Rarely have I been so delighted to be proved wrong. I had suspected that last week's Panelbase poll for the Sunday Times contained a Westminster voting intention question that was being withheld until this week for the purposes of a "blow for Sturgeon" headline, ie. because the results were markedly worse for the SNP than the Holyrood numbers. Well, I was correct about there being withheld results, but not about them being bad for the SNP - in fact they're so wonderful for the SNP that the Sunday Times have seemingly given them only the most cursory of mentions.
Scottish voting intentions for Westminster (Panelbase) :
SNP 41% (+4)
Conservatives 27% (-2)
Labour 24% (-3)
Liberal Democrats 6% (-1)
Greens 2% (+2)
This is the first full-scale Scottish poll of Westminster voting intentions from Panelbase or any other firm since the general election, so the percentage changes listed above are from the actual election result, rather than from a previous poll. I know some people will look at the numbers and think "this looks very similar to the pre-election polls that overestimated the SNP by a few points, so the SNP are probably being overestimated again", but of course this poll has been weighted by recalled 2017 election vote, which should have resolved any skew.
If the poll is right, it genuinely looks as if quite a few voters who switched to the Tories or Labour in June have since come home to the SNP. One of the things that made the election in Scotland so unusual was the large number of seats that were won by knife-edge margins - some of them broke for the SNP (including, remarkably, all four that were decided by fewer than 100 votes), but plenty of others didn't. Labour's six gains are now marginal seats, and most of them are ultra-marginals. Based on the Panelbase numbers, the SNP could expect to regain all of those six seats, with only the extreme oddity of Edinburgh South remaining firmly out of reach. There would also be modest gains from the Tories (Stirling would fall on the tiniest of swings).
In other words, the doom and gloom of the summer is now over. The SNP can stop fearing an early election, and can perhaps even start thinking of it as a golden opportunity to gain seats - although admittedly none of us need any reminding of how suddenly the political weather can change these days. One thing is for sure - if these numbers are spotted in the corridors of power in London, it'll put an end to the Tories' silly notion that they can expect the SNP to abstain on a no confidence vote.
Does all of this mean that the picture painted by Scottish subsamples of GB-wide polls since June (basically that the SNP only had a very narrow lead, and that Labour had surged into a strong second place) was totally meaningless? As this poll has taken me by complete surprise, I suppose I should have the humility to say "possibly", but the flip-side of the coin is "not necessarily". We only have one full-scale Westminster poll to go on at the moment, and it may yet turn out that a 14-point lead for the SNP is 'on the high side'. I wonder if question sequence may have played a part - if Panelbase asked about independence and Scottish Parliament voting intentions first, respondents may have been more likely to stick with the SNP when subsequently asked about Westminster. But there may also be a way of reconciling this poll with the subsamples. YouGov are the only firm who seemingly weight their Scottish subsamples separately - and they suggested in their first few post-election subsamples that there was a tight race between SNP and Labour. More recently, they've shown the SNP with a bigger lead. That could be an illusion caused by the enormous margin of error, but it's just possible there was a Corbyn surge for Labour in the summer that has since subsided as memories of the election have grown more distant. There's no getting away from it, though - to see Labour in third place, and a whopping 17 points behind the SNP, is undoubtedly a big shock.