Sunday, April 20, 2014

Independence rises this Easter, as Yes campaign close the gap to just 3% in earth-shaking new poll from "gold standard" ICM

Santa was rather kind to us by delivering a swing to Yes in the last batch of polls before Christmas, and I'm delighted to say that the Easter Bunny hasn't let the side down either. Actually, that's putting it mildly. ICM are regarded by some commentators as the UK's "gold standard" polling organisation, and their Easter Sunday poll reveals that the pro-independence campaign are now within a hair's breadth of victory...

Should Scotland be an independent country?

Yes 39% (n/c)
No 42% (-4)

With Don't Knows excluded, the position is -

Yes 48% (+3)
No 52% (-3)

These astonishing numbers drive a stake through the heart of the cherished belief of the No campaign that Panelbase are some kind of outlier with their relatively Yes-friendly results - although in truth several ICM and Survation polls earlier this year ought to have already done the trick. Apart from one Panelbase poll conducted last August that used an unusual question sequence, this is the very first poll of the campaign that constitutes what the Americans call a "statistical tie", meaning that even if the figures are accurate to within the standard margin of error, there is still a greater than 5% chance that the side that appears to be slightly behind is actually ahead.

Although the last few polls from Panelbase, Survation and TNS-BMRB have been generally very encouraging for the Yes campaign, the fly in the ointment (which John Curtice was all too eager to point out) was that the results were consistent with Yes consolidating their gains from the winter, but not actually generating any further momentum. This poll is admittedly not conclusive proof that more gains have been made in recent weeks, but it certainly points powerfully in that direction. And it's not simply that there has been a significant 4% decrease in the No lead since last month's poll from ICM. More tellingly, the gap has narrowed by an extraordinary 14% since ICM's first contribution to the campaign last September. If there was ever any such thing as an ICM "normal range" for the No lead, this poll is so far below it that it's almost impossible to believe that what we're seeing tonight is a statistical illusion.

Other than the wafer-thin gap between the two sides, the other thing that leaps out immediately from the headline numbers is the No campaign's unbelievably low share of the vote. The only other pollster that has ever placed No as low as 42% is TNS-BMRB - but of course that generally happens because TNS report a much higher number of undecided voters than other firms. Blair McDougall and co have no such alibi in this case.

ICM have also provided a fascinating breakdown of voting intention numbers by country of birth. Ipsos-Mori have done that in the past, but as you can probably guess the results could hardly be more different. Yes are now in the lead by 42% to 40% among Scottish-born respondents, and contrary to the impression given by Scotland on Sunday's analysis, that finding can be regarded as being more or less statistically credible. Because Scottish-born people make up the lion's share of the overall sample, the margin of error isn't all that much greater when you separate them out.

Strangely enough, though, what I find particularly exciting is the position among English-born respondents. A margin of 58% to 28% in favour of No might look like a handsome lead, but it's actually stonkingly good for Yes among a group of voters that have sometimes been lazily assumed to be a lost cause. It means, for example, that Yes are faring far better with English voters than the pro-sovereignty campaign in Quebec did among non-francophone voters back in 1995. That could make all the difference in a tight contest. Admittedly, though, there is a health warning on these numbers, because the sample size for English-born respondents is indeed very small.

A few eyebrows have been raised over the fact that 15% of ICM's sample are English-born, even though the 2011 census shows that just 9% of the population of Scotland fall into that category. It goes without saying that if English voters are being over-represented in the sample, then the headline numbers should be even better for Yes than they already are. But could there be a logical explanation for the discrepancy? If such an explanation exists, it seems to me that it would have to be a combination of two factors. Firstly, the census figures include children, and it's likely that the share of the over-16 population who hail from south of the border is slightly higher. Secondly, ICM may have reason to believe that a greater number of English people have moved here in the three years since the census was conducted. But for what it's worth, my own guess (and it is only a guess) is that the 15% figure is probably a bit too high. If so, we know that both YouGov and Ipsos-Mori have been making exactly the same mistake in at least some of their polls.

ICM's Martin Boon has penned a rather confusing article in Scotland on Sunday, which could (I only say could) be cynically interpreted as him saying that his company have been actively searching for reasons why they may have been overstating the Yes vote, and have introduced a methodological change in this poll to try to 'redress' the situation. But if I understand the slightly ambiguous explanation of the new procedures correctly, then it's not at all obvious to me that they would actually have the effect of decreasing the reported Yes vote. Perhaps things will become clearer when the datasets are published.

* * *


Unsurprisingly, this update of the Poll of Polls sees the No campaign's lead tumble to yet another new record low - it's now below 12% for the first time ever. With undecided voters excluded, the Yes vote now stands at a new high watermark of 42.8%.

MEAN AVERAGE (not excluding Don't Knows) :

Yes 35.4% (n/c)
No 47.3% (-0.6)

MEAN AVERAGE (excluding Don't Knows) :

Yes 42.8% (+0.3)
No 57.2% (-0.3)

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 was planning to finally bite the bullet and remove Angus Reid in this update (because they haven't reported for a ridiculously long time), but I'm too sleepy to break the mould tonight! However, the advantage of sticking to the familiar method one more time is that it underscores the 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%
Mar 2014 - 13.0%
Mar 2014 - 12.5%
Apr 2014 - 12.6%
Apr 2014 - 12.7%
Apr 2014 - 12.5%
Apr 2014 - 11.9%

Last but not least, because ICM are an online pollster (at least for the purposes of the referendum), we have the updated averages for the four online pollsters that have reported so far this year. These figures do exclude Angus Reid.

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

Yes 38.3% (n/c)
No 46.5% (-1.0)


Yes 45.2% (+0.6)
No 54.8% (-0.6)


Yes 45.6% (+0.6)
No 54.4% (-0.6)


  1. We will have to wait until the morning for the Sunday Post poll. Someone on PB said it was the better of the two polls from info he received from Edinburgh. We shall see. Now off to bed.

  2. SoS says it is 39% Yes (nc), 42% No (-4)

  3. You're quite right - in my rush to get the numbers up I didn't read carefully enough. Corrected now. In many ways that's even better!

  4. 15% of those polled were English. Surely this is too heavy a weighting as I understand that English people are about 8 or 9% of the Scottish population?

  5. Scots were 42% to 40% in favour of Yes.

  6. Yes Andrew, it was amazing reading the prof and the SOS journalists, zooming in on the English voters and how they might swing the vote! As you say they were recorded as 15% of the survey, yet only about 8% of the Scottish population. This is something that has been picked up on in the SOS comments. So this remarkable poll, is still not telling the true story of how well Yes is doing. Pleasing!

  7. One a rare visit to PB today I see that the 'shy No vote' is the latest reason to try and understand what is going on with the Independence referendum. It is when you read that sort of comment you realise their knowledge of politics in Scotland is rather lacking.

    The default position is No not Yes and in the years past (and even to some today) to actually come out as a Yes voter or Scottish Independence was frowned upon and just a political dreamer. You just kept quiet.

    Opposition to Independence is weakening, the speed of which is worrying the No camp and their followers.

  8. In the 2011 census, 9.6% of 16yr+ had England as CoB.

  9. Based on Prof C's numbers for how decided people are, only 37% committed to No in this ICM poll.

    This is on the high side - it was 30% in TNS - and likely down to the sample skew in terms of too high a figure for 'England' with respect to CoB.

    1/3 is the best No can be sure of in terms of fairly solid backing. Doesn't mean they'll all turn out though.

  10. Ah, thanks for the CoB figures for over-16s - I've tried to find those a number of times, but the census website isn't exactly user-friendly. In that case, I'm struggling to see where ICM are getting the 15% figure from.

    Marcia : I totally agree, a Shy Yes Syndrome is far more plausible than a Shy No Syndrome. Presumably our 'friends' at PB are taking their cue from that distinctly odd article from Martin Boon, which entertains the possibility of the latter and completely ignores the possibility of the former. He hints vaguely at anecdotal evidence for a Shy No Syndrome - I'd be fascinated to hear one or two of those anecdotes.

  11. Now that ICM and TNS use the same weighting procedure, we can compare the differences between an online and offline poll more easily. I notice that ICM are showing definite Yes: 30% definite No: 37%, compared to TNS who put those figures at 21% and 30%. ICM also show a 10% higher YES vote and 1% higher NO vote on the headlines.

    Are people less honest with TNS because they are being interviewed face to face? Or are the people on an online panel more engaged with the debate and so more likely to have an opinion? Does the preamble explain the difference between the headline figures?

  12. @Calum Findlay.

    I'd say your answer lies in this:

    2011 SNP voters
    55% Yes / 18% No face to face(rec. TNS)
    71% Yes / 16% No online (rec. PB, ICM)

    2011 Labour voters
    15% Yes / 58% No face to face
    24-30% Yes / 57-58% No online

    Notice how the No doesn't really change but the Yes drops to DK considerably when people are asked in their living room by a stranger rather than when filling in a form alone on their smartphone.

    As for the 'shy no' comments from ICM. Dear dear. Surely if shy No existed, No would do better in more anonymous online polls rather than the opposite? Also, the problem seems to be not enough people admitting they voted SNP in 2011, requiring this group to be up-weighted. Surely this means shy SNP and by implication more likely shy Yes? SNP being of course that 'nasty, anti-English, dictatorial separatist party' :-)