h1

In the only 2018 polls to be tested against real results LAB shares were overstated by 7%+

September 19th, 2018

GE17 LAB polling understatement doesn’t mean that the same will happen next time

One of the things that true believer Corbynistas keep telling me on Twitter is that last year’s general election was a turning point in British politics and that the rules have changed. Thus anything that doesn’t fit into this narrative has to be swept away and dismissed.

A key point here is current polling both voting intention and leader ratings which don’t support the contention that their man is heading for an enormous victory next time. Given how understated Corbyn’s was in most polls at GE2017 it is little wonder that many Corbyn backers raise it in response to less than good current numbers. But is the undoubted polling failure of last year a good pointer to future elections?

Certainly LAB did far better in June last year than most predicted and their leader got most of the plaudits. But for Labour, it is sometimes forgotten, it was a third consecutive General Election defeat and the Tories remained in power.

One of the challenges when trying to assess polling accuracy is that there are very few occasions when election outcomes can be compared with actual pre-election polls. Since GE2017 there has been just when one set of elections when published surveys were put to the test – May’s London Borough elections.

YouGov in partnership with QMUL and the top GE2017 pollster Survation did produce surveys ahead of May’s borough elections in the capital and the party shares together with the GE2017 and the London local election party vote aggregates can be compared.

Those are all featured in the chart above and as can be seen LAB did much worse than any of the polling. The gap on the final polls was more than seven points and suggests that London Labour was being overstated.

Now there were special factors in May. The elections took place a few weeks after the party’s antisemitism crisis broke and the demonstration outside the Palace of Westminster. A feature of the London results was the very high turnout in areas with large Jewish populations and this had some impact on the overall numbers.

In recent years polling of local elections in London has proved to be pretty good. The 2016 Mayoral race was a case in point in 2012.

So I’d argue that the failure of some pollsters at GE17 cannot be taken as a reliable guide to the future in the same way that CON understatement at GE2015 was no guide to what happened two years later.

Mike Smithson