Just before we leave the topic of the ComRes poll entirely, I thought you might be amused to hear about a particularly extreme example of a "control question" that I found buried within the datasets. Some of you will recall a previous discussion on this blog about a question of that sort - a few months ago, YouGov created a fictitious Cabinet minister (I can't remember the name they used) to help put in context the personal ratings of real politicians. For example, if 23% of respondents claimed to recognise the bogus minister, that would make Anna Soubry's 27% recognition rating look considerably less impressive.
ComRes appear to have a rather less subtle approach to control questions, though, and have bizarrely asked each respondent midway through the question sequence whether or not they are "a horse". I can only assume the purpose of this is to make sure that people are actually paying attention to the questions, and aren't just randomly giving any answer to get the survey over with as quickly as possible. But what makes it particularly strange is that this was a telephone poll. Non-attentive respondents are generally much more of a danger in online polls, where there is an incentive of a small monetary reward if the poll is completed (usually 50p). It must have been very awkward for the telephone interviewers to innocently enquire whether or not the person at the other end of the line is a horse - if anything, I'd have thought that would make it more likely that the call would be abruptly terminated.
ComRes poll, 26th-28th March (Labour-held constituencies only) :
Are you a horse?
It's rather alarming that as many as 2% of the sample were dozing off to such an absurd degree, and in a way this calls into question the credibility of the poll's other findings. However, I suppose it's possible that some people may have been taking the attitude of "a stupid question deserves a stupid answer". The most entertaining part of all this is that we've ended up with a unique "horse" crossbreak throughout the datasets, meaning that we're able to see how the attitudes of "horses" differ from their human counterparts. For the most part, there's nothing of any statistical significance, but there is one intriguing exception...
Jim Murphy's net satisfaction rating by group :
So if things don't quite work out for Jackanory Jim in May, all is not lost. Perhaps he should take a leaf out of Cliff Richard's book and recognise that there is so much More to Life than naked ambition. There is, for example, horses.
* * *
SCOT GOES POP POLL OF POLLS
Someone asked me yesterday if I was ever going to update the Poll of Polls again. Your wish is my command (well, within reason). Today's update is based on eight Scottish subsamples from GB-wide polls - three from YouGov, two from Populus, one from Ashcroft, one from TNS-BMRB and one from ComRes. Yesterday's Scottish poll from ComRes is excluded, because it's not a national poll - ie. the fieldwork only covered 40 out of 59 Scottish constituencies.
Making its debut (and quite possibly its swansong) in the Poll of Polls is the new Middleland Integrationists party, which somehow managed to register 1% support in a YouGov subsample last week. It's an eccentric fringe party that was set up last year, having drawn inspiration from Rory Stewart's BBC series about the fantasy kingdom of "The Middleland" (twinned with Narnia). Basically they want Dumfries & Galloway, the Borders, Cumbria and Northumberland to secede from Scotland and England, in order to form a fifth (and naturally ultra-unionist) constituent nation of the UK, presumably with Rory's cairn as the capital city. Rory himself has had to disown them, though, because they're planning to put up candidates against the Tories in May.
Scottish voting intentions for the May 2015 UK general election :
SNP 43.6% (-1.7)
Labour 25.5% (n/c)
Conservatives 16.1% (-1.6)
Liberal Democrats 7.6% (+2.7)
Greens 3.4% (+0.2)
UKIP 3.3% (+0.4)
Middleland Integrationists 0.1% (+0.1)
(The Poll of Polls uses the Scottish subsamples from all GB-wide polls that have been conducted entirely within the last seven days and for which datasets have been provided, and also all full-scale Scottish polls that have been conducted at least partly within the last seven days. Full-scale polls are given ten times the weighting of subsamples.)
* * *
Many thanks to Brian Nicholson on the previous thread for pointing out the most shocking misreporting of a poll that I've seen in a long, long time. Both David Maddox of the Scotsman and Magnus Gardham of the Herald have taken the ComRes voting intention figures for Labour-held constituencies only, and pumped them into the Scotland Votes calculator as if they were national figures, thus producing an utterly nonsensical seats "projection" of SNP 30, Labour 27.
Holy Jesus. Even on a uniform swing, the ComRes poll suggests that the SNP would take 28 of the 40 Labour seats. Here's the thing, Magnus and David - the SNP already hold six seats. So if you add 28 to those six, you've already got 34. Surely that very basic piece of arithmetic ought to have been enough to set alarm bells ringing about your "projection"? In reality, the SNP would also be winning at least seven of the 11 Liberal Democrat seats, plus Eric Joyce's Falkirk seat, so there's no conceivable national projection based on the ComRes poll that would give the SNP fewer than 42 seats.