The Sunday Herald have just tweeted the results of this weekend's new Panelbase poll (with Don't Knows excluded). It shows the Yes vote at an all-time record high for the firm -
Should Scotland be an independent country?
Yes 48% (+2)
No 52% (-2)
I say an all-time high, but of course that's only the case if we exclude the Panelbase poll late last summer which put Yes in an outright lead. The legitimacy of that poll was later called into doubt due to an unusual question sequence (a little like last month's ICM poll). Because this new poll was supposedly commissioned by Yes Scotland before being passed on to the Sunday Herald, we'll just have to be slightly cautious until we get firm confirmation that the referendum question was asked first. But my guess is that it probably was - it was certainly asked first in the two Panelbase polls commissioned by the SNP and Yes Scotland respectively earlier this year.
Assuming that the methodology for this poll does indeed turn out to be sound, what does it mean? Well, the most obvious thing to say is that two regular referendum pollsters (Panelbase and Survation) have both shown Yes at an all-time high within the space of a few days. That could be coincidence but it's unlikely. Oddly enough, my heart sank when I initially saw the Survation datasets, because there were a couple of unusual patterns which made it somewhat more likely that the swing to Yes might have been an illusion caused by random sampling variation. But the fact that Panelbase have independently found exactly the same trend sets my mind to rest on that score to some extent (although obviously it's still possible that ICM might show something different later tonight).
Oh, and it's probably also worth making the small point that if Survation and Panelbase are getting their methodology right, it's now clear that the Yes campaign have every chance of winning this referendum, and Scotland could well be on the brink of becoming an independent country. Even if the more No-friendly pollsters are closer to the mark, there does now appear to be renewed momentum towards Yes, so the somewhat larger No leads reported by the likes of YouGov can't be regarded as in any way safe.
This is the third poll to be published during the campaign that can be called a 'statistical tie', meaning that if the standard margin of error is to believed, there is a greater than 5% chance that the side that appears to be behind is actually in the lead. The previous two were the aforementioned controversial Panelbase poll last summer, and the ICM poll published on Easter Sunday.
The new ICM poll is now out - details of that, plus a landmark Poll of Polls update, can be found in a fresh post here.