How the polls have changed

Just a quick update since there has been a huge flurry of polls since the election was called, and the polling average now shows the abrupt jump in Conservative support at the expense of UKIP.

A fair bit of excitement today over YouGov's best result for Labour for a while, but the uncertainty bands show that it's consistent with the average: most likely YouGov happened to randomly pick an unusually large number of Labour supporters.

In other news, highly-regarded pollster Nate Silver has had a bit of a moan about the accuracy of British polling: https://fivethirtyeight.com/features/the-u-k-snap-election-is-riskier-than-it-seems/

I think he's being a bit unfair, so I've had a go at a similar analysis, taking the polling average 50 days out from each of the last nine general elections and comparing Labour's final result to the prediction (the difference includes real swing and polling error). Having done this, I can conclude that the likelihood of Labour doing better than last time is very small, but not zero - the chance is about 1 in 38 (the chance of them gaining a majority, on the other hand, is <0.02%).