The usual two regular polls from Populus and YouGov this morning. YouGov’s daily poll for the Sun had topline figures of CON 34%, LAB 35%, LDEM 8%, UKIP 13% (tabs here) Populus’s twice-weekly poll meanwhile had figures of CON 32%, LAB 36%, LDEM 8%, UKIP 16% (tabs here).

284 Responses to “Latest YouGov and Populus figures”

1 4 5 6
  1. Labour are on around 36 %.

    This is about 6 standard deviations from their 2010 result.

    This is a massive way off.

    I think any claims that Labour are within MOE if their 2010 result are not robust in any way.

  2. @Mr Nameless

    The data is that varied from previous years, the way to draw a prediction is to have confidence intervals a mile wide (like Mr Fisher).

  3. @Catmanjeff,

    Agreed. I think the misunderstanding comes from treating figures at the low end of oneMOE as the middle point of another MOE. If you get a string of polls ranging from 33-39, that is MOE around 36. You can’t then take one of those 33s and say that the MOE is 27-33 (and therefore that the 33 is “within” MOE of 30).

    That’s the sort of thing newspapers do, not sensible people.

    For me, I’m waiting as always to see what happens to YG. They showed a pretty definite drop in the Labour lead, over many polls. It could well have unwound now, but one poll does not a Summer make. And I don’t think the two non-YG polls provide anything like enough corroboration.

  4. Mr N – polls before 92 were inaccurate and to a lesser extent in the run up to 97, exaggerating the oppositions advantage.

    Of course they still are as it is an impossible task but they should be better now due to adjustment etc.

    Of course Coalition dynamics may well affect genuine last year movements as well even if polls were historically more meaningful.

    That Governing parties improve in the final year as disaffected supporters hold their noses and drift back and that opposition support falls due to greater scrutiny makes intuitive sense of course but by how much?

    Wish I knew.

  5. I think fixed-term parliaments make a difference too. When people didn’t know when the election would be it was less easy to put one’s finger on which party one would support.

  6. Actually I have found it out in Twitter. The referendum poll is in the ‘Sunday Mail’ and is a very good example of bad poll reporting.

    The poll was by progressive opinion but was compared to a previous YouGov as big boost to No. However when compared to the previous Progressive Opinion it actually shows a ‘6% swing to Yes’

    Progressive opinion has traditional shiwn a low yes vote and Since the previous poll was November it seems that it is just bearing out other polls. So nothing dramatic after all. But did generate a but if excitement on Twitter from BT folk

  7. YouGov is now seriously out of line with mainstream polling. Their failure to prompt for the third party is skewing the figures.

    In response to a query about this AW replied words to the effect of; ‘We don’t prompt for Ukip because that would overestimate their support…. we’ll prompt for Ukip when we think that will produce more accurate results.’

    On mature reflection I find this unsatisfactory on a number of counts.

    If prompting ‘overestimates’ then why no downward adjustment for the parties prompted for? And why is the fourth party being favoured with this overestimate in preference to the third?

    ‘we think’ seems awfully subjective. To arbitrarily decide that the third party’s responders somehow don’t really mean what they say is irrational unless backed up by some reasonable criteria.

    In the absence of a compelling case for such a bias I am forced to conclude that YouGov is far from impartial in this matter.

  8. Couper – as far as I know the poll is nothing to do with YouGov.

    In the past we’ve sometimes done fieldwork for Progressive Polling, but I’m not aware of this being such a case. Even if it was, they wouldn’t have used YouGov’s weightings so it wouldn’t be comparable to YouGov Scottish polls – you should treat it as change from the previous Progressive Polling poll, not compare it to YG polls.

  9. Mr Beeswax – I am responsible for doing the YouGov polling, if you want to question my professional integrity and libel me by accusing me of bias you want want to consider doing it elsewhere.



    The only place Ed polls badly is in leadership yet trumps Cameron elsewhere and especially policy. Labour’s fight on real income is obviously a strong one from the polls.

  11. Well at least Mr Beeswax would be happy with today’s Sunday Politics in the East. The three studio politicians holding forth are Conservative, LibDem and UKIP. No Labour voice at all.

  12. @AW

    I think you’re taking Beeswax’ comments a little too personally. I don’t agree with him, but it would be interesting to know how you decide that prompting does, or does not overestimate support.

    After all, you also deny that polling is a prediction of election results, which would seem to be the obvious yardstick by which such a comparison would be made.

  13. @ The Sheep

    Well even I can work out the reasons.

    In the past prompting for UKIP has overestimated their support. There is no hard data on a GE compared to polling since the UKIP surge to go on therefore the obvious thing would be to stick with what has been proved in the past.

    The only evidence to the contrary, which we discuss on here, is by elections and local elections which have their own factors of low turnout geared towards the UKIP demographics (older people more likely to vote).

    I do think there is a chance the UKIP vote could hold up may deserve prompting for but it seems perfectly reasonable to go on past data rather than guess. No-one of here seems to think UKIP will get 20% at the GE so I think AW is in good company.

    I also think the LD don’t knows will not return in the way ICM thinks they will but I can again understand their reasoning and that too I guess is based on historical data.

  14. Just noticed. All four polls of this week. Labour in Scotland is their 4th highest in the five YG regions. If memory serves me correctly, they were generally 2nd (behind the North).

    Perhaps there has been a bit of a shift up this way. It’s just that it takes weeks or months to happen, so it’s subtle.

  15. Mr Nameless
    Your report of leader image polling (thanks) actually displays “Don’t Know” as the overwhelming choice by voters in all but one property (cocky – won handsomely by Nigel Farage). Does this not prove that leader image is vastly overrated as a factor in FPTP general elections?

    If over one third of the country cannot say whether they think a leader is sanctimonious or otherwise, then clearly they do not watch the endless photo ops on TV (could apply to any of them by the way).

  16. TheSheep –

    To some degree it’s a matter of judgement, as well as of experience. The reason I always say polls aren’t a prediction of elections is because they can’t predict changes in the future, they can only be a prediction of how they’d vote in an election tomorrow. That means most of the time they aren’t a prediction of the next election, unless there actually IS an election tomorrow (or in the next day or two) in which case they are.

    The comparison of those eve-of-election polls and actual results is, of course, an absolutely key consideration.

    There are no hard and fast rules as to prompting. Essentially though not prompting for the main parties underestimates their support, prompting for minor parties overestimates their support. The decision is where to draw the line. This is not easy.

    Past experience has led us NOT to prompt for UKIP, and this position is shared by all but one of the other pollsters. Not prompting for them has always got their support about right (including at the 2009 Euros when they came second), prompting for them (or the Greens) at past elections has overestimated their support. We have little evidence to go on at recent elections since UKIP’s support increased – by-elections are a completely different kettle of fish, so that leaves only local elections. The experience of ComRes in those elections (and in our own private test polling) suggests current methods are getting their support about right, so there is no need to change.

    In my own personal considerations, the way the media treat parties weighs heavily on my judgement. If OfCom and the BBC treat a party as a major party they will get a certain amount of coverage on broadcast news, will be reported and treated as a main party and will lead the public to generally think of them as a main party. In that context, it might well produce more accurate results prompting for them than not (and indeed, for European elections they are treated as a main party by OfCom and the BBC and I think all or almost all pollsters prompt for them).

    On top of this, there is the recent experience of 2010 and the polls overestimated the support for a party that was on a roll and had very enthusiastic supporters – possibly due to disproportionate response rates. My experience of 2010 suggests to me there’s a high risk of polls overestimating UKIP support, so that too is something I need to consider.

    The bottom line is that’s difficult to do, there are no right answers and chances to test your models against election results come only once every 4/5 years. A lot of people look a polls and their main consideration is, naturally enough, “Are they being fair to the party I support”. For a few of us accuracy is all that matters, and it is actually an extremely important test for us and our companies.

  17. I do think the subject of prompt questions is worthy of discussion and it’s a pity if people get in a huff about it. Some reasoned discussion would be interesting. I know we’ve had it before, but there are always new members and those with memories like sieves, (er moi?).

  18. Mr Beeswax – I replied to you at length on the matter of prompting a while back, but I don’t know if you saw it:

    Also worth noting that YouGov’s Canute-like aversion to asking people whether they support the third party or not, does depress their figures. Other pollsters have updated their methodologies in the light of events though, so it’s only a matter of time before YouGov follows suit.

    This isn’t true you know. Firstly, as far as I am aware, no “other” online pollsters seem to have “updated their methodologies” by including UKIP in their initial list of Parties (Con, Lab and L/D), rather than in the second list of ‘Others’.

    Survation do do so, but they are a relatively new pollster and seem always to have done so. The other pollsters seem to have continued in the old way. It’s difficult to tell from some pollsters table exactly how questions have been asked as they sometimes combine different things in the same table, but certainly Opinium still ask the two-stage question.

    If you think things have changed, it may be because there have been a number of Euro polls recently and UKIP is included in the main list for those. But that was the case back in 2009 as well (except for ICM phone polls), so nothing has actually altered.

    But you’re also wrong because how you ask the VI question in online polls no longer seems to make any difference. I looked at the average UKIP VI in three different pollsters since last September:

    ComRes (O) (8 polls) 17.25%

    Opinium (16 polls) 16.88%

    Survation (8 polls) 17.25%

    The one pollster that does include UKIP is not significantly different from the two that don’t.

    I haven’t looked at YouGov and Populus because they use other methods from these and in any case Populus changed how they calculate their figures in the middle of this period. It’s also unclear how Populus actually now ask their questions. They gave ridiculously low UKIP VIs until they did alter their daft method of adjusting their figures, but we know that there were other reasons for that than question wording.

    You will also notice that I have specified online polls in doing my comparison. I do wonder if there is a problem with the questions in the telephone polls, which might help explain why there is such a big difference between telephone and online ones. Though I also think that there is a problem with online panels as well, so the truth is probably somewhere in the middle (but exactly where I don’t know).

  19. “Also worth noting that YouGov’s Canute-like aversion to asking people whether they support the third party or not, does depress their figures.”

    In what way is it ‘Canute-like’?

  20. I think there is a very large element of “message-sending” in all opinion polls and much less about GE intentions.

    The most obvious example is those responding UKIP who are aiming to throw a right-wing anti-Europe message to the Tories I would guess. By no means all but a good percentage.

    LDs probably have most to fear from the figures being genuine but there is probably quite a lot of “punishment” going on in those figures.

    Were I asking the question I would say:

    “if you had the casting vote in a GE held tomorrow would you vote and, if so, which party would you vote for; I wouldn’t give any choices.

    If people know so little about politics and the party system I don’t think their vote choice would be that reliable

    It would be interesting to see if that changed the MOE at all and the gurls are prepared to slip AW a fiver of their pocket money [each] if others will also chip in to support.

  21. @ Welsh Borderer

    Thank you :-)

  22. Interesting coalition squabbles.

    Does ANYONE think LDs will get 15% or above?

    I still see them as doooooooooooooooooomed and that is an objective not a prejudiced forecast.

    I feel that their main people put the electorate off them the more they are seen and heard for one thing: this is not good ahead of a GE and saying “it’d have been even worse if the Tories had governed alone” doesn’t help either of them but does help Labour.

  23. R&D –

    There was a time when people didn’t prompt in the 80s, it used to underestimate Liberal support. MORI used to switch from unprompted to prompted mid campaign and it always used to boost Lib support… or perhaps it didn’t, and it was the campaigning/publicity effect of the campaign, or perhaps it was a mix of both.

  24. amber

    You should be out doing your research.

  25. @ R&D

    Paul, How are you today? Is the knee feeling any better? Do you know how long it will be before you get the bionic one?

    I watched a program about acoustic guitar on I-player last night after I quit the boards here. When Johnny Marr (who I greatly admire) was talking about the importance of acoustic to rock/indie music, he cited the Everly Brothers as a big influence. Of course, when he said this, I thought of you! Anyways, thanks for suggesting that I trot off & listen to some music to lift my spirits; your suggestion worked!

  26. ” and the gurls are prepared to slip AW a fiver of their pocket money [each] if others will also chip in to support.”

    Daisie says she won’t as she has her eyes on a sweet new scarf.


    Thanks: I still think stressing the “fact” that the respondee is taking personal responsibility for the overall result might concentrate minds and intention.

    It just seems too easy to say something that has a “pushing” effect as its aim rather than a genuine final choice.

    I could easily respond Green and then, given our daft system, support another party at the GE.

    And I’m sensible.

  27. Amber

    thanks for kind thought. It has been strange: no pain and no crunchy noises as before when I do exercises but a weird excess of mental energy and inability to sleep – hence last night’s nonsense [which I must say I find utterly bizarre and takes pedantry to uncharted levels.]

    However, next time you throw away a little comment make sure you have attachments available which shoe all sources and your working out.

    Thrice LOL as you would say….

  28. @AW

    “To some degree it’s a matter of judgement, as well as of experience.” (enjoyed the rest of the post too, btw)

    Thumping out my spreadsheet, I noticed that the Lab + Con was 69, and generally while less than 70 is low, there have been a few sub-70 results in the past (14 of 30 recent polls, versus sporadic ones other than in May/June of last year, where there was a glut of sub-70s).

    I got wondering about UKIP VI, versus their GE 2015 result. Based on both (judgement and experience), do you have a guess as to how much of Con / Lab / Lib to UKIP folk will flock back, if it’s looking to be a tight election?

    Are we into relatively uncharted territory, and can the EU elections really give us an indication of potential GE 2015 results (my own instinct is not really, hence why I’m asking)?

  29. @STATGEEK: “can the EU elections really give us an indication of potential GE 2015 results”

    Absolutely none (IMO). Its main (sole?) effect will be how the parties react.

  30. “can the EU elections really give us an indication of potential GE 2015 results (my own instinct is not really, hence why I’m asking)?”

    No. The Conservatives won the European elections in 1999 and 2005, UKIP came second in 2009. None of these were particularly good indicators for the general election that followed

    In many ways the local elections will be a more interesting flag, but even then, the Conservatives often “won” the local elections and lost the general. Where UKIP do well in the local may be interesting though (less so in the Euros, as there are a large chunk of voters who consciously vote UKIP at the Euros and Con at GE, which will skew things.

    A couple of people commented on the final lap post about me missing out the locals results – it’s because I don’t think they’ll have the political impact of the Euro results and that’s what the post was about, but in terms of what they tell us they are probably the more interesting.

  31. At the risk of being repetitive may I make a point I’ve made before. There seem to be two ways to interpret any given set of opinion poll figures, (1) as the level of support the parties enjoy at present or (2) as the level of support that is predicted from the raw poll data the parties will get at the next general election. To my mind these are not the same

    It seems certain that at least some of the polling companies are aiming to find the second measure. In other words they find Party A has attracted 20% in one of their polls, but the particular polling company applies an experience / prediction weighting that Party A will actually get 23% (or perhaps only 17%) support at the GE.

    The problem as I see it is that the pollster presents that 23% (or 17%) as Party A’s “true” score at present, whereas non-expert plebs such as you and I take it to be the first of the two interpretations, Party A’s actual level of support now.

    So there are two hazards for Joe Public, making the wrong choice in deciding which interpretation the pollster has decided to present and the subjective assessments that may have been used by the pollsters to adjust their original figures.

  32. If ever the message about not taking too much notice of a single poll could be delivered, then the last two have certainly done the job.

  33. Mr Nameless: thanks for that.

    Of course, the big difference is the lack of swing from Tory to Labour. Most 2010 switchers I meet were Labour in the past, certainly in 2010. They are unlikely to switch back in great numbers (I know we have seen a slight shift, but I think that policies such as tenancies etc will shore them up)

  34. People of all political persuasions support renationalisation of the utilities so its
    No longer correct to describe it as leftwing.

1 4 5 6