Sunday, January 26, 2014

This is the bit where everything changes - sensational ICM poll shows pro-independence campaign just 7% behind

At moments like this you're supposed to punch the air in delight, but I'm so dumbfounded that I think I've only managed to do it in my head so far. ICM, regarded by many as the 'gold standard' of UK pollsters, have released the first referendum poll of 2014, and it shows that the pro-independence campaign have slashed the No lead from 17% to 7%.

Should Scotland be an independent country?

Yes 37% (+5)
No 44% (-5)

I had a funny feeling a couple of hours ago that something special was in the offing. Normally when a new poll is on its way, the anti-independence campaign's troll-in-chief Blair McDougall can't resist a "scene-setting" tweet, but there was only deafening silence from him tonight, which left open the possibility that he'd seen the figures and knew that this time there was absolutely no way he could spin them in his favour. Some of the other usual suspects (notably our old friend Mike "can't be arsed" Smithson) indicated that they had no idea what the poll showed, suggesting an unusual level of secrecy that you wouldn't expect for a poll with a routine result.

So what does this mean? For years now, there has been an ongoing 'intra-nationalist debate' between those who think that polls are basically meaningless ("OK, they asked 1000 people, but what about the other five million?"), and those of us who think that we can certainly be critical of polls and point out their mistakes and limitations, but that there's no point in sticking our heads in the sand and trying to convince ourselves that polls don't matter at all. After all, there is a perfectly straightforward scientific basis for the idea that you can ask 1000 people for their opinion and get a result that is representative of the general population to within a margin of error of 3% (but only if you have your methodology absolutely bang on, which is the caveat that is too often overlooked). And perhaps more pertinently, I remember that when I took part in the For A' That podcast back in September, I made the point that the prevailing media narrative that is so unfavourable for the Yes campaign has been largely driven by poll results, and that everything would be completely different if we could only get those numbers to change. Make no mistake about it - a transformed narrative matters. Voters who previously thought that there was no point in wasting their mental energy thinking about the arguments for independence because Yes were too far behind will think again. Newspapers and other media outlets who previously thought that they were playing it safe and giving the audience what it wanted by churning out relentless anti-independence propaganda will realise that their interests are not quite so clear-cut. More specifically, a spontaneous surge for Yes in the polls is probably a prerequisite for securing unambiguous newspaper endorsements for independence in September.

We can now expect the long-awaited changed narrative to begin to unfold as a direct result of this poll - on one crucial condition. If there is a second poll in the coming days (in theory a TNS-BMRB poll should be due before the month is out), and if that poll contradicts ICM, the media will be all too keen to believe the figures that give them the easiest life. But if the next poll doesn't tell them what they want to hear, all hell will break loose. There's one thing even the most die-hard of unionist scribes will know in his/her heart of hearts - if Yes really have closed the gap from 17% to 7% since the last ICM poll four months ago, they are perfectly capable of closing the gap from 7% to zero over the next eight months. That's especially true given that the official campaign period (with the regulations to ensure balanced broadcast coverage) has not even started yet. The "Yes has already lost" fairy-story so beloved of Ian 'Complacency' Smart and his ilk is now utterly dead in the water.

* * *


Regular readers will remember that I provisionally updated the Poll of Polls after an Ipsos-Mori poll was released on January 12th, and that I then reluctantly had to reverse it after it transpired that the poll hadn't asked a straight Yes/No question, and probably wasn't properly weighted. I say 'reluctantly' because it was rather a good update for Yes, but I'm delighted to say that the first genuine update of referendum year shows much the same story anyway, with the pro-independence campaign closing the gap for the fifth time in succession...

MEAN AVERAGE (not excluding Don't Knows) :

Yes 33.8% (+0.8)
No 48.0% (-0.8)

MEAN AVERAGE (excluding Don't Knows) :

Yes 41.3% (+1.0)
No 58.7% (-1.0)

MEDIAN AVERAGE (excluding Don't Knows) :

Yes 40.9% (+1.3)
No 59.1% (-1.3)

As ever, the Poll of Polls is based on a rolling average of six polls - the most recent one from each of the six pollsters that have conducted properly-weighted nationwide referendum polls during the campaign, and that adhere to British Polling Council (BPC) rules. If any other BPC pollsters enter the fray at some point (for example ComRes or Populus), they'll be taken into account as well.

Obviously, with only one-sixth of the sample changing with this update, it's not possible for the movements to be as dramatic as seen in the headline ICM figures. However, Yes have still reached another new landmark, breaking through the psychological 40% threshold on the median average for the first time. The median was previously calculated as the mid-point between TNS-BMRB and ICM (as the 'middle' pollsters), but ICM have astonishingly just replaced Panelbase as the outright most favourable pollster for Yes, meaning that the median is now the mid-point between Angus Reid and TNS-BMRB.

On the headline average figures, Yes are left requiring a swing of just 7.1% to draw level - but the reality is that it'll almost certainly be much less than that by now if the trend picked up by ICM is genuine. Indeed, on that trend it's quite possible (arguably likely) that Panelbase would be showing either a dead heat or a narrow Yes lead at the moment. If I was a decision-maker in the SNP, I'd be sorely tempted to commission another Panelbase poll right now - the impact on the media narrative could be huge.

* * *

"You're better together with a Tory government you didn't vote for!" say panto stars John Barrowman and the Krankies. "Oh no we're not!" retort the audience.

When the news of this poll broke, I was in the middle of writing a post about the rather more trivial matter of John Barrowman's characteristically self-effacing Burns Night 'Address to the Nation', but I don't think I'll bother finishing it now. Suffice to say I spent my own Burns Night at the Hydro in Glasgow watching pro-independence singers Dougie MacLean, Rachel Sermanni and Karine Polwart performing for Celtic Connections. In particular, Polwart brought a tear to my eye with her beautiful rendition of Green Grow the Rashes. (I mean, obviously it wasn't a patch on Barrowman's timeless cover of Baby Give It Up by KC and the Sunshine Band, but we can only make do with the scraps of entertainment that are actually available to us here in the far provinces.)

Oh, and I think I would have been left with marginally more respect for Barrowman if he'd done his faintly pathetic "kilts, Irn Bru and Salmond is fat" routine in the American accent he uses on every other occasion he's on television. His weird decision to go Scottish for one night only reminded me a bit of the advice Margaret Thatcher apparently received in the 80s with a view to improving her dismal personal ratings in these parts. It was suggested that she adopt a mild Scottish accent on her forays north - a comic spectacle of truly epic proportions that we were cruelly denied from witnessing.


  1. Remarkable poll that should see more than a few of the usual BritNat suspects running around like headless chickens or burying their heads in the sand in complete denial.

    You will be flabbergasted to hear James, that the scottish tory surgers on PB seem none too pleased with this poll and are already reaching for Curtice for excuses.

    Very funny indeed and for those of us who remember their shrieking outrage at Panelbase all the more rewarding.

    Smithson's silence can also be explained by an unsurprising financial interest. Or as he puts it..

    "In the early hours when news of the poll came through I wagered several hundred pounds on Betfair that YES will win."

  2. The first poll showing movement from No to Yes for sometime rather than go temporarily to Don't Know.

  3. James, by my calculations, looking at Poll of Polls back to mid September that's a 6% drop in NO lead over 18 weeks. Dropping at a rate of about 0.33% per week.

    The ICM poll showed a drop in NO lead of 10% over 18 weeks, so a bit higher trend rate.

    On that basis TNS next week should have a No lead of about 12% (40% vs 28% perhaps) to keep this trend rate going.

    If better than that then could be we have seen an acceleration of the trend over the Xmas period, and maybe reached the first 'tipping point'

    If worse than that then could be that ICM was a bit of an outlier.

    Waiting with bated breath.

  4. This comment has been removed by the author.

  5. Great news on the polling front.

    Was rolling around laughing at your put down of Barrowman's piece.

    I loved the irony of his "nostalgia and nonsense" whilst cracking pantomime jokes wearing a tartan jacket.

    And I can't help but wonder how he knew the organiser was wearing shorts under his kilt. Did he look or just ask?

  6. Interestingly, the 37%/44% split is similar to the general picture of opinion polls before 2012. They always shown 35% - 40% of us supported independence. The yes vote fell in 2012 because people wanted more information after the referendum became something that was definitely happening.

    If this poll asked the same question as the September one, people will have been asked "how they think they will vote". Now that this is referendum year people will seriously be considering the question. This poll is evidence of what I've always thought: we would win back those 35%-40% when the time came.

    No other polling company asks people "how they think they will vote" so I don't expect them to show a big swing.

  7. James,

    Is the poll about independence taken at the time of the Cowdenbeath by-election worthy of analysis?

  8. Calum : The question used by TNS-BMRB is closest to ICM's, because it asks how people will vote on the actual referendum date, rather than "tomorrow". But I wouldn't expect TNS to show a swing like this, for the simple reason that (unlike ICM) they've already conducted two polls since the publication of the White Paper, both of which showed a drop in the No lead. So to some extent the pro-Yes swing is already factored into the figures we know about. Polls from Panelbase or Angus Reid would arguably be more interesting, because neither of them have reported since the White Paper.

    Douglas : It's not really possible to interpret the SNP's canvass returns from Cowdenbeath without knowing what the figures were for by-election voting intention. If the percentage saying they were voting SNP wasn't a million miles away from the actual result, then that would add a lot of credibility to the referendum voting intention figures - ie. if people weren't lying about the by-election, there's no reason to think they were lying about the referendum. But there always has to be a health warning on canvass figures - one classic example of the problem is the Liberal Democrats wrongly convincing themselves on the basis of their canvass returns that they had defeated Alex Salmond in Gordon in 2007.