Saturday, March 22, 2014

Pro-independence campaign close gap to just 7% in game-changing poll from "gold standard" ICM

I wouldn't normally dream of taking delight in Kenny Farquharson's befuddlement over his failure (just for once) to keep us in the dark about the results of a Scotland on Sunday poll until the official embargo is over, but on this occasion I simply have to.  These numbers are from ICM, regarded by many as the UK's "gold standard" polling organisation, and they will blow this referendum campaign wide open.

As you may know, a referendum on independence will be held in Scotland on 18 September 2014. Voters will be asked, "Should Scotland be an independent country?"  Do you think you will vote "Yes” or "No"?

Yes 39% (+2)
No 46% (-3)

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

Yes 45% (+2)
No 55% (-2)

It actually takes quite a bit of effort to find a way in which the headline figures can translate into a Yes vote of 45% rather than 46% with Don't Knows excluded, so it must mean that the unrounded figures are somewhere in the region of Yes 45.4%, No 54.6%.  The true No lead without DKs is therefore about 1% lower than it appears.  All the same, the unrounded figures must be fractionally less favourable for Yes than was the case in that extraordinary ICM poll in January that put support for independence at 46% - but of course on that occasion McDougall's mob were able to use the excuse that ICM had drastically upweighted their Yes-friendly sample of 16-24 year olds, meaning that the Yes surge might have been overstated.  There is no such alibi for the anti-independence campaign this time, because ICM have since amended their methodology to ensure that no age group needs to be upweighted by a significant amount.

John Curtice's word of caution about the sensational Panelbase poll on Thursday was that "one swallow does not make a summer".  In some ways that was a tad unfair, because we already had supporting evidence in the shape of a Survation poll showing a No lead of just 8.3%.  And now it looks very much like we've just seen a third swallow.  I can't even remember the last time we had three different BPC-affiliated pollsters all showing a No lead of less than 10% on the headline figures, but that's the position we find ourselves in now.  All three polls are strikingly similar in showing an unusually high Yes vote of between 39% and 40%, representing high watermarks in the campaign so far in two cases.  And what's particularly encouraging about the ICM poll is that the fieldwork is bang up to date - it factors in GERS, it factors in Devo Nano, it even factors in the UK Budget to a limited extent.

One suspicion that was raised about the Panelbase poll concerned the particularly low gender gap, with a No lead that was just 3% higher among women than among men.  ICM haven't replicated that finding, but they do show the No lead falling equally fast across both genders.  In line with the figures from many other polls, women appear to be only very slightly more likely than men to vote No (the difference is a mere 3%).  What really distinguishes the female vote is the much higher number of Don't Knows.

There are also some interesting differences between ICM and Panelbase in the figures from the various age groups.  Panelbase suggested there was now a clear lead for Yes among under-55s, albeit one that was more than offset by a huge No lead among over-55s.  By contrast, ICM are showing a much more balanced picture, with No holding a narrow lead among all three of the under-55 age groups, with Yes and No actually sitting level-pegging among 55-64 year olds, but with No still enjoying a huge advantage among over-65s.  These differences between the two polls may well be down to normal sampling variation - the margin of error for age-based subsamples is always extremely high.

Now that there is no longer any credible doubt that the No lead has dropped (in spite of Peter Kellner's ongoing one-man mission to "prove" otherwise!), the obvious question is "why?"  Is it childcare?  A recent YouGov poll suggested the message that childcare provision would be expanded after independence is being successfully driven home.  Is it the incoherence and relentless negativity of the anti-independence campaign?  Is it the effect of Yes Scotland's advertising blitz kicking in, finally neutralising the media-sponsored fear-mongering about Scotland's economic strength?  Is it exasperation at the failure of Eddie Izzard and his ilk to recognise that True Love Isn't Possessive?  It may well be a combination of all four factors, and more.

*  *  *


The records just keep on tumbling in this blog's Poll of Polls.  The latest update shows Yes on its highest level of support to date, and breaking through the 42% barrier for the first time when Don't Knows are excluded.  The No lead has also dropped to a new low of 12.9%.

MEAN AVERAGE (not excluding Don't Knows) :

Yes 35.4% (+0.3)
No 48.3% (-0.4)

MEAN AVERAGE (excluding Don't Knows) :

Yes 42.3% (+0.4)
No 57.7% (-0.4)

MEDIAN AVERAGE (excluding Don't Knows) :

Yes 42.0% (n/c)
No 58.0% (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, and that adhere to British Polling Council rules. At present, there are seven - YouGov, TNS-BMRB, Angus Reid, 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.)

I've been wondering for some time whether I should bite the bullet and remove Angus Reid from the sample, because their figures are ridiculously out of date compared to all the others, and there must be at least some doubt over whether they will ever return to the fray.  I'll leave them in for the time being to keep the trend figures meaningful, but it's worth pointing out that this is the first time that the No lead on the headline average figures has been slightly lower than in the last Angus Reid poll.  In other words, if current trends continue Angus Reid's presence in the sample will start to artificially flatter the No camp's standing. That's already happening on the median average.

I feel somehow that this would be an appropriate moment to once again take stock of the gradual long-term movement towards Yes...

The No campaign's lead in the Poll of Polls headline figures :

Sep 2013 - 20.2%
Sep 2013 - 20.0%
Sep 2013 - 18.4%
Oct 2013 - 17.9%
Oct 2013 - 17.5%
Oct 2013 - 17.4%
Nov 2013 - 17.5%
Dec 2013 - 17.1%
Dec 2013 - 16.3%
Dec 2013 - 16.2%
Dec 2013 - 15.8%
Jan 2014 - 14.2%
Jan 2014 - 14.8%
Feb 2014 - 14.8%
Feb 2014 - 14.7%
Feb 2014 - 15.1%
Feb 2014 - 13.6%
Feb 2014 - 14.0%
Mar 2014 - 14.0%
Mar 2014 - 14.3%
Mar 2014 - 14.3%
Mar 2014 - 13.6%
Mar 2014 - 12.9%

And of course as ICM are an online pollster (at least for the purposes of their Scotland on Sunday polls), it's also time for an update of the average figures for online pollsters that have reported this year. This set of figures does exclude Angus Reid.

MEAN AVERAGE OF ONLINE POLLSTERS (not excluding Don't Knows) :

Yes 38.3% (+0.5)
No 48.0% (-0.8)


Yes 44.4% (+0.8)
No 55.6% (-0.8)


Yes 45.4% (+1.5)
No 54.6% (-1.5)



  2. Thanks, David. It's surprising that the Yes figure with DKs excluded is 45% rather than 46%. That probably indicates that this time the rounding has favoured Yes, and that the No lead on the unrounded figures is closer to 8% than to 7%. But it'll still do to be getting on with!

  3. That's what I thought - when I did the rough sums I got 46% Yes as well.

    I think it's just going to be 46% for Sassenach consumption. It feels right!

  4. From a quick look at David's info, there seems to be a similarity in the changes in the demographic crossbreaks between this ICM and the Newsnet Panelbase poll.

    if so, then this suggests that both are measuring real shifts in these groups.

  5. As far as the gender gap goes last TNS poll said that men and women were equally intending to vote NO but the YES vote was higher among men.

  6. I've updated the plot - tried doing it in google docs this time but for some reason the "publish" feature doesn't work, so here's the image instead. Still a steady decline in the "no" lead, all that's necessary for the moment until the campaign intensifies. (Given that the linear trend so far isn't likely to continue I've omitted that from the plot)

  7. Thanks a lot, Sandy. I've added your graph in a new post.