h1

Why we need more polls of the marginals

January 26th, 2010

Why survey the 5/6ths that don’t matter?

As we look forward to the coming battle why is it that we have our fixation with national polls when what really matters is the state of opinion in the 100 seats that will determine the outcome?

Both big parties, though they won’t admit it, are no longer pumping extra resource into seats held by Labour where the 2005 majority is less than 4%. It’s the seats above that which are key and where, surely, polling efforts should be focussed.

But in a standard 1,000 sample national survey barely 140 of those interviewed will be in the key targets and if they are viewing things in a different way then it’s hardly going to register.

We know, for instance, that voters in Labour marginals are more likely to be married than those in safe Labour seats – so the demographic make-up is not the same and that might be reflected in voting intentions.

Yet somehow the media seem reluctant to commission surveys of the marginals. Since hostilities resumed after October’s conference season there have only been two and both showed quite significant variances compared with the closest standard poll from the same firms.

Many on the site get the wrong idea about them. Their purpose is not to make predictions in specific seats but to point to trends in defined categories of seats. The starting point will always be the 2005 outcome and we then compute a swing based on the outcome of the poll and compare it with the swing in standard polls.

From the two surveys we have it looks as though Labour is struggling more in LAB-CON marginals than the national polls suggest but we need more evidence. We also need to know about the LD-CON confrontations where there’s been nothing specific since last September.

There is an PB/Angus Reid marginals poll which includes a significant LD seat segment in preparation.

Mike Smithson