We finally have an end to the mystery...
Should Scotland be an independent country?
Yes 48% (+2)
No 52% (-2)
Yes 45% (+2)
No 55% (-2)
The datasets aren't out yet for either poll, so I don't know the unrounded figures, or if there have been any further methodological changes. Most importantly, I'm wondering if ICM have joined the new orthodoxy of weighting by country of birth? If they haven't, there's every reason to assume that they are underestimating the Yes vote, because on previous occasions when they have asked for respondents' country of birth, it's turned out there are significantly too many English-born people in their sample, and of course that is a strong predictor of a No vote.
There's no such problem with Panelbase, the firm that in their last poll started the long-overdue trend towards country of birth weighting. However, they coupled that with a No-friendly methodological change (a degree of weighting by European Parliament recalled vote), which in my view may have more than offset the benefit for Yes. That's just a guess, though, and it will have to remain so, because Panelbase have indicated that they will not provide figures using the old methodology for comparison. If I'm right, it means that Yes have effectively reached a new all-time high with Panelbase.
Incidentally, there's a qualitative difference between the two methodological changes Panelbase have made. The Yes-friendly change (country of birth weighting) has logic behind it that is absolutely inescapable - we know roughly how many English-born people there are in Scotland, so leaving too many in the sample is bound to skew the result. By contrast, the No-friendly change of weighting on a combination of recalled Holyrood and European Parliament vote seems merely to be based on inspired guesswork about what 'feels right'.
Question : If Panelbase had ONLY introduced country of birth weighting, and hadn't changed the weightings for recalled vote, is it possible that Yes would be in the lead in this poll, or level? It's an intriguing thought...
Unlike Panelbase, ICM aren't showing Yes on an all-time high - although the Yes vote is up markedly on the last poll, that takes it back to a position that is more or less average for the year so far. So there are two possibilities - either the Yes vote has increased in spite of (or perhaps because of) what happened in the leaders' debate, or the change is margin of error noise masking a static position. Either way, we now have sufficient evidence from multiple polls to effectively extinguish the delusion that there was a pro-No bounce following the debate - it looks highly likely that the Survation poll led us up the garden path on that one.
Although the No campaign can draw some comfort from the fact that undecided respondents break more for No when pressed by ICM (a radically different finding from many other surveys), I'd caution you to ignore John Curtice's nonsensical claim that three-quarters of undecideds are on the No end of Yes Scotland's preferred 1-10 scale. All he tells us is that those people are on 5 or below - and doesn't say anything about how many of them are on exactly 5. As I've pointed out many times before, Yes Scotland do indeed use a 1-10 scale, but it's highly unlikely that they're stupid enough to interpret a 5 as being on the No end of that scale - most people who say they're a 5 regard themselves as being exactly in the middle.
Curtice has made a song and dance about how both this poll and the last one from ICM have failed to provide evidence for Yes Scotland's claim that public opinion is more favourable for Yes when measured by the 1-10 scale. The issue of how to interpret 5s obviously muddies the waters here, but either way Curtice's analysis ignores the findings of the Economic and Social Research Council study, for which ICM conducted the fieldwork, and which used a modified form of the scale. That unambiguously showed that there were more people leaning Yes than you'd expect to find based on the results of the headline question.
UPDATE : Ignore the suggestion from an anonymous poster in the comments section below that the Panelbase poll was tainted by the asking of leading questions prior to the main referendum question - that was quite simply not the case. Professor Curtice has confirmed that the methodology was identical to the one Panelbase used for their last Sunday Times poll.
* * *
Swing required for 1 out of 6 pollsters to show Yes ahead or level : 2.0%
Swing required for 2 out of 6 pollsters to show Yes ahead or level : 4.5%
Swing required for 4 out of 6 pollsters to show Yes ahead or level : 6.5%
* * *
SCOT GOES POP POLL OF POLLS
MEAN AVERAGE (excluding Don't Knows) :
Yes 43.0% (+0.6)
No 57.0% (-0.6)
MEAN AVERAGE (not excluding Don't Knows) :
Yes 37.3% (+0.8)
No 49.5% (n/c)
MEDIAN AVERAGE (excluding Don't Knows) :
Yes 42.5% (n/c)
No 57.5% (n/c)
(The Poll of Polls is based on a rolling average of the most recent poll from each of the pollsters that have been active in the referendum campaign since September 2013, and that adhere to British Polling Council rules. At present, there are six - YouGov, TNS-BMRB, Survation, Panelbase, Ipsos-Mori and ICM. Whenever a new poll is published, it replaces the last poll from the same company in the sample. Changes in the Poll of Polls are generally glacial in nature due to the fact that only a small portion of the sample is updated each time.)