Dealing with the polls’ Labour bias

Dealing with the polls’ Labour bias


    Creating a tool for General Election betting

One of the great challenges with General Election betting is the opinion poll bias to Labour. At the last three General Elections this has been an average 6.5%. At the Euro Election in June even the internet pollster, YouGov overstated Labour and understated the Tories.

To help with this we are proposing what we are calling the Politicalbetting Polls Variance Calculation which weights and averages poll results according to their proven record in actual elections.

Thus we take,say, the 14.9% that the Lib Dems got at the Euro Elections and divide it by the 13% that YouGov predicted to produce a figure with which we can weight Lib Dems totals in new YouGov polls.

We do this for each of the parties from two pollsters and produce an average.So far only YouGov’s Daily Telegraph poll and Populus in the Times are included in the calculation because they were the only firms which carried out Euro Election surveys against which they could be tested.

These Euro polls should have been more accurate than usual because a large part of the electorate had already voted by post so the polls took on the form, for this group at least, of an exit poll.

The effect is to boost the Tory figure and depress Labour, which is not surprising given what’s happened in the past. The “errors” of 1997 and 2001 were obscured by the scale of the Labour victory. The 9.5% bias in 1992 caused most people to call it wrongly.

This is the first Politicalbetting Polls Variance Calculation . Make of it what you will.

CON 34.2%: LAB 30.2%: LIB22.6%

This reinforces our strong and longstanding CALL on Labour to win the next General Election because even with the number crunching Michael Howard is still some way off where he needs to be.

Anybody contemplating a bet on the Birmingham and Leicester by-elections or the General Election might find this new calculation helpful.

Comments are closed.