Scottish YouGov poll

There is a new YouGov poll of Scottish Parliamentary voting intentions tonight, commissioned by the Scottish Greens and YouGov’s first Scottish polling since last October. Topline figures are:

Constituency: CON 15%, LAB 41%, LDEM 8%, SNP 32%
Regional: CON 15%, LAB 40%, LDEM 7%, SNP 26%, GRN 6%

The last Scottish polling we saw was a rather surprising MORI poll earlier this month. Until then Scottish polls had been pretty consistent in showing Labour ahead, often by quite hefty margins. The MORI poll either reflected a remarkable turnaround for the SNP – showing Labour’s lead dropping by 11 points in the constituency section since the previous MORI poll – or a rogue poll. The hope was that the next Scottish poll would give us a clue.

Obviously we can’t be confident what way the trend is moving with YouGov, since this is the first poll for over three months, but in the constituency section it certainly shows a Labour lead consistent with the other polling prior to MORI’s last one. The regional vote shows an even larger lead for Labour – it’s unusual to have such a difference between the regional and constituency votes, so that might be a bit of a blip.

84 Responses to “Scottish YouGov poll”

1 2
    I do think overestimating the SNP and underestimating Labour – however small – at the last 3 elections is relevant and not just the overestimating of the SNP and underestimating of Labour in 2007.

    Perhaps you are right for the wrong reason in being concerned about 2005 & 2010 elections which, at best, have little relevance to the 2011 General Election.

    In an ideal world, YouGov would separate entirely polls and weightings for Holyrood and Westminster elections. But, clients being what they are – and clients for political work are often dealt with by the pollsters on a “loss leader” basis, if we can believe what we read – they want as many questions answered as possible.

    Accordingly, they have a tendency to ask Holyrood and Westminster [or vice versa] questions consecutively, which must at the very least generate some confusion in those polled and perhaps needs different weightings for each political context.

  2. Barbazenzero
    I agree with you in saying that we should be wary of polls commisioned by clients with an agenda – in this case the Scottish Greens. I hope that when the SNP commision a poll (as they have of You Gov in the past), we will also hear your skepticism.

    However, the SGP have said they asked just for a standard Holyrood VI poll – so no westminster questions to muddy the water.

    I hope that when the SNP commision a poll (as they have of You Gov in the past), we will also hear your skepticism.

    I shan’t even be sceptical of this one until I have had chance to view and analyse the full tables as well as the weightings on Newspaper Type and Political Party Identification, neither of which are always published for Scottish polls. But I confirm I will be wary of all such polls before I have had the same viewing opportunity, irrespective of who commisions them.

    However, the SGP have said they asked just for a standard Holyrood VI poll

    Would you have a link to that statement? Their news release here which makes no such assertion, although it does link to what it describes as the “Full tables” which turn out to be just the first four pages of weighted data for constituency and regional voting intention but without the usual weighting information.

  5. I see that a lively debate has arisen on the YOUGOV political weightings.

    As I understand it the position goes far beyond any normal weighting of samples to correct for bias.

    It is said that YOUGOV have political identifiers for Scotland which put Labour loyalty at twice that of the SNP – 36 to 16. It is against his impossible benchmark that their sample results are transformed from a big SNP lead to a big Labour one.

    Is this true ? If it is then they are wrong and until they are changed the organisation should – for the sake of its reputation – simply stop polling in Scotland.

  6. Barbazenzero

    Thanks for that link to the link.

    One wonders in what demented universe “a standard Holyrood VI poll” starts off with a question on Westminster voting.

    The most significant point of the discussions following the MORI poll was the prism (Westminster or Holyrood) through which politics are being seen.

    If YouGov are going to start all their polling on Holyrood by asking about Westminster then they are (at the very least potentially) distorting the entire exercise.

  7. Anthony

    Just to avoid any doubt. when I was talking about “what demented universe”, I was indeed meaning YouGov. :-)

    I realise that you don’t do the Scottish polling, but would you care to ask your colleague who did it, why the Westminster question was asked/ Was this a YouGov initiative or at the request of the Greens?

  8. Yougov seem to have substantially different figures from IPSOS Mori and (for Westminster) from Angus Reid’s subsamples. One of Yougov’s recently had 62% support for Labour (after weighting!) and 7% support for the SNP. I am sorry that is ridiculous.

    I have a substantial concern about the degree of impact of weighting in this case too, which is quite extraordinary. With the SNP well ahead before weighting and well behind afterwards. If these weighted figures were true, we would not be seeing the kind of swings to the SNP where I live, instead we would be going backwards. That is the opposite of reality on the ground. The only thing it has right is the scale of drop in the Lib Dems.

    Time to invest in shares in salt, as we need a huge pinch of the stuff with this particular poll.

    Anthony – I respect your knowledge, but this is just plain wrong. Yougov are losing much of their hard won credibility as a polling organisation in Scotland, in my opinion.

    Weight by demography and geography – people will trust that. Weighting by past poltical affiliation or linking Westminster to Holyrood voting or even newspaper readership…give me strength.

  9. oldnat
    Thanks for that link to the link.
    One wonders in what demented universe “a standard Holyrood VI poll” starts off with a question on Westminster voting.

    You’re welcome, but be careful of putting AW’s nose out of joint. As you’ve said before, he’s one of the good guys but my own response to your rhetorical question would mirror Francis Urquhart’s. Off out for the afternoon now, but I hope to be back tonight to look at the full tables including weightings.

  10. Much of this debate on YouGov’s weightings is hypothetical.

    The only sure way to gauge the accuracy of their weighting is to compare with actual results. Here is a repost of the 2007 Scottish election results plus Scotland’s polling in the 2010 UK election:

    2007 Scottish Elections:
    Constituency Vote:
    Actual, ?YouGov, ?Party
    32.9%, ?37%, ? SNP
    32.2%, ?31%, ? Lab
    16.6%, ?13%, ? Con
    16.2%, ?14%, ? L-D
    2.1%, ? 5%, ? Others

    Regional Vote:
    Actual, ?YouGov, ?Party
    31.0%, ?32%, ?SNP
    29.2%, ?27%, ? Lab
    13.9%, ?13%, ? Con
    11.3%, ?10%, ? L-D
    14.6%, ?18%, ? Others

    2010 UK Election – Scottish results
    Actual YouGov Party
    42% 37% Labour
    19.9% 22% SNP
    18.9% 21% LD
    16.7% 17% Con

    On all three polls, YouGov underestimated Labour support and overestimated SNP support.

  11. I’m a bit perplexed about what is going on with the weighting in YouGov. According to this poll the unweighted figures gave the SNP a 13% lead but have ended up with a 9% deficit when weighted.

    In comparison with previous polls it is unusually large – unfeasibly so I am thinking. I have pasted below the differences from previous polls going back a year and I fear this poll may be overweighting Labour.

    I am wondering if this methodology is correct. Are too many voters being allocated to Labour on the mistaken basis that they will be voting Labour when they could be abstaining in polls as they may be moving their vote to the SNP or even abstaining in this election?

    It just does not seem right that a 13% lead can evaporate by 22 points to a 9% deficit.


    Latest YouGov / Green Survey Results

    Samples ~~ SNP Lead
    Weighted ~~ -9%
    UnWeighted ~~ +13%
    Difference ~~ -22%


    YouGov / Scotsman Survey Results
    Sample Size: 1405 Scottish Adults
    Fieldwork: 18th – 20th October 2010

    Samples ~~ SNP Lead
    Weighted ~~ -5%
    UnWeighted ~~ +9%
    Difference ~~ -14%


    YouGov / Scottish Mail on Sunday Survey Results
    Sample Size: 1010 Scottish Adults
    Fieldwork: 31st August – 3rd September 2010

    Samples ~~ SNP Lead
    Weighted ~~ -11%
    UnWeighted ~~ -5%
    Difference ~~ -6%


    YouGov / SNP Survey Results
    Sample Size: 1212 Scottish Adults
    Fieldwork: 17th – 18th August 2010

    Samples ~~ SNP Lead
    Weighted ~~ +0%
    UnWeighted ~~ +9%
    Difference ~~ -9%


    YouGov / Scotsman Survey Results
    Sample Size: 1507 Scottish Adults
    Fieldwork: 3rd – 4th May 2010

    Samples ~~ SNP Lead
    Weighted ~~ +3%
    UnWeighted ~~ +5%
    Difference ~~ -2%


    YouGov / Scotland on Sunday Survey Results
    Sample Size: 1520 Scottish Adults
    Fieldwork: 28th – 30th April 2010

    Samples ~~ SNP Lead
    Weighted ~~ -1%
    UnWeighted ~~ -1%
    Difference ~~ +0%


    YouGov / Sky Survey Results
    Sample Size: 1001 Scottish Adults
    Fieldwork: 21st – 23rd April 2010

    Samples ~~ SNP Lead
    Weighted ~~ +4%
    UnWeighted ~~ -2%
    Difference ~~ +6%


    YouGov / Scotland on Sunday Survey Results
    Sample Size: 1002 Scottish Adults
    Fieldwork: 24th – 26th February 2010

    Samples ~~ SNP Lead
    Weighted ~~ -6%
    UnWeighted ~~ -7%
    Difference ~~ +1%

  12. Andy B – no, YouGov don’t do any reallocation of don’t knows or won’t says. If people say don’t know, we treat them as a don’t know.

    After spending the morning digging through the data of the Scottish poll (sick of the site of the damn thing now!), I’ve found the reason for the large weighting factors. It wasn’t unusual response rates as I speculated it might be earlier – it was simply not sending out the correct proportion of emails to differnet party IDs on the panel in the first place. At least it is easy to solve next time.

    As I’ve said further up, the effect of having to do a lot of weighting to a poll is to decrease the effective sample size, and therefore gives a poll a larger margin of error than if less weighting had been needed. It should not have any effect on the levels of support for respective parties, other than to make them a bit more volatile.

  13. Anthony

    I know you are not responsible for this poll, but surely this renders the credibility of the results here as being highly suspect in this case, i.e. it should be treated as a bit of a rogue poll given the substantial weighting applied?

    Given the significance others have made of this poll, it is a concern that this sampling issue was not more prominently reported in the news releases especially as the margin of error is affected. For the record what do you assess the margin of error to have been increased to?

    Andy B has made a very helpful analysis there and this poll and the one in October stick out like sore thumbs….

    All the best.

  14. barbazenzero,
    I was going on what James had said over at Better Nation. Since he works for the SGP I think he may know a bit more about it than me.

  15. Borderer – all margin of error figures are bit spurious – the 3% figure we normally quote for example is for a perfectly random 1000 sample with no weighting at all!

    It would probably increase it to about 4% rather than 3%.

  16. Barbazenzero

    Good point (and I’ve been singing Anthony’s praises elsewhere as yopu point out)/


    Sorry. As You have pointed out many times polling companies do their best to answer their clients questions

    The Scottish Greens would have known exactly what the effect of including the Westminster question would be.

  17. oldnat

    What really irks me about the poll is neither the additional question which can only skew the results nor the curious weightings, which at least can now be analysed but the hay being made with the “topline” numbers in the MSM. The Hootsmon’s Opinion poll joy for Labour as voters put party back in front is a case in point. Making bricks without straw, of course, but giving unmerited comfort to people who should know better. The one good thing about the “report” is that the hapless Greens, who commissioned the darned thing, get no mention in it whatsoever.

  18. Barbazenzero

    There might be mileage for the Greens yet – if they refuse to pay their YouGov bill! :-)

  19. oldnat

    Francis Urquhart’s response would again seem apposite.

    Family duty calls, followed I hope by some analysis of the weightings, but back later, I hope.

  20. Thanks for the revised mragin of error info, Anthony.

    All the best.

  21. oldnat

    After a long and tiring day, I’m afraid I need to sleep on this but I have posted an interim note in another place.

  22. Andy B

    “It just does not seem right that a 13% lead can evaporate by 22 points to a 9% deficit.”

    Apart from the weightings, not only can one poll be an outlier but both can be in the opposite direction. Until we have more polls, we know noting.

    We need an explanation for a change of 22 points.

    Anything of that scale would be so obvious as to be unmissable. Nothing an opposition could do could ever come near it. It would have to be a staggeringly stupid mistake by a governing party which repelled their own supporters.

    There are many on here who will draw comfort from, or be depressed by a 1% change in their party’s support at Westminster. Not only that, but they will attribute the change to policy announcement or performance at PMQ. If you have ever believed such explanatons, try to imagine what the cause must be of a swing of 22%.

    Shooting hundreds of demonstrators would be one way you could do it. Something like that.

    Also, in a system where gains and losses in a constituency are often, even usually, compensated by losses and gains on the list, the scale of change inferred from this poll is just not possible.

    Between the two largest parties, the parliamentis finely balanced now. It will remain finely balanced now, whoever wins and it must be more unpredictable than any Westminster election in my lifetime at which there was a change of government.

    As an accountant, I was taught that just before I signed off anything I should ask the question “Is this consistent with everything else I know?”

    Clearly, considering the scale of the predicted changes alone, this is not credible. No party is going to gain oe lose more than a handful of seats. There must be some explanation.

    The weighting is suggested and that may be the answer. If not that, then is there another expalnation for a polling error or an explanation of the swing?

    That is not to say that Labour may not yet have a lead over the SNP, but you can bet that neither party will
    exceed the other by more than a handful of seats.

  23. Okay, a few points now:

    Firstly, if I get this right this poll started with questions on Westminster voting intention before going on to Scottish Parliament voting intention? But it is well understood that this has the direct consequence of skewing the results for the Scottish Parliament voting intention, and if SGP asked for the usual poll on Scottish Parliament voting intention then YouGov should have constructed the poll differently.

    Secondly, the poll did not weigh for likelihood of voting. But it really should be clear that to any political observers that we are now in an election period since the budget debates (and the related news coverage which always put things into election perspective). I mean, if we weren’t, why would the Greens commissioning a poll in the first place?

    Thirdly, YouGov made a humongous error when sending out emails which had consequences for weighting and margin of error – not an unimportant issue if your client is a party for whom the difference between +3 or +4 would make a big difference.

    And that’s before we get to the question whether the YouGov method of party ID works well in a Scottish context.

    So, if I’d be the client I would not be happy with my contractor and would ask for a thorough explanation before I would ever commission more work…



  24. The Scottish Green party paid for the poll, so lets look at what their press release says.

    “If these vote shares were repeated in a Scottish election, Weber Shandwick’s Scotland Votes site predicts the following result.

    Lab: 59 (+13)
    SNP: 35 (-12)
    Con: 19 (+2)
    Lib: 9 (-7)
    Green: 6 (+4)
    Ind: 1 (n/c)”

    I think ALL of these changes in seats are too high.

    The first two I dealt with in my previous post.

    Is there some reason – inspiratonal leadership from Westminster and hugely popular Conervative policies – for example, that has moved thousands of Scots not only to halt a steady and inexorable decline of two generations, but to reverse it?

    Step forward those who believe that. I must be missing something. Tell me what it is.

    If they hold on to what they have now in seats and lose no more han 1.5% of their votes as they did last time, they are to be congratulated on their sucess.

    That would be a well deserved recognition and reward for a responsible and skilful performance in opposition to a minority government. The LibDems have not done well in this regard as might have been expected given their self image, and as junior coalition partners the LibDems are in competition with the Cons. Con’s win on flexibiity and style. LibDems could do better.

    +/- 0 is a good result, -1 a disappointment, but not unexpected. Nothing more or less is likely and +2 is fanciful.

    The LibDems are losing votes. There is no doubt about that.

    Where they have constituencies, they can afford to lose many votes because the majorities are large, and they have no list seats to lose. Elsewhere, they can lose list seats but if you deduct from the existing total, those that individually seem safe, then 7 is an incredibly high proportion of the remainder.

    The Greens will improve their position. I have predicted that since the last election when they lost five out of seven seats because of the expected closeness of the result, AS for FM, and loss of “Second vote Green.”

    They have a natural core support which is growing, socialists no longer in contention, and environmental concerns are no less clamouring for attention than they were.

    In the long view, parties have come into existence and flourished as a result of the more general dominence in society of the interests they represent. First it was the landowner (Tories) then the successful businessman (Liberal) then with the extension of the franchise, the employee (Labour) and now the importance of environmental issues and awareness of them is increasing daily so the Greens should progress.

    They have all that going them in the long run, but May is not the long run.

    They should certainly get back some of the 5 they lost by falling just below the threshhold for list places. Fewer than three would be disappointng for them, and three or more would allow them to function fully as a party. Four is optimistic. It’s possible, but at the upper limit of what is likely and this is the least unlikely of the predictions above. Compared with the improbability of +2 for Cons or +13, you would have to offer +7 for the Greens, and nobody would bet on that.

    The least likely prediction is for one independent.

    The Margo is retiring. There is only the one. There are no others on the horizon at the moment.

    With the exception of the merely unlikely and overstated but just about possible +4 for the Greens, ALL of the predictions for change are excessive. That all of them might be correct is even more unlikely and requires a minimum of 20 uncompensated changes.

    That in itself is unbelievable.

  25. Just a note. Margo isn’t retiring.

  26. Bearing in mind compensation for losses, can we spot the LibDem losses?

    The certain one is Ross Skye and Inveress East.

    A long serving independent minded local LibDem more popular than his party is on the list and with a huge personal vote, is retiring. I do not think this constituency can be passed on to a less well known LibDem candidate. The SNP challenger shadowing him is a high profile list MSP who has made a name for himself in road and rail safety.

    An SNP gain in the constituency will almost certainly be balanced by a loss on the list, but since LibDems are overrepresented, and some lost LibDem votes will go to the Greens, this is perhaps the most likely of the list places the Greens can get back with a slight increase in votes taking them over the threshhold.

    So LibDem -1 Grn +1

    For the LibDems to get down to 12, you need to pick three more. Are any in Highland? LibDem losses there will not be made good on the list. Which might they be?

    I might agree with you that four losses are possible overall, if you can show me where they are, and then we can move to your “more likely than not” choices and see where we each draw the line.

    Your pick next.

    We can also look at it the other way. You are not going to suggest that a majority of half those voting can be overturned in one election other than as a consequence of criminality. Shetland is a safe seat, and Orkney not far behind.

    There will be no change in Western Isles. In Argyll, the LibDems having lost incumbency, will continue to leak votes which break in favour of the SNP. That should secure the 815 SNP majority, though it isn’t going to be enough to produce another SNP list seat.

    Likewise, Fergus Ewing is secure in Inveness East, Nairn and Lochaber because he is up against a LibDem challenger.

    In Moray, Richard Lochhead has a 26% majority and is no longer up against the sensible and decent Conservative Mary Scanlon. In a Glasgow supermarket this quiet and unassuming minister would not be recognised. In a Fish market, he would be the centre of attention. The Westminster SNP results were geographically skewed, and most favourable in consttuencies where fish are landed. It’s all due to Richard Lochhead.

    Look for the SNP’s largest improvement in this constituency.

    In Highland region that leaves Caithness, Sutherland and Easter Ross. The sitting MSP with a 10% majority is retiring and that might just be enough to lose to the SNP and not be compensated, though the SNP gain would be.

    That wouldn’t be any contribution to the predicted SNP total losses from Highland Region.

    It is difficult to see what party would gain a second list seat if the SNP convert one of theirs by ousting a second (uncompensated) LibDem.

    So in Highland, there will be no change in 2 LibDem and 4 SNP constituencies. LibDms are sure to lose 1 uncompensated which the Greens will pick up on the list. Just maybe, a second LibDem loss if the SNP do well in Caitness, and that could result in them vacating a list seat for Labour.

    Least change for the region: Libem -1; Grn +1
    Most change for the region: LibDem -2; Grn +1; Lab? +1

    That’s only one region, but it is the one which has likely uncompensated LibDem Losses. With changes on that scale, where are the other 17 going to be possible? Not in Glasgow for sure.

    Where can another dozen Labour gains come from, if not in Glasgow or Highland?

  27. Daniel

    Then that’s another change not happening, and the only certain prediction.

    I hope that means that her doctor has given her cause for encouragement to carry on.

  28. Having looked in some detail at all of the weightings in the poll, it’s Political Party Id that still worries me most.

    Recognising that all pollsters take some steps to "downgrade" those unlikely to cast valid votes, and that they are arguably less likely to
    sign up to an on-line pollster, I have examined YouGov's weightings excluding those who identify with None/Don't know, producing the following table:
     YouGov Scottish Weighting Data – Political Party Id (excluding the 21.0% None/Don't know)
     GE2010, Target, Weight, Grouping
       42.0%,   48.1%,   1.64,    Lab
       16.7%,   16.5%,   0.73,    Con
       18.9%,   12.7%,   1.88,    L-D
       19.9%,   20.3%,   0.69,    SNP
         2.5%,   2.5%,     0.37,    Others
     100.0%, 100.0%,     n/a,    Total

    We can now see that the figures for Con, SNP & Other support (in a Westminster context) are broadly similar to their actual performance in 2010 but YouGov provide no explanation for the compensating Lab increase of 6% and L-D decrease of 6%. We know in that an English context, the huge drop in support for the L-Ds – currently apparently being ameliorated a little – is being reported as largely to Lab and have one English by-election to lend support to that. We have much less evidence that in Scotland any movement away from the L-Ds is toward Lab.

    In any event, we have harder evidence that voting behaviour in Holyrood general elections is different to behaviour in Westminster general elections, so we know that even if 48.1% of voters would vote Lab for Westminster, a significant proportion of them will vote differently for Holyrood. The problem then becomes a question of whether we have any assurances that the “mix” of that 48.1% between Lab “loyal” and Lab “disloyal for Holyrood” in any given sample of them will be correct. The simple answer is that we don’t, and absent evidence we should not try to second-guess the proportion but simply regard the 48.1% as, at best, not proven.

    If I'm right, then it is no surprise that Lab responders to the survey had to be weighted upward by a factor of 1.64 as fewer of them exist than YouGov are expecting.

    For any future Scottish polls, I suggest looking very carefully at the "forced" weighting upward of Lab votes.

  29. A Brown

    I don’t understand what this list is meant to be.

  30. A Brown
    Don’t worry, I follow. You have to tell them they are winning or they don’t like it

  31. Barney

    I only just spotted that your good self and Ross Thomson, the Tory are doing a double act around the North East!

    Leaving the vexed question of 1st and 2nd place aside (I think we may differ there!), how about 3rd & 4th?

    The LDs will obviously have lost some support after the creation of the UK Government, and losing their candidate won’t have helped either. Seems to be a distinct possibility of the Tories getting 3rd place in Donside?

  32. Old Nat
    I sense ever more will vote either SNP or Labour. When Labour won in 1999, the Lib Dems were a close third. The Lib Dems must however be staggering given the current position and there will be no campaign. Locally though the Tories are in disarray with two warring groups on the local council. But yes the Tories may squeak third

  33. STUART DICKSON on the Populus thread has flagged YouGov’s new Welsh referendum poll for S4C at

    Interesting to note that the poll perfectly reasonably asked an Assembly VI question – after all the whole poll is about the Assembly but apparently DID NOT ask a Westminster VI question.

    It’s also interesting to note that 17% overall want the Assembly abolished and more interesting still that 17% of L-D supporters for the Assembly want the same. Federalism really would seem to have been eradicated within their ranks.

    Sadly, it does lack weighting information, however.

  34. Old Nat
    post script
    Met one of Aberdeen’s leading Lib Dems just now. He will vote for me FPTP and Green on list (top of the Green list is expelled Lib Dem Martin Ford)

1 2