Wednesday, November 12, 2014

SNP soar to a Britain-wide vote of 8% in astounding Ipsos-Mori poll

This afternoon has seen the publication of one of the most remarkable Britain-wide polls for many a year.  Ipsos-Mori are now showing the SNP on an 8% share of the vote - in spite of the fact that Scotland comprises just 9% of the population of Great Britain!  Strangely, the SNP are 'only' on 59% in the Scottish subsample - the apparent anomaly comes about because respondents in Scotland are more likely to say they are certain to vote.  And it's perfectly possible that phenomenon will carry through (at least to some extent) to the election itself, due to the surge in public engagement caused by the referendum.

The Liberal Democrats are just 1% ahead of the SNP across Britain - and that's in spite of the fact that their 9% share is a touch higher than their recent average.

Britain-wide voting intentions (Ipsos-Mori) :

Conservatives 32%
Labour 29%
UKIP 14%
Liberal Democrats 9%
SNP 8%
Greens 7%
Plaid Cymru 1%

The combined share of 61% for the two 'major' parties is desperately low by historical standards, and with the Conservatives now holding a slight lead (which neutralises Labour's in-built advantage from the constituency boundaries), this could be another welcome sign that the UK political system is becoming more like India's - ie. hung parliaments and coalitions being the norm without proportional representation even having been introduced.

*  *  *

SCOT GOES POP POLL OF POLLS

This will be the last update of the Poll of Polls before the Panelbase poll drops out of the sample, so unless a new full-scale Scottish poll comes along soon we'll revert to being reliant on subsamples from GB-wide polls only.  The Panelbase numbers make up exactly half the sample, with the other half drawn from ten subsamples - four from YouGov, two from Populus, one from Ipsos-Mori, one from Ashcroft, one from ICM and one from Survation.

Scottish voting intentions for the May 2015 UK general election :

SNP 43.6% (-1.6)
Labour 26.1% (-0.9)
Conservatives 15.9% (+0.3)
UKIP 6.3% (+0.4)
Liberal Democrats 4.7% (+0.4)
Greens 2.4% (+0.4)

(The Poll of Polls uses the Scottish subsamples from all GB-wide polls that have been conducted entirely within the last seven days and for which datasets have been provided, and also all full-scale Scottish polls that have been conducted at least partly within the last seven days. Full-scale polls are given ten times the weighting of subsamples.)

31 comments:

  1. Stunning stuff. People have been claiming that Labour are dying in Scotland. This seems to be confirming their view. Hell mend them.

    ReplyDelete
  2. Maybe it is time devolved and regional parties stood up for their electorate in Westminster. By my rough calculations, the South East/London only represents 25% of the population. And that assumes that 100% of them are content for London/Financier-centric parties to rule the UK roost.

    ReplyDelete
  3. Going back to the previous article about the SNP's right to be heard in the UK TV debates: I think its now totally untenable not to allow the SNP a place wherever the LibDems have one - 1% difference in the polls (as that's one of the things the BBC say they take into account) is not enough to deny them a place.

    ReplyDelete
  4. Unbelievable. Really strong figures for the Greens too. Just an idle thought - does anyone know who actually does Ashcroft's polling? Are they British Polling Council members? I presume it's telephone based. Surely he hasn't gone and set up a call centre from scratch?

    ReplyDelete
    Replies
    1. As I understand it, it is Populus, but done by telephone. Also, the Ashcroft method uses 2010 vote recall weighting instead of the 2010 party ID method populus employ in their own, online polls.

      MORI, as per the 8% SNP poll, use landline telephone with no past vote weighting.

      2010 past vote weighting or 2010 ID weighting hurts the SNP share.

      Delete
  5. We need to plan responses for the eventuality that we don't get on the debates. Giant free Yestivals on the nights of the debates, protests outside the studios, even civil disobedience.

    Sounds childish, but if we're excluded we must do everything to make those nights about US and not THEM. And so well - planned and large scale that the MSM simply can't ignore it.

    ReplyDelete
  6. Even if Labour were to perform significantly better in Scotland than current polls suggest, I'm finding it increasingly difficult to envisage any scenario where the SNP wouldn't at least quadruple their current representation in Westminster.

    ReplyDelete
  7. I have a feeling the fact IPSOS-MORI are using telephone polls means they are picking up higher numbers of SNP voters.

    I think people who will vote SNP will be much more likely at this time to want to respond to a random phone call asking about politics; they are enjoying the current political debate the SNP are riding high, Labour in a mess and despite losing the referendum it's looking somewhat positive for even further political divergence between Scotland and rUK.
    So this will mean that the poll has a self-selection problem of people who are willing to answer are more likely to be SNP voters.
    Given that MORI only weight by demographics this won't be corrected for Labour/Tory i.d. voters who won't respond at this time are not accounted for by party affiliation weighting.

    Still though, even if this is the case the SNP looks in a very very good position. Roll on May 2015

    ReplyDelete
    Replies
    1. There may be at least something in that. In my experiences campaigning for Better Together, Yes supporters tended to be much more eager to engage in political discussions than No or Undecideds. It's perfectly possible that you might see a similar dynamic going on with SNP supporters right now. Though that obviously couldn't entirely explain the huge leads that the SNP currently enjoy, so whatever the case I think we expect a very strong showing from them whatever the case.

      Delete
    2. If so, then an explanation as to why MORI showed the lowest Yes and SNP shares for pretty much the entire iref campaign. Only in the last few weeks did MORI catch up with other (mainly online) pollsters and never did it get Yes ahead. What happened there in this theory?

      The fact that MORI doesn't weight by 2010 vote as other UK-wide pollsters do is likely an advantage. The BBC's prof C has come down heavily on this for well-established reasons.

      Ask 1000 Scots how they voted in 2011 and they remember well. Ask them in the same poll about 2010 - even reminding them that was when DC came to power / the Tories returned - and apparently, the SNP won comfortably... Really hurts SNP VI in such polls due to down-weighting to match results.

      Anyway, that quirk shows a whole lot of regret about tactically voting Labour and the Libs in 2010; this of course being made up for in 2011.

      Delete
    3. That's a good point about MORI, Skier. Though as I said, even if there was something in it it couldn't explain the huge leads that the SNP currently enjoy.

      Though I'm quite taken aback that people recall their 2010 Westminster vote as being an SNP win. I'm trying to rack my brains around the implications of that. Any ideas?

      Also, what does SNP VI mean? I'm not quite familiar with that piece of polling terminology.

      Delete
    4. Voting intention, presumably.

      Delete
    5. @stoat
      Have a look at e.g. recent tables from panelbase where respondents were asked about how they voted in 2010 and 2011 in the same poll. The unweighted sample (which is targeted to be demographic although is of course never quite so needs a tad of weighting) gives a 2011 vote very close to the result. However, the same respondents say they gave the SNP a win in 2010 by a clear margin. This is why 2010 weighting hurts SNP - massive downweighting follows as pollsters up Labour + Lib and down-weight SNP.

      False recall / confusion with 2011 is possible. However, when respondents are asked about 2010 and 2011 in the same poll this becomes less plausible. When 'David Cameron coming to power' is added to the 2010 question for clarity and it still happens (while noting the 2011 result was a reaction to 2010), simple false recall becomes even less plausible.

      When we look back at polls ahead of 2010 another possibility emerges... SNP came close briefly to Labour in Westminster VI in 2009, taking from both them and the Libs. Nowhere near like now, but a much closer potential result than came about. This was because a large section of SNP ended up voting tactically for Labour and Lib to stop the Tories 'one last time' (maybe). This of course failed and if you e.g. voted Lib, you actually voted Tory. I suspect people are blocking this stupid action (desperate tactical vote) out a bit and actually saying who they supported or strongly were leaning towards (SNP) for 2010 rather than who they ended up voting for (Labour + Lib), which is who they then voted for in 2011.

      Anyway, we can't be sure. However, I find simple forgetfulness difficult to believe. Instead, another explanation is needed.

      Delete
  8. Out of interest James, do you average the yougovs to get a single figure for them before adding that to the pot? Otherwise of course you are giving bias to Yougov's methodology.

    ReplyDelete
    Replies
    1. No, I give each subsample equal weight. It's not really feasible to give equal weight to each firm, because sometimes all there is to go on is four YouGovs and one Populus.

      Delete
    2. Fair enough; we could spend all day deciding what is the best way to aggregate subsamples! Readers should note however that your PoP will tend to be biased towards Yougov who are giving lower SNP values compared to other pollsters like MORI, survation and Ashcroft.

      Delete
  9. My oldest friend has worked for years at Ipsos Mori in Leith. Solid Yesser, but all summer and autumn he kept saying the polls were overstating Yes by 3-5% for the reason mentioned above, that Yes voters were simply too keen to stay on the line when Mori called, compared to Noes.

    My friend drilled this into me so often and convincingly that with the exception of that Saturday before the vote, I never expected us to win.

    I can imagine that this polling issue has only been exacerbated since then.

    Still, happy days indeed when outliers have us at 59%!

    ReplyDelete
    Replies
    1. I'm very dubious about the idea that Ipsos-Mori of all firms were overestimating Yes - until the sudden convergence that happened very late on, they were consistently one of the most No-friendly pollsters, and had Yes as low as 36% in early March. If your friend's theory is right, that would mean Yes were actually trailing by as much as 69-31, which seems unlikely. The final Ipsos-Mori poll overestimated Yes by 2%, and I'd be more inclined to put that down to a combination of late swing and differential turnout, rather than genuine error.

      Delete
    2. His point was that the polls in general were overestimating Yes, for the reason given above. Pollsters including Mori were weighting to account for Yesser willingness to stay on the line (ie not just slam the phone down) but not by enough in his opinion. And sadly he turned out to be dead right.

      There was nothing paticularly controversial about his theory. It was just Shy No Syndrome experienced at first hand.

      As I'm sure you know, the vast majority of Mori's work pre-indyref was unpublished work on behalf of BT asking loaded questions about pensions etc. In other words Mori became a wing of the BT campaign. This made my friend suspicious about the mis-weighting of the polls.

      However, as I keep saying, the current polls are a distraction IMO (no offence intended to your work here). The key question for next year that hardly anyone is discussing is how do we keep the Yessers among the missing million engaged and voting.

      Just terrible strategy to ignore those hundreds of thousands and return to our pre-indyref politico goldfish bowl of obsession g over the SNP's latest %.

      Delete
    3. "His point was that the polls in general were overestimating Yes"

      But that's exactly my objection - there wasn't a picture painted by "the polls in general", because the Yes-friendly and No-friendly pollsters were light-years apart prior to the very late convergence. Were Ipsos-Mori really overestimating Yes when they showed 64% No, 36% Yes in early March? I very much doubt it. I think they were underestimating Yes, and possibly by quite a substantial amount.

      Incidentally, if Ipsos-Mori really were applying some kind of specific weighting to counter what might be called the "exhibitionism" of Yes voters, that would be completely new information - to the best of my knowledge they were only weighting to demographic targets.

      Ipsos-Mori were of course the only phone pollster during the campaign anyway (except at the very end when ICM and Survation jumped in). TNS used face-to-face fieldwork, and all the others were online.

      Delete
    4. MORI average OCt 2012-March 2014:

      65% No
      35% Yes

      May 2014
      40% Yes
      60% No

      August
      42% Yes
      58% No


      Man those Yesser's couldnae just put the phone doon.

      I used to dread the latest MORI poll. Blair McDougall would be tweeting how how he was 'looking forward to it' hours in advance.

      Delete
    5. Maybe I'm not explaining this properly.

      Firstly I was only speaking to him about this through the summer and autumn so can't comment on March polls.

      As I remember it his thinking went like this: The poll of polls currently says Yes is at X%. Yessers are too keen to stay on the line when I call, so I believe the true poll of polls figure for Yes is X minus 3-5%. [How he arrived at that 3-5% figure I have no idea, but presumably he'd built up his own private methodology over ten years at Mori. Remember that getting people to stay on the line is the biggest issue Mori pollsters face, so he's had plenty experience with this].

      I can't remember if he actually used the word "weighting" but he certainly said something equivalent to "Mori are accounting for the enthusiasm of Yessers to answer calls, but NOT BY ENOUGH." He said that repeatedly.

      I've known this guy since 1989 (in fact I slept on his extremely uncomfortable floor the week of the referendum) so I see no reason to disbelieve him. And he was so adamant about this that I would return from manning the Yes stall in Portobello fired up by everyone else's enthusiasm, and then within minutes of chatting to him would be back down to earth with a bump. It then became tough not transmitting that pessimism to other Yessers.

      We have two different ways of accounting for the disparity between the final polls and the eventual 45/55 vote. There's my friend's and there are those others suggested here and elsewhere. I'm happy to stick with the guy who told me in no uncertain terms that the final vote would be in the 43/57 to 45/55 range, and therefore saved me from the colossal disappointment that other people suffered.

      He's not a blogger so I doubt I can get him to come here and explain all this more fully, plus he also has his anonymity to protect, but I'll see if he has anything important to add to the above and whether any of the above is relevant to the current polls.

      Btw do you believe the Yessers among the missing million will vote for indy parties next May? As yet I'm not convinced, and this, I believe, is still the strongest argument for the Yes Alliance. What a waste it would be to fritter away those hundreds of thousands of votes because we were too focused on the old party political way of doing things.

      https://www.facebook.com/groups/yesallianceaction

      Delete
    6. Fair enough, but MORI very much underestimated Yes a month before the vote.

      No reason not to trust your mate's sincerity about his own theories, but MORI got it very wrong - in general ridiculously so - until the last couple of weeks.

      Your mate might work for MORI, but that doesn't make him, nor MORI, a polling genius. That's not how it works.

      The online pollsters, bar yougov, called it right much earlier. MORI were way off and predicted a huge No throughout the campaign. End of story.

      They're likely out on the stupid SNP VIs they are showing too, being on the upper end of the scale. Yougov and populus on the lower.

      Delete
    7. Remember, his point wasn't specifically a Mori one. He was saying the PoPs were out by 3-5%. This isn't an issue about Mori's accuracy compared to other pollsters, but about pollsters *in general* overestimating Yes due to overenthusiasm. It's a very simple point, and a plausible explanation for the discrepancy between those final 48/52 polls and the eventual result.

      This isn't "concern trolling", by the way, just in case any of our Indy Inspector Clouseaus are getting excited. I'm not even bothered if anybody agrees with my mate. In fact I'd be happier if he were wrong, because then I wouldn't have to keep subtracting 3% from the SNP's showings. I just thought a Mori employee's view might be of interest here.

      Delete
    8. Didn't a large numbers of yes voters not make up their mind in the final month, especially after second TV debate, which was shown in the narrowing of polls a with a month to go, as number of don't knows started to decrease in these polls.

      Delete
    9. This isn't "concern trolling", by the way.

      Ok.

      Final polls were 2% off (average 47% Yes).

      I did an analysis - the 'Skier correction' - and this should be allocated 1% to Con, 0.5% to lib, 0.25% to UKIP and 0.25% to Labour. My theory for this is that the former three are more enthusiastic about the union than Labour voters based on polling evidence.

      Delete
  10. UK-wide, it now looks quite probable that the Tories will win on votes, but Labour will win on seats. Sadly, any pressure this might cause for PR will be suppressed by a Labour government which is only in office because of the unfairness of FPTP.

    ReplyDelete
    Replies
    1. Right now maybe. There's a long time to go yet.
      It seems obvious to me that the closer we get to election day, the more will swing back to the safety of 'better the devil you know' and the Tories.

      Delete
  11. Here,s an idea, Why don,t the SNP put up a candidate in the Corby constituency in the 2015 GE ?

    ReplyDelete
  12. Keatons, it looks likely, but still a long way to go.....I still think these UKIP voters will go back to their original parties, there are more Tories who have gone UKIP than Labour, so odds are the Tories will have much more stronger numbers than Labour, even with the boundary changes.

    I think we'll be looking at 4 or 5 UKIP MP's, maybe taking one in the North of England, Farage seems to be appealing to tired Labour voters, with his pledge to work with Miliband so UKIP might sneak one in up there? In other areas, i.e. south, I think the UKIP support will go back to the Tories again, based on farage miliband pledge. They won't like that down there and Cameron seems to be turning slowly purple.

    So the chances of the SNP holding power with 20+ seats is one we can be confident of campaigning on going into the election. It's funny, we had a once in a lifetime chance in the ref, now we have a once in a lifetime chance to have a real say in what government we have.

    Once the Smith Commission is out of the way we'll see what the polls are saying, but I'd imagine them getting worse for Labour and better for the Tories in Scotland.

    ReplyDelete
  13. Tories will NEVER get a voice in Scotland...they are not trusted and never will be. Once the Smith commission delivers, it will open up a huge can of worms.......just watch.

    ReplyDelete