Eleven weeks to go

Here are this week’s polls

YouGov/S Times (13/2/15) – CON 32%, LAB 35%, LDEM 7%, UKIP 15%, GRN 7%
Populus (14/2/15) – CON 31%, LAB 33%, LDEM 10%, UKIP 15%, GRN 5%
ICM/Guardian (15/2/15) – CON 36%, LAB 32%, LDEM 10%, UKIP 9%, GRN 7%
Ashcroft (15/2/15) – CON 30%, LAB 31%, LDEM 9%, UKIP 16%, GRN 8%
TNS BMRB (16/2/15) – CON 28%, LAB 35%, LDEM 6%, UKIP 18%, GRN 7%
YouGov/Sun (16/2/15) – CON 32%, LAB 32%, LDEM 6%, UKIP 16%, GRN 8%
YouGov/Sun (17/2/15) – CON 33%, LAB 34%, LDEM 6%, UKIP 15%, GRN 7%
YouGov/Sun (18/2/15) – CON 32%, LAB 34%, LDEM 8%, UKIP 14%, GRN 6%
YouGov/Sun (19/2/15) – CON 32%, LAB 33%, LDEM 9%, UKIP 15%, GRN 6%
Populus (19/2/15) – CON 31%, LAB 32%, LDEM 9%, UKIP 17%, GRN 6%

With the exceptions of the rather anomalous looking ICM poll with its four point Tory lead and the TNS poll showing its typically larger Labour lead (two outliers I discussed here), the polls have returned to the same picture we’ve had for the whole year so far – a very close race with Labour just ahead. The UKPR polling average stands at CON 32%(nc), LAB 33%(-1), LDEM 8%(+1), UKIP 15%(+1), GRN 6%(-1) – none of those figures has moved more than one point away from that so far this year. The race is pretty much static.

Scottish and Constituency polls

Survation put out their monthly Scottish poll for the Daily Record this week, giving the SNP a slightly reduced but still very healthy 17 point lead over Labour. Lord Ashcroft released four polls of Conservative held UKIP target seats, showing UKIP just a point behind in Castle Point and only three points behind in Boston and Skegness. I wrote about both here.

Week 7

  • The Conservatives started the week promising to cut sickness benefits for fat people or addicts who refuse treatment. This is an interesting example of policy and how to look at public reaction to them. At one level such policies are popular – by 57% to 28% people support stopping sickness benefits for overweight people who don’t seek weightless treatment, by 64% to 23% people support stopping sickness benefits for addicts who don’t seek treatment. However, the potential downside for such policies, especially for the Conservatives, is if it reinforces the party’s image problems of being seen as heartless or unconcerned for the less well off. The same polling found 40% also thought the policy was uncaring and heartless.
  • Labour started the week talking about economic policy and had their policies endorsed by Lord Mandelson, the former business Secretary. There was a YouGov poll in the week asking if the endorsement of various retired politicians was an asset or a liability – 52% thought Mandelson’s backing a liability, only 7% an asset. Tony Blair’s endorsement was seen as little better – 14% an asset, 61% a liability. The reason for both is the same – most other politicians were seen an asset by supportwes of their own party, a liability by their opponents. Blair and Mandelson (and Michael Howard) were seen as liabilities by both their opponents’ supporters and their own parties’ supporters. According to today’s news Peter Mandelson is now warning Labour against their tuition fees policy, so perhaps his criticism will be an asset!
  • On Thursday party donations for the end of 2014 were announced. The Conservatives received just over £8million, Labour just over £7m, the Liberal Democrats £3m (the party had a record year of donations, despite their precipitous drop in support since 2010), UKIP £1.5m and the Greens a quarter of a million. There was some polling on party donations last weekend, showing people pretty cynical about both main parties – by 48% to 30% people think Labour should try and reduce Union funding, by 52% to 25% people think the Conservatives should try and reduce their business funding. Around two thirds of people would support a cap on business and trade union donations, 51% would support a cap on individual donations to political parties, only 19% would support taxpayer funding.

Projections

The latest forecasts from Election Forecast, May 2015 and Elections Etc are below. Elections Etc and Election Forecast both have Labour and the Conservatives pretty much equal in predicted seat numbers, May 2015 are projecting Labour to have more seats, but not by very many.

Elections Etc – Hung Parliament, CON 281(nc), LAB 282(+1), LD 23(nc), SNP 40(-1), UKIP 3(nc)
Election Forecast – Hung Parliament, CON 282(+2), LAB 280(-3), LD 25(-2), SNP 40(+3), UKIP 2(nc)
May 2015 – Hung Parliament, CON 266(-3), LAB 275(+1), LD 26(+2), SNP 56(nc), UKIP 4(nc)


159 Responses to “Eleven weeks to go”

1 2 3 4
  1. Having said that OPINIUM have just issued their first Tory lead.

  2. @ Guymonde

    The problem I see is that some of the pollsters, ICM for example, are starting to try and apply a classical swingback model, through their poll weighting models.

    This is not borne out by what is happening on the ground, in that taking Hackney North and Stoke Newington and Hackney South, as an example, the Green party obtained 4.6% and 3.4% in the 2010 GE election respectively and 17.2% in the 2014 European election.

    So from 2101 to 2014 they went from 4th or 5th to 2nd in an election. Now the Green Party is reporting on their website that they have 1,000 members in Hackney.

    Doing the math that would mean that 8.6% of the voters from the 2014 European electon have not just voted Green but joined the Party.

    Now I read this is happening all over the UK and I assume not just for Green, but I assume for SNP and UKIP as well and then I see no real reflection of this in the polls or seat projections, except in relation to SNP in Scotland.

    So I see a very real disconnect between polling, seat projections and the reality of what might be happening.

    I remember the very infamous newspaper headline from the 1948 US Presidential election:

    “Dewey Wins”

    In which Truman actually won the election , and so I think newspapers like the Guardian should actaully pay attention to what they are publishing, as it could be basically tripe and nonsense.

  3. Opinium showing some classic swingback. Tories up 2, Labour down 2.

    Ooops!

    :-)

  4. I see a number of comments on twitter like

    “Trend chart from Opinium which has just recorded its first ever CON lead in its Observer polling series”

    “Opinium looking gd for Tories & pretty dreadful for Ukip (usually one of their stronger polls)”

    But really all that has happened is that Opinium recently changed their methodology that they said at the time would lower the UKIP score. (And I guess the flip side of that is increase their conservative score).

    Spot the date when their methodology changed on this graph?

    https://twitter.com/MSmithsonPB/status/569196810675343360

    And spot the large upweighting of conservative voters in their party propensity weighting, which was the change they made earlier this year.

    Conservative 378 ==> 482
    UKIP 207 ==> 188

    If they didn’t make that change there would be no first conservative lead tonight, and no apparent UKIP drop.

  5. And…no swingback!

  6. @AndyvShadrack

    What annoys me is we are told polls are a snapshot but if pollsters are weighting on the basis of past votes they are making assumption and therefore predicting behaviour which a poll is not meant to do.

    The other annoying thing is ‘party propensity’ or ‘party id’ which Opinium use. We have had 2 extra polls in Scotland than the rest of the UK -Scottish Government & Referendum. A person who last voted Labour in 2010 but has voted SNP since & Yes in the referendum is hardly likely to identify as Labour depressing the SNP score as you can see by the large down weighting of others in Opinium

  7. Smithson tweeting this Opinium is showing first ever Con lead.

    That might only be true if one ignores the other times, such as in 2012, they have shown a Con lead…..

  8. Why does swingback to Cons have to have occurred already, failing which it doesn’t exist?

    Isn’t the essential point about swingback , that it manifests itself in the GE result?

    And if that is its key identifier, it’s appearance in the last week -or the last n days is as much proof of its existence as from any other date -isn’t it ?

  9. @Couper

    Agree. Just look at the Scotland crossbreak in Opinum

    SNP 31, Cons 28, Lab 25

    We know from recent Scotland polls that is nonsense

    And look at SNP

    62 downweighted to 39!

    If they hadn’t done that downweighting they would have had a much more reliable poll.

    They need to re-look at their party weighting, it is not working…

  10. CROSSBAT11
    I think we’re on course to see the lowest combined Labour and Tory aggregate vote of all time in May.

    Even if you meant “big 2” rather than specifically Con & Lab that cannot be true, if only because of population growth.

    It could well turn out to be the lowest percentage share since equal suffrage arrived in 1950, with the abolition of plural voting and the university seats, but sadly not plurality voting.

  11. @ Bramley

    It’s the first time they’ve been in front since they were behind? LOL. :-)

  12. @James Peel

    I was curious enough your comment to spend half an hour trawling through the archives for late October 2013 and you are right that there were one or two critical comments about the Fisher model (interleaved between about ten times as many wishing @Amber happy birthday!). For the most part, contributors were just expressing a sense of disbelief about the projections and I’d be interested if you could pinpoint posts which set out reasons like those you list today.

    One comment that did identify a series of objections appeared at 6.53 pm on October 25th, 2013 when @ Phil Haines ended his post by writing:

    Personally I think that a hung parliament outcome (with the LDs being left with too few seats to call the shots) is the most likely

    Now I have to say that I was highly impressed to see a statement like that from all that time ago. How about a list of more recent updates, Phil?

  13. @ON

    Your post re Fisher may have been a mistake ( too much tablet again, presumably) but it was a useful one.

    I said at the time that I viewed it as a test of a hypothesis and we don’t yet know just how well the test will do – his margins of error were so wide that they encompassed both a Tory and Labour landslide, so I suspect he’ll only fail, by his own lights, up your way.

  14. UNICORN.
    Good Evening to you; I too have felt the Lib Dems have been facing a disappointing GE result in May 2015, sadly.

  15. @Richard

    There are two many significant methodological changes happening far too often that it’s ever becoming difficult to compare polls from the same pollster, let alone to see the general trend.

    It seems to me that Cons are being up weighted to seek to provide a projection of the final result (including swingback) rather than a snapshot of the poll actually taken.

  16. @Unicorn

    I remember my criticism of Fisher which may have been in the comments on the site or on Twitter. The swing back was too high. What he did was average the swing backs from previous governments but in fact the present government has not fallen so far as previous so should not get the same size of swingback.

  17. @Richard

    So the question then becomes if UKIP are not really at 15%, where are they at and are the polls for Green and LD accurate either?

    The pollsters could not suppress the rise of the SNP in Scotland, but could their modelling for England and Wales be doing that for UKIP and Green?

  18. @”It seems to me that Cons are being up weighted to seek to provide a projection of the final result (including swingback) rather than a snapshot of the poll actually taken.”

    If we add this to “the evil Tory Press have destroyed EM’s character” & ” the devious SNP have seduced Labour voters with their lies”-does it count as :-

    “cries of “foul and it’s all beastly and unfair because the voters should be more grateful etc etc”,

    lol :-) :-)

  19. Now I have to say that I was highly impressed to see a statement like that from all that time ago. How about a list of more recent updates, Phil?

    And I was highly impressed by all the people wishing me Happy Birthday. :-)

    But Phil’s comment was jolly good too!

  20. Happy Birthday Amber – sorry to be late ;-)

    My birthday is 4 days after the GE, so I’m hoping for an early present. If I don’t get it in the form of the right result, I suppose I can console myself with the thought that I got another year older.. or perhaps not.

  21. Guymonde

    They say that, with age, you gain wisdom.

    Today, I have demonstrated the fallacy of that proposition.

  22. @ Guymonde

    Thank you :-)

  23. @ Guymonde

    Lol … But not in the Cameron sense…

  24. @Colin

    ““cries of “foul and it’s all beastly and unfair because the voters should be more grateful etc etc”

    Very amusing.

    Whilst I feel you have a point to a degree that individuals may be allowing ‘loyalties’ to intrude on their judgements with regard to the outcome of some polls, there do seem to be some issues at the moment re:

    (a) fairly regular methodological changes which directly affect the results produced by certain polling organisations – effectively intentionally altering their own ‘house effect’

    (b) whether ‘voter propensity’ as deployed by some pollsters is too crude and is therefore creating some very odd results.

    The second of these is fairly obvious in Scotland, everyone (apart from the leader of LiS) maybe wrong and the SNP vote may collapse in May, but this seems unlikely. Not to factor this in seems awry.

    Far more tricky is the question of UKIP. Isn’t it simply the case that we don’t really know how many and where these current VIs will jump.

    I wonder whether all this second guessing might leave some with egg on their face – presumably exactly what they are trying to avoid.

  25. @Couper 2802

    You are absolutely right, I was so furious when I calculated the ICM poll that I wrote the British Polling Council from Canada.

    And what you see in these polls is a manipulation of what election results have been showing, because weighting is based on an election, 2010, that no longer exists.

    But I think the voters are already moving on in that last night I read a post of an article on Labour cavassers in Edinburgh finding Conservative support swinging to Labour.

    This is classic to what I observed in Canada when the Bloc Quebecois vote rose in Quebec in 1993, in that non-nationalist voters started to line up behind the party they saw as most likely to beat the Bloc.

    In Edinburgh during the 2014 European election SNP tied with Labour at 23.1%, Conservative were third at 19.4% and Green fourth at 16.1%. So I assume that if Conservatives are moving to Labour, that Green will move to SNP. At least that is what I read when I look at the debate on the SGP website.

    The LD vote in Scotland has already collapsed and UKIP and Green will not be a factor, but it is quite predictable for the Conservative vote to collapse in support of Labour vs SNP.

    Pollsters using a weighting model based on the 2010 GE will not pick that up.

  26. @ Couper

    Yes, some of your comments were there on the same page as Phil’s above and it is clear the you and others were right on the ball at the outset.

    What I was objecting to was not the criticism of Fisher and his model (you probably know that I am no fellow-traveller myself). I just didn’t think it helpful to refer to it as ‘rubbish’.

  27. Andy Shadrack

    I have posted before that tactical voting by Tories for Labour was always quite likely.

    In fact the more that LiS can position themselves as slightly pinkish Tories, the more successful they are likely to be in achieving that.

    At the same time, such an approach can alienate more of their traditional base, unless Jim has calculated that it is already down to its hard Unionist core.

  28. @Colin

    Why does swingback to Cons have to have occurred already, failing which it doesn’t exist?

    I think a common answer to this would be that swingback is normally construed as a process that operates over an extended period of time. Historically, it has started to operate as much as two years out from elections and has typically gathered pace over the last few months. I imagine the sense is that if little has happened to date why should we still expect the later components to kick in now.

    It’s a bit like most people’s reaction to all those prophets predicting the end of the world. Once the predication has failed a few dozen times you begin to get a bit cynical about whether it’s ever going to happen.

  29. From Opinium

    If a referendum were held on the UK’s membership of the European Union with the options being to remain a member or withdraw, how do you think you would vote?

    Leave 44%
    Stay 41%
    Unsure 14%

    Now compare to historical polling on that question here:

    http://en.wikipedia.org/wiki/Proposed_referendum_on_United_Kingdom_membership_of_the_European_Union

    So that is in line with their last poll on this issue, but Yougov and Opinium are getting completely different results

    Yougov generally get 50%+ to stay, 25-27% to leave and 16-19% undecided.

  30. Sorry ignore that last post – I was comparing to the Yougov intention if terms were re-negotiated….different question in Opinium.

  31. @ OLDNAT

    I am not sure Labour have factored that in as I think they might be becoming as desperate as the Conservatives.

    The London Labour leaflet on immigration is simply going to reinforce, for some voters, why they had already decided to not vote Labour.

    In fact I think that there is a real possibility that when the writ is dropped that the campaigns of Labour and Conservative will reinforce why neither party should be given voter support, just remember that in this pre- election period up 40% of Labour and Conservative voters have said they could change their minds.

    Looking at campaign reports from the Green Party in communities as diverse as Haverfordwest, Wales, and Harrogate, Yorkshire I sense that some voters are about to throw tactical voting out of the window.

    I am getting a sense that just like Labour has lost hegemony in Scotland that might also be starting to happen in England and Wales, except in England it is a two front war , UKIP on the right and Green on the left .

    The Conservatives face the same prospect with UKIP, and the narrative this week was that they faced the same fight on the left with LD, but then I think that is more wishful thinking than reality.

    But then what do I know, as I am not on the ground anywhere in the UK.

    I certainly think if people on this list are unhappy with pollster methodology they should email the Bristish Polling Council and explain why.

  32. ANDY SHADRACK
    & OLDNAT
    I have posted before that tactical voting by Tories for Labour was always quite likely.

    Left to their own devices I would expect the same, but I wonder if they will listen to their leadership.

    I didn’t notice until recently, but the BBC’s Brian Taylor posted yesterday evening on the Scottish Tory conference,oddly concentrating on Mundell rather than his betters, albeit echoing Cameron’s morning speech, ensuring the “vote Labour, get SNP in shared power” theme was raised again.

    Can’t see that this will be helpful to their cause, somehow, but equally I can’t see how most of the MSM will argue against it.

  33. @ Unicorn
    ‘Historically, it has started to operate as much as two years out from elections and has typically gathered pace over the last few months’

    But it has not usually extended to the final month leading up to polling day. Of the last 14 General Elections a swing to the incumbent in the course of the official campaign has only occurred on 3 occasions – 1979 – 1992 – 1997. Both 1979 and 1997 were rather exceptional in that the government trailed by big margins of 12 -20plus% at the outset of the campaign and despite making up some ground went on to lose decisively.

  34. @Andy

    I certainly think if people on this list are unhappy with pollster methodology they should email the Bristish Polling Council and explain why
    __________________________________________

    I don’t think the Polling council has any rules on methodology. Members are free to use whatever methods they want, but just need to disclose properly what was done.

    It is then for users of those polls to make a judgement call whether those methods are appropriate or not.

    example:

    http://www.britishpollingcouncil.org/minutes-of-the-annual-general-meeting-2013/

    “Two pieces of communication have been received concerning members’ practices of prompting or not prompting for UKIP at voting intention questions. It was agreed that this was entirely for members to decide, but that it should be clear in any publication of polls whether UKIP had been prompted or not. It was further agreed that members would advise the Secretary of their current practice, and a statement would then be added to the BPC website giving the norm, while making it clear that members are free to use alternative approaches.”

    it also has the rules on that site, and if you read them they are all about disclosure rather than methodology.

  35. I agree with Andy and others critical of methodology now being used by some pollsters. The upcoming GE appears to be triggering some of this. I think we may need a Naughty Step for them – from the last week or so I would put ICM, TNS and now Opinium on it. Opinium’s Scottish results were as amusing and presumably bonkers as the Con lead posted by TNS. On top of this we have Ashcroft and the Libdems selectively posting constituency polls which also looks to be influenced by GE.

    Thank goodness for YG – their daily diet of dead-heats may not excite, but is much more likely to be right.

  36. UKIP say they are to fight (I use the term loosely) 40 Scottish constituencies.

    Excitingly, we learn who is to seize Gordon from Salmond’s grasping hands.

    http://www.buzzfeed.com/aidankerr/ukip-scotland

  37. Anyone know who gets the proceeds of lost deposits?

    Some level of government, I presume.

  38. YouGov/Times:

    CON 33 (+1)
    LAB 34 (+1)
    LIB 8 (-1)
    UKIP 13 (-2)
    GRN 6 (=)

  39. @ Welsh Borderer
    ‘Opinium’s Scottish results were as amusing and presumably bonkers as the Con lead posted by TNS’

    I presume you meant the Labour lead posted by TNS. Re- Opinium – amusing though the Scotland figures may be they are but a crossbreak, and whilst I do note them myself we have been told repeatedly by AW and others that in themselves they are not meaningful. In my opinion, that apparently strange crossbreak is not sufficient to dismiss the veracity of the poll as a whole.

  40. re: OldNat

    I’m sure that Misty Thackeray will have impressed (Celtic fan) Aidan Kerr with talk of “dodgy land deals” in the east end of Glasgow. The idea that Celtic have benefited from generous deals with Glasgow City Council is a particular paranoid delusion belonging to Rangers fans (“#StateAid” on twitter).

    Somewhat interestingly, UKIP seem to have latched on to this. Going after the Orange vote? Won’t win them much but could hurt Labour and the Tories a bit.

  41. And now we have YouGov!
    Prima facie tonight’s two polls do not contradict one another at all – well within MOE

  42. Graham @ Welsh Borderer

    “that apparently strange crossbreak is not sufficient to dismiss the veracity of the poll as a whole.”

    The problem is that it is a systemically strange crossbreak with Opinium due to their weightings.

    Downweighting SNP support from 62 to 39 and Plaid from 8 to 4 will make little difference to the overall GB figures, but it casts doubts on their choice of weightings.

    Professionals should seek to get such things as accurate as possible. Analogously, some pollsters are playing in the Bundesliga while Opinium might struggle to make the first team with Forres Mechanics.

  43. For all that Tories are keen to see crossover, I can’t say that anything interesting has happened in the polls since the SNP increase / Labour decrease in Scotland finished circa mid-October.

  44. Well according to:

    http://ukgeneralelection2015.blogspot.ca/p/latest-opinion-polls_25.html

    Labour are now I5 candidates short of having a full slate in GB:

    The Conservatives are on 96.3% , assuming they are not running in NI

    UKIP if they run candidates in NI are at 76% or 78.1% for GB.

    LD are on 75.6% in GB

    The Green are at 71.5% if you include NI or 72.9% for GB.
    They are at or have exceeded their 75% target in England and Wales and the SGP are at 55% so far.

    And the Green Party raised 350% more in donations in 2014 than 2013, at 661,000 Pounds.

    When was the last time five parties ran full slates in an English GE and six near full slates in Wales and Scotland?

  45. @ Unicorn

    I spent some time on your calculations today. I really appreciate your intentions and I think it is in line with accepted methods, and they add value.

    However, we are talking here with a situation where systemic and random influences cannot be a priory separated (and ex post only in the narrative). Do you think it would be feasible, useful to run a Bayaian pre-test analysis, and which would allow us to use the usual p values, or if not (which I think the case if we accept the usual approach) rejecting everything with less than p=0.005?

    (The reason for the 0.005 is, because if in pre-test there is a 50% probability and you attribute p=0.05, no more than 13% is the feasible change?)

  46. @Amber

    Happy Birthday!

    Hope you had a great day.

  47. UNICORN

    @” I imagine the sense is that if little has happened to date why should we still expect the later components to kick in now.”

    I might equally ask-why should we not?

    Some of these “rules” seem arbitrary to me-and the worse for being used to support the spurious accuracy of “forecasts”

  48. Yes,happy birthday Amber.

  49. @ Laszlo

    I am not sure which calculations you are referring to. My comment earlier on this thread was concerned with detecting departures from trends. If you raised the bar in the way you suggest then you would stand the risk of never being able to pick up a shift away from trend. But perhaps you were thinking of a different caclulation.

  50. @ Amber

    Official birthday now in February, it seems.. ;-)

1 2 3 4