General election polls: How they’ve changed since Theresa May made her shock announcement

What lies behind the rise and rise of Labour over the past seven weeks?

Sources: UK Polling Report, The Independent

When Theresa May called the election, I asked if Labour might be heading for its worst defeat since 1931. It has not worked out like that. Although Conservative support rose in the days before the Prime Minister’s announcement and remained high for the next four weeks, Labour’s support has grown steadily throughout the entire seven-week campaign.

Are we too obsessed with polls?

The one thing everyone agreed after the 2015 election was that journalists had become too obsessed with opinion polls, and had allowed them to dictate the news headlines too much. One of the big stories of the last election campaign was what Ed Miliband would do if he needed Scottish National Party support in a hung parliament: that was pushed by polls pointing in that direction that turned out to be wrong.

I’m not sure about that criticism, because the only way Miliband was likely to win would have been in a hung parliament, so it was fair to ask what he would have done about it. But, yes, an obsession with who is up and down in each poll is a distraction from the question of which team is better placed to govern the country.

That said, the trends in the polls have been so strong in this election – the trends hardly moved during the 2015 campaign – that things are clearly happening, so let us try to work out what they might be.

What happened when Theresa May called the election?

Theresa May, like a bolt out of the blue, called the election on 18 April. Oddly enough, Tory support was already rising to above 45 per cent before she did so. I have no idea why. After her announcement, Labour support strengthened. I assume this was a partisan reaction to an unexpected election. People who had been unenthusiastic about Jeremy Corbyn’s leadership, telling pollsters they didn’t know how they would vote in an election that was probably three years away, suddenly reverted to their Labour allegiance.

Did the manifestos change anything?

The polls started to move again after Labour’s draft manifesto was leaked on 11 May, five days before its official launch. That was when Labour support really started rising and the Conservatives started to decline. A lot of attention was paid initially to the plans to renationalise the Royal Mail, railways and energy, followed up after the official launch with abolishing tuition fees and free childcare.

These all proved popular, and the Conservative counter-attack on how it was all going to be paid for was feeble. Then Theresa May got into trouble of her own with the launch of her manifesto on 18 May. The plan to force pensioners to pay for care visits in their own home out of the value of their property went down badly. Even if only one in six pensioners will ever need home visits, six out of six fear they might and they all have adult children who worry about it too.

All the while, the three minor parties drifted downwards, Ukip showing the most dramatic decline, but with the Lib Dems and the Greens also squeezed.

Labour’s rise continued, even if the decline in Tory support slowed, and the terrorist bombing in Manchester on 22 May did nothing to stop the gap between the two parties narrowing. Some of the polls suggest that the Tory lead has stopped shrinking as we entered the final week of the campaign, but the average of the most recent polls suggests that May’s lead is no greater than the margin of David Cameron’s victory two years ago, which gave him a slender majority of just 12.

Which polls should we believe?

The big question now, though, is which of the polls, now spread between a Tory lead of one point (Survation’s online poll on Sunday) and one of 12 points (ComRes for The Independent on Saturday), is a better guide to the vote on Thursday.

On this it is worth looking at this blog post by Nigel Marriott, a statistician, who is currently predicting a majority for the Conservatives of 100 seats.

Chart by Nigel Marriott, showing difference between modeller polls (blue) and self-reporters (red)

Marriott divides the polls into two groups. One, which he calls the self-reporters, including Survation, which give more weight to whether voters say they will vote, and another, called the modellers, which downplay what voters say about their likelihood to turn out and assume they will turn out roughly as their age and social group did last time. The problem is that Survation assumes that young people will turn out at about the same rate as older people (because that is what they tell Survation), which is unlikely, although their voting intentions are given less weight through other adjustments the company makes.

Wherever the parties end up on Thursday, the one thing that is clear is that Labour has had a good campaign and the Tories a poor one, and that Corbyn is unlikely to secure the worst result for the party since 1931.