As I noted in the comments section from yesterday's post, I was a bit concerned after looking at the Ipsos-Mori datasets that Labour might have taken a narrow lead in Holyrood voting intentions. As it turned out, I needn't have worried, because it's just been revealed that the SNP are still ahead on the headline figures. In fact there's only a very marginal difference with the last Ipsos-Mori poll from September, amounting to the equivalent of a 1% swing to Labour (in other words the change could easily be an illusion caused by that legendary "margin of error stuff").
SNP 36% (-3)
Labour 34% (-1)
Conservatives 15% (+3)
Liberal Democrats 8% (+1)
The figures are for the constituency vote - for some reason Ipsos-Mori don't seem to bother asking their respondents for list vote intention.
The fact that the SNP's lead is relatively narrow, though, may provide a clue to the reason for the huge differences we've been seeing between the pollsters on referendum voting intention. We know, for example, that Panelbase (the most favourable pollster for Yes) has tended to show very large SNP leads for Holyrood, while Newsnet Scotland discovered after some sleuthing that Progressive Scottish Opinion, the non-BPC pollster that has been producing the most inflated No leads of all, was privately showing a thoroughly implausible Labour lead for Holyrood of 8%. So there does seem to be a clear (and admittedly unsurprising) relationship between the divergence on referendum voting intention figures, and the number of SNP voters each pollster has in its sample.
That's not to say Ipsos-Mori are necessarily getting it wrong, of course, although that will certainly be the suspicion of many of us - especially given that they are one of the only pollsters who still refuse to weight their sample in line with the 2011 Holyrood result.