There were two schools of thought in the wake of the extraordinary recent polls from TNS and Ipsos-Mori polls showing a majority for independence. Along with a number of others, I felt it was most likely that there hadn't been any recent change in public opinion, and that what we were seeing was a divergence opening up between online and non-online pollsters. But some people felt that TNS and Ipsos-Mori were picking up a shift that had occurred over the summer, and that the next poll from an online firm would confirm that. Unfortunately, it looks like I was right. In their new poll, YouGov are reporting much the same state of play that they've been showing since their major methodological change a few months ago.
Should Scotland be an independent country? (YouGov)
Yes 48% (+1)
No 52% (-1)
So the probability is that the increase in Yes support shown by TNS and Ipsos-Mori was exactly the same increase detected by the online firms in the very early stages after the referendum, and that the position has remained relatively stable since then. If TNS and Ipsos-Mori had been polling on independence regularly over the last year, the chances are that we would have seen a steady Yes lead from them, while the online firms were showing (on average) a very narrow No lead. What's changed over the last couple of weeks is that we're now seeing the whole picture, rather than only half of it. The new information increases the chances that Yes have been in the lead for quite some time without us being aware of it - but we certainly can't be sure of that. It just depends which firms and methodologies you trust most.
The YouGov datasets don't seem to be out yet, but I would advise people to be enormously sceptical of the spin being put on the results of the supplementary questions. This seems to be a classic case of an anti-independence client (the Times) asking a pollster to use leading question wording to produce the results they want. For example, a multi-option question was asked on nuclear weapons to nudge people towards the middle, 'moderate' option, so that backers of the pro-Trident and the middle option could be lumped together to claim that there is a majority in favour of the retention of nuclear weapons. Utterly cynical, and I expect we'll find that equivalent tactics were used to produce an artificial majority against a second independence referendum taking place in the short-to-medium term (a finding that suspiciously contradicts what we've seen from many other polls).