ICM’s latest poll is apparently due out tonight, but in the meantime Ipsos MORI have released their monthly tracker, and there’s no sign of any Labour recovery here. The topline figures, with changes from their last poll, are CON 43%(+7), LAB 26%(+2), LDEM 19%(-6).

Their previous poll was carried out just after the Liberal Democrat conference, and got a lot of attention since it showed Labour in third place. Not surprisingly this one shows the Lib Dem conference boost unwinding, but unlike the YouGov and ComRes polls at the weekend there is no narrowing of the Conservative lead – in fact, the 17 point Tory lead is exactly the same as the last pre-conference MORI poll.

Full tables are here.


23 Responses to “Ipsos MORI show 17 point Tory lead”

  1. Anthony,

    What is different about YouGov methodology that tends to make their Labour scores 2-3 points higher than other pollsters ?

  2. Anthony, perhaps you can explain why the headline figures are for those certain to vote, 528/495 of them? Yet the weighted figures are C 40% L 31% LD 18%, and the unweighted figures are C 39% L 31% LD 18%. Am I reading table 1 wrong?

  3. One more thing Anthony, you have Labour as being down 2%, whereas they are up 2%

  4. This poll couldnt be less useless from a Scots perspective.

  5. Wayne – My guess would have been likelihood to vote – if Labour supporters have suddenly become significantly less likely to vote than they were earlier in the Parliament (or there has been a swing towards “Labour” amongst those who are unlikely to vote in practice), it could result in YouGov showing a different pattern.

    However, looking at the actual figures, while there is a difference between ComRes/MORI and ICM/Populus/YouGov, between YouGov, Populus and ICM there doesn’t seem to be any difference beyond normal sample error.

    Taking figures from the whole of 2009, YouGov are in the middle. They produce an average Labour score of 27.3%, compared to 27.5% from Populus and 27.6% from ICM. Below them are ComRes and MORI, with average Labour scores of 25.4% and 25.6% respectively.

    Taking figures from just the second half of the year, since the perceived difference appears to have surfaced quite recently, the figures are Populus 27.7%, YouGov 27.1%, ICM 26%, Ipsos MORI 24.7%, ComRes 24.9% (I’ve taken out the daily tracker figures since they’d have skewed YouGov’s averages towards conference season levels of support).

    The difference between ComRes/MORI and ICM/Populus/YouGov is probably down to “shy Labour voters”. ICM and Populus reallocate voters who say don’t know on the assumption that around half of them will actually vote for the party they did last time, YouGov don’t have human interviewers – in both cases (don’t know reallocation, and the lack of interviewer bias) we’d expect this to boost the support of unpopular parties, and hence lead to higher Labour support.

    The difference between ICM and YouGov I think is illusionary – ICM and Populus’s methodologies are very close to being identical, and they are either side of YouGov, so I think any difference is just normal sample variation. Given there is no meaningful difference between Populus and ICM’s methods, my guess is that ICM have just had a couple of polls showing Labour towards the lower end of the normal margin of error.

  6. CLAD – you are reading the tables correctly.

    The raw unweighted figures were 39/31/18
    Weighted by demographics they became 40/31/18
    Filtered to exclude those less than certain to vote they ended up 43/26/19

  7. The more meaningful figures IMHO are for those with 6-10 likelihood to vote . These give Con 41 Lab 29 LD 19 . The Labour figure is in the same ballpark as Yougov on this measure .

  8. Thanks Anthony. Are all pollster’s headline figures for those absolutely certain to vote?

  9. CLAD-

    Nope, the pollsters have differing views of which figures are the best and how to factor in turnout. MORI are alone in imposing such a strict filter. ICM, Populus and ComRes all have milder filters, and YouGov do not filter or weight by likelihood to vote at all.

    I address it in more detail than you probably want to know here:
    http://ukpollingreport.co.uk/blog/faq-turnout

  10. Anthony,

    Thanks for the reply, very informative !

  11. ICM poll out on Guardian.

    44/27/18 – lovely stuff!!

    http://www.guardian.co.uk/politics/2009/oct/20/party-conferences-turn-off-voters

  12. Why does the predicted seat maj say 44 ,when the % point to a tory maj of 120 ?

  13. @Anthony

    Much as I’d love to jump about, claiming that the mammoth Tory leads are back to stay, it seems all too volatile for now.

    But I’m wondering about “shy Labour voters”; are there data to compare the idea with the “shy Tories” of 1992, or even the ex-“shy Tories” of 1997, where the Tory vote collapsed? What demographic are they? Are they concentrated in changeable seats, or will it be mainly solid Labour seats where the voters feel ignored but would refuse to vote for the BNP, so making little difference to the numbers of Labour MPs?

  14. Guardian ICM poll

    44/27/18

    So same as ipsos more or less. Seems the Tories have regained the leads as before the conferences.

  15. @ Mr Mackinnon – Thomson
    But it is very pleasing to the English eye. Perhaps you should look at poll that deals with Scottish politics.

  16. Interesting – Labour’s vote drops from 31% to 26% when excluding people not certain to vote. That’s quite a significant difference for them.

  17. This is the first poll that I’ve seen that identifies the England only numbers. 330 likely to vote in GB, 438 in England. Clearly the 108 Scots and Welsh tell us nothing about voting in these nations, but is the sample of 330 big enough to give a reasonable picture of England?

    (Anthony – help with this please)

    English figures are Con 47% : Lab 24% : LD 21%

  18. Sorry I got the numbers the wrong way round

    438 likely to vote in GB, 330 in England.

  19. Are your English percentages still correct?

    If so, that would be a swing in England of about 11.5% compared to 10% in GB.

  20. Andy Stidwill

    Double checked – those are the numbers for England. I simply mistyped the Eng/GB sample sizes.

    People will know that I have argued that including the data from Scotland (and probably Wales) distorts the data for the political system that provides the vast majority of MPs.

  21. Obviously this means that any Labour seat in England with the Tories in second place where their majority is 23% or less is in danger – which is pretty sobering news for the government. The other poll today is also giving a 17% lead.

  22. This is a strange set of results. I am inclined to think Ipsos MORI may have been unlucky and got a couple of rogue samples. Let’s see what happens in the next few opinion polls before drawing too many conclusions from this one.

  23. “Frederic Stansfield

    This is a strange set of results. I am inclined to think Ipsos MORI may have been unlucky and got a couple of rogue samples. Let’s see what happens in the next few opinion polls before drawing too many conclusions from this one.”

    I disagree… I find MORI far more plausible than some of the others.