Today brings word of a full-scale Scottish poll from Panelbase, and not for the first time it illustrates beautifully the yawning chasm between the actual state of public opinion, and the fictional version of public opinion that the unionist media would rather we heard about. Ludicrously, the Times (who commissioned the poll) claim there is "little support" for a pre-2021 independence referendum, even though the poll actually shows that a whopping 42% of the electorate - the sort of percentage that governments are elected on - want a referendum within around twelve months, let alone within three years. 17% want it to be held while Brexit is still being negotiated, meaning within less than one year, and an additional 25% want it at the end of Brexit negotiations, meaning in about a year's time.
As I've noted in the past, the format of Panelbase's question on referendum timing isn't ideal. There is no obvious option provided for people who want a referendum in two or three years' time - anyone in that position is effectively forced to be more negative about a referendum than they really feel (by choosing the third option of "no referendum in the next few years") or to be more bullish about timing than they really feel. Which way such people are jumping in the poll can only be a matter of speculation. What I would point out, though, is that the relatively even split of 58% against a very early referendum, 42% in favour, has occurred in spite of a prolonged spell in which the SNP have not been openly making the case for a vote. If they had been, it seems at least conceivable that the numbers would be even more favourable.
Just as was the case in the Ipsos-Mori poll a few weeks ago, there is no sign whatever of Pete Wishart's so-called "indy-gap" - a claimed phenomenon of support for an early referendum running significantly below support for independence itself. In reality, support for an early referendum (42%) is once again essentially identical to support for independence (43%).
The Yes vote continues the trend of recent months by remaining static. Some pollsters have shown Yes essentially static in the mid-40s, some (like Panelbase) in the low 40s, and some in the high 40s. These are simply 'house differences' between the various firms, and it's impossible to know who is closest to the truth. It's a remarkable turnaround from the long indyref campaign that Panelbase online polling is now on the No-friendly end of the spectrum, and that Ipsos-Mori telephone polling is on the Yes-friendly end.
There are also Westminster voting intention numbers -
SNP 36% (-5)
Conservatives 28% (+1)
Labour 27% (+3)
Liberal Democrats 6% (n/c)
Greens 2% (n/c)
The drop in the SNP vote may look alarming, but the 41% recorded in the previous Panelbase poll was the highest in any poll from any firm since the general election, so it may have been an inflated number caused by the margin of error. This is only the second post-election poll (out of nine) to put the SNP below the 37% recorded on election day, but there has been no reduction in the eight-point election gap between SNP and Tory, and only a statistically insignificant one-point reduction in the gap between SNP and Labour. So even if this poll was accurate, it's not clear that the SNP would be losing seats in an early election.
More details and analysis to follow...