YouGov London poll

I’m a little late with this, but just to flag up that yesterday’s Times had a new YouGov poll of London. Topline voting intentions there are CON 34%, LAB 42%, LDEM 8%, UKIP 9%, GRN 6% (tabs here).

I sometimes hear a sort of London as the weathervane of the country argument, as London goes so the country goes. It doesn’t, despite a Tory mayor it’s Labour leaning compared to the country as whole, and it doesn’t necessarily swing in the same way as the rest of the country either. In 2010 while the Conservatives were seven points ahead in the country as a whole, Labour won in London by 2 points, and the Conservatives managed a smaller swing in London than elsewhere.

This poll suggests a 3 point swing from Con to Lab in London, the equivalent of a 1 point Conservative lead in national polling, so actually a little less than national polls are currently showing. It could be the Tories are doing a little better in London, or it could just be ordinary sample error – as ever, it’s just one poll and shouldn’t be overinterpreted. A three point swing in London would net Labour three or four gains from the Tories – Hendon, Brentford and Isleworth, Enfield North and, right on a knife edge, Croydon Central.

396 Responses to “YouGov London poll”

1 4 5 6 7 8
  1. @ Little Red Rock

    I’ll try to answer your questions about swingback that is sometimes introduced after polls are published, leaving others to comment on the effects of ‘don’t-know-resllocations’ before a poll is even published.

    And, if there is less time for swingback to occur might we expect less swingback to be built into projection models?

    Yes – swingback/reversion-to-mean reduces as the election approaches. Stephen Fisher’s model should now be using historical swingback figures for about 10 weeks before previous elections. As we approach the election the amount of historical swingback gets smaller and smaller. Eventually, on the eve of May 7 swingback reduced to zero and his model effectively becomes a Nowcast.

    In the EF model the weighting assigned to the regression-to-mean component is reduced steadily towards zero as the election approaches.

    If that is right (someone please tell me if it is or isn’t) Then, without a single person changing their mind we ought to see a slight Labour increase at the expense of Lib Dem but a decrease against SNP. And a slight Conservative decrease to the benefit of UKIP.

    Correct. That is exactly what I expect to see (unless real swingback comes riding to the rescue). Predicated upon the non-appearance of swingback, the EF model will row back from its current calculations of a 4-5 point Ukip drop, a smaller LD gain and so on. If and when this happens it will look as if they are suddenly getting more optimistic about the prospects of Labour, Ukip etc. but what will actually be happening is that they will just be switching off there current calculated pessimism.

  2. I think it would only be correct to say the Guardian’s model is biased towards the Lib Dems in Conservative vs Lib Dem marginals.

    In contests vs Labour and SNP the model appears to be similar to other models and/or follow Constituency data.

    If you take constituency polling like that in Eastleigh, Carshalton, Colchester and project it to other Lib Dem vs Con seats with largish leads and popular incumbents then I don’t belive the numbers they are projecting are wildly inaccurate.

    Without seeing their numbers it’s difficult to make judgments but I don’t think it’s impossible that Lib Dems could do better in incumbent fights vs Cons than national polls suggest at present.

    The “hold your nose” Labour voter who’ll support Lib Dem vs Con is difficult to identify in polling.

    The big question is will those people still hold their nose after 5 years of Coalition government?

    Answer – maybe – if the election looks close and Lib Dems make hints about doing deals with Labour…

  3. Couper,

    I think a 30 seat drop to around 27 LD seats is a fair judgement for LD retentions based on specific seats.

    Most of the contrary to UNS holds will be in Tory targets due to ABTs holding their noses. Not all bit enough in many seats.

  4. Mike N

    Do all pollsters currently include a propensity to vote factor? Or, do they, like YouGov, include it in the month or so before the GE?

    I think YouGov are the only ones who don’t. Though I would expect them to start doing so very soon (as they did with the Euros). Whether they all get them right or not is another matter – last night’s Survation looks a bit odd, though that may be due to weighting back to sample size.

    Whether it makes a difference or not is something else. It always used to said that Conservative voters were keener than Labour ones, but most polls seem to have them showing similar LTV profiles at the moment.

  5. Roger Mexico

    It would have been “the Polis”, but the Swedes claimed copyright on the name.


    The Lib/Dems as someone said further up are opportunists. The answers Danny Alexander gave on Marr were absolute waffle.

    My ole papa in the deep south siad…..”There’s a tree stump in a Louisiana swamp with a higher IQ and boy If leather were brains, he wouldn’t have enough to saddle a junebug”

    Populus cross break…




    LIB/DEM..10%…………Far too high!!

  7. JASPER22
    Good piece in The Times this morning concerning Police Scotland (an awful name).
    Since the merger into one police force covering the whole of Scotland in 2013, response times and police morale has, apparently, declined.

    As that article is behind the Murdoch paywall beyond which many of us will not venture, perhaps you’d enlighten us on what it says on cost and co-ordination, which were just two of the reasons put forward for the change.

    Surprised AW hasn’t applied the snippers to much of the rest of your comment.

    Re your as if the location of a building makes any difference!!, I don’t think anyone is claiming the building is important. IMO all parliamentary business outside the Arctic Circle would be better conducted in the open air, if only to keep the debates short and on topic. What’s relevant is who makes the decisions for any particular nation.

  8. Polis?
    Not one person from the north east of Scotland has ever used the word polis to describe the police. Such usage is anathema in this region which one writer has describes as the Catalonia of Scotland.

  9. @ Bill Patrick

    It’s reported that Populous are phoning people in Ross Skye & Lochaber this week so the assumption is it’s an Ashcroft Poll (hopefully to be released next Thursday).

    Other Scottish Ashcroft Polls in the field rumoured to be Kirkcaldy & Cowdenbeath and East Renfrewshire.

    Not clear yet whether these are part of a wider batch of Scottish polls or a few Scottish polls as part of a UK wide spread ahead of his polling seminar on 4th March:

  10. @JP

    “Looking at your previous posts I suspect your query is an example of looking for reassuring data to fit pre-existing bias but I think you will probably find the findings disappointing.”

    Steady on, that’s a bit ad hominen and waspish, isn’t it, if not somewhat presumptuous too? Still, moving on to the data you pointed me towards on the Ipsos website, there is some interesting information on there, but there is a subtle difference in the YouGov and Ipsos questions, I think. YouGov ask respondents to say whether they approve of the Government’s record whereas Ipsos ask whether they are satisfied with the way the Government is running the country. Much common ground there, I admit, but approval is a more judgemental term than satisfaction. Is anyone ever totally satisfied in this demanding world?

    That said, I was surprised at the seeming lack of correlation between Government satisfaction and voting behaviour at elections. If I’m wrong, and Government approval is indeed the same question as Government satisfaction, then we might as well totally ignore the approval rating measure in YouGov between now and May. The Ipsos data suggests that it is almost totally irrelevant in terms of providing a reliable guide to voting behaviour. Blair’s governments were winning elections in 2001 and 2005 with very negative net satisfaction ratings in the run up to those elections, although I suspect they were well ahead in the VI ratings at those stages though.

    As for today’s Populus poll, plus ca change, plus la difference. MOE fluctuation around what still appears to be a microcosmic Labour lead.

    I think the detail was stuck behind a paywall, but Populus had a headline suggesting that Labour had now nudged marginally ahead in the election forecasting models with Miliband having a “79%” probability of becoming PM in May. Has anyone seen this in more detail? Populus suggested it was based on new polling, but the article kept disappearing behind an ad screen before I had a chance to read more.

  11. Barney

    “Not one person from the north east of Scotland has ever used the word polis to describe the police”

    But I just did!

  12. Little Red Rock

    On the issue of double counting swing back:

    Do pollsters only reallocate Don’t Knows back to their 2010 vote? And, might we expect the number of Don’t knows to decrease as the election draws nearer?

    Theoretically pollsters also tend to reallocate people who refuse to say how they will vote as well. Though as such people normal won’t say how they voted in 2010 either[1] there’s usually nowhere to allocate them too.

    Don’t knows should decrease and have a little. Spearmint is the one who keeps a check on this. Whether they have dropped as much at this point as in previous election cycles is another matter – though this is the first when we’ve known the date for certain in advance.

    [1] Which I suspect often means they didn’t vote at all but won’t admit it.


    Have you spoken to every person from the North East? With over 480,000 people living in the North East (Moray, Aberdeenshire and Aberdeen) I doubt it very much.

    You don’t represent the whole region but just one little bit of Aberdeen and it was in that city the SNP out polled Labour at the last local council election but due to the voting system Labour managed to win more seats.

    So please stop coming on here and giving the impression you are Lord Aberdeen.

  14. It’s been said to me that the Polis pronunciation in Scotland reflects the historical development of Police Burghs in Scotland taking on responsibilities for roads, lighting, water supply, rubbish collection, public health and of course what we’d call modern policing via the Burgh Police (Scotland) Act, 1833.

    In this context Polis is harking back to the Athenian concept of Citizenship rather than just the modern police context and some of this history is still present in modern usage.

    Or it could just be that folk talk funny like the Swedes up here.

  15. @ Richard

    I don’t know if things have changed in respect of registration law, but as a student in the 1980’s I was urged to register in both my home town and my university town, this was in order to vote in both local elections as I would be affected as a resident in the university town and my grant provision could be affected in my home town. This was general advice to all students but it was made absolutely clear that in a general election a vote should be exercised only in one or the other and not both. How that was monitored I know not.

  16. @ Crosbat11

    I just love the accidental but apt neoligisms that appear here “microcosmic” for a labour lead in the polls: priceless!!

  17. If we’re using Barney’s hypothesis that the word “polis” is not north east Scotland terminology then surely one must apply the same formula to the surname “Crocket” and therefore not north east in origin.

    However my own surname Christie (sub-clan of Farquharson Royal Deeside) is from the area and therefore I reckon I can speak for the north east on that basis.

    It’s called the Barney Barmy application. ;-)

  18. Lord Aberdeen does have a certain ring to it but I think it is taken.

  19. They don’t like me do they?

  20. @Barney

    “Not one person from the north east of Scotland has ever used the word polis to describe the police.”

    As an Aberdonian, I can say that I and my friends sometimes referred to the name ‘Polis’, if for no other reason than references to Oor Wullie.

    Its that broad-brush approach that is alienating voters. Is it deliberate?

  21. Scottish crossbreak in today’s Populus

    SNP 43% : Lab 20% : Con 12% : LD 10% : UKIP 10% : Grn 6%

    Mean of Scottish crossbreaks in last 6

    Populus polls SNP 43% : Lab 24% : Con 14% : LD 10% : UKIP 6% : Grn 3%

  22. @Northumbrianscot

    Whatever we think of the LDs they are not proper Tories so anyone who is ABT would prefer them to a Tory.
    They have been at some pains to tell people that it would have been worse if the Tories had won an OM and whilst the difference is no doubt overstated I don’t think it’s entirely non-existent.
    Hence I would expect quite a few occluded noses and strategic X’s

  23. Even more on polis?
    One striking feature about north east Scotland is how very few people move to the area from other parts of Scotland, no matter how great the economic opportunities. It is somehow seen as alien territory. Notoriously it is so much more common to hear of men (almost entirely) from the north east of England working in the oil industry than those from central Scotland. It is also, I believe, one reason why so many people from abroad live in the city as all over Eastern Europe it is known that jobs are relatively easy to come by even in an era of low oil prices. So influence on our (strong) local dialect from other parts of Scotland is zero.


    That’s Aberdeen’s strength, diversity and the ability to attract a skilled workforce.

    For a industry such as oil I don’t think Aberdeen is in a unique position in that it can and has to attract people out with the local area and indeed the Country to work in such a complex industry as oil.


    “They don’t like me do they?”

    I wouldn’t chuck in the towel quite yet, the local elections are at least a year away.

  26. Anent polis

    “Polis” is also used in Norn Iron and presumably came over with the Scots, many of whom arrived in17th and 18th centuries.

    Barney, I’ll be yer pal. Crockett, Earl of Montrose?


  27. After a week off Prof Curtice has now published two blogs in one day!

    An interesting snippet from his latest missive that I suspect almost entirely explains the current Scottish political situation in 4 sentences:

    “What divides voters is the importance they attach to the constitutional question. Amongst SNP supporters 53% mention independence and 48% the devolution of further powers to the Scottish Parliament.

    In contrast the equivalent figures amongst those backing Labour are just 11% and 16% respectively. That suggests Labour’s (enhanced) promises of more devolution are cutting little ice amongst those for whom the issue matters – and that the party is unlikely to make much progress for so long as that remains the case.”

    I wouldn’t chuck in the towel quite yet, the local elections are at least a year away.

    Quite so. Back to polling, if we get more Scottish constituency polls, as seems likely, it would be interesting if they ask whether all the local Lab/Con coalitions formed in 2012 are liked by their electorates. My guess is that the BT coalition has done most damage to the LiS vote but could the 2012 elections also be a factor?

  29. @Guy Monde

    I agree but I suspect the nature of the incumbent (if there is one) will affect the amount of Nose holding going on.

    What these numbers highlight is that the post election Lib Dems are likely to be mainly made up of those fighting the Tories locally.

    If they had 25-30 or so survivors then it could be that 20-25 of them are in Lib Dem vs Con marginals with maybe John Hemming, Simon Hughes, Greg Mulholland, Charles Kennedy and Alistair Carmichael as those not primarily fighting Conservatives.

    If that sort of block of MPs is what remains they may see a strategic long term advantage in enhancing their anti Con credentials by supporting a Labour administration next time whether that be via Coalition or C&S.

  30. What were the Scottish cross-breaks in the London Poll?

  31. TNS found 44% who both say that they are certain to vote and express support for a party.

    Among this key group of voters, 46% back the SNP (41% last month), 30% Labour (31), 14% Conservative (16%), 3% Liberal Democrat (4%), 45% Green (6%) and 3% UKIP (2%).

    Tom Costley said: “It is worth noting that a quarter of those who indicate they are certain to vote are undecided as to which party to support.”

    Unfortunately, that last finding is pretty useless unless we know who their 1st and 2nd choices are!

  32. Jim Jam

    Inexplicably, there were no geographic crossbreaks in the London poll, so we don’t even know the variations between different parts of London.


    Two reason I think the BT lot on Aberdeen council suffer and in particular Labour.

    One…Local factor. The mess the council have made over Aberdeen city center l will almost certainly lose them control of the council in spectacular fashion, in fact I will put a bet on that the SNP will take control of the council at the next election.

    Two..Labour will lose support at the GE to the SNP because of the BT campaign with the Tories and the fact that the two are running the council.

    Labour jumped into bed with the Tories on the council 4 years ago for a quick gain of power but it’s going to cost them big time in the long run.

    As far as I’m concerned the SNP will complete the ha-trick… take control of the council, win both Westminster seats and hold the 3 Holyrood seats in 2016…Game…….set……….&…………match.

  34. Sorry for the typos in my lost post. I’m using a tablet and it’s horrific trying to type.

  35. As usual TNS provides crossbreaks of Westminster voting by Holyrood regions.

    While, inevitably, they do pick up regional strengths to some extent, all the usual caveats apply – the biggest (Glasgow) is only 86, and the smallest H&I only 36.

    FWIW the breakdown is –

    Party: H&I: NE : M&F: Lot: Cen: Gla: West South
    SNP : 56%: 43%: 42%: 51%: 46%: 48%: 51%: 37%
    Lab : 12%: 26%: 17%: 36%: 28%: 37%: 39%: 30%
    Con : 0% : 18%: 22%: 9% : 12%: 3% : 8% : 21%
    L_D : 21%: 6% : 8% : 3% : 6% : 2%: 0% : 1%
    Grn : 11%: 4% : 10%: 0% : 1% : 8%: 1% : 4%
    UKI : 0% : 1% : 0% : 0% : 7% : 2%: 1% : 6%

    Obvious nonsenses in there for the minor parties.

  36. If no Scottish Cross-breaks the poll can’t be meaninful.

  37. Jim Jam

    “If no Scottish Cross-breaks the poll can’t be meaninful.”

    Ah! That must be why those in London can’t be bothered to post many comments about it!

  38. ON – Many have disappeared over the last 6-12 months whilst other post less often.

  39. Roger Mexico
    Thanks for reply re propensity to vote.

  40. @ Unicorn

    Support for SGP was 8% in the European election and now, depending on what day of the week or pollster, ranges from a low of 1% to a high of 8%.

    How much does Scotland impact the UK wide polling result? Futher in the European election the Green Party attained only 4% in Wales, and does not appear to be ranging much above that now.

    Of five England only polls conducted in January the average was 9.1% Green and 7.7% LD, but of three undertaken so far in February the average is 8.3% Green and 8% LD.

    So is it possible, from a statistical point of view, that crossover occurred in England in January, but is now slipping back in favour of LD?

    The margin between the two parties is still very close, and from my perspective the real issue is where have the two thirds of voters who have deserted the LD party gone and will they stay there once the writ is dropped.

    The last time I checked LD still had the lowest level of stickiness of any party able to obtain seats. So the pollsters can continue to project a better outcome for LD, all they like (based on past glories) , but it don’t mean it will happen as their “snapshots” predict it will.

    The Green Party, on the other hand appears to have converted a 2014 European election vote in England of 7.9% into a Westminster vote of between 7% to 8%, a potential seven to eight fold increase over 2010.

    UKIP likewise look poised to increase their support five fold over 2010. So England, to some extent, looks like Scotland, except the vote for “other” is split into two parts.

  41. @ Jim Jam
    “ON – Many have disappeared over the last 6-12 months whilst other post less often.”

    Yeah! Like me. Although I still check it all every couple of days. Reasons are – all seems pretty static as we coast towards the GE date – without more movement nothing to add to my previous efforts!

    But I still enjoy reading all your exchanges……

  42. Apologies if someone has already published the link to the TNS tables but I haven’t spotted it upthread.

    They’re at:

  43. @Unicorn

    Hmmmm. It keeps happening, doesn’t it? Sometimes there. Sometimes not quite there. I think we can probably all agree that we are teetering on seeing a Ukip drop since Jan 1.

    I agree.

    If one the critical processes I was monitoring at work (one that should stay the same) was giving the results of the UKIP VI in 2015, I would be concerned and watching it like a hawk. Being so close to the tipping point, one of too data errors could be all the difference.


    I’ve been really busy, and thus the delay in the control chart information.

    As regards to the p value, it’s probability. The highest value is 1, which means 100%.

    In the case of UKIP VI, Jan 2015 vs Feb 2015, it runs like this.

    If you want to test the data, you start by creating two things. You need a null hypothesis and an alternative hypothesis. The null hypothesis is usually the status quo (basically there is nothing to see).

    The null hypothesis would be :

    “YG UKIP VI in Feb 2015 is the same as it was in Jan 2015”

    The alternative hypothesis would be:

    “YG UKIP VI in Feb 2015 is different to what to it was in Jan 2015”

    How confident do you want to be which hypothesis is correct? By convention a 95% confidence interval (CI) is used.

    You can now create the p values that determine whether your null or alternative hypothesis has been proved to the confidence level you want.

    Given a 95% CI, the alternative hypothesis is considered proven if the p value derived from the test is 0.05 the null hypothesis is assumed.

    If you wanted a 99% CI, the alternative hypothesis would proven with a value of < 0.01.

    We can now say that if we want to compare the mean of UKIP YG Jan Vi vs Feb (to a 95 CI), if the p value is 0.05 they are the same.

    When comparing with a T test, the P value was less than 0.05 so says there is a significant difference (to a 95%). When tested by another method (one designed for data not normally distributed) the p value is 0.06, so not significant (to a 95% CI).

    It shows that according to different methods there is a difference or it’s very close to being different.

    This is hopefully helpful, but there plenty of links on the t’internet that may explain better than me.

  44. The last post has had a number if brackets dropped.

    Probably down to html tags.

  45. @ Richard

    “If p0.05 means that the sample doesn’t support the hypothesis. Note that it’s only valid for the sample and with the construct that is created by the hypothesis. It doesn’t mean that the hypothesis is not true.

    CMJ has a very important remark: not normal distribution. Statistics has problems with non-bell shaped distributions (other than student’s distribution – strong skew to the right or to the left of the curve), especially when there are very strong outliers (regression fallacy) – for example the claim that government debt exceeding 70% is bankruptcy was such due to New Zealand. Effectively you measure the majority of the sample against the outliers. A common error in economics.

  46. Common misconceptions about the p value (I don’t know where I copied it from, but must be some medical related website)

    1. If P = .05, the null hypothesis has only a 5% chance of being true.
    2. A nonsigni?cant difference (eg, P ?.05) means there is no difference between groups.
    3. A statistically signi?cant ?nding is clinically important.
    4. Studies with P values on opposite sides of .05 are con?icting.
    5. Studies with the same P value provide the same evidence against the null hypothesis.
    6. P =.05 means that we have observed data that would occur only 5% of the time under the null hypothesis.
    7. P =.05 and P ? .05 mean the same thing.
    8. P values are properly written as inequalities (eg, “P ?.02” when P = .015)
    9. P = .05 means that if you reject the null hypothesis, the probability of a type I error is only 5%.
    10. With a P = .05 threshold for signi?cance, the chance of a type I error will be 5%.
    11. You should use a one-sided P value when you don’t care about a result in one direction, or a difference in that direction is impossible.
    12. A scienti?c conclusion or treatment policy should be based on whether or not the P value is signi?cant.

  47. For some reason this tablet plays up with copy and paste.

    The response to Richard should have had if p>0.05 … As well as the sentence in his post to which I responded …

  48. Twitter alive with news that Ashcroft doing local polling in the Outer Hebrides and Balamory.

    Key battlegrounds in the upcoming GE, according to the noble Lord.

  49. What’s the story in Balamory?
    I’m excited to hear how PC Plum and Eade Macreadie will be voting.
    Ok, yes, I’m just three years old…

  50. Northumbrian Scot,

    Thanks. It’s not totally improbable that Kennedy could lose his seat (the Highlands were a relatively good area for the Ayes, even though they lost) but his personal vote and lack of association with the Coalition should surely give him a very big boost to his already large majority. In the long run, I expect the seat to go SNP, because it’s potentially good country for them, but not before Kennedy leaves.

1 4 5 6 7 8