I know that a number of us have separately come to the conclusion that Blair McDougall is a thoroughly objectionable and rather thuggish individual, at least when he dons his online persona. In short, it's hard to think of a more fitting supremo for the anti-independence campaign. But give the man his due, he does offer some precious moments of entertainment along the way. One game I've become fond of playing is "spot the good poll for Yes" - all you have to do is look out for the things that McDougall doesn't say about a poll on Twitter. This was his offering yesterday...
"No bounce for Yes after SNP conference. TNS poll finds support for leaving UK stuck at historic low of 25%."
From which I instantly deduced that the No vote had fallen for a second successive TNS-BMRB poll, which in turn meant that there had also been a net swing in favour of the Yes campaign for the second successive poll by the company. Here are the figures -
Should Scotland be an independent country?
Yes 25% (-)
No 43% (-1)
When compared to the TNS poll conducted in late August, the No vote has slipped by four points from 47% to 43%, and the overall No lead has also dropped four points, from 22 points to 18. As you know, I can't help but roll my eyes at the mainstream media narrative that a small increase in the No lead is always "another blow for Alex Salmond", whereas a small decrease in the lead is always "another no change poll". In spite of the laughable double standard, there might sometimes be a justification for the latter point, because small changes are often just normal fluctuations that will cancel each other out. But I have a severe problem with anyone peddling the "no change" line in this particular case. Not only has the No lead reported by TNS-BMRB dropped by four points since August, but this is also the lowest No lead that TNS have shown for almost two years. At its peak in the autumn of 2012, the lead stood at 25 points, so a drop of seven points since then is clearly statistically significant (ie. it's not a mirage caused by the standard margin of error). Furthermore, the proportion of undecided voters has climbed to a new high watermark for the third successive TNS poll, now standing at an astonishing 32%. How the hell does that constitute a "no change" position, guys?
Another curious thing that the mainstream media seem to be doing with their reporting of this poll is ignoring TNS' own designated headline figures, and instead zeroing in on the figures for those certain to vote. Presumably this is an attempt to put the maximum possible glitter on the figures for the No campaign ("nearly half of Scots support the union" is the predictable line warbled by one or two publications), but in this case they may be shooting themselves in the foot, because the No lead has in fact slipped even further among definite voters -
Yes 29% (+1)
No 47% (-3)
That's a net swing of 2% to the Yes campaign since last month. With the lead down to 18 points on both measures, TNS have effectively joined ICM and Angus Reid in the group of pollsters showing a middling lead. YouGov are also not too far away from that group as well, assuming their last poll wasn't a blip. And at this point, I must break the habit of a lifetime, and actually give YouGov some credit. I was away on holiday when their last poll came out, and consequently I didn't realise until a few days ago that they had finally started to get their act together, by replacing their notoriously biased preamble, and by weighting the figures by recalled Holyrood vote. I use the word "started" advisedly, because the Scottish sample in their daily poll still seems to be weighted under the old system, producing the usual fantastical results. But as far as referendum polling is concerned, let's hope they've permanently laid to rest the spectre of cretinous Cochranism. The methodology in their September poll was absolutely fine as far I can see, and in those circumstances it was scarcely surprising that the No lead tumbled by a whopping ten points.
So that leaves Panelbase as the outlier on one end of the spectrum showing very modest No leads (and in one case an outright Yes lead), with Ipsos-Mori now alone as the equivalent outlier on the other extreme showing a large No lead. I exclude Progressive Scottish Opinion from that equation because I don't regard them as a credible pollster (their 'mad as a bucket of frogs' polls for the 2007 Holyrood campaign will explain why). Ipsos-Mori aren't so easily dismissed, and it is slightly troubling that they are now showing a No lead much larger than in their poll in early 2012, which was a notable high watermark for Yes. There's no obvious explanation for that movement, unless the company have introduced a significant methodological change over the last eighteen months. But I would be much more worried about that if other pollsters, and especially ICM, had detected the same trend - and, quite simply, they haven't. We do still have to be a little cautious, because it may yet turn out that Ipsos-Mori are right and all the others are wrong - but it's just as likely that will turn out to be true of Panelbase.
To finish off, here are a few interesting nuggets from the details of the TNS poll -
* 35-44 year olds are the age group most likely to back independence.
* The 'West' (excluding Glasgow) is the region of the country with the biggest proportion of Yes voters, although Glasgow itself is where the No lead is at its lowest - just seven points.
* Liberal Democrat voters are the least likely to be loyal to their own party's constitutional stance - just 51% of them currently plan to vote against independence.