By Dave Taylor

TNS-BMRB must be deeply regretting their launch of the “Scottish Opinion Monitor” (presumably to try to rival Ipsos-MORI’s “Scottish Public Opinion Monitor”) with the first of their planned monthly polls on the referendum. Every pollster produces a rogue poll from time to time, but to have such an obvious rogue as the first of your series must be embarrassing.

Whether the Herald is embarrassed by its headlining of the poll may be less clear!

The deep deficiency of their methodology with regard to political polling has been widely noted by critics ranging from Wings over Scotland to John Curtice.

I’ve written about the problem of false recall before. Some pollsters, including TNS-BMRB and Ipsos-MORI are so convinced that people don’t accurately remember how they voted, that they don’t weight by political leanings at all. Fortunately, they do record the recalled vote, so that we can measure the likely accuracy of the poll, by how remote from reality their sample is.

If future TNS-BMRB polls produce a sample which is more like political reality, then such a poll should be taken seriously. The level of rogue sample in this case is shown by the extent to which it varies from what actually happened. If their sample of the population had been accurate, then Labour would have won 43 constituencies in 2011, instead of the 15 that they actually did, the LDs would have won 8 constituencies instead of 2.

The Panelbase poll for the SNP has released the detail of questions 6 & 7. (Will we see Q5 later – or did the SNP not like the answers to it?)

People don’t vote in referendums solely on the issue at question. Political attitudes on completely different issues influence how people vote. While those most engaged with the issue will dislike that, it is reality. Who those in England decide to elect as the government in Westminster will influence some.

Given the prospect of a Tory led Westminster Government, the likely Yes vote rises from 44% to 50% : if Labour, it rises by 3%. Surprisingly, not all of the shifts within party voters are predictable!

There’s no change among Tory voters thinking they’ll get a Tory Government at Westminster, but 7% move to Yes if they think Labour will rule in London. Labour voters have an almost exact mirror image.

Asked to think about London Government by any party seems to turn off lots of LD voters. Prospective Tory (or Tory/LD) rule moves 13% to vote Yes, while Ed as PM turns an extra 11% to Yes.

7% of SNP voters (mainly referendum Don’t Knows) move to Yes faced with Tory led London rule, while 4% of the SNP voting Don’t Knows split equally to Yes and No if they think London Labour will be in charge.

Among Greens, 11% move from No to Don’t Know if they have to think about either Tory or Labour in No 10.

“Others” show no shift faced with a Labour Government, but half of them (presumably Socialists) move from No to Yes.

Question 8 asked “Are you satisfied or dissatisfied with the way the following political leaders are doing their job?” The overall ratings shouldn’t surprise anyone (Salmond Positive Rating: +11% : Cameron Negative Rating: -45% : Miliband Negative Rating: -46% : Clegg Negative Rating: -53%). However, that Salmond has a higher percentage satisfied with him than Milliband can find – among Labour voters shows some of the complexity of Scottish politics – and maybe why the pollsters have such difficulty determining how we think.

The YouGov/Devo Plus poll asked for voting intention for both Holyrood constituencies and Westminster in their poll – but the weighted results are redacted. No explanation for this has been forthcoming, but the suggestion by one Labour activist seems likely “The results were crap for the parties commissioning the poll.”

If the weighting for Westminster 2010 is applied to them (and it’s not something which one should do – but it’s fun!) then you get – Holyrood – SNP 50% : Lab 28% : Con 14% : LD 8% (which doesn’t seem too far away from other polling). Westminster – SNP 44% : Lab 31% : Con 16% : LD 8%.

Such a pattern would give the SNP 41 MPs to Labour’s 14. That would actually fit with the theory of some commentators that Scots already see the SNP as a defence against London power, so there’s no need to vote for independence.

It would be little comfort to the SNP if the Yes side lost the referendum, but got lots of MPs who could do little except influence which UK party governs England.


2013-09-05 21:36


Panelbase have issued a response to their critics, while accepting that no pollster may have the right methodology to test a referendum decision.…/…

Meanwhile, I’m told that TNS-BMRB are having internal discussions to decide whether they need to make changes to their methodology.

That’s positive thinking from both pollsters. Sadly, others seem content to carry on as if their methodology requires no examination.

You must be logged-in in order to post a comment.