Wednesday, March 25, 2015

Aye, aye, ICM : Holyrood poll sees SNP extend lead to 20%

The datasets from Monday's ICM poll have finally been released, and along with confirming that the "spiral of silence" adjustment did indeed reduce the SNP's lead slightly, they also provide voting intention numbers for next year's Scottish Parliament election, and for a hypothetical second independence referendum.

Scottish Parliament constituency ballot :

SNP 46% (+2)
Labour 26% (n/c)
Conservatives 13% (+1)
Liberal Democrats 6% (n/c)
UKIP 5% (-1)
Greens 4% (n/c)

Scottish Parliament regional list ballot : 


SNP 42% (n/c)
Labour 26% (+1)
Conservatives 14% (+2)
UKIP 6% (-1)
Liberal Democrats 6% (n/c)

Greens 5% (-3)

What's fascinating here is that, although we see the familiar pattern of the SNP lead being somewhat lower on the list, that isn't tied to a big list vote for the Greens.  The small number of SNP constituency voters who opt for a different party on the list are breaking for a variety of parties, with as many going to Labour as to the Greens.  That doesn't help Labour, though, because they lose almost as many votes on the list as they gain.

If there were a fresh referendum on Scottish independence tomorrow, then how would you vote?

Yes 46%
No 54%

It's difficult to make much sense of these figures, because we have no baseline to work from - this is the first time ICM have asked the independence question since referendum day.  We can't even make a direct comparison with their pre-referendum polls (which on the face of it often produced very similar numbers to these) because the methodology has changed - in line with other firms, ICM have now introduced weighting by recalled referendum vote.  This has led to people who recall voting Yes being significantly weighted down in this poll, from 469 to 423.  No voters have been upweighted, albeit by a more modest amount.  So, as we've seen from all the firms, it looks like Yes have made substantial progress from ICM's pre-referendum polls (with the exception of the rogue 54% Yes poll on the final weekend) once you take account of the new methodology.

However, the headline numbers as reported are obviously less good for Yes than we've seen in any other poll since September.  That difference may be caused by ICM's methodology, or by random sampling variation, or simply by the political composition of ICM's panel being slightly different.  To put it in perspective, there have been nine independence polls since the referendum - four have shown a Yes lead, four have shown a No lead, and there has been one tie.  It simply isn't possible to know whether Yes or No are in the lead, although we do know that it's a very tight race.

*  *  *

I think I may have to scream - Alex Massie has tried it on with his barmy "jury trial" wheeze yet again.

"As it happens, I think the BBC did scrutinise the Yes campaign’s claims with greater vigour than it did those made by the No campaign. I’m neither sure that could have been avoided, nor that it was necessarily the wrong instinct. It was the Yes campaign – and the Scottish government – that were proposing a significant change to our way of life, after all...the burden of proof was on the Yes campaign and the jury – press and voters alike – were entitled to ask that the case for independence be made beyond a reasonable doubt. That, in other words, it be subject to a criminal, not civil, standard of proof."

Memo to Mr Massie - if you want to spout this kind of nonsense, please at least attempt to back it up by identifying the portion of electoral law which states that fair and balanced TV coverage is not required or even desirable in a referendum campaign "because it's kind of like a jury trial". You may be searching for some time. As for the remark about bias being "unavoidable", that reads a bit like someone saying : Yeah, OK, your pet poodle was shot. It wasn't a pleasant thing to watch, but hey, what could the gunman do? He just felt his finger squeezing the trigger, it must have been gravity or something. Bummer.

Massie also claims that Salmond implied in his new book that Henry McLeish had betrayed his country, which - not to put too fine a point on it - is a pretty bloody outrageous distortion. Because Salmond said that McLeish had been torn between loyalty to party and country, we're invited to believe that McLeish's eventual decision to side with his party must automatically mean that he's being branded a traitor to Scotland. I suspect Massie would struggle to make that tenuous logic stand up in a court of law (how ironic).

"Scottish politics is a faith-based business these days...How else to explain the fact that, according to a recent YouGov poll, 56 percent of SNP supporters think plummeting oil prices are neither good nor bad for Scotland?"

There's a pretty straightforward answer to that question if you're remotely interested in listening, Alex - some Yes voters simply aren't as obsessed with oil as the average right-wing unionist commentator, and won't have been interested in the question. Others will have spotted a mile off that they were being fed a leading question, and weren't prepared to play along with that little game.

"Of course Salmond was so convinced Yes were going to win that it comes as some surprise to discover that Scotland actually voted No. It certainly seems to shock him."

Perhaps it did, but that pales into insignificance when compared to the shock of the discovery that "undecided voter Alex Massie" - who took part in a high-profile BBC referendum debate on that specific basis - had in fact been a dyed-in-the-wool unionist all along. You could have knocked me down with a feather.

62 comments:

  1. There's a few reasons why we shouldn't dwell too much on the referendum polls;

    1) The Yes campaign no longer exists in official form. Where as the No campaign in the form of Unionist parties and MSM, continue to try to discredit independence without any defence or attack by the Yes campaign. Note, the SNP are pushing forward with the task at hand.

    2) Despite the above mentioned No campaign groups, the support is growing or staying static. What would an official campaign do now for Yes?

    3) The demographics that voted No were very much of the older generation. They are still not being reached by new forms of media and are still hearing the same messages.

    ReplyDelete
    Replies
    1. I don't think there's any particular reason not to want to dwell on the independence polls - we've had nine since September, and this is the first I would consider to be mildly disappointing. Even this one suggests progress for the Yes camp.

      Delete
  2. The Green number seems bizarre and is the lowest they have polled for well over 12 months. Given it is around 40% below the rolling average, there's going to need to be at least one more poll before this is anything other than a rogue figure.

    The Indyref outcome is still within a 3% MoE if the "correct" current figure is 49% which I would be very comfortable making the assumption it is.

    ReplyDelete
  3. " This has led to people who recall voting Yes being significantly weighted down in this poll, from 469 to 423. No voters have been upweighted, albeit by a more modest amount. So, as we've seen from all the firms, it looks like Yes have made substantial progress from ICM's pre-referendum polls "

    I don't fully understand this point. Surely weighting by referendum votes simply seeks to make weighting procedures more accurate. Isn't it possible that the downweighting in this poll reflects an unusually high number of Yes voters in the sample, but that the overall Yes/No position is more or less unchanged? Or in other words, why does downweighting per se mean there has been progress made by Yes?

    ReplyDelete
    Replies
    1. I'm not talking about progress since referendum day (although even on that measure Yes are up 1%), but progress since ICM's pre-referendum polls. If ICM were still using their old methodology, that big downweighting wouldn't have occurred, and the Yes vote would obviously be significantly higher.

      Delete
  4. Surprised at the regional voting numbersf or the Scots Greens. Seems quite a low base for them? Are they the 'new' SNP by being squeezed due to lack of media presence.

    Impressive stuff as an SNP member.

    I agree regards the Indyref question. It's non-relevant at present. The post-referendum polls have been very encouraging on the whole, even if this one is a bit more 'as you were' with 41-47.

    The support for Independence will continue to rise throughout the years, and as we see the slow ebbing of the British state. I'm convinced also that there will not be a referendum until a wide range of polls are showing at least 55% support for Indy.

    ReplyDelete
    Replies
    1. More likely 60%. At that stage would the 'no' campaign even bother? Gordon Wilson's idea of an Yes alliance economic forum, seems like the way forward to achieve this.

      Delete
    2. With optimal conditions - and with a very left wing approach - the Yes campaign reached just under 45%. Gordon Wilson has suggested reaching out to the right, to make it up to over 50%. But if you try that, you erode your support on the left.

      Maybe we should concentrate on making the UK a better place, now that the sovereign will of Scotland is known.

      Delete
    3. Snigger. What was so "optimal" about the conditions last September? Evidently you slept through the wall-to-wall propaganda from the unionist media.

      I am absolutely determined to make Britain a far better place than the right-wing nightmare it currently is. A crucial part of that renewal process is Scottish independence. I urge you to set aside tribalism and join me in this noble endeavour - let's make Britain great again.

      Delete
    4. Tories in power. Bad economy. Charismatic SNP leader. A dullard as no campaign leader.

      The no campaign may have dominated traditional forms of media but I think the one thing we all can agree on is that the yes campaign dominated the new media - facebook, twitter, blogs like this one and WoS. So the no campaign didn't entirely get a free ride.

      I can't join your cause as I believe in a united Britain. I think that unity benefits everyone. I also think we diverge strongly on what our idea of a "right wing nightmare" is. To me, it's Germany circa 1939 - not free market economics tempered by liberal democracy.

      Delete
    5. Fine - if you think what we've got at the moment is political centrism in action, you've just identified why I believe the system is broken.

      I'm struggling to see how "didn't entirely get a free ride" constitutes optimal conditions, but nice try.

      Delete
  5. A bit O/T but readers should be aware of this.

    I think it is important to clarify the 1951 Westminster results and totally refute the lies being told by Jim Murphy and Labour.

    The 1951 Westminster results were as follows:

    Labour 295,
    Conservative 293,
    Liberal National Party 19,
    Ulster Unionist Party 9,
    Liberal Party 6,
    Independent Nationalist 2, and
    Irish Labour 1

    .... a hung parliament.

    The Conservative, Liberal National and Ulster Unionist Parties formed a coalition and governed with 321 seats.

    Labour actually won more of the popular vote (13,948,883) than the governing coalition
    (Conservatives 12,384,784 , Liberal National 1,58,138,and Ulster Unionist 274,928 for a total of 13,717,850)

    Now for the cherry on top, the Labour Party gained MORE seats than the Conservative Party.

    In 1951, the party with the largest number of seats in Westminster DID NOT FORM THE GOVERNMENT!

    ReplyDelete
    Replies
    1. It has been brought to my attention that in 1951, the Scottish Unionist Party won 29 seats. I was aware of the Scottish Unionist breakdown, but for ease of understanding, I included the Scottish Unionist seats in the Conservative column.

      Without that inclusion, the seats would be as follows:

      Labour 295,
      Conservative 264,
      Scottish Unionist Party 29,
      Liberal National Party 19,
      Ulster Unionist Party 9,
      Liberal Party 6,
      Independent Nationalist 2, and
      Irish Labour 1

      The party that was the largest by over 30 seats DID NOT form the government.

      That seems strangely familiar…..;-)

      Delete
    2. The 1951 example doesn't really work, because MPs from the Scottish Unionist Party, Ulster Unionist Party and the National Liberals all accepted the Tory whip in parliament - in other words the Tory parliamentary party really was the biggest. However, the 1923 example is plenty enough to be getting on with, because it means that one out of the last four hung parliaments resulted in the second-largest party forming the government.

      Delete
  6. Today's or sorry, tomorrow's YouGov poll has
    Lab 35 Con 34 UKIP 12 Lib 8 Green 6.

    Although no one has Tweeted is that 5% SNP/PC? So, back to our more 'normal' numbers in YG after a few days of Labour closing the gap.

    ReplyDelete
    Replies
    1. For reasons known only to YouGov and possibly The Sun, we never get the SNP/Plaid figure until 6am.

      Delete
    2. SNP 34%, Labour 27% - but an even bigger downweight of SNP and Plaid identifiers than yesterday. They've almost been cut in half. It's starting to get a bit silly now.

      Delete
    3. and UKIP on a laughably high 14%.

      Pretty obvious that YouGov have changed their methodology this week to something like Opinium or what Populus did until January. The much greater down-weighting started with the poll released on Tuesday morning.

      Delete
    4. I'm not sure - the target figure looks much the same as before.

      Delete
    5. Could be YouGov attempting to make political weather rather than simply measure it? The question is - will they succeed? If a full scale Scottish poll is released in a few days showing an SNP lead in single figures and then steadily reducing after that, will it dampen expectations on the nationalist side and actually cause some of them to revert to previous voting habits or just stay at home?

      Delete
    6. The methodology is different for full-scale Scottish polls, so this downweight wouldn't happen. It may just be that more people are identifying with the SNP than before, but are being scaled down to this rigid target figure.

      Delete
    7. Anon: you're getting way ahead of yourself. It's just a strange pattern of figures in sub-samples of <200 people, which are inevitably unreliable. There's no firm reason to suppose any full-scale poll conducted this week would show any significant difference from the two conducted last week.

      JK: sure, the target figure for identifiers is the same (~2%, based on 2010 GE). But that doesn't explain the sudden and consistent rise in the number of party identifiers. It could just be a statistical fluke, but with each day it happens the likelihood of that decreases.

      Delete
    8. There are some companies showing SNP on 56% with labour down beside the tories fighting over the scraps. At least the YouGov figures have generally been realistic and don't show the combined SNP+Green score towering above the 44.7% the yes campaign achieved.

      Delete
    9. Yougov have SNP up into the 50's now as a rough calc. Before they down-weight them massively that is. So more out of line than the other pollsters.

      Delete
  7. I'm wondering where the full yougov scottish poll I completed the other day is. Did not look like an internal party poll. Fair enough questions etc, so more like for a paper such as the Times.

    ReplyDelete
  8. 48-41 in favour of "no" with "don't knows" included?

    Given what we know about "don't knows" i.e. they actually, in the end, vote "no" overwhelmingly, I'd have no qualms about rerunning the independence referendum right now - the nationalists would still lose.

    ReplyDelete
    Replies
    1. I agree with you - the British nationalists would lose. Let's do this. Right now.

      Delete
  9. We could rerun it tomorrow. We'll need your pal Brian Soutar to help us out.

    ReplyDelete
    Replies
    1. I am so, so glad that you've seen the light over your pal Cammo's "it's over for a lifetime" arrogance. There is more joy in heaven over one sinner that repenteth, and all that.

      Delete
    2. The question over whether a second referendum will be held will dog the next Scottish parliament if, as is surely the case, it turns out to have an SNP/green majority. The radical section of the SNP will want a referendum. Sturgeon will be wary of one for obvious reasons - lose twice and where do you go from there? Winning would, arguably, bring even greater problems.

      In other words, it wont be the tories or labour who veto referendum #2 - it will be the leadership on your own side.

      Delete
    3. "Winning would, arguably, bring even greater problems."

      Yes, I'm sure Nicola Sturgeon lies awake at night wondering if she really wants to achieve her number one political objective.

      You chaps really do live in a world of fantasy.

      Delete
    4. The current generation of SNP leaders joined the party when it was polling well below 20% and devolution was a far off prospect. Is it possible that at least some joined up looking for "a joab" and aren't really that bothered about independence - possibly viewing it as more trouble than it's worth?

      Delete
    5. That's a bizarre point. When they were polling badly, and there was no Scottish Parliament to provide a career, the SNP would hardly be an attractive option for someone seeking "a joab".

      Delete
    6. Local councillors, jobs within the party...

      Or maybe you start off with this big ideological commitment but then as time goes on you get cosy having lots of money and power without responsibility. If Sturgeon called and won a referendum by a slim margin in a few years' time, she would no longer be in charge of a devolved chamber - but a country, where half the people are seething and the economy / public finances are in a poor state. Who needs that?

      Delete
    7. Ah, apologies. Until you said that, we were making the mistake of taking you seriously.

      Delete
    8. A wee anecdote, from a friend of a friend (pre-referendum):

      'We were dining in a restaurant in Edinburgh when Nicola Sturgeon and some of her colleagues sat down at the table next to us. They were talking openly and brazenly about the campaign for independence, with Nicola Sturgeon declaring "I can't believe people are going for it in such huge numbers. I mean, we literally have NO policies!" '

      Might be bs - or Chinese whispers. But if true it's not exactly indicative of someone who would sacrifice everything for the cause. I reckon they had no choice but to hold a referendum as it is their very reason for being - but many would also have found it to be an unexpected inconvenience. Now that the answer is "no", the SNP have the perfect excuse NOT to have one! But they'll string the 45 along anyway, hinting that they MIGHT have one - until eventually everyone gets bored and drifts away.

      Delete
    9. It's funny you should say that, because I have a mate who overheard David Cameron in the queue at the Post Office - apparently he thinks it's hilarious that all the No voters fell for his cunning plan. Apparently he only wanted Scotland to stay in the UK so he could flog off the golf courses to the Japanese, and pay off the national debt. "I knew the Jocks were dim, but even I didn't think they were that stupid!" he was heard to chortle.

      The story might have been embellished slightly, but it certainly makes you think.

      Delete
    10. Ah - but my source is reliable! ;0)

      Delete
  10. Yougov have a bigger vote for UKIP in Scotland this morning at 14% than it is across the UK (12%). Another massive SNP(+PC) down-weighting at 47% (x0.53).

    Someone's having a laugh.

    Slightly annoying to have to now remove Yougov from my running averages it seems. SNP down-weighting I'm used to, but Scotland daan the pub with Farage?

    ReplyDelete
    Replies
    1. Oh, I see this has already been discussed. Aye, can only be a method change. And a ridiculous one at that.

      Delete
    2. I'm in agreement that UKRAP on 14% in Scotland when the national average is 12% is rather silly. But you guys are perfectly willing to accept polls and subsets that put the SNP at nearing 60% support - which is clearly bunk. If you couldn't get 45% in the indyref then I think 50%+ political support can be ruled out as statistical nonsense or "someone having a laugh" (as someone said in relation to YouGov).

      Delete
    3. There have been subsets showing the SNP at almost 60% - but not polls. The best polls for the SNP have been the two from Ipsos-Mori putting them at 52%. Would you describe those two polls as "statistical nonsense"? If so, you're braver than I am.

      Delete
    4. What's this 'you couldn't get 45%' stuff?

      Yes is not the SNP, and the SNP is not Yes. There is clearly not perfect crossover between the people who voted Yes and who say they will vote SNP, so there is no reason that some polls should not show that SNP can get more than 45% even if that's not the final outcome. You only have to look at the polling in some No areas to see that more people there will vote SNP than wanted to vote Yes.

      Delete
    5. Why would a no voter vote SNP? It makes very little sense. And even if they were inclined to do so on the morning of September 19th, 6 months of vitriol have probably put paid to any such notion.

      I firmly believe that the support for nationalist parties is effectively capped at 45%. If you assume 2% support for the Greens, that puts the SNP on an absolute upper limit of 43%. They will not score higher than this.

      Delete
    6. "Why would a no voter vote SNP?"

      You're evidently befuddled by the phenomenon, but you can't credibly deny that it's happening - the evidence is there in the polling datasets.

      Delete
    7. BBC's Prof C calculated 53/59 seats for the SNP based on 43% of the vote. I'd take that.

      Delete
  11. But that doesn't explain the sudden and consistent rise in the number of party identifiers

    Yep. Chances are Scotland is either swinging more to the SNP, or Farage is setting the heather on fire. That or Kellner's finally lost it.

    ReplyDelete
    Replies
    1. Or maybe the politically illiterate, anti establishment protest vote in Scotland is simply swinging wildly from SNP to UKIP to Green to UKIP...

      Delete
    2. Brilliant spoof post, Anon. More please, and don't forget to put your name to it!

      Delete
  12. SNP identifiers has rocketed over the course of the week. Now something 40%+ of the Scottish sample, which would mean massive SNP shares when you exclude DK (without massive down-weighting).

    Short of a some bizarre method change, could be all the anti-Scottish 'exclude the jocks' from Government stuff.

    ReplyDelete
  13. My recollection is that you gov ask which party you most identify with when you first register rather than before each poll. Or have I got that wrong? Seems strange that you would see a big shift in party identifiers unless there were lots of new people signing up as the panel should otherwise remain fairly constant.

    ReplyDelete
    Replies
    1. My recollection is that you gov ask which party you most identify with when you first register

      I understood this to be the case. I can't recall being asked myself, but I suppose I must have been asked. Joined back in 2011. Either way, we have seen a wholesale shift in party ID (see Comres for example) in Scotland since 2010 and notably since the iref. Therefore, using past party ID is likely to make your results very unreliable; the older the data the worse it will be. Best bet is using current / recent party ID if you are going to do that.

      All we know is the number of SNP identifiers (not specifically people saying they plan to vote SNP, but people who identify most with them as a party) appears to be rising sharply. At the same time, these are being heavily weighted down in response. This has been happening steadily in Yougov UK since the iref as SNP share of VI has increased in all polls to new highs. Past week seems to have taken a sudden upsurge.

      Delete
    2. YouGov periodically asks which party you feel closest to - it's tacked on to other surveys. I'm not sure whether they use your most recent party ID for the weightings, or stick with what you originally said. If the former, that might be a partial explanation for what's happening.

      Delete
  14. Perhaps more relevantly than the effects of party ID weighting in a sub-sample of a standard GB opinion poll, there are four real elections in Scotland today.

    26th March

    Fife UA, Glenrothes West & Kinglassie - Labour died - 8 candidates: Con, Lab, LD, SNP, UKIP, Green, 2 x Ind

    Moray UA, Buckie - Independent died - 3 candidates: Con, SNP, Ind

    Western Isles UA - Beinn na Foghla agus Uibhist a Tuath - Labour resigned - 2 candidates: SNP, Ind

    West Lothian UA, Armadale & Blackridge - SNP disqualified for non-attendance - 5 candidates: Con, Lab, SNP, Green, Ind

    Of these, the Western Isles one isn't relevant to party preference because the SNP are the only party fielding a candidate. With Moray, there are only two party candidates (SNP and Tory), and the independent seems to have some public profile (he was a contestant on Great British Bake Off apparently). The other two (Armadale and Glenrothes) might be more interesting though.

    ReplyDelete
    Replies
    1. Isabel Hutton who was the councillor in Armadale is seriously ill. My understanding is that she is suffering from a life-threatening condition and has been undergoing treatment for it.

      Delete
  15. Something really weird has happened with lost last two Yougov's.

    If you plot unweighted SNP/PC as a function of Scottish sample size you get a general trend that the higher the SNP/PC in unweighted, the higher it is in the weighted cross-break.

    Not our two most recent cross-breaks however; they are very out of line. If they fitted the general pattern, they should be at least 45%, more 50%+ SNP.

    ReplyDelete
    Replies
    1. That should be "unweighted SNP/PC as a function of Scottish sample size versus SNP share in the weighted cross-break"

      Does not seem to be random fluke noise anyway; there's big surge in SNP identifiers all of a sudden and these are being shredded by Yougov. Such a surge in a very short period of time would be very hard to explain by new entrants. You would need a sudden mass flood of them all in the space of a few days. As mass mind control hasn't been invented yet, I think this unlikely.

      Delete
    2. So just to clarify, are you suggesting that there are a number of people who until recently might have planned to vote SNP but still identified as something else, but have now begun to actually identify as SNP, thus messing up the weightings?

      Delete
    3. That could be one explanation. That has happened since the iref - SNP ID has risen sharply. Another boost in this might cause the pattern seen in Yougov. However...

      I honestly don't know as I'm not 100% sure how yougov get their ID data from. They say they asked people just after 2010 and weight to that. The idea would be there's no false recall as it was so close to the election. However, that won't work potentially for new entrants, especially in Scotland. And new entrants are essential as people do leave panels. Also, the bigger your panel the better.

      Whatever is going on, the last 2 polls definitely do not fit a trend of falling SNP support relative to previous. If they followed the pattern of past polls, they should have yielded very good cross-breaks for the SNP.

      This suggests something big happening with the panel and ID, either naturally or due to Yougov doing something.

      Or maybe someone made a spreadsheet error and everything will go back to normal!

      Delete
  16. Sorry for going off on a tangent, but I wondered if anyone ever has problems posting here with particular browsers? My posts were just disappearing for a bit and I normally use Firefox. Wanted to post today and so tried IE, which worked. But I would prefer to use Firefox. Can any whizzes tell me if there is anything I can try to hopefully get posting through my default browser again? Once again, I apologise James....I realise this is way O/T.

    ReplyDelete
    Replies
    1. I've had disappearing posts using Firefox too. Not tried any other browsers.

      Delete
  17. I've had a Poll of Polls going for months which has the YouGov samples as it's base due to their frequency.
    The last 3 have seen SNP support nose-dive.
    37-31 34-32 34-27
    If this is due to new methodology it's fcuked my poll right up!

    ReplyDelete