But British Electoral Survey has Tory lead of 2.8% amongst those “certain to vote”

After the euphoria in the Tory ranks following the February ICM poll it’s is down to earth with a bang after the publication of this month’s survey. What seemed an attainable gap of just three points in February is now eight points and we are back in Labour landslide land.

But the only blot on the horizon for Tony Blair’s party was news last night of the British Electoral Survey (BES) which was reported as showing the following vote shares amongst those “certain to vote” – CON 38.8: LAB 36: LD 17.4.

What these latest figures reinforce is the point that many observers have been making – turnout is going to be crucial. Currently Tory supporters seem more motivated to vote than those of other parties.

So what is the BES and how much weight should be given to its results? It is funded by the Economic and Social Research Council (ESRC), based at the University of Essex. This presentation gives a good outline of its objectives and methodology. It’s broken down into several parts including face-to-face interviews which are being carried out by NOP; a major pre-campaign survey of 8,000 people carried out by YouGov; and a rolling series of daily questioning of 200 people using the internet.

The BES is a major academic study and is about much more than just party choice. In 2001 the internet survey elements proved to be pretty accurate. It would appear that the figures announced last might are from the “first wave” study which appears to have involved 8,000 people being questioned – which is a substantially bigger sample than the standard 1,000 – 2,000 of conventional polls.

The daily surveys start 30 days before the election and will create a rolling poll. The data should appear on the website. So far there is no information about the last night’s reported results on the site.

If we are correct then this latest survey was carried out by YouGov but with questions designed by the academic team. Normally the headline figures from YouGov polls do not involve an element of whether people are likely to vote. The BES study did.

© Mike Smithson 2005