After a year of forecasting, what’s changed?

Stephen Fisher, 24th October 2014

I first published a forecast of the 2015 general election result in October 2013. After taking on board comments and testing more candidate models, in February I revised the method from the one in this working paper to the one in this working paper. Both use opinion polls and election results going back to the 1950s to tell us what is likely to happen in this electoral cycle, and importantly, how sure we can be that it will happen. The historical pattern suggested governments tend to recover from mid-term blues while oppositions suffer a set back. Also the polls have tended to overestimate Labour and underestimate the Conservatives. Both factors suggested a Conservative lead at the next election. But also the variation in previous cycles was plenty enough to suggest very different outcomes were also possible if less likely.

Using a polling average for 8th October 2013 of Conservatives 32%, Labour 39% and Liberal Democrats 10% the revised method suggested a 42% chance of a Conservative overall majority. Over the last year the probability of a Conservative majority dropped steadily to 24% now. Why? Continue reading After a year of forecasting, what’s changed?


What the Scottish Independence Referendum results tell us

Stephen Fisher, 19th September

Alex Salmond claimed that there were no No votes, just deferred Yes votes. They were deferred too late for the independence cause. But taking that attitude is possibly the best way of explaining the overall No vote in yesterday’s Scottish Independence referendum.

It is much easier to ask what factors led so many people to vote Yes than it is to ask why more people voted No. Up to a month ago the polls showed large leads for No. Voting to stay in the union was the default position. The key question is how the Yes camp managed to come so close even if they still did not actually manage to win. Continue reading What the Scottish Independence Referendum results tell us

Election analysis and forecasting