ComRes show a 13 point lead in Crewe and Nantwich

Tuesday also sees what is probably the last poll on the Crewe and Nantwich by-election, this time from ComRes. This must be the most heavily polled by-election for many, many years (perhaps ever - though the days when by-election polls were the norm are beyond my memory) perhaps reflecting both the extreme rarity of a by-election the Tories might win, and the importance the media are suggesting it has for Gordon Brown's future.

The full figures for the ComRes poll show a Tory lead of 13 points (the shares of the vote are not yet available). This would be a crushing Labour defeat on a swing of about 15% and if media rumours about the Crewe and Nantwich result being a trigger for moves against Gordon Brown are any more than media fluff, he will be in trouble.

While this is much larger than ICM's lead, it's actually what we would expect to see - remember that ICM's figures are based on the projection that many of the don't knows, who are mostly former Labour voters, will end up voting Labour come the actual election. Their unadjusted figures showed Conservative leads of 12 points and 15 points before the re-allocation. ComRes in contrast don't reallocate don't knows in the same way, so their findings are broadly in line - they just made different assumptions about how this will translate into votes at the ballot box.

UPDATE: The full figures are CON 48%, LAB 35%, LDEM 12%. Forget my earlier explanation about the difference between this and the ICM poll being down to ICM re-allocation of don't knows. Unusually for them, ComRes also reallocated their don't knows on the assumption that they would vote for the same party they did in 2005. The unadjusted totals would have had a Conservative lead of 15 points.

UPDATE 2: Just for fun (well, actually not just for fun, for £500) there's a competition on PoliticsHome to predict the majority in the by-election, enter here