Wednesday, January 8, 2014

New ComRes poll of referendum voting intentions in the Borders and Dumfries & Galloway

The anti-independence campaign have spent the evening making an unconvincing attempt at looking thrilled to bits with the news that a regional ComRes referendum poll has shown them ahead in the Borders and Dumfries & Galloway, which is about as surprising a result as a constituency poll showing that the Tories have the lead in Windsor.  It of course tells us absolutely nothing at all about the national picture - these two local authorities have a combined population of just 265,000, or roughly 5% of the population of Scotland as a whole.  They're also just about the only parts of the country where the uber-unionist Tories can be seriously regarded as a major force, and in the 1997 devolution referendum the Yes vote there was approximately 12.5% lower than the nationwide figure.  So it's not remotely unexpected that the Yes vote in this poll is some 9% lower than the current national average of 33.0%, and that the No vote is just over 10% higher than the national average of 48.8%.  These differences are mirrored in the finding that Scottish national identity is somewhat weaker in the region than across the country as a whole...

ComRes figures for national identity in the Borders and Dumfries & Galloway (with YouGov's most recent figures for national identity across Scotland as a whole in brackets) :

More or wholly Scottish - 45% (54%)
Equally Scottish and British - 35% (27%)
More or wholly British - 19% (14%)

So although the most popular identity is Scottish, there is considerably more of a British loyalty in these two local authorities than there is across Scotland as a whole, and the extent of that difference is uncannily similar (and almost certainly directly connected) to the headline differences in voting intentions.

This poll also further muddies the waters with respect to the state of play among young people, because 16-24 year olds seem to be both the second-most likely age group to vote Yes, and the second-most likely to vote No.  The apparent paradox comes about because there are fewer undecided voters in that age group than in any other - an entirely counter-intuitive finding that can probably be explained by the small sample size.  The findings are more clear-cut in terms of national identity - a bigger percentage of 16-24 year olds feel mostly or wholly Scottish (51%) than is the case among any other age group.  Not much comfort there for Jan Eichhorn and his wildly implausible theory (which was earnestly reported as hard fact by the Telegraph) that exposure to Facebook may somehow be making young people more "British" than their elders!

It's worth noting that the poll's billing as the "first full poll on attitudes to independence in the south of Scotland" is grossly misleading (if admittedly not technically inaccurate).  It gives the false impression that the poll was conducted throughout the entire South of Scotland electoral region, whereas in fact the Borders and Dumfries & Galloway comprise significantly less than half of that region in terms of population, and are by far the most Tory-oriented part. So it's not even possible to make a direct comparison with the regional breakdown for the South that has been provided in recent TNS-BMRB polls.

Basically, the overall summary is "nothing much to see here".  It may seem distinctly peculiar that the first voting intention poll of referendum year was confined to respondents in just two out of Scotland's thirty-two local authorities, but in fact there is a kind of logic to it.  The 5% of the population who live there are the only people in Scotland who don't receive STV, and instead have to put up with an appalling service from "ITV Border" (basically the London-based ITV network with a tiny number of regional opt-outs, most of which are broadcast from Gateshead!).  The poll was commissioned by ITV Border to launch their new late-night Representing Border show, which will presumably be a rough equivalent to STV's Scotland Tonight.  This marks the station's attempt to up their game in referendum year - which won't exactly be difficult, given that they're essentially starting from a position of zero!  The only slight cause for concern is that their political editor is former Labour spin doctor Peter MacMahon, although to be fair he isn't exactly John McTernan.

Probably the only people who might prick up their ears at the results of this poll are political dinosaurs like Lord Kilclooney and Jim Wallace, with their crazy dreams of London enforcing an Irish-style "partition" of Scotland in the event of a Yes vote...


  1. "former Labour spin doctor Peter MacMahon"

    I had no idea he was a former Labour spin doctor. There's an awful lot of them going about. Has anyone in the Scottish media not got some sort of history with the Labour party?

  2. Yes, it was way back in Henry McLeish's day, so it was a relatively short stint.

  3. Having mainly English TV and no STV does influence opinion.

  4. To be strictly correct ComRes have not asked or reported a nationality question. They asked and reported the Moreno question which is really an enquiry about how we reconcile our Scottish & British identities.

    In this regard, the best data set is that produced by Curtice's Scot Social Attitudes Survey (available on What Scotland Thinks website) which askes both the Moreno and (a ‘forced’) Identity question of the same sample populations over more than a decade. Interestingly the What Scotland Thinks website allows us to investigate the cross breaks and interaction between the respondent’s answers to both questions. We can therefore, with a little bit of work, establish a viable relationship between Moreno and national identity and vice-versa.

    So far so interesting, and it wouldn’t take us far except that the Census has given us definitive data for national identity (reported both nationally and by council area). I should say at this point, that Curtice has rubbished the Census data because its “use of the singular form, potentially leading people to give a single answer”, but in so far as this is a flaw then it is no worse than the way that Moreno channels respondents toward a duel identity response. Neither question produces the right or wrong answer, but we can safely say that both force a contrived answer. That said, Curtice’s critic of the Census is all the more surprising as its national identity result broadly reflects the results of his own surveys.

    Curtice, of course, is a TV darling and rarely does a week pass without us seeing him opining on matters wide and various. I suppose, given the British/Unionist sympathies of Scotland’s broadcast media, it should be no surprise that in his analysis Curtice nails his colours to the Moreno mast but his questioning of the Census results feels like little more than a professional pollster pleading the probity of his preferred analysis over an actual outcome (remember the census enquires of every household and requires answers for every citizen).

    Curtice’s self-serving critic aside, the Census does give us a body of ‘control data’ against which we can compare the results of polls that also report a result for either National Identity or Moreno. Referring a poll’s Identity/Moreno result back to the Census data and/or the Scot Social Attitudes cross break data allows us to gauge how accurately the polling sample describes the character of the broader population whose opinion the pollster is investigating.

    Being a regional poll, latest from ComRes presents some difficulty in making this comparison as the Scot Social Attitudes Moreno data is Scotland wide. However, we can identify National Identity for the Dumfries & Galloway & Scottish Borders council areas from the Census data, and while that data does show a slightly higher than average number reporting that they are ‘British Only’ it is counterbalanced by a reduction in the number reporting a duel ‘Scottish & British’ identity. Given that the Census found that respectively 59.6% and 57.7% of the residents of Dumfries & Galloway & Scottish Borders consider themselves Scottish Only, one wonders how ComRes have managed to find a sample population 78% of which identify themselves as British?

    Of course we should not be surprised at such a result. Most polls (Panelbase being the exception) that include an Identity or Moreno question seem to find sample populations that include a higher proportion of respondents claiming a partly or wholly British identity than can be justified in comparison to either the Census or even the Scot Social Attitudes data. In short there is increasing evidence of a serious (and in terms of results) probably misleading flaw in the polling methodology currently being used to assess how Scots may vote on the 19th of September.

  5. I got polled in this one. The first ever. I got a phone call from a very friendly chap with a south of England accent. My response to voting intentions was a definite "yes", and "More Scottish than British". Geographically I'm British, and I'm quite comfortable with that. My national identity is Scottish, but the question didn't specify this distinction. It struck me too that to say to this nice south of England chappie that I was Scottish and NOT British might seem a trifle rude. Similarly, if I was less definite about my voting intentions, I might have thought stating a definite "yes" might have seemed a bit curt too - "yeah, we want to cut ourselves off from you guys down there", and all that. When you are involved in a pleasant telephone conversation I do believe your self-censor kicks in somewhat.

  6. EricF.
    You have made some very interesting observations there. Pollsters must be aware of the reluctance to seem rude in telephone polls and would use 'nice' polite people to help achieve their clients' objectives.
    Also, the fact that you think you're British on a geographical basis again would aid BT's objecttives.
    Like Alasdair Stirling I believe there are major errors in most of the polls on Indy published so far and your comments suggest soem of the ways errors are introduced intentionally or not.

  7. Thanks for this James. Like polling, identity also leaves me a bit confused probably because I'm not particularly bothered by my own.

    But the discussion above is fascinating. I'm on the east coast in Berwickshire and for various reasons I don't have the courage to wear my yes badge in public - I'm an incomer from England. I wonder how many people are also reticent. I have noticed though in this small town an increase in the flying of satires and am wondering if people are expressing their preference this way rather than through overt symbols like badges.

    I am also finding that in conversations close Scots friends are voting yes. I'm quite pleased that at least 24% are likely to vote yes. We now have to convert most of the 17% to take us to mid 30%.