Blair Freebairn helping us to understand more the battles ahead
â€œThe curse of dimensionalityâ€ is a term used by statisticians to describe the exponential increase in complexity as you add variables (aka dimensions) to a problem.
So, in the interests of simplicity, political commentators rarely venture beyond one dimension; this is most limiting in the traditional reporting of swing, changes in vote share are reported using a single number.
However one of the defining themes of post-war elections has been the steady growth in the non-Labour/Conservative vote. In order to accommodate three blocks of vote we ought to add an extra dimension. Instead of change moving along a line, it now travels over a surface.
Using the conservative vote share as the x-axis and the Labour share as the y-axis generates the surface. On the first graphic the 10% winning margin contours are shown together with the two recent by-election results and Nick Palmerâ€™s seat.
Using a surface opens up visualisation options. For example the evolution from neat line in 1992 to blob in 2005 illustrates the growth in Lab/Lib marginals. Meanwhile visualising swing at general election on a seat-by-seat basis would produce a â€˜vector fieldâ€™ similar to the wind strength and direction maps seen on weather forecasts.
Recently the BBC graphics department has produced a series of rubbish electoral visualisations. Adapting this 2-D approach could revitalise the â€œPeter Snowâ€ segment, it has to be better than Jeremy Vineâ€™s cowboy antics.
Blair Freebairn is an occasional poster but long time reader of pb.com. For a living he helps various organisations make good decisions where location is important (new stores, wi-fi hotspots, hospitals that sort of thing)