Yes campaign surge to new record high in sensational poll from traditionally No-friendly pollster Ipsos-Mori

Has the Glasgow 2014 effect worked its magic?  Since the campaign started, Ipsos-Mori has never shown a gap anything like as close as this -

Yes 40% (+4)
No 54% (n/c)

With Don't Knows excluded, it works out as...

Yes 42% (+2)
No 58% (-2)

The previous record high for Yes from Ipsos-Mori with Don't Knows excluded was 40% - and in all but one poll from the firm, the Yes figure has always been in the 30s.  When you bear in mind that Ipsos-Mori are one of the two most No-friendly firms, and traditionally show a much lower Yes vote than the average, this is a huge breakthrough for the pro-independence campaign - and could well translate to a neck-and-neck race among the more Yes-friendly firms.

*  *  *

A couple of hours on, and I've now had a look at the datasets.  The first thing to say is that the Yes campaign missed out on being rounded up to 43% by the absolute tiniest of fractions -

Yes 42.48%
No 57.52%

But even that is likely to be an underestimate, because for reasons that I simply cannot understand, Ipsos-Mori continue to ask respondents for their country of birth and then fail to weight by it - or if they are weighting by it, the target figures clearly bear no relation whatever to the 2011 census results, and it's hard to fathom where the extra English, Welsh and Northern Irish people are supposed to be coming from.  A full 14.3% of Ipsos-Mori's weighted sample were born in a part of the UK other than Scotland, which on the face of it is significantly too high.  The percentage of the sample who report that their national identity is either exclusively or predominantly Scottish is also suspiciously low at just under 50%.

It's important to stress that the methodological errors that Ipsos-Mori appear to be making had a bigger impact tonight than simply suppressing the Yes vote in this poll.  You see, the firm also selected the audience for STV's debate.  Did it seem to you that there were a disproportionate number of No voters in the audience?  Were you surprised that Darling sometimes seemed to receive greater spontaneous applause than Salmond?  Did it strike you that significantly more No-friendly questions were asked?  That's because the audience were hand-picked to be representative of what Ipsos-Mori think is the demographic balance of Scotland.  I saw a Yes supporter rather narkily suggest on Twitter that there were an awful lot of English accents - this is obviously a very delicate subject, but at the end of the day it's a simple and plain fact that if the audience were supposed to be demographically representative of the Scottish population, there should have been no more than about 10% who were born in England.  Yet due to Scotland apparently having completely different demographics over on Planet Ipsos-Mori, the figure is likely to have been considerably higher - and that will have been a deliberate and conscious choice.  Why do they do it?  Answers on a postcard, folks, because it's beyond me.

I gather that an instant poll conducted by ICM suggested that 44% of respondents thought that Salmond won the debate, and 56% thought that Darling won.  That gap isn't particularly significant, especially given the unusual difficulties of putting together a representative sample for a poll like that (ie. you have to choose between being representative of the Scottish population as a whole and being representative of only the people who watched the debate, because you can't do both).  Nevertheless, it does obviously have a little bit of importance in terms of bragging rights, and for my money there's no way on Earth that those numbers would have been produced if it hadn't been for the 'social proof' of the support Darling was getting from the hand-picked audience.  So Ipsos-Mori have got a lot to answer for - and so have STV for commissioning them.  It's been a long while since I praised the BBC, but their approach of using audiences that are strictly 50/50 divided between Yes and No, with a sprinkling of undecideds added in, has a lot to commend it.

Incidentally, the Yes campaign can take huge heart from one particular aspect of the ICM poll - of respondents who were undecided about their referendum voting intention before the debate started, more thought that Salmond won (54%) than Darling (46%).  The only reason that Darling had a slight lead in the overall numbers is that people who are already on the No side were more likely to have been impressed by their man than Yes supporters were by Salmond - and that may simply have been the problem of Yes voters having higher expectations of the First Minister.

Hand on heart, I burst into uncontrollable laughter when Colin Mackay said at the end that the real winner of the debate was Bernard Ponsonby.  STV's moderator is undoubtedly a man of considerable talent, but surely it's obvious to anyone that he was a big part of what was wrong with the debate tonight.  Claire Stewart said that she didn't hear enough vision - well, does it ever occur to her that her own colleagues might be predominantly responsible for that?  You can only set out a vision if you're given the space to do so, and tonight's format was only a very mildly watered down version of the bear-pit debates that have been repeatedly served up on Scotland Tonight.  I can only hope that the BBC aim for a more thoughtful and considered pace in their debate.  As for whether Ponsonby was actually guilty of outright imbalance, it's hard to say, but I do think he was bang out of order at the start.  It's perfectly reasonable to ask Alex Salmond about "poll after poll" showing a No lead (even though a great many of those No leads have been extremely slender).  But if you're being fair and even-handed, what you then do is challenge Darling about the fact that the No lead has dramatically narrowed in the new Ipsos-Mori poll to a record low.  Instead Ponsonby repeated the "poll after poll showing a No lead" line, and put it to Darling : "you can't lose, can you?"  I mean, what?  WHAT?  Had Darling hypnotised him or something?

*  *  *

This post is meandering around in a slightly chaotic way, but there are still two other important points to make about the Ipsos-Mori poll. Firstly, when undecideds are pressed further, they break for Yes by 29% to 23%. It's not possible to calculate what the headline figures would be if these undecided leaners were added in (because Ipsos-Mori use only definite voters for the headline numbers), but it seems a reasonable guess that it would increase the rounded Yes vote to 43%.

Secondly, the sample size for the poll was a completely normal 1006. That means the mystery of why so many people have reported being interviewed by Ipsos-Mori recently has not been solved. My guess is that more than one poll has been in the field, and that the other one is an internal poll for a No-supporting client - quite possibly the UK government using taxpayers' money. Perhaps there's another explanation, but we'll see.

* * *

REQUIRED SWINGS

Swing required for 1 out of 6 pollsters to show Yes in the lead or level : 3.0%

Swing required for 2 out of 6 pollsters to show Yes in the lead or level : 3.5%

Swing required for 3 out of 6 pollsters to show Yes in the lead or level : 4.5%

Swing required for 4 out of 6 pollsters to show Yes in the lead or level : 5.5%

* * *

SCOT GOES POP POLL OF POLLS

There's been a lot of interest in the fact that ICM's instant poll shows voting intention numbers of Yes 47%, No 53%, which on the face of it represents a whopping 8% decrease in the No lead since the last ICM poll. That's certainly encouraging, but unfortunately I can't include it in the Poll of Polls update because it wasn't a full-scale poll (it was demographically weighted, but by definition anyone who didn't watch the debate was excluded). However, even just taking into account the new Ipsos Mori-figures, the updated Poll of Polls still shows the No lead slipping back to just 11% when Don't Knows are included.

MEAN AVERAGE (excluding Don't Knows) :

Yes 43.5% (+0.4)
No 56.5% (-0.4)

MEAN AVERAGE (not excluding Don't Knows) :

Yes 37.0% (+0.7)
No 48.0% (n/c)

MEDIAN AVERAGE (excluding Don't Knows) :

Yes 43.4% (n/c)
No 56.6% (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.)