2019 UK General Election Opinion Polls Kalman Filter

2019-12-12

This uses as input the table of opinion polls on Wikipedia's Opinion polling for the 2019 United Kingdom general election. The polling companies adjust the raw data of voter intentions by making various assumptions about likelihood to vote. Unfortunately, I don't have the raw data the polling companies use, and have to use the figures they release, so that introduces an unknown error. Each poll is spread out over several days, so I split the data evenly over those days, combining it with any other polls taking place at the same time. I then run a Kalman filter. For the days on which no polling takes place, I use the previous day's estimate in place of polling data. I've added the major political news stories which may have influenced voting intentions.

A Kalman filter estimates the current value, which lags behind the most recent measurements during trends. So I've used weighted least squares best-fitting straight lines during the current trends for the three main parties, to give predictions for the final votes. How this converts into seats depends on the extent of tactical voting, but I've given numbers suggested by Electoral Calculus's User-defined Prediction, which suggests a Tory majority of 36.

I'll try to keep this up to date with the latest polls.

Up

© Copyright Donald Fisk 2019