Can the pollsters really end their pro-Labour bias?

Can the pollsters really end their pro-Labour bias?

v

How would more accurate polls affect gambling?

With the first post-May 5th opinion poll due out tomorrow there are encouraging signs that the polling industry is coming to grips with the issue that has dogged it for nearly half a century – the ongoing over-statement of Labour’s position which Martin Baxter calculated was 2.7% at the General Election.

At a seminar last week organised by the Adam Smith Institute a panel consisting of myself and representatives from MORI, YouGov and Populus discussed both formally and informally the General Election outcome, the pollsters’s performance, and likely future developments.

Michael Simmonds of Populus reported on work following up those who had told interviewers before the day that they were voting Labour and then said in the follow-up interview that they had not actually done so because they thought the party was safe. How will the pollster deal with this which obviously has a distorting effect? We don’t know but at least they are aware there is an issue and have sought to quantify the problem.

Roger Mortimore of MORI acknowledged there was tendency for pollsters overstate Labour which all came down to turn-out. Declared Labour supporters, they now believe, are more likely than others to say they are certain to vote when they do not in fact do so. As a result MORI is now discounting the proportion of Labour supporters when they answer this question. This is something we’ve argued for in the past and it would be great if other pollsters followed suit.

Peter Kellner of YouGov said that they had questioned more than 40,000 people on their panel immediately after May 5 to find out if and how they had voted and this will be stored and form the basis of future surveys. Like other pollsters who weight the previous vote YouGov have in the past had to make an estimate of those who say they voted Labour but did not actually do so. With their new data they should be able to produce more representative samples.

On the election itself Kellner noted that the Tories had dropped about 2% during the campaign and Labour would have had a very comfortable victory if it had not been for the emergence of Iraq as an issue in the final fortnight – a move that boosted the Lib Dems.

I thanked all the pollsters for over-stating Labour – particularly before the final polls – because it had kept the spread price of the party’s vote share artificially high and allowed me to make a very big profit!

Mike Smithson

Comments are closed.