Time to party like it’s 2017? YouGov’s seat-by-seat projection of the general election result has predicted a 68-seat Conservative majority if voters went to the polls today – their biggest win since 1987. Of all the surveys of the 2019 campaign, none will generate quite so much excitement as this one: ahead of the last election, YouGov’s model correctly called a hung parliament when few MPs and commentators were willing to entertain the prospect.
The predicted scores on the doors see the Tories up on 359 seats, up 42 from 317; Labour on 211, down 51 from 262; the SNP on 43, up seven; and the Liberal Democrats on 13, up one (despite doubling their share of the vote). Plaid Cymru would stay on four, while Caroline Lucas would retain Brighton Pavilion for the Greens. None of the independent candidates in the running would retain their seats – and nor, for that matter, would any Liberal Democrat defectors.
What do the numbers tell us? As Stephen argues here, it is, of course, worth remembering that the most accurate answer to that question is that they tell us what one pollster thinks the result will be. YouGov’s multilevel regression and post-stratification model – or MRP, which extrapolates constituency predictions from a large national sample of 100,319 people – might have been mythologised as a result of its headline accuracy in 2017, but its seat projections underestimated the Conservatives and overestimated Labour and the SNP, and as a result were some way off the final result.
But it just so happens that this time it paints the picture we would broadly expect from most other polls. Labour’s 44 losses to the Conservatives almost exclusively come in the Leave seats in the north, midlands, and North Wales – and include redoubts like Bolsover in Derbyshire, home to Dennis Skinner, Caroline Flint’s Don Valley seat, West Bromwich East, recently vacated by Tom Watson, and Leigh, Andy Burnham’s old stomping ground. Elsewhere, the story is of Conservative resilience in Scotland – where, far from inviting extinction, Johnson would hold on to 11 of 13 seats – and of woefully inefficient distribution of the 14 per cent Lib Dem share under first-past-the-post, which means only three Tory seats turn yellow (while three others, all of which voted Leave, are lost).
For all the excitement, however, the MRP cannot and is not attempting to predict the future. It assumes a Tory lead of 10 points nationally, and all of the polls show that gap narrowing. There are 67 seats in which the Conservatives or Labour lead by just five points over one another, and two weeks to go. Whether or not these numbers are right, they will undoubtedly change the race. Those in the shadow cabinet and Labour leader’s office who believed that the party’s electoral vulnerability was in Leave constituencies rather than among Remain voters feel vindicated. We can expect Labour’s campaigning posture to change accordingly. For the Liberal Democrats, meanwhile, the numbers are exactly the squeeze message they want to send to Remainers in its Conservative battlegrounds: a vote for us can’t put Jeremy Corbyn into office. In other seats, like Kensington, it will allow them to pitch themselves as the strongest anti-Tory force.
It’s for that reason that many Tories think the hype over Johnson’s hypothetical majority could well be bad news. As the recently demobbed Dominic Cummings warned on his blog last night, that many will now chalk up the race as a foregone conclusion could impact turnout and voter intention in statistically significant ways. That YouGov was broadly right in 2017 is no comfort. Instead, they fear that the elevation of the MRP to holy writ could end up producing a similarly unexpected result. Should that scenario materialise, we might once more find ourselves looking to 18 seats in the part of the UK that neither this pollster nor any other in Great Britain has touched: Northern Ireland.