ICM April Poll

ICM’s April poll suggests a narrowing of the Tory lead. The headline figures with changes from ICM’s last poll are CON 37%(-4), LAB 30%(-1), LDEM 21%(+3). The Labour and Lib Dem figures are pretty much par for the course from ICM polls, and are similar to the levels of support ICM have been recording this year (though the level of support found for the Lib Dems tends to be more erratic than Labour), but the Tory vote is down slightly, having been at 40% or above in the last three ICM polls. The poll was conducted between April 20th and 22nd.

The hypothetical question naming party leaders and asking how respondents would vote with Brown as Labour leader as usual shows a larger Conservative lead, though not to the freakish degree of 15 points that was found last month and splashed across the media. Voting intention with Brown as leader stands at CON 40%, LAB 28%, LDEM 20%.

In comparisons with Gordon Brown, David Cameron’s position has strengthened (or perhaps more likely, Brown’s has weakened) – he has a one point deficit to Brown when respondents were asked who was most likely to make the right decisions when the going got tough compared to a seven point deficit last September; Cameron also has an 11 point lead over Brown as the man most likely to take Britain in the right direction compared to a 5 point lead in September. A large majority of respondents (78%) said they would like to see a contest when Tony Blair stood down as Labour leader.

Finally, 54% of respondents want to see a change of government after the election, with only 21% of people wanting Labour to remain in office. This shouldn’t be a huge surprise – remember the headline voting intention figures exclude don’t knows and won’t votes, so the 30% of voters supporting Labour is probably roughly equivalent to the 21% of people overall. This is however another suggestion that there is a public mood of “time for a change” – Labour must hope that the transition from Tony Blair to his successor, presumably Gordon Brown, will diffuse some of that desire or it will be a huge benefit to the Conservatives come the next election.


43 Responses to “ICM April Poll”

  1. “Time for a change” won’t just be a “huge benefit” to the Conservatives – it’s the political equivalent of Domestos: kills all long-term governments…dead.

    Therein lies the tragedy of David Miliband’s failure to challenge for the Labour leadership – because only someone of his generation (and he’s the only one with sufficient seniority to be able to run) could be the change from within.

    These polls seem to me odd in that neither Labour or the Conservatives have done anything to see such a slippage to the Lib Dems; nor has there to my knowledge been any of the oxygen of publicity that normally boosts the figures of the third party.

    I suspect that part of the issue here is that we are in a sort of twilight zone with the drift caused by a leadership vacuum. Politics is like wading through treacle at the moment, so no wonder that people are opting for the all-things-to-all-people party.

  2. Adam – since we’re in election time, the 3 parties now get equal media coverage, hence the oxygen of publicity for the Lib Dems you’re wondering about. In particular, each party is showing a Party Political Broadcast on TV most evenings, which has probably boosted the LD share.

    Anthony – thanks for the excellent analysis as usual! Slightly off-topic, but I was doorstep-polled by MORI this evening, and they didn’t prompt me by party name. I thought they were one of the companies that did prompt, or have I got that wrong?

  3. Not being funny Catherine, but are you really arguing that sufficient people watch, let alone are swayed by PPBs to account for a 3% increase in Lib Dem support?! That really would be a radical argument!

    As for the news – sure there is some coverage of the council elections going on at the moment but nothing comparable to general election campaigns and arguably less than in years gone by – and it’s the sustained blanket coverage forcing people to think about politics for more than two seconds that I suspect produces whatever bounce there might be for the Lib Dems.

  4. Perhaps not PPBs particularly, but Lib Dem election leaflets coming through the doors, canvassing, and just generally people thinking about who they might vote for next Thursday might be enough to push the LD vote up by 3%.

    I’d say twilight zone is a good definition of where we are. Polling now will all seem pretty irrelevant once we have a Prime Minister in place who intends to fight the next election. Not saying they’ll necessarily be hugely different, just that they’ll mean something!

  5. I’m currently running the following figures (though before factoring in this new ICM poll):

    Con 38.5 +6.1%
    Lab 30.2 -4.9%
    LDm 18.1 -2.1%

    Obviously the figures might change before the end of the month as I look in detail at the ICM poll.

    It seems to me that the Lib poll rating is currently the most volatile of the three main parties. Trying to work out reasons for that.

  6. This is interesting.

    It shows people are heartily sick of Labour and want a change in government, but aren’t yet convinced by the Conservatives.

    If you listen to people’s view on the streets, there isn’t any real enthusiasm for a Conservative government yet.

    If Cameron can crack this problem, he’s got it made.

  7. Catherine – that’s interesting. MORI do supposedly always prompt by party name these days, but currently they are integrating their fieldworkers with those from Ipsos. Perhaps that has involved some change to the scripts. Was it a computer-aided personal interview – i.e. did they have a laptop with them that they used to show you the questions and possible answers?

    (It also possible that they prompt by party name in their main monthly political surveys, but don’t prompt by party name in some other surveys and you got one of them. For example, in YouGov’s Telegraph polls we always prompt by party name, but IIRC when we did fieldwork for the British Election Survey we sometimes didn’t).

    It’s perfectly possible that all the leaflets through the door, canvassing and so on has given the Lib Dems a boost. They seem to get far more limited media coverage these days than under Charlie Kennedy, so a sudden boost in personal contact from the Liberal Democrats might well remind people they exist and increase their level of support. That said, I’m not sure there has actually been one – ICM had a poll for the Sunday Mirror last month that also had them at 21% and had them at 23% in January. I think, as Peter O suggests, that ICM are just showing more volatile figures for them than the other two parties. There’s no comparable Lib Dem advance in YouGov and Populus (though they are both due soon, so there’s time!) and Communicate we don’t really know anough about how they are weighting yet to draw a conclusion.

  8. I have read all of your comments and have to say I believe that there is a element of all the factors brought up by all concerned. Yes there is added publicity not just from PPBs on LD policies these are helped by the comparisoms made with the other main parties, Conservatives going greener like the new logo morphing from blue to green with the oak tree. I agree that Chales Kennedy kept the LDs in the spot light better than Ming but that was his style of leadership.

    Is the volitility in the LD share due to policy hijacking as the conservatives and may be labour have taken on environmental issues and raised them as campaign issues. I live in a rural district council area which is one of the poorest performer on green issues and in the recent election manifestos they are trying to say that they are performing well especially as they have just increased thier recycling by finally starting to implement new Kerbside collections with segregated waste a policy most other Borough and District councils have been working on for a couple of years. The County Council is also trying to beef up its comitment to recycling by starting a new system of waste management at it refuse tips.

    So the poor green libdem of old now has a mountain of issues raided by the other parties.

  9. Anthony,

    You said YouGov’s Telegraph polls always prompt by party name. What about the Scottish polls, and if so, which many parties are prompted?

  10. Anthony, OK, let’s test the theory that election activity rather than specific media coverage of elections is the reason the Lib Dems have bounced up (again). If that’s true, given that there are elections every year somewhere (admittedly not on the scale of this year’s set) there should be a record of Lib Dem upturns at this stage of every year, shouldn’t there?

    Or, if the argument is that this round of elections, because of their scale, is the significant factor then there should have been similar upturns in April 2003, 1999, 1995, 1991 etc, shouldn’t there? Don’t think you’ll find evidence to sustain either theory.

  11. Thanks, Anthony. He did have a laptop with him, but only started showing me the screen about half-way through the poll, so maybe he just forgot to show it to me earlier or something.

    There was a fair amount of non-political stuff about phone + internet companies etc, similar to YouGov’s brand index thing. But it started off with all the political questions that I’ve come to expect after reading this site for a while :) i.e. how did you vote last time and likelihood to vote on a scale of 1-10.

    There was also a question on satisfaction with party leaders, but no hypothetical voting intention with Brown as leader.

  12. I would guess that most Labour supporters would be pretty happy with this poll, all things considered.

    For a government to be only 7% behind after 10 years in office, and for the main opposition party to be on only 37%, isn’t too bad from their perspective.

  13. Andy Stidwill. Yes it could be worse, BUT as readers of this blog well know, modern polls are not like the 1980’s/1990’s ones where leads were massively overstated. 27-30% in a modern poll is pretty dire for a party in office. 36-37% for the opposition is just as good as a much higher figure in the older polls. Remember the rise of the “others”. 37% is not 37% of 100, it’s 37% of 90 (=41).

  14. Well the Weighted Moving Average is 38:32:18 and the lead is 7.6% – pretty stable. It was admittedly 8.2 last month, but Communicate has a very high standard deviation (3.8 on WMA, 3.6 historic) so really not much weight should be attached to a small movement.

    Andy: don’t kid yourself.

  15. NBeale:
    I think the government should be at least 10% behind on the WMA after everything that’s happened: cash for honours, Iraq war, pensions fiasco, 10 years in office, leadership wrangling, etc.

    I’m not a Labour supporter myself, but things certainly aren’t as bad as they should be for this government.

  16. Adam – personally I don’t think there has been a Lib Dem increase anyway (though I’m perfectly happy to be proved wrong if YouGov and Populus do show the same trend as ICM), but if there has been the local elections seem a plausible factor given the party’s low profile recently. I wouldn’t have expected a boost in the 2003 poll because the Lib Dems were getting lots of publicity in the mainstream media then anyway and campaigning wouldn’t necessarily have boosted that, now they seem to be getting much more scant coverage.

    Greenpousse – because they are done online YouGov polls nearly always prompt by party name, its a lot easier than analysing answers that are typed in! In the most recent Scottish polls the constituency section was prompted with Con, Lab, LD, SNP and Other (the text of the question specifically pointed out to people that the Greens, SSP and Solidarity were not putting up constituency candidates). The regional vote prompted with Con, Lab, LD, SNP and other, and then prompted people who selected other with the SSP, Solidarity, Greens, Senior Citizens Unity or ‘other other’

  17. Andy: ‘I think the government should be at least 10% behind on the WMA after everything that’s happened’

    As all we have to go on to judge where parties ‘should be’ are old poll results that are as reliable as First Great Western isn’t a comparison pointless?

  18. Whether it’s pointless or not and without referring to anyone in particular, I would not welcome the sort of thing you get on Mike Smithson’s blog sometimes when every new poll is greeted by supporters of all three parties swearing blind that it is actually a good poll for them and awful for the other two. I don’t normally get it here, but before it sneaks in it’s not welcome! Its very tiresome, very annoying and quite stunningly transparent :)

  19. Andy,

    On the question of prompting for party – when one enters the polling booth, one is prompted by party name on the ballot paper (it is the law after all!)

    So, why would prompting in a opinion poll be considered flawed? Indeed could one not argue that prompting is a better method because it does what a ballot paper does, I.e. remind the voter of the options?

  20. I Loved reading the arguments on party naming and say that this practice can cause fluctuations which are not always true as a supporter for one party may skew the results by purposefully stating the opposite to thier voting intention and one they get to the ballot box they vote either how they traditionally would or tactically crating possibly some of the effects we saw in the eighties. I must say though as a Politics and History graduate the swings of the last few decades of the last century in public opinion polls.

  21. Anthony,

    Without the aid of Deep Thought do you think there is a way of determining how far out older polls were so comparisons can have a point? The pollsters after 1992 must have looked into it, did they publish anything?

  22. Is there any evidence that the mid-term polls in the eighties/nineties were actually wrong?

    Yes, obviously, the polls in the mid-term did not reflect the result in the election – but could that not be explained more by a party/event-specific recovery by the government/failure by the opposition rather than some sort of systematic failure in the polls?

  23. There can never be evidence that mid-term polls are wrong – there is no reality check to compare them to. We do know that the polls were sytemically wrong come the 1992 election though and that most of them (with the honourable exception of ICM) were wrong in 1997, so it seems reasonable to assume they were also wrong in the mid terms before the elections since the methodological failings weren’t suddenly adopted in the election campaigns.

    Ralph – there is no simple formula you can plug into old polls and say, add x to party 1 and subtract y from party 2 and you get what you would have got using the new methodologies. Different methodologies favour different parties at different times – for example, the adjustment that ICM pioneered in the early 90s to account for the spiral of silence used to favour the Tories and decrease Labour leads, now it favours Labour and reduces Conservative leads. Since ICM did go back and re-adjust their polls from 1993-1994 to show the figures they would have produced under their updated methodology, you can compare the sort of figures ICM found in the 1992-1997 parliament with the sort of figures they get now.

  24. Simon – all the pollsters now prompt by party name, so I think the argument is won :) The main difference is that not prompting my party name normally produced a Lib dem figure lower than polls that did prompt, normally by about 2%.

  25. I don’t understand the argument being made that the huge Labour leads in the opinion polls during the 1990s were not an accurate portrayal of the state of the parties at the time they were recorded.

    To follow on from this, there seems to be an argument which says that the fact that the Tory lead at present is only in single figures cannot be properly compared to those large Labour leads a decade ago. I don’t see why they cannot be so compared.

  26. Are these polls asking about actual voting intentions in the local elections or a theoretical “GE tommorow” scenario?

  27. Andy, Anthony’s argument is that the polls during the 1990s did not accurately calculate the spiral of silence (people ashamed to admit they were supporters of the very unpopular Tories) which over-inflated Labour’s lead and, consequentially made the Tories appear even less popular and therefore Conservatives even less likely to volunteer their allegiance (hence the ‘spiral’).

    It is true that Labour’s real vote leads in council elections during the 1990s, while substantial (in 1995, their best ever year, Labour was something like 18% ahead), never mirrored their opinion poll leads.

    But that assumes that peoples’ council elections voting intention was the same as their national intention – in fact Labour usually underperforms in council elections compared to national elections (both because they find it much harder to turn out their vote for non-national elections and because of the damage the ‘loony-left’ tag did to Labour in local government).

    It could be argued that some of the spectacular national by-election results that Labour achieved during the 1990s was evidence that their national opinion poll lead was bona fide, but equally the evidence was that the huge swings were simply the earliest indications of the scale of tactical voting away from the Tories (hence the Lib Dems also smashed the Tories while Labour’s vote usually – though not always – collapsed in seats they were in third) that was seen at the 1997 general election.

    Where I challenge Anthony is to produce a solid argument to support his assertion that the spiral of silence is now affecting Labour. I maintain that the spiral of silence will always impact the Conservatives to a varying extent because right-of-centre parties are (fairly or not) seen as efficient but unlovable; whereas left-of-centre parties are seen as cosy but more wasteful – of the two characteristics, people will always prefer to be identified with cosy rather than unlovable.

    An interesting consideration – that links back to my points above about the bounce of the Lib Dems – is that it’s entirely plausible for both Labour and the Tories to currently be suffering a spiral of silence – and that it is this that is inflating the Lib Dem vote. After all, Anthony’s argument that the spiral of silence is now affecting Labour is (presumably) based on the fact that the government’s unpopular and the fatigue factor; coupled with my argument that the Tories will always suffer an element of spiral of silence and that there may well be some residual anti-Tory feeling left over from their last period in office in some quarters helps explain the surge in support for minor parties and the Lib Dems.

    And that theory IS actually borne out by council election results.

  28. Adam – yep. There could be both reluctance for people to admit voting Tory and reluctance for people to admit voting Labour. The net result of ICM and Populus’s adjustment these days is normally to boost Labour, but that doesn’t mean there isn’t a smaller boost to the Conservatives from it. (Just to add – I entirely agree with your theory that there may be reluctance to admit voting Tory as being left wing is seen as more altruistic. People who refuse on principle to tell people their voting intention are interesting too, obviously no one knows for sure how they vote, but demographically they look most likely to be Tory than anything else.)

  29. Keith – these are all “general election tomorrow” polls, AFAIAA there haven’t been any polls on local election voting intention, and to be honest, there aren’t normally any.

    Andy – at a simple level, in 1992 the polls overestimated the Labour lead by around 8 points, so we know that polls around that period were wrong. Unfortunately, we can’t just deduct 8 points from the type of lead Labour got then and say that was the real lead because the changes were far more complex that that.

    There are 5 main relevent changes between the methodologies back then, and the methodologies now.

    1) Spiral of silence adjustment. One of the reasons the polls were wrong in 1992 was that people were embarrassed to admit to voting Conservative. ICM and Populus now adjust their polls to account for this, YouGov’s mode of surveying without an interviewer should reduce it in their polls. Essentially this should reduce the lead of the most popular party by increasing the reported support for less popular parties, though it doesn’t always work that way. Around 2001 ICM and Populus’s adjustment was favouring Labour as the party people were now embarrassed about supporting, even though Labour were still in the lead.

    2) Sampling – the quota samples and the way fieldwork was conducted in the late 1980s and early 1990s no longer provided representative samples. Samples contained too many Labour inclined people to be representative, partly because quotas were out of date, partly because of problems with giving interviewers too much leeway with where to interview (and it’s quicker to knock on terraces than big houses with long drives!). Since then most pollsters have switched to phone polling and/or…

    3) Political weighting. To try and get samples that are politically representative most pollster now weight by past vote or similar, making samples more consistent and less biased to Labour.

    4) Prompting by party name. These days everyone does it, back then they didn’t. Referring to the three main parties by name in the question increases the level of Lib Dem support found, normally at the slight expense of Labour.

    5) Likelihood of voting. If you look back at the mid-term MORI polls in the 1980s (and they are the most readily available) they don’t take into account people’s likelihood of voting. Back then it didn’t make so much difference of course, because turnout was higher, but this again tends to favour Labour as Labour voters are less likely to actually vote than Tory ones.

    Essentially, there were five differences in methodology back then that probably all increased the size of reported Labour leads.

    (That said, the Labour leads shown by ICM post-1992 are after ICM amended their methodology and should be comparable to today’s polls).

  30. One very interesting question, to which we will get a partial answer in a week, is how the spiral of silence affects the BNP. A very limited number of seats at the moment affected, but well worth watching.

  31. People are generally very reluctant to admit to pollsters that they support an extremist political party. You can see the general effect by comparing the level of BNP support in YouGov polls, where there is obviously no interviewer, and ICM/Populus and MORI polls where respondents have to give their answer to another human being.

    It’s a far more serious problem for pollsters in France where the Front Nationale are actually a serious force – there the polls seriously underestimated support for Le Pen at the last Presidential elections, thought this time round the pollsters seem to have over-adjusted in his favour and all overestimated his eventual level of support.

  32. The BNP spiral of silence is indeed interesting – because if you look at somewhere like Barking & Dagenham, it was very much diminished last year: many more people than normal were quite openly acknowledging their intent to vote for the party.

    That’s in part because the BNP was pretty much the only opposition to Labour throughout much of the borough – so a large share of their vote was protest at both Labour and the other parties for not standing sufficient candidates themselves.

    But it also suggests that there is a “critical mass” issue for parties like the BNP – when it stops being a stigma to announce your support and starts becoming almost a badge of pride because you know you’re not isolated in your voting intention.

  33. Maybe, as well as the “spiral of silence” there is a group of people who have withdrawn from voting for a particular party because they’re basically fed up with them, but don’t like any other party enough to vote for them (ie, abstainers)

  34. Adam. It is getting like that in parts of my neck of the woods – Bradford Met District. Watch the Bradord South council seats next week.

  35. John, this is going off on a tangent, but it’s also interesting to consider the different motivation for voting BNP in an area like Bradford against Barking.

    It’s just that in Bradford, which (as you know) has a very large Muslim community, the BNP support is basically about race relations, multicultural polarisation and the interaction – or lack thereof – between the communities.

    Don’t know what the perception of Barking & Dagenham is outside London, but in fact that borough actually has a very small BME community: it’s overwhelmingly white and the principal minority community is afro-caribbean.

    I think the motivation in both cases is fear of change; but in Bradford it’s change that’s already happened – in B&D it’s potential or imagined change. However, of those two scenarios, I’d have thought that the BNP vote has a far greater chance of being entrenched – and people being “proud to be BNP” where there is a large BME community.

    To me, the fact that B&D has such a large BNP presence is almost accidental: far more about the absence of mainstream party competition than an inherent race problem there (though Jon Cruddas, who represents Dagenham and has been outspoken on this issue, may well disagree).

  36. The level of discusion on here over the past couple of days seems very interesting. Congrats!

    Maybe I can therefore ask a question relating to the dilemma that our ‘national’ polls have in comparrison with if we were to have just English polls. Take for example the recent average polls where nationally the split is 38-C, 31-La, 20-Li. If you were to put this into any of the ‘seats calculators’ they all will suggest a Tory largest party but short of a majority [305], while labour is down only a moderate amount [270], and Liberals mwn a lot [40]. Yet my point is, that we know that in Scotland, the polls are very different with Labour above this av, and Lib / Cons well below theirs. As a result, this will make the average for the cons and Liberals lower, by 1.25 in my calculation.
    Whether these figures are right or not, my point I am trying to make is that where as a 10% point gain/loss in Scotland makes only a small difference to the number of Cons or Lib seats at a general election. However, if we were to adjust both these parties up by 1% [after removing the drag of the scottish polls] so that it is applied just to England, then this 15 can shift the number of cons by dozens of seats! So surely all these seat calculations everybody does [including the learned contributors to this site] are understating the Liberals and Cons- I have not calculated how much the sensitivities are.

  37. Actually there is quite a large non-white population in Barking these days; as of the last census only 46% of the Abbey ward is White/British.

  38. re my previous question as to whether these were general election polls or not, I was wondering whether we ought to take GE-related opinion polls or actual voting percentages in real local elections as a pointer to which way things are going? Or average polls as above(?)

  39. Keith; I’m just not sure what you’re asking – because it’s apples and pears. Opinion polls tell us how people would vote in a general election held tomorrow; local elections tell us how people vote in local elections.

    You may have seen tonight’s Newsnight which forecast next week’s council election results as Con 38%; LD 29%, Lab 24%. That’s by no means an outlandish prediction (though the very small number of council by-elections the forecast is based on makes it more unreliable than usual) but it’s not a vaguely likely general election outcome.

    So it depends what you’re looking for: if you want “a pointer” to the way people are feeling about national politics look to opinion polls (or indeed, polls-of-polls); if you want a pointer to party strengths in council elections look at council by-elections or all-out elections; while always bearing in mind the types of areas the elections are being held in.

  40. Adam. Sorry. slightly off the topic again (but of interest vis-a-viz how pollsters treat “other”) you are spot on about “fear”. BNP strength is not in “mixed” wards, but in white wards whose voters fear, rightly or wrongly, that they are next to be “swamped” as they put it. That is why I said watch Bradford South wards. Mostly white, but with a large south asian Muslim population in nearby wards. I’m sorry if this seems very “colour conscious” but too many dodge the issue and are not objective. That is folly.

  41. Two separate polls out for WA – NOP/ITV Wales and Beaufort/Western Mail – both saying basically same thing – Labour on 32/35%; Plaid on 26/27%; Conservatives on 19/20%; LDs on 16/17%.

    Both polls were sampled before latest row about coalitions which can only have damaged Labour further. Looks like being a very interesting week ahead.

  42. John, yes, absolutely. Don’t know if you have any experience with MOSAIC type consumer-branding programmes that parties increasingly use for targeting, but one of the groups they describe (and the naming is both poncey and may be different now) is “bohemian melting pot” – typically young professionals, mainly renting but occasionally first time buyers who actively opt to live in diverse, multicultural communities. Of course, my more cynical outlook is that these areas are also much more affordable and this may have at least as much to do with their residence, but that’s beside the point…

    Brent East, for example, is swarming with this demographic; and while it’s probably very much a London-oriented definition, I suspect the “mixed” wards in Bradford you mentioned probably also fit this description.

    The point being: that there isn’t a high BNP vote in “mixed” wards isn’t coincidental: if this MOSAIC profile is correct then the white voters here would be actively opposed to segregationist and confrontational platforms, whereas the BME communities obviously get nothing from supporting the BNP.

    On Alun’s point about Barking earlier, yes, I’m not saying that there isn’t a BME community in Barking (especially) – in fact in the wards like Abbey that border Newham and Ilford it would be amazing if the high BME percentages in those two areas stopped at the borough boundary.

    But equally, the BNP didn’t stand in wards like Abbey – they were concentrated in the much whiter centre and east of the borough – towards Romford; it’s here that the fear factor of “being swamped” is both strongest, and of least foundation in reality. And as I’ve argued above, the BNP needs both the fear of a problem and the lack of it actually being one to thrive.

    It does go beyond fear of different ethnic groups though, and Jon Cruddas has been right to identify lack of affordable housing in particular as another critical ingredient in the BNP’s ability to present white communities as being neglected by the government/council.

  43. Adam Don’t want to be boring, but the “mixed” wards in inner Bradford are what I’m sure you would describe as “poor”. People live there in great density in mainly older houses in fairish condition. They are trying to attract the type of young professional you mention to loft type converions – don’t know how successful they will be. Many of the Europeans still there are older people – maybe they are more “resigned” than those in the “white” wards. Anyway, enough of my socialogical theories.