Stay up to date with notifications from The Independent

Notifications can be managed in browser preferences.

The Independent's journalism is supported by our readers. When you purchase through links on our site, we may earn commission.

The new YouGov poll means this election is going to the wire

The much-vaunted YouGov MRP poll – the one that got it right last time – suggests Boris Johnson is heading for a small majority

John Rentoul
Wednesday 11 December 2019 00:25 GMT
Comments
General Election 2019: Opinion polls over the last seven days

The election is back on, with tonight’s YouGov MRP poll suggesting Boris Johnson is heading for a majority of 28 seats, with a 9-point lead in share of the vote. This would be good enough for the Conservatives if it were the result, but as a prediction, it is too close for their comfort.

A hung parliament is within the margin of error: YouGov says that its numbers mean that, if the survey design is good, there is a 95 per cent chance that the result could be between a hung parliament, with Johnson one seat short of a majority (even with the Democratic Unionist Party supporting him, which it might not), and a Tory majority of 84.

Of course, because the true picture is likelier to be in the middle of that range than at the extremes, and because YouGov’s MRP was so close to the result at the last election, optimistic Tories might be tempted to assume that Johnson is going to make it with a reasonable mandate for a five-year parliament.

And it is true that Survation, the conventional pollster which was closest to the result in 2017, is now showing a 14-point Tory lead.

But there are good reasons for thinking that this election is still too close to call. One is that this poll shows a narrowing of the Tory lead from the first YouGov MRP poll just two weeks ago. That showed an 11-point lead and projected a Conservative majority of 68 seats.

The main reason, though, is statistical uncertainty. Polls can never be totally accurate, not even MRP ones, which use huge samples, complicated maths and powerful computers to produce seat-by-seat estimates for the whole country. Just because YouGov’s MRP poll got it nearly right last time does not mean it is certain to do so again. That is why the company publishes the range of projections, from the Conservatives winning 311 seats to 367, with a central estimate of 339.

These figures are sensitive to small changes in shares of the vote, which is why projections for Johnson’s possible majority vary so widely. A rival MRP poll by Focaldata, which has also produced the seat-by-seat estimates for the Best for Britain tactical voting website, gives the Conservatives a majority of just 24. It too shows a big reduction from an earlier survey, carried out between 15 October and 24 November, of a majority of 82. Another MRP poll by Datapraxis, for The Sunday Times, put the majority at 38.

This compares with the picture from conventional opinion polls, which range from ICM, showing a 6-point Tory lead that would give Boris Johnson a majority of about 10, to Opinium, with a 15-point lead that would produce a majority of 130. The average of the most recent polls from the 10 active polling companies is an 11-point lead, suggesting a majority of 68.

Labour should be worried that their support has failed to rise as dramatically as it did last time, and that Jeremy Corbyn’s personal ratings still lag so far behind the prime minister’s. The Conservatives are still ahead on managing the economy, by 15 points.

But a poll suggesting a majority of 28 is virtually no majority at all, as it would take just a tiny amount of poll error, shifting opinion or statistical noise to wipe that out altogether. This election is going to the wire – by which I mean a Professor Sir John Curtice exit poll at 10pm on Thursday night.

Join our commenting forum

Join thought-provoking conversations, follow other Independent readers and see their replies

Comments

Thank you for registering

Please refresh the page or navigate to another page on the site to be automatically logged inPlease refresh your browser to be logged in