Monday, March 23, 2015

Ice-cool ICM poll indicates impressive 16% lead for SNP

This morning has seen the publication of only the second full-scale Scottish poll since the referendum from ICM - regarded by some as the UK's "gold standard" polling organisation.

Scottish voting intentions for the May 2015 UK general election (ICM, 13th -19th March) :

SNP 43% (n/c)
Labour 27% (+1)
Conservatives 14% (+1)
UKIP 7% (n/c)
Liberal Democrats 6% (n/c)
Greens 3% (-1)

The percentage changes are from the previous ICM poll, which was (weirdly) published on Boxing Day.  So ICM are showing exactly the same pattern as YouGov, Ipsos-Mori and Survation - ie. no long-term change at all beyond random margin of error fluctuations, which reinforces the very strong impression that the SNP's enormous lead has remained absolutely stable since roughly mid-October. 
Where the firms do disagree is on the exact extent of that lead, and oddly enough ICM - who were one of the more Yes-friendly firms during the referendum - seem to have emerged as one of the less favourable firms for the SNP in the general election campaign.  The 16% gap they are suggesting is only just over half as big as reported by the last Ipsos-Mori poll in January.  It's also 3% and 5% smaller than reported by the most recent YouGov and Survation polls respectively.  (As is noted in the comments section below, some of this difference can probably be explained by ICM's highly questionable "spiral of silence" adjustment, which artificially allocates some undecided respondents to the party they voted for in 2010, thus boosting the reported vote for both Labour and the Liberal Democrats.  It may not be immediately obvious that the Lib Dems have benefited, but in fact 6% is better for them than other firms have suggested.)

However, as with the December poll from ICM, the Guardian have pointed out that, beneath the surface, the results are every bit as unforgiving for Labour as anything we have seen from any other firm.  The SNP surge appears to be much more dramatic in Labour heartland seats, and less so in seats where a smaller swing is required due to a split unionist vote.  On the face of it, this would translate to exactly the same type of apocalyptic wipeout for Labour as suggested by Ipsos-Mori, or by the Ashcroft constituency polling.  However, the breakdown by 'type of seat' is based on small sample sizes, and is therefore not hugely reliable.

As an aside, I must point out how intensely irritating it is that the Guardian's report on the poll has done what so many other media reports have been doing for weeks, and distorted the SNP's latest comment on the potential for a deal with Labour in order to paint it as something dramatic and new.  The claim is that Alex Salmond ruled out anything beyond an informal vote-by-vote arrangement at the weekend, whereas in fact he did the complete opposite - he explicitly told the Andrew Marr show that a formal confidence-and-supply deal was "possible".  He added that vote-by-vote was "probable", but that's exactly the same formulation we've been hearing for weeks.

*  *  *

UPDATE : Today's Britain-wide Ashcroft poll has the SNP on an exceptionally high 6% of the vote - ahead of the Greens, and just 2% behind the Liberal Democrats.  The Scottish subsample figures are : SNP 56%, Labour 20%, Conservatives 17%, Liberal Democrats 4%, Greens 1%, UKIP 1%.

38 comments:

  1. Interesting that ICM shows around 5% higher support for UKIP than the other polling companies.

    ReplyDelete
  2. Best bit is when it is reduced down to regions the SNP swing is biggest where Labour are Strongest. Seats are predicted at 53 SNP 2Labour 1Tory and 3 Lib Dem

    ReplyDelete
  3. Lib dems 6%

    LOL

    ReplyDelete
  4. As Calum mentioned, UKIP seem to poll a lot higher than other Scots only polls. Of course not all UKIP's support in Scotland is from rUK residents, but it'll be interesting to see the numbers when released - if rUK reisdents are over poportional.

    Still, pretty impressive. The SNP vote is holding up.

    ReplyDelete
    Replies
    1. I somehow doubt the last 'scottish' tory Mundell will be delighted to see the kippers at that level since it is blatantly obvious they will be taking votes off the tories far more than anyone else.

      'Mon the Pandas! :-D

      Delete
    2. We are going to need a Red Panda count soon.
      I know there are 2 on the Highland Park near Kingussie. Does anyone know how many are there across Scotland.

      More Red Pandas than Labour MPs - that would be good.

      Delete
    3. I make it 2 Pandas and 4 Red Pandas.

      Now that is a target to aim for!

      http://www.highlandwildlifepark.org.uk/global-news/2014/07/double-celebrations-for-red-panda

      Impossible ?

      Not with Murphy and Mcternan in charge.

      Delete
  5. The main thing about this poll is that UKIP/Greens/LibDem have 16% between them, whereas other recent polls have had them scrabbling for the last 10%. As a result the major parties' shares are proportionally a little lower. Is there any reason for this I wonder. A difference of methodology maybe? So I'd interpret this as no real change, everything looks rock solid. The interesting question is what will happen when the campaign finally hots up, given especially the hosts of new SNP activists?

    ReplyDelete
    Replies
    1. ICM usually have a "spiral of silence" adjustment, which aggressively reallocates people who say they are undecided. This usually helps parties that are polling worse than the last election and hurts parties that are polling better. That would explain Labour and Lib Dems being a bit higher than in other polls and the SNP being a bit lower.

      It doesn't explain why they have UKIP higher though. I think that ICM are just plain wrong about that, with the effect that they are probably over-rating UKIP by ~3 points and under-rating the Tories by a similar amount.

      Delete
    2. Composition of ICM's panel could be UKIP friendly for some reason.

      Delete
  6. Populus subsample:

    47% SNP
    19% Con
    18% Lab
    6% Lib
    6% UKIP
    2% Green

    ReplyDelete
  7. Populus sub-sample: SNP 47, Con 19, Lab 18, Others <6.

    Broadly in line with their sub-sample results since their methodological change in early February. Populus sub-sample results are now slightly more favourable to the SNP than YouGov.

    I think it's somewhat interesting that there have been a few sub-samples lately that have put Labour behind the Tories. Certainly more of those than the odd ones that have put Labour ahead of the SNP (just Opinium really).

    http://www.populus.co.uk/wp-content/uploads/OmOnline_Vote_23-03-2015_BPC.pdf

    ReplyDelete
    Replies
    1. Aw man. Only just got there before you!

      :-)

      You can have Ashcroft later.

      Delete
    2. On the Lab / Con thing.

      I've noticed an apparently swing from Lab to Con in Scottish subsamples recently. Not something that's evident in full Scottish polls. Well, not that I can see.

      Delete
    3. Beginning to like this tactical voting thing Labour are on about.

      Looks like the Tories may overtake them - how ironic!

      LOL!

      Delete
    4. Seems like the mantra vote SNP get tories has put the tories in labour voters mind. The unintended consequences or Murphy's law, take your pick. Lol

      Delete
  8. I guess if you're a Tory ueber-Unionist who was considering voting Lab to keep the SNP out, all this talk of SNP-Lab deals might be a bit off-putting?

    ReplyDelete
  9. I can imagine a large rump of Conservative supporters seeing through the tactical voting flimflam and sticking resolutely with their own party if they can envisage parity with or even a lead over Labour on the horizon.

    Notwithstanding the above statement I am highly cynical about the real world accuracy of these subsamples regarding Con:Lab. Encouraged, mind you, but still cynical! ;)

    ReplyDelete
  10. My dream scenario is more Scottish LibDem MPs than Tory or Labour MPs. Of course with the SNP at over 50 MPs. It appears that Professor Curtice is now claiming that this is entirely possible. Now that would be a night worthy of a massive celebration. Come on May 7.

    ReplyDelete
  11. Ashcroft sub-sample: SNP 56, Lab 20, Con 17, Others <4.

    http://lordashcroftpolls.com/wp-content/uploads/2015/03/ANP-150323Y-Full-data-tables.pdf

    ReplyDelete
  12. Something that was missed almost entirely from the Ashcroft numbers in February is that Scottish Labour are looking down the barrel of at least 10 lost deposits. In 2010 their worst showing was 10.2% but that will be a fantasy number for a lot of their candidates this time around.

    ReplyDelete
    Replies
    1. Is that really very likely, and any inkling of which seats in particular? I'm not suggesting it won't happen, but for example Election Forecast has Labour's lowest vote share at 9% in Banff & Buchan and Berwickshire (although they also show them at 9% in Na h-Eileanan an Iar vs SNP at 84%, which absolutely will not happen, and points to flaws in a lot of their calculations). The only way I can see them getting into lost deposit territory is if the whole #SNPout tactical vote nonsense gains traction and they switch to Tory/LibDem.

      Delete
    2. That doesn't seem very likely. The only Ashcroft poll where Labour were under 10% was Ross, Skye & Lochaber (9%) and even there they were on 15% on the standard VI question.

      The point Curtice was making earlier is that Labour are doing worse in the seats they hold. The corollary of this is that they are doing relatively better in the seats they don't.

      Delete
    3. Better in this case is a VERY relative term. They didn't have as many votes to lose in those seats in the first place. Remember Ashcroft polled a relatively small number of constituencies.

      Delete
    4. He's polled nearly half, and six of the ones he didn't poll are already SNP anyway, taking it to well over half.

      Delete
  13. I agree to a point but looking at the modelling from various other sites there does seem to be some real risk potential for lost deposits. Electoral Calculus repeatedly offers up that hope for one, but they do model on a uniform swing basis. However if the Tory core stays loyal and there is any tactical placement by fearful Nat-loathing Labour voters where their own party has no hope at all then I can see a few sub-5% performances on the horizon. Labour discipline in the rural fringes is lax to non-existent.

    Anyway, let's not waste time on this issue here and now; we can revisit on May the 8th and compare notes.

    ReplyDelete
    Replies
    1. Labour would happily lose 10 deposits if it meant them doing a bit better in the seats where they have a realistic chance of winning.

      Delete
  14. If Labour hold 20 seats MSM will call it a triumph

    ReplyDelete
    Replies
    1. In my opinion a win is a win. If SNP get more votes and seats than the second party that's more than good enough to claim the authority to speak up for Scotland in London.

      Delete
  15. I'm going to go with this as my prediction:

    SNP: 35-45 (~40)
    Labour: 10-20 (~15)
    Lib Dem: 1-3 (~2)
    Conservative: 1-3 (~2)

    ReplyDelete
    Replies
    1. Headlines on May 8th.

      Conservative Party, Double their Seats in Scotland.



      Delete
    2. 'SNP Struggle to hold off Resurgent Labour'

      Delete
    3. "Lib-Dem breakthrough in Mainland Scotland"

      Delete
  16. Does anyone have thoughts on why media commentary seems to keep suggesting 30-40 SNP seats even as the "worst" polls are suggesting 50+? I'd have thought pushing that number might be hysterical enough to get the stray unionists voting Labour at least in the likes of Glasgow and Kirkcaldy and could give them an blow for Salmond (not Sturgeon obviously, since she's just some wee lassie)" line if the SNP don't quite storm the nation like that. I just don't see why they keep estimating downwards...

    ReplyDelete
    Replies
    1. Most of the forecast models (e.g. Election Forecast) have SNP at around 40 seats because they assume a certain amount of movement in vote shares back towards what the parties got in 2010. The bookies odds also imply SNP winning just over 40 seats, both in the spread betting and individual constituency markets.

      Delete
  17. Very nice anyalsis of Murphy in the National, James. ;)

    ReplyDelete
  18. Where is the link to the tables?

    ReplyDelete