Scale of Davidson flop becomes clear as "extraordinary" Panelbase poll reveals that 42% of public want an independence referendum within LESS THAN TWO YEARS

Panelbase have released the datasets for the full-scale Scottish poll that was published (and so comically misrepresented) in yesterday's Sunday Times.  Here are a few points that leap out -

1) As expected, Panelbase have introduced weighting by recalled 2017 general election vote.  In one sense that's very good news, because it means that there's much less reason to be sceptical about the SNP's double-digit lead over both the Tories and Labour.  If, for example, there had been weighting by 2016 recalled Holyrood vote but not by 2017 vote, there would be a danger that respondents might have got the two elections mixed up, which in all likelihood would lead to the SNP's vote being wrongly adjusted upwards (the exact reverse of the effect that was seen when YouGov used to weight by 2010 vote, rather than 2011).  Instead, the SNP have actually been significantly weighted down on recalled vote, with the 385 respondents who said they voted SNP in June being reduced to 346.

The downside is that 2017 weighting was also applied to the independence question.  We know from YouGov polling that the reduction in the SNP vote in June can be partly explained by people staying at home, rather than defecting to another party.  And yet it's not unreasonable to assume that a lot of those people might well turn out to vote in a second indyref, and would be more likely to vote Yes than No.  The problem with Panelbase's new approach is that to some extent it treats those missing voters as if they don't exist - which could, theoretically, lead to the Yes vote being underestimated.  It certainly means that the new poll is not directly comparable with Panelbase's previous independence poll, which was conducted just before the general election.  That factor alone might explain the small (and statistically insignificant) drop in the Yes vote from 44% to 43%.

2) The only possible reason I can think of for still being a little cautious about the SNP's handsome lead is that the independence question was asked before the Holyrood voting intention question.  It's arguable that this might put pro-independence voters in a frame of mind where they'd be more likely to favour the SNP rather than Labour.  That's pure speculation on my part, but I don't think the possibility can be totally ruled out.

3) There's a preamble to the independence question: "If the referendum was held again tomorrow, how would you vote in response to the question..."  There's no way of knowing whether that makes any difference to the result, but I think the wording is unwise, because it invites people to think about how they would vote in a re-run of a referendum they have already voted in - perhaps nudging them back towards their original choice, rather than inviting them to think of the next referendum as a completely fresh vote taking place in a different context (ie. Brexit).

4) Judging from the numbering in the datasets, there are clearly some results from the poll that haven't seen the light of day yet.  In the past, the Sunday Times have sometimes held results back for a week so they can get two weeks' worth of headlines out of the same poll.  If a Westminster voting intention question was asked, and if the results were less favourable for the SNP than the Holyrood constituency numbers (as they probably would be), I'm wondering if they're being held back for a good old "blow for Sturgeon" effort next week.  It does, however, look like there was no question about Holyrood regional list voting intention, because there's no obvious reason why that would have been withheld.  The omission makes it impossible to use the poll to project seat totals in the Scottish Parliament.

5) Although the wording is a bit slippery, there's a finding that seems to imply that 49% of respondents think that Scotland will be independent within less than 20 years, compared to 42% who do not.  And only 32% think that Scotland is not likely to become independent "at any point in the next few decades".

6) An astonishing total of 42.3% of respondents want the next independence referendum to take place within less than two years.  Admittedly that's down on the roughly 50/50 split we've seen on that question in previous Panelbase polls, but nevertheless it's an absolutely stunning slap in the face for the commentators who have spent the last three months trying to convince themselves that the SNP's decisive victory in June somehow rendered the issue of independence "dead".