Monday, May 19, 2014

ICM poll : SNP on course to deny UKIP a seat in the European Parliament

The Scotsman have published further details from their ICM poll, and there must be a fair chance that this will be the final full-scale poll of Scottish voting intentions for the European Parliament prior to the election on Thursday.

SNP 36% (-1)
Labour 27% (-1)
Conservatives 13% (+2)
UKIP 9% (-1)
Liberal Democrats 7% (n/c)
Greens 7% (+3)

That would translate into three seats for the SNP, two for Labour and one for the Tories - effectively meaning that the Lib Dems would lose their only seat to the SNP, who currently have two.  Under this scenario, it would be the SNP who take the last of the six seats and thus perform the almost mythically heroic role of "the party that stopped UKIP".  There are other scenarios in which it might be Labour or the Tories who assume that role, but barring something miraculous it won't be the Greens or the Lib Dems.  So a "tactical vote" for the Greens in the hope of stopping UKIP still looks utterly counter-productive.

In any case, the number one priority for any SNP supporter must surely be to ensure the SNP win the popular vote, because failure to do so might drain momentum away from the pro-independence campaign at a critical stage.  The party may look assured of top place on these numbers, but that could yet prove deceptive - YouGov are showing a much less clear-cut position, and previous European elections have often confounded predictions due to very low turnouts.

It's fascinating that the SNP's vote is virtually identical to last month's in spite of the change in referendum voting intentions.  There are two ways of looking at that - on the one hand it means that we can't put the referendum figures down to there being too few SNP supporters in the sample, but on the other hand it underlines that the referendum race is becoming increasingly decoupled from regular voting intention.  That suggests there's no automatic reason to be suspicious of a Yes-friendly referendum poll just because the SNP are doing particularly well in supplementary questions.  That said, it would still be a cause for some concern if the SNP's performance on Thursday doesn't live up to the forecasts of the broadly Yes-friendly pollsters.


  1. I'm very keen to see the ICM tables and what they did to methodology.

    What I found odd is that those believing independence would be bad economically shot up. This had been steadily falling under a barrage of business leaving etc scare stories yet suddenly rises for no apparent reason?

    The level was too high to begin with based on long term SSAS values which is in line with ICM being not demographically representative (e.g. CoB data).

    As we know, ICM seem to wish / feel that Yes support should be lower, which is a very worrying thing for a pollster to be saying. Pollsters are there to try to measure public opinion; not put forward their own.

  2. Daily Record have realeased new Survation polling for the Euros too:

    LAB - 26% (- 3%)
    CON - 13% (NC)
    SNP - 37% (- 2%)
    LD - 6% (+ 1%)
    UKIP - 10% (+ 3%)
    Green - 6% (+ 3%)
    BNP - 1% (NC)
    Another Party - 1% (NC)

    Eerily similar to ICMs I see.

  3. Well, ICM not demographically representative again.

    75(-6.5)% Born in Scotland
    15.9(+5.9)% Born in England

    Interestingly ladies, labour voters and DKs some of the least comfortable about being asked what they plan to vote.

    High levels for an online pollster; 20% of women uncomfortable being asked.

  4. That should have been:

    15.5(+5.9)% Born in England

  5. Correct that, it's even worse. I made the mistake of looking at unweighted base. Post weighting the base is even more unrepresentative.

    75.0(-6.5)% Born in Scotland
    16.4(+6.8)% Born in England

    If ICM asked the comfortableness question ahead of the intention question then we really need to write off the poll. The order in tables suggests this was the case.

    You can't make people start to question how comfortable they are being asked about a sensitive issue them proceed to ask them about that issue in a poll and not expect the result to be affected. You've already got them thinking about motivations for them being asking.

    Oh and quelle surprise, the Yes reduces most from last month in those who are most uncomfortable giving a view on indy. So, looks like ICM just made lots of shy Yes in this poll.

    I also note ICM are saying they're getting a political bias in their sample after demographic weighting. They're having a TNS issue here and up-weighting the DNVs massively so creating imaginary views.

    This poll is very interesting, but in terms of Y/N, it's for the bin.

  6. They did ask the comfortableness ahead of the indy Q.

    Poll's therefore no good. Leave it out of your poll of polls James.

    Any Q's before the indy one should not have anything that might make people answer differently to indy.

    Funny how prof C doesn't mention this yet was all over the SNP one last year where they had some simple questions which might lead you more to Yes.

    ICM looked for shy No and found shy Yes. I suspect this bit of the poll will therefore be kept quiet.

  7. James; S-S;

    Any pointers for what to look for in terms of the result on Thursday to help us 'calibrate' the pollsters ?

    What I'm specifically thinking about is the situation where we have a clutch of more Yes friendly pollsters (P/base; ICM; Survation) and a clutch of more NO friendly ones (TNS; YouGov; I-Mori).

    First question : are we seeing substantially different results from each of these pollsters for Euro VI in Scotland ?

    If so then what kind of result on Thursday would show the Yes friendly pollsters as being the most accurate, and what kind of result would show the No friendly ones as being more accurate.

    Understand lots of caveats need to be applied in terms of relationship between Party and IndyRef VI, and also in terms of turnout which should be much higher in IndyRef than it will be on Thursday, but would be nice to have a broad-brush overview to give us something to watch for on Thursday.


  8. Ivan : YouGov are showing the SNP at a lower level of support for Thursday than other pollsters, and as they're a No-friendly firm it might be a matter of some concern if they ended up closest to the truth. I suppose what the SNP would be looking for to prove the Yes-friendly pollsters right is mid-30s, which is a tall order because last time around they only got 29%, at a time when the UK Labour government was extremely unpopular, and the SNP government was still enjoying an extended honeymoon.

    I have a sneaking suspicion the SNP's vote might be a touch lower than we expect - the combination of pure list PR plus a very low turnout seems to produce that effect.

  9. Turnout will be the key - I expect it will be fairly low but SNP should top the poll. Don't think the Greens will be polling well enough for the last seat.