Sunday, October 27, 2013

Boost for pro-independence campaign in new Panelbase poll

I'm delighted to see that RevStu has included a straight voting intention question in the latest Wings over Scotland poll, partly because it takes away any alibi from Alex Massie and others for making any of the thoroughly daft comments they did last time, but also because it will help enormously in the pursuit of the poll's secondary objective of bolstering the profile and credibility of alternative media. The fact that the poll was apparently discussed on this morning's Headlines programme on BBC Radio Scotland (with full attribution given to Wings) tells you all you need to know.

The figures themselves make encouraging reading for the Yes campaign, with the No lead decreasing by two points since the last Panelbase poll in September -

Should Scotland be an independent country?

Yes 37% (-)
No 45% (-2)

I know those are slightly different from the headline figures used by RevStu, but the only figures that are directly comparable from last month's poll are the ones filtered by certainty to vote.  In any case, the difference isn't significant - with or without the filter, the No lead stands at a modest eight points, although understandably the number of undecided voters falls if less probable voters are removed from the equation.

Incidentally, although the usual suspects will doubtless ignore the two-point drop in the No lead, and dismiss this as "another no change poll" (isn't it amazing how a decrease in the No lead is always invisible, whereas any increase is invariably a "blow for Alex Salmond"?), it's worth pointing out that those same people were all too keen to discredit the Panelbase poll conducted in August showing a lead for Yes.  Well, if we take their views at face value and treat that August poll as illegitimate, then today's No lead is in fact smaller than in both of the last two Panelbase polls, conducted in July and September respectively.

I was also very excited to see that Panelbase have called the bluff of their more moronic critics, and used another polling company's panel for half of the sample in this poll.  The No lead is actually slightly smaller (eight points) in the other company's sample than it is in Panelbase's (nine points).  Not only does that go a long way towards disproving the bizarre Twitter smear that Panelbase have been "infiltrated by Cybernats", but more importantly, it lends credence to Panelbase's own theory that the significant divergence between different pollsters' results is not caused by differences in the raw responses of the voters interviewed, but rather by the filters and weightings that the pollsters themselves apply after the interviews.  That raises the fascinating possibility that polls by companies such as YouGov would be showing exactly the same type of close race that Panelbase polls typically show, if the filtering and weighting of the raw data was done in the same way.  Indeed, it suggests that YouGov and co might even be showing an outright lead for Yes by now, if they also asked the main question after the same two innocuous questions that Panelbase did in their "illegitimate" August poll!

It would be really interesting to know which pollster's panel was used in this poll - the findings would be highly significant regardless of the pollster's identity, but they would be even more significant if it was YouGov, Ipsos-Mori or TNS-BMRB, which are the three pollsters that have tended to be most favourable for No.  (ICM and Angus Reid are somewhere in the middle.)

There was a great deal of theorising in the comments section of Wings this morning about the reason why 18-24 year olds are the least likely to support independence (while, paradoxically, 25-34 year olds are the most likely to support it), but in my view that may have been based on a false premise.  It's only been a few months since an Ipsos-Mori poll showed a huge lead for Yes among 18-24 year olds.  The sample sizes for that age group are extremely small, and when we have such widely varying results, all we can meaningfully say is that we just don't know what the true state of play is.


  1. In the biggest shock of all time, PB ignores the new Scottish Poll in favour of lardy Ashcroft's 6 month old survey.
    And the tims are proving themselves even more insane with their claims that ed millipede is the most popular politician in Scotland.

    They should go back to supporting genital mutilation for babies.

  2. I've managed to keep my life free of PB contamination for several weeks now, and I'm not planning to break the spell, but if Smithson really has run a thread on an ancient Ashcroft poll while ignoring the bang up to date Panelbase poll, then words fail me. Another new low (if such a thing is possible).