Tuesday, March 2, 2010

SNP still polling strongly with ComRes

In the light of the utter confusion over the last 48 hours about the polling situation in Scotland (the entirety of this evening's Newsnight Scotland was even devoted to it), just about any scrap of additional information is potentially useful. I must admit my pessimistic working assumption has been that YouGov's bleaker figures for the SNP are probably closest to the truth, but a little bit of contrary evidence has been provided by the latest ComRes subsample. It appears that the previous figures from the same firm were not the blip we might have assumed, and the SNP are still well ahead of where YouGov say they are, although admittedly they have once again slipped marginally behind Labour. Here are the full figures -

Labour 31% (+4)
SNP 29% (-5)
Conservatives 20% (-2)
Liberal Democrats 17% (+6)
Others 4% (-2)

Set against this relatively good news for the SNP, we have to bear in mind that the Scottish subsamples from YouGov's daily GB-wide polls have typically been showing somewhat lower figures for the party, broadly in line with their full-scale poll in Scotland on Sunday.


  1. 101% is a pretty impressive total percentage (not quite sure how they figure that 16.36 rounds up to 17% for the Lib Dems...) I'm cautious of giving too much credence to a poll of under 60 people; after all, if just two more people had said SNP instead of Labour, you've suddenly got the SNP on 33% and Labour on 27% (although they'd probably round 27.27 up to 28%), so it's easy to see how these +/-4% changes happen. Then again, the YouGov sub-samples are equally questionable, so at least it provides a counter-argument to their results.

    It's a cliche, but there's only one poll that matters. All these other ones up until the election can only serve to distort the electorate's idea of the political landscape, trying to convince people that voting for party X is a waste of your vote because they're only getting Y% of votes, compared to party Z who are getting twice as many votes. Hopefully people will just vote for whoever they intended to vote for anyway (unless they weren't voting for the SNP, that is!)

  2. Absolutely, Doug, a sample of 60 is utterly meaningless in isolation - the margin of error is astronomical, although looking at the pattern of several subsamples over time might give a rough clue as to what is going on. Incidentally, a combined share of 101% is not at all unusual - even in full-scale polls - because of rounding up or down, although I must admit I hadn't noticed the issue with the Lib Dem share!