Glenrothes set for 6th November
It hasn't been formally announced, but it is being reported that the Glenrothes election will be on the 6th November, which may go some way of disrupting the media hype in the build up to the by-election now it will be overshadowed by the US Presidential election.
The full results of ICM's constituency poll in Glenrothes are now on their website here. The full voting intention figures are CON 5%, LAB 43%, LDEM 8%, SNP 43% so without doubt a two-horse race.
After previous by-elections I've expressed some scepticism about the use of the normal re-allocation of don't knows (the "spiral of silence adjustment") in by-elections. There's good evidence that people who say don't know before general elections disproportionately vote for the party they backed last time, but the reallocation made the polls less accurate at recent by-elections: perhaps "don't knows" behave differently in by-elections. Without the reallocation this poll would have shown the SNP 3 points ahead.
Another question people have asked about this and the Glasgow East poll is about weighting by past vote. The raw samples here and in Glasgow East had far too many people who said they voted SNP last time, might people accidentally be saying how they voted in the 2007 Scottish elections, rather than the 2005 general election, and consequently the SNP getting weighted down too much? It is possible, but there's no obvious way of proving it. Glasgow East was a very unusual constituency, so let's wait and see how polls do in Glenrothes before we start discussing why they were wrong.
Moving on, the polls we are waiting for now are the first reactions to Gordon Brown's speech: will it have had any effect on Labour's position in the polls? With luck we'll have something in tomorrow's newspapers, so should have some figures tonight.
