General Election Forecast based on Uniform Swing by Region

On May 15, YouGov produced a General Election opinion poll broken down by the eleven Regions of Great Britain. It seems impressive with 14395 GB Adults, plus 1040 London Adults, 1017 Scottish Adults and 1018 Welsh Adults. However, with fieldwork on 24 April to 05 May, it might be a little out of date. By combining this with the General Election Results by Constituency (available for the British Election Study) I have been able to produce a crude forecast for the General Election on June 8th.

The starting point in the General Election Results of May 2015, shown in Figure 1. Since then Con has gained 1 seat from Lab (Copeland), and lost a seat to LD (Richmond). The sole UKIP MP, Douglas Carswell, left the Party in March to become an Independent. These are ignored.

Note that the 18 Northern Ireland constituencies are not included. The 331 Con seats are against 326 for a majority in the House of Commons.

I made the following assumptions.

  1. Within each constituency, for each Party I have assumed the change in the vote is the difference between the regional share of the vote in 2015 and the opinion poll share from YouGov.
  2. If the constituency vote share in 2015 was less than the regional drop in vote share between 2015 and current opinion poll, then the vote is nil.
  3. A party may have a predicted vote despite not having a candidate. There are two instances where this is possible. First is that UKIP are not standing candidates in every constituency. Second is that the Progressive Alliance of Lib Dems, Greens, Labour and the SNP are standing down candidates to maximize the impact of the anti-Tory vote. Guido Fawkes’ summary of 16th May is here.

This simple model produces the forecast in Figure 2.

Implied Conservative majority is 128, up from 12 in the previous Parliament. The Lib-Dems also increase there number of seats, whilst SNP lose 9. UKIP’s “gain” is in Buckingham, the seat of the Speaker. This is due to a flaw in the crude model.

The Party gains by region are in Figure 3

Of note is that Labour do not gain a single seat, as YouGov estimate that their popularity has dropped in all but two regions. In the South East and the South West Labours’ presence is quite low. The SNP in Scotland lose seats to both the Conservatives and the Lib-Dems, but the loss of 9 seats is would still mean they have 47 of 59 seats.

These switches in seats are shown in detail in Figure 4.

My very crude forecast can be compared to the current forecast by Martin Baxtor at Electoral Calculus of Con 409, Lab 167, LD 7 & SNP 46. The Conservative majority is 168, 40 more than my own.

Another comparison is the mid-point of the spread betting at Sporting Index. This Con 399, Lab 159, LD 15.5, SNP 45.5. The implied Conservative majority is 148, bang in the middle of my own and Martin Baxters’.

In subsequent posts I intend to

  • Clear up the obvious errors.
  • Refining the forecast for Scotland based on the local election data of May 5th.
  • Look at the forecast for Wales, where I believe YouGov might be out of line with popular opinion.
  • Update in relation to more recent, but National, polling. For instance the recent strengthening of the Labour poll share and the fall in the UKIP share.

Kevin Marshall

 

Leave a comment

1 Comment

  1. Revised General Election Forecast gives Tories a Majority of 25 | ManicBeancounter

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: