30 March 2015

Forecasting GE2015

It isn’t necessary to go into the detail of this chart from the Economist to see how the two main parties’ market share of the UK electorate has declined since 1945:

This has meant that forecasting the outcome of UK general elections has become more difficult although the tools for doing so have become more powerful. Various academic political scientists have constructed models making use of polling, census and past election data to forecast voting at constituency level and then the number of seats each party might win. On 27 March the forecasters met in London and shared their current estimates as to how the 650 UK parliamentary seats might be distributed, as tweeted the following day by Matthew Goodwin:

At first sight, there is quite a large spread in these figures but the chart below might put it into context. The number of Conservative (blue) and Labour (red) seats for the last five general elections is shown and then the forecasts (and spread thereof) taken from three sources. I have chosen electionforecast.co.uk because it is the UK partner for FiveThirtyEight (the Editor-in-Chief of FiveThirtyEight is Nate Silver, author of The Signal and the Noise). The other two appear to me to be the most credible as they are attempting to address the number of seats for all the parties and are avoiding obvious outliers like the SNP with only nine seats. As in 2010, a hung parliament looks likely and there is currently much discussion of various coalition-type arrangements and combinations.

Outside the UK (and quite possibly inside), not everyone may be aware that the four “nations” of the United Kingdom are disparate in terms of the Westminster parliamentary seats they have currently (there were more Scottish seats before the Scottish Parliament was established):

(England, Scotland, Wales, Northern Ireland; Conservative, Labour Liberal Democrat, Others includes the Speaker). 

The Conservatives are almost confined to England and Wales and, if they are to form a single-party majority at Westminster, that is where they have to secure the seats to do so. If the Scottish Nationalists (SNP) do as well as is expected (see the second chart above) they will have about 2/3 of the Scottish seats and Labour will be in a similar situation. The next chart shows the relationship in England and Wales for the last five general elections between the votes cast for Conservatives and Labour and the seats the parties obtained, both expressed as percentages.

The first-past-the-post voting system tends to reward parties in that as their share of the vote increases they gain disproportionately more seats (otherwise the dotted grey line would apply). However Labour does even better than the Conservatives (the thin red line through their recent election results as opposed to the thin blue one) because of the distribution of the electorate across constituencies.

As long as the two main parties have similar low 30% voting support in polls, it suggests that Labour will be the largest party in E&W although probably not achieving an overall Westminster majority. Because the five Sinn Fein MPs do not attend Westminster, the theoretical minimum for an overall majority is 323. Labour would also have support on from the remaining Scottish Labour MPs, on many issues from the SNP, and, as the largest party, on many issues the LibDems.

If the Conservatives break through to the upper 30% level – this would require squeezing UKIP, there are only hard-core LibDems left to squeeze – leaving Labour behind, they become the largest party, and as such may turn again to the LibDems for support. Whether they would get more seats together than Labour and SNP combined is difficult to say and best left to FiveThirtyEight & Co. 

My hunch is that the LibDems will do better in their established areas than the national polling would suggest, and also that Labour will do better in Scotland than some of the forecasts are indicating.

No comments:

Post a Comment