Sunday, January 12, 2014

Ipsos-Mori's 'state of the nation' poll provides dramatic new evidence that the No vote is soft

The think-tank 'British Future' (who, judging by both their name and their analysis, are anything but impartial on the matter in hand) have published the results of an Ipsos-Mori survey that asked voters in Scotland and the rest of the UK a variety of questions on a broad range of topics - including the independence referendum question itself. Oddly, however, respondents were offered an extra 'Neither' option, in addition to the standard 'Yes' (or apparently 'Agree' in this poll), 'No' ('Disagree') and 'Don't Know'. So the outcome isn't directly comparable to the last Ipsos-Mori poll, but is fascinating nonetheless.

Should Scotland be an independent country? (Scottish-domiciled respondents only)

Yes ("Agree") 32% (+1)
No ("Disagree") 49% (-6)
Neither 10% (n/a)
Don't Know 9% (-5)

(The percentage changes listed above are based on the assumption that this poll was not filtered by certainty to vote. If it turns out that such a filter was applied, then the No vote is down a whopping eight points, the Yes vote is down two, and Don't Knows are down one.)

Now a couple of health warnings here - the Scottish sample size was unusually small at just under 400, although that certainly isn't so low a number as to be meaningless (for example, two of the three Angus Reid referendum polls from last year had sample sizes of about 500). Also, we don't know if the results were only weighted to be representative of Great Britain as a whole, or whether the Scottish sample was weighted separately. The latter possibility seems much more likely given that British Future are making quite a song and dance about their Scottish results, but we may find out more information when Ipsos-Mori publish details of the poll on their own website.

If we assume for the moment that this result is robust, though, what does it tell us? Just like the Ipsos-Mori poll for the Law Society, it provides further evidence that the No vote is softer than the Yes vote. All it seems to take is for respondents to be provided with an additional de facto 'undecided' option, and the No vote instantly plummets, while the Yes vote remains steady.

The poll also confirms the unsurprising finding of recent surveys that Scottish independence is considerably more popular in Scotland than it is in the rest of the UK, which may reflect nagging fears south of the border about the true cost of losing Scottish oil revenues. This is certainly something of a blow for the tedious Tory trolls who used to taunt us with the line "Scottish independence is less popular in Scotland than in England". Those days are over, chaps. Oh, and talking of trolls, this was the brave face that No campaign troll-in-chief Blair McDougall was trying to put on the poll earlier -

"New poll in morning with pretty clear message for us from rest of UK - please don't leave."

Hmmm. So if the implication is that we should always listen to "pretty clear messages" from south of the border, I presume we should also be listening to the pretty clear message from voters in the rest of the UK that David Cameron must stop running away from taking part in a TV referendum debate with his opposite number Alex Salmond? (As of yet, we have no polling evidence on whether English and Welsh voters think that Vladimir Putin, Mariano Rajoy or any of the other foreign leaders that the UK government have begged for help in keeping the Jocks in their place should be allowed to deputise for Cameron in the debate.)

Cynics who reckon that Scots don't really give a monkey's about independence one way or the other will be dismayed to learn that Scottish respondents say that the referendum is by far the most important event of the coming year to them personally. The grotesque 'celebration' of the start of the horrors of World War I limps in at a distant fourth place with 13% saying it is important, only just ahead of the football World Cup - at which Scotland won't even be represented. Incidentally, a mere 13% of respondents in Scotland will be supporting England at the World Cup, which is less than the 15% of respondents who will be supporting whichever team England happen to be playing. 20% regard themselves as genuine neutrals, 10% will be supporting a specific team other than England, and the remainder won't be watching the World Cup at all. Not much comfort for the "UK is a big happy family" worldview there, and I must admit I'm really surprised by those numbers - particularly given that almost 10% of the population of Scotland are English themselves!

The other interesting finding from this poll is on national identity...

More or wholly Scottish 49%
Equally Scottish and British 31%
More or wholly British 19%

I know that Scottish Skier will be concerned that this once again indicates that Ipsos-Mori may be getting their Scottish sampling wrong, because both the census results and the Scottish Social Attitudes Survey suggest that Scottish identity is stronger than that (much, much stronger in the case of the census).

* * *


I've been swithering over whether I should take this poll into account for the Poll of Polls. At first, the answer seemed like an obvious 'no', because the methodology is so different from past (and presumably future) Ipsos-Mori polls, with a smaller sample size, fieldwork conducted online rather than by telephone, the peculiarity of offering two distinct 'Neither/Don't Know' options, and the likelihood that no certainty to vote filter was applied for the headline numbers. We also don't know for sure whether the Scottish sample was weighted properly, and I have one or two doubts over the question used. (The British Future report lists the actual referendum question with no alterations, but places a very suspicious asterisk next to it, for which no explanation is provided.)

The counter-argument, though, is that the purpose of the Poll of Polls is to calculate a straightforward average of the latest headline findings from all British Polling Council pollsters, regardless of the methodology they choose to use at any given time. So I'm going to provide a provisional update that takes account of this poll, but I'll revert to the previous update if it turns out that the Scottish sample wasn't separately weighted, or that the question asked bore no relation to the referendum question. Fortunately, one thing we do know for sure is that the fieldwork for this poll took place later than the fieldwork for the last published Ipsos-Mori referendum poll, albeit only just (6-11 December).

Unsurprisingly, the provisional update shows the pro-independence campaign closing the gap still further...

MEAN AVERAGE (not excluding Don't Knows) :

Yes 32.7% (-0.3)
No 47.5% (-1.3)

MEAN AVERAGE (excluding Don't Knows) :

Yes 40.8% (+0.5)
No 59.2% (-0.5)

MEDIAN AVERAGE (excluding Don't Knows) :

Yes 39.6% (n/c)
No 60.4% (n/c)

If this update is confirmed, then the 1.3% drop in the headline No vote is by far the biggest individual change we've seen since the Poll of Polls started - as only one-sixth of the sample differs with each update, movements in the figures are usually much more glacial in nature. The No lead has dropped from 15.8% to 14.8%, and the Yes campaign now require a mere 7.4% swing to draw level.

* * *

There's an editorial in today's Scotland on Sunday that references the Ipsos-Mori poll, and if the familiar phraseology is anything to go by it seems to have been wholly or partly penned by Kenny Farquharson. It also indirectly mentions the anti-independence campaign's wizard plan to bombard Scotland with nuisance phone calls from English people trying to persuade us to vote No (they'll almost certainly be activists from the London parties posing as 'ordinary people'). It concludes with this utterly bizarre example of logical gymnastics -

"The SNP makes reassuring noises about the social union with the rest of Britain continuing after independence. But what happens when the people in that social union – our friends, relatives and colleagues from Liverpool, Newcastle, the Welsh valleys and the rolling hills of Antrim – ask us not to sacrifice the UK’s political union? Such a moment will be the ultimate test for the kind of Yes voter who says 'I’m not a nationalist, but...' Because it is very difficult to characterise the ignoring of that plea as anything other than act of political nationalism."

OK, let's think this through, step by step. Imagine you're a member of the Green party who is definitely not a nationalist, but you've nevertheless reached the pragmatic conclusion that the kind of transformative change you want to see in Scottish society is far more likely to come about through independence than by sticking with the Westminster system. However, you then receive a telephone call from someone who wants you to vote against independence. According to Kenny (or whoever wrote the editorial), the simple fact that this person has an English accent is sufficient to mean that your own opinion has become invalid - and if you want it to become valid again you have to henceforth call yourself a nationalist. Er, why? How does that work? How does an English political activist telling you "we should stay united because of EastEnders and the Queen and stuff" suddenly mean that your own logic about Westminster's inability to deliver transformative change no longer applies?

Unless the activist is offering you some kind of vote swap (I'll vote Green at Westminster if you vote No), there's no reason whatever to suppose you would be given even the slightest pause for thought. Indeed, the fact that your support for independence is founded on policy objectives rather than nationalist emotion means that you're even less likely to be open to that kind of heartstring-tugging persuasion. If this 'love-bombing' wheeze reminds me of anything, it's the Guardian's idea of getting its readers to write to Ohio residents in 2004, asking them to vote for John Kerry - and we all know how that went down.

* * *

UPDATE (2.50pm) : More details of the poll have now been released, and on that basis I'll be removing it from the Poll of Polls - see the comments below.


  1. All of the data tables are online

  2. Well, on the basis of those tables, this poll will have to be removed from the Poll of Polls. Contrary to the misleading presentation in the British Future report, the actual referendum question was not used, and there were six possible responses not four (ie. it can't really be said to approximate to a Yes/No question). Ipsos-Mori's summary also fails to mention whether the Scottish sample was properly weighted, which makes it much more likely that it wasn't.

    However, the point about the No vote proving to be soft when the question is posed in a slightly different way still stands. And it turns out that only 35% of the Scottish subsample "strongly disagree" that Scotland should be an independent country, which if correct leaves the referendum wide open.