Is this the first polling move in the post-Ken world?

In a number commentaries ahead of the May 1st London mayoral election I questioned whether the telephone pollsters were over-stating Ken’s support because of the way they calculated voting certainty. In fact it was this element that caused me to maintain a big betting position on the Tory even when the phone polling firms had Ken ahead.

For it was clear from the turnout history of previous mayoral races that the phone pollsters had far too many in their samples saying they were certain to vote and this, I argued, was skewing their figures. And the tendency, as was shown in the final results, was to overstate Labour and Ken.

For the problem is always that when asked how certain it is that respondents will vote a number like to show they are good citizens by rating themselves 10/10 when they probably won’t bother.

As was touched on here a couple of days ago when looking at the detail of the ICM Crewe and Nantwich poll the firm had adopted a different way of asking about voting certainty and of calculating its suggested vote shares.

In the new approach ICM asked about people’s attitude to voting itself and then weighted their turnout score in line with those answers. So three options were offered – whether they considered that it was their duty to vote; whether respondents should only vote if they cared who won, or whether it was really not worth doing at all. So those only who rated themselves 10/10 and agreed with the first option were given the highest turnout rating.

This increased the Tory share by 1% and decreased the Labour one by 1%. So the Tory lead of Labour was 2% up on what it would have been using ICM’s previous approach.

The interesting thing now is whether ICM will continue with this and what the other pollsters will do. We know that Ipsos-MORI is examining its methodology. ICM has always shown itself ready to adapt in the light of experience. After the 1992 polling debacle it brought in past vote weighting which now all but MORI follow.

So could the attitude to voting test become the standard? Possibly – and the effect is likely to be to increase marginally Tory shares and to depress Labour ones.

For those who follow the polls in their gambling this could be critical. A two point difference could have a big impact on the number of Commons seats won.

Mike Smithson