So, right on cue, the Survation poll with the confusing question that I mentioned in my previous post has been trotted out by the Scotsman newspaper, and they've done exactly the same thing that the Sun newspaper did in its reporting of the previous poll in the series - they've told their readers a downright lie about the trend shown by the results. The headline reads "Poll: Scottish independence supporters switching to back the Union". In fact the poll shows the complete opposite of that - it shows that voters have swung towards independence since the previous poll.
The percentage of respondents who say that they "completely support Scotland becoming independent" has increased from 24% in the previous poll (conducted in March) to 26% now. By contrast, the percentage of respondents who say that they "completely support Scotland staying part of the UK" has fallen in the same period from 40% to 37%.
If you also take into account respondents who are not on one of the two extremes, the pro-Yes trend is even stronger. The percentage of respondents who are on the pro-independence side of the 0-10 scale has increased significantly from 35% in March to 40% now, while the percentage of respondents on the anti-independence side of the scale has declined sharply from 58% to 51%.
It's rare that we can simply say that a newspaper has lied, as opposed to just misleading its readers or telling half-truths, but this is one of those rare occasions. The headline contains no quotation marks, and it doesn't say "according to Scotland in Union" or something like that (although I think we can make a fairly safe guess that's how this story came to appear). It simply tells a direct lie.
As for the question of why this poll format produces such different numbers from standard Yes/No polls, there's something of a mystery. We know from David Halliday's screenshot that at least some respondents were confusingly asked to regard the number zero as being the pro-independence end of the scale, and the number ten as being the anti-independence end of the scale - which may well explain why a wildly implausible 16% of Yes voters from 2014 are now supposedly expressing "complete support" for Scotland remaining in the UK. (The equivalent figure from the March poll was almost identical.) And yet the Survation datasets suggest that the opposite was done, and that the number ten was in fact the pro-independence end of the scale. Are the datasets inaccurate? Was there a dummy poll conducted in a different way for research purposes? Either way, it does seem very surprising that Angus Robertson's Progress Scotland have persisted in commissioning polls using this format after the results produced in the spring proved to be so totally out of line with the results of conventional independence polling.
UPDATE: I see The Herald have done much the same thing as the Scotsman - their headline is "New poll suggests shift in support away from Scottish independence". I don't think any of us are going to faint with amazement if it turns out that both papers have just lightly rewritten a Scotland in Union press release without bothering to check whether its factual claims are accurate.