Sunday, December 21, 2014

Less than limitless levity looms for Lousy Libs as yummy year-end YouGov yields a 'Yikes!'

After the false alarm on Friday, it looks like this morning's Britain-wide YouGov poll really is the last from the firm for this year.  It's the Winter Solstice today, and the Lib Dems might just be wishing the darkness could swallow them up - once again they find themselves 2% behind the Greens, and just 1% ahead of the SNP and Plaid Cymru.  And yet the broadcasters still haven't announced a change to their initial proposal to exclude the SNP, Plaid and the Greens from the leaders' debates.  Just how much longer can this nonsense go on?

Britain-wide voting intentions (YouGov, 18th-19th December) :

Labour 34% (-1)
Conservatives 32% (+2)
UKIP 15% (-1)
Greens 8% (n/c)
Liberal Democrats 6% (n/c)
SNP/Plaid Cymru 5% (+1)

Even more importantly, the Scottish subsample gives the SNP an enormous 44% to 24% lead over Labour.  So while Jim Murphy may think he's had a "good week", the polls beg to differ.  Still, it's perfectly possible that proper, full-scale Scottish polls will pick up a change in attitudes that the Scottish subsamples in GB-wide polls have thus far failed to detect, and the first test of that will come in a Survation poll that is apparently due in the Daily Record tomorrow.  The last Survation poll gave the SNP a ridiculously big lead, so if that eases down only a little it might well be due to normal sampling variation.  A big drop might be more significant (unless there's another methodological change, of course).

I'll hold off for Survation, and then calculate the final Poll of Polls update of this tumultuous year.


  1. Do you think Ed Miliband is doing well or badly as leader of the Labour party?

    12% Well
    80% Badly

    Nick will be annoyed (17% W / 76% B); Ed's beating him here.

  2. SK, Ed has demanded a recount. Cannot be true? :)

  3. I think there's another YouGov (GB-wide) tonight/tomorrow.

    Opinium have SNP on 5% across GB. Presumably they would have the SNP well ahead in a Scottish sub-sample, but they don't show regional / national splits in their tables.

  4. For Opinium, Scots planning to vote SNP in the final weighted numbers are 40% of the total demographically weighted Scottish sample, including DKs and WNVs.

    Total DK+WNV for the UK is 24%.

    If this was 20% in Scotland, it would therefore be 50% SNP north of the border.

    15% would be 47% SNP...

    So, mid to high 40's at least in Opinium.

    1. Scottish Skier, please forgive my ignorance. When you've been saying recently that some of the SNP downweighting has been the heaviest ever, what is the justification for this from the polling companies concerned?

    2. It's not that the pollsters specifically need a justification as its not intentional, more it is a product of their methodology.

      They are weighting to varying extents by 2010, be that in VI or by party ID at the time. The problem is, if you ask Scots what they voted in 2010, they tell you they gave the SNP the most votes by a clear margin over Labour, i.e. SNP won the election. They even do this when you ask them what they voted in 2011 at the same time (which they tell you, correctly, that they gave the SNP 45% etc), so it’s not purely confusion / false recall.

      There was a very large tactical vote for both Labour and the Libs in 2010; up to 25% of the vote, and coming from both SNP supporters and leaners. When these people are asked what they did, they tend to lie about it. After all, they tried something desperately and it failed, with their vote either letting the Tories in (a vote for Labour), or they actually voted Tory (by voting Lib).

      Anyway, the pollsters see one or two % too many SNP for the UK in 2010 and they down-weight to match the 2010 result. Seems trivial as its only a couple of % UK-wide. so hey ho, nothing to see here. Makes all the difference in Scotland though!

      If they are down-weighting progressively more and more – and this is happening in all polls that weight to 2010 to varying extents (in Yougov and Populus it’s very clear / they provide this data) – it might imply that they are getting an increasing number of respondents saying they voted SNP in 2010 when they voted Labour or Lib and these people are saying they plan SNP, as they did vote in 2011. They thus get down-weighted.

      We can’t be sure how much this is affecting polls, but it does explain why SNP are lower in UK subsets than they are in Scotland-wide polls. In Scotland wide polls, only 2011 weighing is normally used, 2010 proving erroneous. In MORI polls, no past vote weighting is used so there’s no SNP down-weighing at all; it has the highest SNP share.

    3. Couldn't ask for a clearer explanation. Thanks.

  5. Let's all sing a rousing chorus of "In the bleak mid winter" in their honour.

  6. OT, but was any research done into the postal voting figures by region in the referendum?
    I was wondering if the regions with the biggest No votes also had the biggest postal vote.

  7. This has probably been asked before, so apologies in advance: Can a newspaper which hires an opinion poll company influence or tweak the findings in its publication of the results?

  8. Anonymous

    Strictly it should a no. The opinion poll companies now ask to see the articles based on the poll finding before publication although it doesn't stop the newspaper cherry picking bits from the poll to suit their political bias.

    1. e.g. last month the Record highlighted the findings in the Survation poll that there was a majority (53-47) against independence and that a minority (I forget the exact %) had "fallen out" over the referendum. They only reported the finding that the SNP were well ahead of Labour on Westminster VI (the main point of the poll!) on a second day of reporting it.

    2. Thanks to those who answered my question.

  9. News about the Survation poll has been retweeted by Kevin Pringle, so it looks like a good result for the SNP.

    1. SNP 48, Labour 24 according to Aidan Kerr on twitter.