung parliament. In the event, the Conservatives won 42 per cent of the vote to Labour’s 34 per cent. This discrepancy was put down to “Shy Tories” being unwilling to disclose their voting intentions to pollsters. When polling errs, it tends to under-predict Conservative success—no consolation for Labour here.

The 1997 Labour landslide was well-predicted, as was Tony Blair’s second term in 2001, with an unprecedentedly low turnout of 59 per cent. Labour’s lead was over-estimated for the 2005 election, but its victory came as no surprise.  In 2010, Britain’s first hung parliament in decades was predicted almost to the exact seats, with only the over-estimation of Liberal Democrat support preventing a perfect score in the exit polls. In the run-up to the election, David Cameron’s polling lead had been decreasing but it never dissipated completely, and it was still enough to make the Tories the largest party in parliament.

Interestingly, even the 2015 election, where pollsters were maligned for their forecasts, wasn’t as bad a performance as popularly believed. In the last week of the campaign, relatively few polls predicted a Labour victory, more had the major parties tied and the final polls by SurveyMonkey, Ipsos Mori and ComRes had the Conservatives ahead.

So Jeremy Corbyn can take heart from his improving poll ratings, which have defied the low expectations of the media. Nonetheless, given the track record of the polls, the prospect of a shock result should be taken with a pinch of salt.