By Peter McLeod
Yes, the polls are going haywire. At the same time as YouGov and Survation put the race back to the 2015 result of a 5-7 point Tory win (disastrous for Theresa May, as I wrote a few weeks ago), ICM, ComRes and Panelbase have the Tories cruising to the 100-or- so seat majority that would make this election a success for the prime minister. Now YouGov, based on a complex multi-level regression model using constituency-level polling data, predicts a hung parliament – despite the fact that their own poll of 50 marginal Labour-held seats taken in early May would have seen the Tories win every single one of them on a uniform swing. Poll watchers casual and professional alike are increasingly baffled. Meanwhile all the pollsters who have numbers out there will be increasingly nervous.
Tonight: we reveal YouGov's first seat by seat projection of the campaign - suggests Tories fall 16 seats short of overall majority pic.twitter.com/8ouPRHTZ7m— Sam Coates Times (@SamCoatesTimes) May 30, 2017
In defence of the pollsters, they are firing at a moving target at the same time as they try to correct for what happened in 2015. This is a genuinely unusual campaign featuring a shocking amount of movement: Sir David Butler, who has been covering election polling since 1945, says the polls have changed more over the course of this campaign than in any he recalls. Although the pollsters differ in the particulars, all of them show Labour gaining significant vote share while the Tories' share deflates after an initial jump.
Beside what appear to be real shifts in opinion over time, the different approaches the pollsters have taken to adjust their turnout models since 2015 are driving the huge variation in what they're currently reporting. Essentially what they disagree about is whether we should believe young people when they say they are going to vote: if we do, Labour polls fairly well, if we don't (and history tells us we shouldn't), Labour polls badly.
Personally I'm encouraged by the variation across polls: it shows innovation is happening. A variety of new ideas should produce a variety of different outcomes: if everyone was still producing the same results, you'd worry about herding and intellectual stagnation. The variation means several firms will inevitably miss the final outcome, but those that call it right will give everyone an idea of the best way forward next time. This is natural selection in action and the whole pollster population should get healthier for it.
Methodological drama aside, we've witnessed a strange May. Back at the beginning of the month I pointed out four campaign unknowns that should have been keeping the PM up at night: the pollsters overcorrecting their models in favour of the Tories; the CPS charging Conservatives over their 2015 election accounting, throwing the Tory narrative out the window; the Lib Dems bouncing back; and Labour coming up with a surprisingly good campaign.
It's fair to say that the Lib Dem bounceback hasn't materialised. But the pollsters are no longer providing the assurance they were a few weeks ago and, while the CPS didn't charge anyone, Theresa May wilfully tossed away her control of the narrative with her U-turn on the social care manifesto commitment.
It's difficult to overstate how much of a misstep that was: not only did it undermine the Conservatives' story about strength and stability, it clobbered the most reliable and enthusiastic part of their coalition, the over-65s. It had a real impact too: ICM's poll of 12-14 May gave the prime minister a 29%-28% lead over Corbyn as the leader most trusted on "Protecting the interests of pensioners" in a poll that ran 24-26 May, Corbyn led 41%-24% on being trusted to "Look after the future of our pensioners." Among voters over 65, May went from a 43%-21% lead to a 30%- 39% deficit (yes, the questions are worded differently – presumably because the two polls were done for different clients – but I don't believe for a second that that difference is enough to cause a change of this magnitude).
And Labour have indeed surprised with their campaign, underpinned by a manifesto promising enormous, universal giveaways like free university tuition. In that early May column I noted that Corbyn was back in his natural environment and this was clear in this week's two TV set pieces.
Especially on Monday night's May vs. Corbyn programme on Channel 4 and Sky News. Corbyn proved to be a more confident and fluent speaker than May, he was really effective in pivoting back consistently to his core message without sounding like a robot.
When a small business owner challenged him on tax-and-spend, he talked about building a country where all kids can get a good education, when Paxman said he was weak because he couldn't get all his core beliefs into the Labour manifesto he affirmed his commitments to democracy and fighting for social justice. These play enormously well with Corbyn's base and, although he has done the opposite of creating the reassurance on fiscal responsibility that Labour needs in order to win the centre, it appears to be enough to bring 30%+ of the electorate with him.
Hammering home a prepared message while sounding like it's just occurred to you is one of the difficult, counter-intuitive skills that politicians have to master, and Corbyn's decades of vigorous campaigning have prepared him well. The relative success of Monday night cleared the path for him to appear on the BBC leaders' debate last night.
So a large part of the Tory campaign nightmare seems to have come true. Does that mean a hung parliament – or even a Labour win? No. May's non-appearance on the BBC and a series of awkward campaign encounters like the instantly-notorious non-interview with the Plymouth Herald have undermined her for sure. But it's difficult to envisage these setbacks amounting to the sort of disaster that would close the remaining gap between the two main parties.
Instead, take Amber Rudd's approach last night as a template for the rest of the week: a concerted attack on Labour's credentials on the basics of leadership, fiscal responsibility and Brexit. That ICM polling I mentioned showed that the three issues the electorate cares most about are the NHS, the economy, and Brexit. On the latter two, May has a 17 to 26 point lead over Corbyn even after the tribulations of the past few weeks. The Conservative hope will be to nail the final week and that, like with most stories, the audience will only really recall the beginning and the end of this campaign. That should lead to an increased Tory majority – but most likely not the blowout we expected a month ago.
Peter McLeod is a vice president at polling firm GQR Research. GQR was pollster to Labour at the 2015 election but does not work with the current leadership.
The opinions in politics.co.uk's Comment and Analysis section are those of the author and are no reflection of the views of the website or its owners.