Or at least, it would were there an election called next week, which Gordon Brown is expected to confirm any second now that there will not be.

ICM’s News of the World poll in 83 Con/Lab marginal seats – the 49 most marginal Labour seats with the Conservatives in second and the 34 most marginal Conservative seats with Labour in second place – shows the Conservatives on 44% and Labour on 38%. According to the News of the World this would result in Labour losing 49 seats if replicated at a general election, wiping out their majority.

The poll didn’t include any Liberal Democrat marginals, so we can’t project these findings onto the whole House of Commons, but depending on how the Liberal Democrat marginals fell Labour would be left with something around 306 according to the News of the World. That implies the poll was based on the old boundaries, not the new boundaries that a new election would actually be fought upon (The number of seats Labour won 355 minus 49 = 206; 348, the number of seats Rallings and Thrasher have Labour winning minus 49 doesn’t), though that could just be the News of the World subtracting it from the wrong figures.

Notably the fieldwork was done between the 2nd and 5th, so some of this poll would have been carried out prior to David Cameron’s speech.

UPDATE: Peter Kellner has just confirmed on Sky News that the YouGov poll in tomorrow’s Sunday Times will show the Conservatives continuing to advance. I’ll update once the figures for that are available.

42 Responses to “ICM poll in marginals points to a hung Parliament”

  1. It seems to me that the seat numbers quoted are innaccurate. Basically the poll is saying that Labour would lose all 49 seats. Presumably they would lose a whole load of others that haven’t been polled (every seat had a majority of less than 4% ,but the Tories had a 6% lead). So the Labour total of 306 is a maximum, and the Tory total a minimum.

  2. It looks like the poll has been done on the basis of old boundaries, as well.

  3. Scrap part of what i said. Of course the 6% lead will include the Tory seats.

  4. Very interesting indeed.
    Maybe the swings at the next election will be largest in marginal seats (particularly in the south) – a similar scenario in the 1992 and 1997 general elections. A uniform swing in 1992 would have given the Tories a majority of over 70 – in the event it was 21.
    If this research was done on the basis of the older boundaries, then that is probably even worse news for Labour as the new boundaries are slightly less favourable towards them.
    I would be very interested to see opinion polls broken down into regions.

  5. As I said in the previous post, I have a suspicion that most of the Brown Bounce since June has come mostly and superfluously from Labour’s heartlands, ie. in seats they held in the 1980s?

  6. Richard – it says they all needed a swing of less than 4%, so a majority of less than 8%.

  7. Well I’ve Done It !! My predictions – though ridiculed for 3 months on here have come true – there is a Tory majority of 6% by the end of the first week of October and Brown will not call an election !!

    So all the ticking clocks can stop now .

    Taken at it’s base level – thew POLL suggests a hung parliament – but as written above , only working on the marginals and a low Liberal vote – neither of which gives an accurate figure – as the Liberals will poll a figure more like 16 to 17% in an election .

    44% is an amazing figure – not seen in the whole of David Cameron’s leadership .

    ANTHONY :-

    Will you be updating the graph soon to show these figures ?

  8. Ah right. So the key thing is to know what the base line was.

    But you agree the seat figures are clearly misleading?

  9. Mike – it is a poll only of marginal seats, so it doesn’t go on the graph of GB voting intention. This doesn’t mean that the Conservatives are 6 points ahead overall, only in the subset of seats sampled. You’ll have to wait for a Tory lead in a national poll (though from Peter Kellner’s comments, maybe not for long).

  10. No, Mike, you said they would lead at the end of September and were totally wrong so you changed your prediction to October.

    Even this poll is a poll of marginals not a national poll so you are stil not right anyway.

    That said, it has been a very bad day for Brown and a good one for the Tories who were desperate to avoid an election.

    We will now have to wait two years for an election and anything could happen before then, even a Tory victory.

  11. Also, T Jones has a lot to answer for, for saying that he had news that the Sunday polls would show a boost for Labour. That gave me false hope for a day!

    It will be fascinating to see what the polls say on or around November 1. I have a fear that in not calling an election Brown is repeating the same mistake as Callaghan in 1978. I think he should have backed his chances of blowing holes through the Tory tax policies (sic) in the white heat of a 3 week election campaign rather than filling his pants at the sight of a few bad polls after the Tory conference. If he had called an election after this ICM poll, it would have been a bold move and one that would have been respected by the public. Instead he must take the pain of a deserved beating from his opponents and the media.

    While this will have some lasting damage to Brown’s image as the resolute and decisive strong man, I do not believe it will fatally damage him in the way that certain overexcitable pundits may believe. In May 2009 this episode will be a distant memory and other “events” will have happened by then, some predictable, some not.

    Sadly it looks like we are now back to five polls per month

  12. My calculations show that the vote share in these 83 seats at the last election were Con 38.5 Lab 39.6 LibDem 16.2 Others 5.7 so Con + 5.5 Lab -1.6 LibDems – ?

  13. No Arnie I didn’t say that,I said initial findings did,however they were subject to change,go back and have a look.From now on I will keep my information to myself and then comment on them when you all eventually have the information.

    By the way unlike Mike R,mine isn’t just guesswork.If it was it was I very lucky in getting the C4 figures exactly right 20 minutes before they were broadcast and put on the internet.

    Keep up the good work!

  14. Arnie-I would also like to point out it wasn’t this poll I was actually refering to.


  15. YouGov is 41 38 11

    Cameron personal ratings up 19 to 54

  16. AndyD,

    I think it’s very possible that Labour’s ‘bounce’, largely, came within it’s heartlands. I’ve always maintained that it was difficult to conceive Labour not losing seats to the Conservatives irrespective of what the polls were telling us.

  17. THe lesson from all this frenetic polling activity is not to even think of having elections during the conference season. There is also a lesson for those who predict results based on uniform swings and ignore differing attitudes and predelictions in different parts of the country.
    When the psephologists’ frenzied discussions subside I think that Brown and entourage will reflect on the fact that they have flushed out Conservative policies – some of which are distinctly questionable. We shall see more of the Brown/Cameron duelling and learn how substantial a politician Cameron may prove to be – or otherwise – when pitted against a PM of long successful experience in a major parliamentary role.

  18. Dave,

    Or the sheep vote, ‘Oh Labour’s doing well I’ll say I’m going to vote for them’.

  19. Richard,

    “Cameron personal ratings up 19 to 54”.

    Musn’t be able to see the, proverbial, pig for the lipstick! He’s not fooling me.

    Is that Con 41, Lab 38 and LD’s 11?

  20. Truly amazing stuff.
    TV coverage riveting.
    Brown torn to shreds by Adam Bolton & Michael Portillo.

    Thank you Neil Kinnock-time to fall in the sea again now!
    Thank you Ed-aka Balls Up
    Thank you Andrew-aka Alexander The Great.

    Can’t wait to see the Marr interview tomorrow-bet he asks lots of tough, searching, fearless questions (not!)

  21. The notion that Brown has backed down and the Tories are ahead could have a distinct effect in Scotland.

    The treasury has also apparently agreed to let the SNP government have the £900m of unspent money to spend over the next three years so all in all it’s good for us.


  22. With regards the above poll ,when you take away all the excitement of the findings the differences from the notional 2005 result is are such:

    Conservatives – 44% – +5%
    Labour – 38% – -1%
    Liberals – 16% – -2%
    Other – 4% – -2%

    So there has been actually no big move from Labour to the Conservatives but more from the Liberals and others,which begs the question would these people then revert to the tactical voting they have done in previous elections when the General election comes.

  23. Times:

    Conservatives 41%
    Labour 38%

  24. Looks like it’s officially off. Not surprising with polls like these and winter weather closing in. Try again next spring and see what happens, eh?

  25. Hmmm I was just sample giving figures to illustrate my comment in a previous article (http://ukpollingreport.co.uk/blog/archives/1046) without trying to be accurate at all… but I was pretty close there, hey! ah ah ah looks even more confirming my words about the craziness off such voter swings in such a short period of time.

    PS: Gary Gatter, what was your comment about me jumping the gun? ;-)

  26. *******Just read the small print of the Marginals Poll.It was based on a telephone poll where the first 1026 people who answered the phone where questioned.Obviously ex-directory and people who didn’t answer do not count.Which means on average 12 people per the 83 of the marginal constutencies where asked.


  27. T Jones, your percentage don’t add up to 100 !

    44 + 38 + 16 + 4 equals 102
    Oh how sad am I ?

  28. These are quite remarkable polling results. Out of the 80-odd marginals studies it was found that the Tory lead of 6% would reveal a 49 seat gain. As they were done on the old boundaries you could add 15 or so seats to that total.

    The poll was not looking at the LD/Con marginals too, and it is a fair bet that the 30 or so seats the Tories are looking at taking would return Conservative gains, this total could be boosted further.

    Add into this equation the great unknown; how may fairly comfortable Labour seats could switch Tory? Perhaps the party could regain 15-20 seats it lost in ’97 from Labour.

    All in all the Tory total could edge towards the 300 seat marker. That would see the be the largest parliamentary party and represent a total failure of Gordon Brown’s prime ministerial career.

  29. Sorry,must have felt sorry for the Liberals.They should read 14% which is down 2%.

    Anything to help them along.

  30. TJones – phone polls are conducted by randomly selecting phone numbers, then randomising he last digit of the number – this ensures ex-directory people are included. People who do not answer the phone are rung back up to 8 times to try and minimise non-contact bias.

    You’ll get 12 people or so per marginal seat, but this poll does NOT make individual calls per seat. It is treating all these marginals as one big group, and assuming there is a uniform swing across them all.

  31. For someone who allegedly has loads of inside (but incorrect!) information on opinion poll results days before they are published, T Jones appears to show a worrying lack of understanding about how polling works…..strange that!

  32. Anthony – So to say it’s a 6% lead in the said marginals really means it’s a 6% lead for people who actually answered the phone in these marginals.

    So there are no weighting in these types of polls.So for instance if there was a freak choice of numbers,they could be ringing up say a large proportion of Lib Dem voters which could make interesting reading,but not actually a true estimate?

  33. They will be weighted in the same way as a normal poll- but to the collective profile of the marginal seats being polled, rather than the country as a whole.

  34. Would it be a little bit forward to suggest that TJones has been well and truly rumbled? ;)

  35. Yes Richard – I am Really Tony Blair.

  36. Richard – surely not!!

  37. David and Richard – You cads!I am in stitches here.

  38. Can I just ask: is it possible to oblige ICM under the Market Research Code of Conduct to publish their detailed poll findings by constituency – they must have these figures, surely?

    And do we know what the overall sample was? Appreciate it makes the individual constituency results subject to a large margin of error unless the overall sample was close to 100,000 (!) but it would still be interesting to see.

  39. Adam, this was not an aggregated poll made up of polls of each constituency any more than a poll of GB is a poll of each town and village. The poll was a poll of the constituencies as a group and there was a standard 1k people polled in total as there would be in a GB poll. Just that the people polled were from and representative of the marginals and not of all constituencies.

  40. Adam – what Philip says. On top of that, no, ICM wouldn’t have to show what the figures were for individual constituencies unless they’d published voting intentions for individual constituencies.

  41. Anthony has already corrected T Jones.
    Random digit dialling is used.

  42. There was one poll of marginals around 18 months ago Populus ?? which gave individual constituency figures . They were of course worthless giving LibDems 90% in 1 constituency and 20% in another .