A new ICM poll for the Guardian has topline figures of CON 37%, LAB 31%, LDEM 21%. Changes since the last poll are Conservatives minus 6, Labour minus 4, Liberal Democrats plus 6 (and by implication, the ‘others’ up four). From the last ICM Guardian poll, the changes are Conservatives minus 3, Labour minus 4, Lib Dems up 3.

This is a strange result and certainly doesn’t tally with YouGov’s finding yesterday, or what we’d really expect to find. It should come as no surprise that Labour are down, neither should we be shocked that the Liberal Democrats are up – the recent coverage of alleged negative campaigning by Chris Huhne might on first sight be a potential negative, but I suspect the Liberal Democrats main problem over recent months has been a lack of coverage full stop, so even this was probably good for them. Vince Cable’s response to the missing benefit data affair also seems to have been well received, so I can quite imagine a big boost for them. The mystery is the significant drop in Conservative support, which has no obvious explanation.

There is a rule in market research called Twyman’s law: “anything surprising or interesting is probably wrong”. While not going that far, I would always advise that if you find a poll result that seems somewhat counter-intuitive, that seems to have no obvious explanation, treat it with caution until other polls support the findings. Statistically there is no more reason for this poll to be wrong than the last poll or the poll before that, and we may indeed find that this is a genuine trend and everyone starts showing the Tories down, but it is a bit odd.

It was only a couple of days ago that I was saying how the polls finally seemed to be presenting a consistent picture and now this comes along to upset it! Still, we have at least two timetabled polls due in the next week or two, plus whatever extra polls newspapers have commissioned on the back of the benefit data loss, so hopefully together they will give us a clearer picture.

UPDATE: It may be an even bigger recovery. Later reports in the Guardian have the Lib Dems on 23%…but report that as being three points up (which would put them on 21%, not 23%). No idea if they are actually on 21% or 23%.

UPDATE 2: Confirmed via Mike Smithson: it was indeed 21%.

30 Responses to “Big Lib Dem recovery in latest ICM poll”

  1. Anthony, the Guardian’s own article on this poll(http://politics.guardian.co.uk/labour/story/0,,2216363,00.html) puts the Lib Dems on 23%, and describes them as “up four”. Naturally I would always trust your good self over any hastily written journalistic summaries (!), but can you confirm that they have indeed got their own poll wrong?

  2. Toby – I can’t confirm until something goes up on the ICM website. The Guardian’s orginal report here had them on 21% (at least when I typed this it did!), which was indeed 3 points up on the last ICM/Guardian poll, which had them on 18%.

    The link you’ve provided says (at the moment) that they are up three on 23%. One way or another that’s wrong, since if they are at 23% then they are up five points. I guess we won’t know for sure until the Guardian correct one of their articles one way or another, or until the ICM tables turn up on their website.

  3. ICM are starting to challenge for the title of joke pollsters (though ComRes still edge it for me): I just can’t take their Liberal Democrat figures seriously anymore. A 6 or 7% difference in share of vote between ICM and where almost all other pollsters are represents literally millions of votes.

  4. I see no reason why the Libdem vote cannot go up (but as usual, we need a trend) Maybe it is a “Vince Cable effect”?
    We need to see who wins the Libdem leadership.

    Perhaps the lack of a imminent election has taken some steam from the Conservatives. Then again with the current “weekly disaster” government we have now, perhaps they should be doing better?

  5. I’m also a bit suspicious of this poll but if the Liberals have recovered this much then it’s not really that suprising the Tories are hurt badly.

    There aren’t many soft Labour voters to change sides now we’re down to our core support so it’s inevitable the Liberals are going to be taking votes off the Conservatives.

  6. Is it possible this Lib Dem surge is a result of the much publicised spat between Clegg and Huhne on the Politics Show last weekend giving them some rare widespread media coverage?

  7. I was actually polled by ICM and (a) it was a week ago and (b) there were an enormous number of questions about double glazing at the end (and Everest in particular). I very nearly put the phone down. It is possible that lots of C supporters did and their votes were disregarded, and if they have been sampling over a week they will be behind the curve on the latest fiascos. Although they have the WMA C lead pretty well spot on (it’s now 5.8) the shifts in support look very dubious to me. Having said which, their historic Standard Deviation is 2.5% which is exactly average.

    Stuart: a vast fraction of Lab’s “core support” will be on these child benefit disks. There is plenty of scope for further falls: we had polls showing 27% earlier this year, and this was when Labour were seen as competent.


    If you think that 31% is your core support – you need to look at previous elections and POLLS for Labour over the months and years – they have been known to drop as low as 26% – keep watching , they are heading that way right now .

    I certainly don’t take the Liberal figure of 21% seriously at all – the best they are on is about 16% to 18% / definately a rogue POLL

  9. “they have been known to drop as low as 26%”

    Local election results bear little resemblence to a national parties standing, particularly if that party in in government.What both yougov and ICM show is no Tory surge.And it will take a Tory surge to push Labour into the mid twenties nationally.

  10. Populus showed both Labour AND Conservatives down on various Trust and Honesty ratings.

    This poll is not that inconsistent with those findings.

  11. I very much agree with Anthony’s statement that we should look for other evidence when you have 2 figures as divergent as 14 and 21 for the LibDems which clearly cannot both be correct . We are fortunate that in this case we have it in Thursday’s local election results which support a standing nearer 21 rather than 14 . We need to see the detailed data to see if there is a reason why the Conservative figure is lower than expected but again is consistent with this week’s local election results .
    More polls needed of course .

  12. If Vince Cable’s efforts in the commons have helped the LibDems, then if the new leader doesn’t perform as well te LibDems could find themselves in the position of being seen as having overlooked the best candidate.

    Anthony, ant idea on why the higher Others vote, or do we need to wait for the tables.


  13. Wait until the dust settles over the LibDem leadership contest. The Northern Rock fiasco and others are not necessarily good for the Conservatives: many of the problems can be attributed to big business and pre-Labour days. Additionally the mood is not good at the moment with Englands exit from the football (really this can affect sentiment – remember if you can the 1970 election).

    There may be a mood against the big partiies in all this. Wheher or not it is temporary we will have to wait and see!

  14. Mark, if local election results were a good reflection of public opinion we would have a Tory Government and Labour would probably be the smallest party in the House.
    They are not. They won an election not that long ago.

    I live in a seat which has been Labour for years. The Local and District Councils are all Tory, Tory, Tory. Labour lose those electios hands down and its not worth getting out of bed if you are a Liberal Democrat.

    I also see no reason why there might not have been dramatic changes but not in 24 hours when everything “dramatic” had happened before both polls.

  15. Actually a weighted average of the ICM and YouGov polls (both of which were completed on the 22nd) produces a plausible 39.5-31.6-16.7. Filtering this data produces a hung parliament.


    IMHO both the polls produced questionable results with YG suggesting an unrealistic shift to minor parties and ICM suggesting too high a Lib Dem figure. This is not a criticism of YG’s or ICM’s reliability (in fact I believe that they are the two most reliable pollsters), just a reflection that current events have reduced Labour’s popularity rather than boosted the Tories.

  16. I always find it odd that newspapers, TV etc seem to change the polling organisation they use with such frequency.
    Why don’t they buid up a relationship with one organisation and keep to it?
    My memory may serve me incorrectly, but I seem to recall that in the past, some papers did tend to stick with one organisation.
    Do they tender for the work?
    Are they overstretched???

  17. I simply do not believe that Vince Cables’ commons statement could have had any effect on the LDs poll standing.How many people actually watch PMQs? Any LD recovery would be largely due to the extra general media coverage of the LD leadership contest.

    As for council by elections.From experience the LDs are immune to falling/rising poll ratings when it comes to council by elections.This is because their local campaigns are invaribly highly localised and utterly centred on one particular local issue.

  18. One can understand the drop in support for Labour, but the level of support for the Tories and LibDems baffle…. As no election (local, national, or supra-national) is due for a few months these polls serve little purpose (other then allowing “us pundits” to spout their own theories).

    That said, England’s failure to qualify will hurt Labour. With sub-prime, NorthernRock, and HMRC the confidence of 51,000,000 people must have been damaged. Add to this the Defence Lords “sounding-out” over Rusty Brown’s lack of sympathy to those whose purpose is to die for his legacy, and I think that the next election will return a comfortable Conservative government majority.

    I wonder what Rusty wrote in his Christmas card to his predecessor. You know, whats-his-name….

  19. We always seem to be waiting for the next poll to hopefully explain the last one. In the present fevered atmosphere we should not be surprised if we get one or two quirks. At 32% the share of votes allocated to the Lib Dems and others does appear hard to reconcile with other recent polls but ICM are a fine polling organisation so I certainly am not inclined to simply dismiss this poll as a rogue. Besides which the general trend against the government is the one consistent thread running throughout all these polls. Oppositions don’t so much win elections as governments lose them and that appears to be the case here.

  20. As I’ve said, I don’t believe this poll. I think the Lib Dems might plausibly be at 18% from their most favourable pollster- but not higher. Not in the current climate. Cable doing well at PMQs is something only people like us will notice- not Joe Public.

  21. Although I think Vince Cable has done a very good job as acting leader I can’t see a reason for the shifts in this poll apart for the Labour one.

  22. Sally C – some papers, mainly the broadsheets, have long standing contracts with pollsters. So the Telegraph and the Sunday Times will always use YouGov, the Times will always use Populus, the Sunday Telegraph and the Guardian will always use ICM. The Independent will always use ComRes, though that relationship is less long standing.

    Other newspapers chop and change depending on who is available, so for the Sun (for example) you might get polls by YouGov or by MORI, the News of the World used to use Populus a lot, now they often seem to use ICM. MORI’s monthly poll doesn’t have a regular contract, so sometimes it is published in the FT, sometimes the Observer. It all changes.

  23. A look back at the polls in 1995 shows regular Labour leads of 25 to 40 points (a particular favourite from December ’95 shows the Conservatives on 23%, Labour on 62% and the Lib Dems on 12%). Whilst this may be a happy trip down memory lane for Labour supporters like me after an incredibly bad week for the government, were I a Conservative supporter I would be wondering why the Tory lead is still not in double digits at this point.

    Labour are still only four or five points below their winning percentage from the last election, and a few points above their low water-mark in the final days of Blair. They may fall to the high 20s, but the Tories really should be doing a lot better than 40% and single digit leads mid-term if they are to be cushioned against any subsequent Labour recovery.

  24. Thank you Anthony – again!

  25. Warren,
    The polls you refer to are not comparable – the methodology has changed since then.

    The best thing to do with that era is to take the ICM polls as reference. With the other pollsters of the time (gallup in particular), you’d be looking at Labour winning the 97 election by 20 points or more and pushing 50% overall (looking at the eve of poll predictions)

    With ICM, the Labour Party got to a pretty consistent 15-17 point lead in the mid-term and ended up winning the election by 12.5%.

  26. Warren, I understand what you mean but please don’t forget polls held from 1993-1997 were all precursors to the biggest landslide of any party since 1935 – I wouldn’t be surprised if there is no victory of this scale for another 50 years and so I wouldn’t expect the Conservatives to be that far ahead.

    It’s also worth remembering that in the last 18 months of the Callaghan government (1977-1979) the Conservatives often struggled to be the lead at all let alone be ahead in double figures, yet they managed a comfortable majority at the end of that period – after the unforseen “Winter of Discontent”.
    Who is to say what is round the corner?

  27. I agree that this poll may be a bit of an “outlier” and we should whether others in the near future confirm the trend. However, it would appear that the Lib Dems are on the up.

    I don’t want to make a point against a nice bloke like Ming Campbell, but Vince Cable’s good performance does indicate that an older politician – in historical terms one could say a more normally aged politician – can do well. That is an important point, because I am concerned that in falling over themselves to pick leaders who will do well in the polls, both opinion polls and actual elections, the parties are looking to candidates who may not have sufficent experience for a very senior executive role. I am sorry to say it, but Campbell’s poor, and specifically hesitant, image and performance may not have been so much due to age as to the aftereffects of his health issues. Why is it acceptable for a party to be concerned about negative impact of their leader’s age on the polls, when I doubt the relevance of age anyway, but not of their health?

    Perhaps the real issue is that the electorate are becoming more sophisticated about the effectiveness of leaders and, as Brown is finding out, are becoming ruthless at taking against any indication of incompetence. If this is so, is physical presentability, including age, as important for popularity as is assumed in political circles?

    Just a thought, but one reason for decreasing importance of presentability may be that physical appearance is particularly important for television. But now people are turning away from “the box” in favour of computers and the internet. And whilst of course we do look at images on the internet, there is a greater place for textual information, including indeed UKPollingReport. What politicans need, for instance, is favourable mention in the news snippets we read as we log onto email, and such snippets are driven by real achievement or otherwise, not what the politican looks like. If Vince Cable gets a good paragraph on the internet about Government incompetence over Qinetiq, it is no longer offset by his baldness.

  28. Andy Cooke – I’m not sure that the methods which were called into question hadn’t been substantially addressed by pollsters by 1995/6, but I will take Anthony’s guidance on this historical point! And no, Labour didn’t win in 1997 with anything like that, but I remember plenty didn’t bother actually voting because they knew we had it in the bag.

    Andy D – yes, events, particularly economic, may well still seal the government’s fate, but I’m struggling to believe that there are still “shy” Tories out there, or that as others have pointed out, the nearer Labour gets to it’s core vote the less need there is to adjust downwards Labour’s poll ratings based on the “likelihood to vote”.

    Whilst issue and leadership indicators look bad, even a pessimist like myself looks at 5-7% deficits mid-term after awful news stories and thinks, we might just claw that back.

  29. Generally speaking ICM were the only pollster to really reform their methods during the 1992-1997 Parliament. They switched to what is substantially their present day methodology (though it has evolved along the way) sometime in 1994 IIRC (though they recalculated their previous figures on their website from 1992-1994 using the new methodology, so we’ve got comparable ICM figures for most of that Parliament). If you look at the polls between 1992 and 1997 the reformed methodology of ICM consistently shows lower Labour leads than everyone else.

    MORI did amend their sampling technique to try and solve the problems of 1992, but they only made major changes that brought their final figures in line with ICM, Populus and so on in February 2003 when they added their likelihood to vote filter. Prior to that their polls continued to report larger Labour leads. The only companies polling in 1992 gradually disappeared from the scene one way or another to be replaced by new companies that used methodologies more like ICM’s.

    Andy is quite correct – if you want to look at trends in 1992-1997 look at ICM and ignore everyone else. ICM’s polls turned out to be far more accurate come 1997, and their methodology isstill broadly comparable to what they produce today.

  30. “That said, England’s failure to qualify will hurt Labour.”

    Why’s that?England comparative success in Euro 1996 did not help the Tory government.As for comparing polls from 92 to 97 Anthony has made this point several times, and again in this thread.Those 30% Labour leads would have been whittled down substantially using today’s, more accurate, methodolgy.That said, ICM were still producing Labour leads of around 15%+ ,comparing to the curent Tory lead of around 5%. My hunch is, based on nothing more than 28 years of political activisim, poll watching and gut feeling,that we have returned to the ‘usual’ situation of unpopular mid term governmemts, and the subsqequent poll ratings.

    All these polls show is it’s all to play for.The media pick up a story and bludgeon it to death until the next news story comes along.At some stage the Tories will be in the firing line and their poll ratings will suffer.