This event will introduce and explain a wide variety of different election forecasting methods for UK general elections, including aggregate opinion poll modelling, multi-level regression and post-stratification (MRP), citizen forecasts, betting markets, elite surveys and combined forecasts. The historical track record of the different methods are reviewed, including accuracy of opinion poll methodology.
The event is scheduled for the aftermath of the 2024 UK general election. Forecasts from before that election will be reviewed and analysed. There will be some discussion of possible lessons from the 2024 election for forecasting elections still to come.
As well as pre-election forecasts, the day will cover exit poll methodology and seat prediction from early constituency results.
There are no pre-requisites for this event except a broad familiarity with the basic ideas of random sampling, statistical significance and regression. The ideas and methodology will be presented as much as possible in a way that would be accessible to a general reader interested in elections, but there may be a few simple equations presented. Methods will be explained and discussed on a conceptual level. Some of the practical issues with implementation will be discussed, but not detailed computer coding. Note: this is not a hands-on practical workshop.
Please note: this event will close to enrolments at 23:59 UTC on 26 March 2025.