A few hours ago, ICM released a constituency poll from Sheffield Hallam, suggesting that Nick Clegg has a 7% lead over his Labour opponent. On the face of it, this contradicts a string of Ashcroft polls in the same seat, the most recent of which gave Labour a slender 1% advantage. It's being widely assumed that the divergence can be explained by ICM providing respondents with the names of the leading candidates, something which Ashcroft never does. The Lib Dems are obviously keen to spin this as meaning that the Ashcroft polls are unreliable in general, including the ones that have shown them on course to lose seats in Scotland.
That may sound plausible, but there are a number of caveats. First of all, it's not actually clear that the ICM and Ashcroft polls are even in disagreement. The two results are within the margin of error of each other, and it's particularly noticeable that ICM have Labour fractionally ahead before applying the turnout filter and the 'spiral of silence' adjustment. Secondly, it can't automatically be assumed that naming the candidates down the phone produces a more accurate result. That certainly doesn't replicate the polling booth experience, because while some voters may scan the ballot paper carefully for the names of the candidates, others will just quickly look for the party emblems before marking their cross.
Most importantly, the Lib Dems mustn't get away with claiming that their 'comfort polling' - including the poll purporting to show Jo Swinson narrowly ahead of the SNP in East Dunbartonshire - is now vindicated simply because it names the candidates. The main reason those polls have no credibility is that they use a question sequence designed to get respondents thinking positively about the local Lib Dem MP just before they are asked the voting intention question. Needless to say, ICM haven't done anything as silly as that.
But if we assume for the sake of argument that the ICM poll in Sheffield is a useful guide to how much Ashcroft may be underestimating the Lib Dem incumbency bonus, what does that mean for the SNP's prospects in Lib Dem-held seats north of the border? Ashcroft has polled seven such seats, including six in which the SNP have emerged as the main challengers. In those six, the smallest SNP lead he's found is 11% in East Dunbartonshire. The difference between the Ashcroft and ICM polls in Sheffield is equivalent to a 4% swing - so if the same applies in Scotland, the SNP would still be ahead in all six seats. Admittedly, it would be a close run thing in a couple of them - but not in Inverness. Unless Danny Alexander turns out to have a bigger personal vote than Nelson Mandela, he really is toast.