The YouGov poll is not the end of the election – there is still huge uncertainty ahead

YouGov's MRP does not stand for ‘Magically Realistic Polling’: all surveys can be wrong, and there are still two weeks of the campaign to go

The YouGov MRP poll achieved a semi-mythical status at the last election by suggesting, 10 days before polling day, that Theresa May might lose her majority.

Wednesday night’s version of the same poll lacks that shock value, projecting pretty much what the normal opinion polls have been saying.

Its projection is a Conservative majority of 68, comfortably in the middle of the spread of outcomes suggested by the standard opinion polls, which range from a majority of about zero to one of 200, with the average around 40.

Since YouGov’s success with the MRP model in 2017, there have been some early rival MRP polls this time round. The tactical voting website Best for Britain launched one at the start of the campaign, and issued an update today, based on polling between 15 October and 24 November, which projected a Tory majority of 82.

A Datapraxis MRP poll in The Sunday Times suggested a majority of 48.

Wednesday’s YouGov poll used the most recent information, basing its model on 100,000 interviews over the past seven days, ending at 4pm on Tuesday (before Jeremy Corbyn’s interview with Andrew Neil, not that that was likely to move opinion significantly on its own).

The reputation of YouGov’s MRP is richly deserved, but MRP does not stand for Magically Realistic Polling. It is a way of using large and expensive samples to generate seat-by-seat estimates of how people are likely to vote. It depends, just as conventional polling does, on the quality of the information fed into the machine, and on the quality of the adjustments made to it.

We should remember that there were two other MRP polls that didn’t perform well in 2017, and one conventional poll that did just as well as YouGov’s MRP – Survation suggested the Tories had a one-point lead in its final poll, when the average of the conventional polls put them eight points ahead (Theresa May ended up with a 2.4-point lead).

That is why Labour supporters became known for the catchphrase “I’ll wait for Survation,” whenever an adverse poll was published. But the most recent Survation poll gave the Tories an 11-point lead – exactly the same as the national lead in tonight’s YouGov MRP poll.

If it is any consolation to Labour supporters who were hoping that YouGov would finally reveal the truth of Corbyn’s secret popularity to a nation blinded by “Tory-biased polls”, it could be that YouGov were lucky with their MRP poll last time. It could be that, this time, it is ICM and Savanta ComRes who have got it right. Their most recent polls both show the Conservatives just seven points ahead of Labour, an outcome that could leave Boris Johnson teetering between a hung parliament and the slenderest of majorities.

Unfortunately for Labour supporters, it is equally likely that most of the polls, whether they are the “magic” MRP kind or cheaper standard surveys, have got it wrong in the other direction, and that Opinium (19-point Tory lead) or The Independent’s BMG (13-point lead) are closer to how people currently intend to vote, suggesting a possible Johnson landslide.

We shouldn’t discount stuff happening in the final two weeks of the campaign – and YouGov will be running another MRP poll before 12 December. And we should recognise that opinion polls change behaviour as well as trying to record it.

If the story of this campaign turns into one of hubristic Johnson heading for a comfortable win, it may be that many voters will change their vote as a reaction against giving him too much power – much as they reacted against Theresa May appearing to take a large majority for granted.