Is this why Labour was so far ahead?

After my moan this morning about Progressive Scottish Opinion, the pollster that carried out last week’s Scottish Daily Mail Glasgow East poll, the information has been released to me within the past hour and two things stand out:-

Although respondents were asked how they voted at the general election this does not appear to have been used to weight the sample. The stated weightings were that it should be “be representative of the Scottish population by age, gender, social grade and regions.”

No certainty to vote question was asked – something that is standard practice amongst many pollsters who weight the response so that the views of those saying 10 out of 10 are regarded as being more important than those who don’t.

This element can be critical especially as declared Labour supporters in other surveys are much less certain to vote than other parties. Ipsos-Mori, it will be recalled, only include in their headline numbers those who are 100% certain and this has the affect in almost every survey of reducing Labour shares by a considerable amount.

So from the PSO figures we read that the number saying they will vote Labour on Thursday is a staggering 68% of the number who declared that they had voted for the party in May 2005. I don’t find that plausible.

My overall conclusion is that if the certainty question had been put and the sample had been past vote weighed then Labour’s lead, if any, would have been nothing like the published 17%.

By-election betting – live latest prices

Mike Smithson