A quick update on the latest voting intention polls. This morning’s YouGov poll for the Sun has topline figures of CON 33%, LAB 34%, LDEM 6%, UKIP 14%, GRN 8% (tabs) – more typical figures than YouGov’s Tuesday poll.

There is also a new TNS poll with topline figures of CON 31%(+3), LAB 31%(-4), LDEM 8%(+2), UKIP 16%(-2), GRN 7%(+2) (tabs). Up until now TNS have tended to release their polls incredibly slowly, often a week or more after their fieldwork was finished, rendering them out of date by the time we see them. Fieldwork for this one however finished on Monday so it’s more timely than usual. The last couple of TNS polls showed Labour leads of 7 points, significantly and consistently larger than any other company. Today’s looks more like other polls. (UPDATE – thanks to those in the comments section who missed something obvious in the TNS tabs that I overlooked! TNS have always weighted by 2010 recalled vote, but they are now weighting by 2010 recalled vote AND 2014 European vote – whether or not this is responsible for their figures coming into line with other companies’s is unclear.)

Finally Ipsos MORI put out their latest Scottish poll yesterday. Westminster voting intentions in Scotland were CON 12%(+2), LAB 24%(+1), LDEM 4%(-2), SNP 52%(nc), GRN 4%(-2). We’ve now had three Scottish polls conducted in 2015. The first one from Panelbase showed the SNP dropping four points and raised some speculation about whether their huge post-referendum surge was fading away again. The second from Survation also had the SNP down, but only by two points and this one from MORI has the level of SNP support holding steady. (On top of that, when tables for the Panelbase poll appeared it turned out that the voting intention question wasn’t asked first, it was asked after a question about whether or not falling oil prices damaged the economic case for Scottish independence, so the SNP fall in that first poll may be a question ordering effect rather than a genuine change)

419 Responses to “Latest YouGov, TNS and MORI polling”

1 5 6 7 8 9
  1. @ OldNat

    “Con, 12, 19, 16, 17, 17, 16, 16, 12, 13, 20, 21, 17, 18, 18, 22.

    I can see a hint of an upward trend there, but the statisticians may well tell me it is a total illusion!”

    I could take a look if you like. Presumably you start with “12, 19..” etc because these were the figures from the start of 2015. I ask because these were preceded by a couple of VIs of 19, and that might change the verdict.

  2. neil a

    As 34% is my rough [top] estimate I can’t oblige.

    IAnyway I don’t trust policemen……………]

  3. Unicorn

    Yes please. That’s the series from 5 Jan to today.

  4. The Populus poll is another with an increase in the Green vote (+2). On this occasion, the Labour vote remained the same.

    The ICM poll showed the same pattern.

    Is it possible that some of the Green vote is from voters who would have previously expressed “none of the above” in polls? Polling isn’t a zero-sum game and there is a large poll of “non of the above” voters.

    The evidence on this is inconsistent, but there would be a big difference in the electoral effect of a Green increase if they were introducing new voters rather than taking them off other parties.

  5. I do hear Kippers say that they find it easier to win votes off SLAB that the Scottish Tories, and that many of their voters are Orange voters, in the broad sense. And the only Scottish libertarian I’ve ever met was from a small family business background in Airdrie, the kind of voter the Tories once could win over in Scotland but not anymore.

  6. Lurker

    Populus Online massively weight down “rising” party IDs.

    UKIP 216 -> 82
    Green 98 -> 50
    SNP 68 -> 38

  7. @OldNat

    A crude regression analyses on those figures just misses showing a rising trend for the Tories (p=0.066).

    ‘Crude’ because I have indexed the entries by sequential order and not by polling date (As you know YouGov normally polls four days in succession followed by a gap of 3 days).

    Given my previous comment that the Tories were higher just a couple of days before the start of your sequence I would say that it is currently safer to treat this as routine variation around a fixed figure.

  8. Bill Patrick

    Yet again, we need to see constituency polling to see if the c 5% UKIP vote is concentrated in the once Labour heartlands.

    If Con VI is actually rising and not under threat from UKIP in rural Scotland, that could be good news for the Tories in the likes of Berwickshire, Roxburgh and Selkirk.

  9. Unicorn


  10. Oldnat

    I am not getting at the weighting this time (and for what it is worth, it seems pretty clear they are underestimating the SNP and thus overestimating Labour).

    I am questioning what might be happening on the ground rather than talking about polling effects. If the Greens are picking up people who previous had no party preference (i.e. people who would otherwise have not voted) then the effect on other parties (in practice Labour) would be zero. If they are picking up people who until a few months ago would have voted Labour then they would have an effect.

    With the SNP, they seem to be doing both.

  11. Why on earth would QT choose a question on the SNP with no SNP member on the panel? It’s bizarre It has certainly riled up social media. But for @Unicorn and statisticians everywhere: Ester McVey ‘It was unanimous No won by 10%’

  12. JIM JAM

    lol :-)

  13. Re Cons VI in Scotland ,

    The areas where the Cons might stand a good chance are in the same areas where UKIP will perform above average . There’s a strong Anglo vote in the Borders , Edinburgh and Tayside

  14. Chasglas

    I can’t see that being an “Anglo” would make anyone more likely to be a Kipper.

    I’d assume (but let’s wait for actual evidence) that their support was more likely to be met in Orange Halls and Ibrox!

  15. Lurker

    Yes. The effect of “new parties” in energising those who had turned off from politics is an interesting one.

  16. Sven Hassel Schmuck

    I’m inclined to agree with that and I think the debates will be hugely important in convincing people. EM doesn’t come across well in soundbites or in the horrible environment that is Prime Minister’s Questions but when he’s allowed to speak properly he comes across quite well. He has (IMO) a lot more going on intellectually than most of the other party leaders so if he can show that & also look like a normal human being at the debate things will go well for him.

    Does anyone know if they’ve announced what order the debates will be in yet? I think if the head to head one is last that will work in Miliband’s favour.

  17. Funypippin,

    Most logical format for the debates is 7-7-2-2.

  18. Mr Nameless,

    4-4-2 would be a tried and tested option.

  19. For what it’s worth I’ll share what I’ll dub my ‘Mayfly’ trend model for SNP, the soubriquet referring to the fact that I don’t expect to last more than a day or two. This is especially that case as few will be impressed by its projection for Election Day.

    Those of you with a quantitative bent can check its performance and tear it apart.

    Trend analysis reveals statistically reliable linear and quadratic trends in the SNP VI series since August [1], suggesting that the best curve to try to fit is a quadratic equation with a decent-sized first-order coefficient. The best I could manage was the following equation:

    SNP VI=22.13+4.76*T-0.27*T^2

    ..where T is a time measure in fortnights since August 3rd 2014 [2]. (T=1 for the polling period between Aug 3-4 and Aug 14-15, comprising two lots of five weekly YouGov Polls. T = 2 for the next 10 YouGov polls and so on)

    You can obviously convert a past or future polling date into T-index figures by calculating the number of days that have elapsed past Aug 3 and then dividing this figure by 14.

    A table of correspondences and predictions is given below.

    Fortnight : Mean YouGov VI : VI predicted
    1 : 29 : 26.6
    2 : 29 : 30.6
    3 : 32 : 34.0
    4 : 36 : 36.9
    5 : 39 : 39.2
    6 : 42 : 41.0
    7 : 41 : 42.2
    8 : 43 : 42.9
    9 : 43 : 43.1
    10 : 43 : 42.7
    11 : Xmas fortnight (only one poll published): 41.8
    12 : 42 : 40.4
    13 : ?? awaiting polling data ?? : 38.4
    14 : ?? : 35.9
    15 : ?? : 32.8
    16 : ?? : 29.2
    17 : ?? : 25.0
    18 : ?? : 20.3
    19 : ?? : 15.1
    20 : ?? : 9.3% – Election Prediction ?!

    The fit was based on data from fortnights 1 – 8 and the rest of the figures are predictions. A plot showing all individual YouGov polls since August reveals that the quadratic equation provides quite a good fit for the overall pattern of data. In particular its projections for Fortnights 9 -12 are reasonably accurate.

    This exercise highlights the difficulty in making projections at a time when everything is in a state of turbulence. But at least I feel I now have a framework for interpreting future SNP VI figures. It should quickly become apparent when they leave this unrealistic trajectory and when that happens it should be possible to reassess what is going on.

    [1] I used August 2014 as may starting date because it is pretty clear that the referendum campaign changed everything after that. Prior to that point the SNP VI had been going nowhere in particular for many, many months.

    [2] I batched the figures in this way to avoid breaking up the regular pattern in YouGov polls (4 polls in succession followed by a gap of – normally – three days).

  20. @Unicron / Keith

    “seat totals for Conservative and Labour – between 279 and 283 for both”

    On track for my 275 / 275 prediction. Within 3% actually. That’s a decent MoE.

    “In that last sentence my wise spellchecker turned ‘exceptions’ into ‘escape potions’.”

    I doubt it will be long before the technology takes over. All they’ll have to do is measure us a nation by the leaders we have and they’ll take pity on us all.


    I prefer 4-1-3-2, but only if Paul Scholes is included. :))


    I was going to start with “I can’t believe that the BBC would allow such a question in such a manner.” However, I can believe it. I just forgot about the way the BBC handled much of the reporting back in August and September.

    Don’t go Scotland, we need you, and get back in your box.

  21. Addendum

    The equation ‘thinks’ that today’s SNP VI should be precisely 39.7%

  22. Oldnat ,

    English voters are obviously more likely to vote UKIP so I think it’s a fair assumption UKIP will damage the Cons slim chances of seats in Scotland . The Orange type voters in parts of Central Scotland will certainly increase this possibility .

  23. “Why on earth would QT choose a question on the SNP with no SNP member on the panel? ”

    Particularly farcical given it is only a few months or so ago that at a Question Time from Scotland, Dimbleby was determined to shut Sturgeon up on the grounds that questions relating to Scotland were inappropriate as QT is a UK programme.

  24. Unicorn

    Brilliant spoof. If it is not a spoof you need help!?

  25. Populus:

    LAB – 36% (=)
    CON – 32% (-3)
    UKIP – 13% (=)
    LDEM – 9% (+1)
    GRN – 6% (+2)

    Some seem to have taken “Vote Green, Get Blue” to heart.

  26. @Old Nat; what evidence there is in who supports the Kippers in Wales suggests that being “an Anglo” is a big factor…..and that the “Rise of the Right” analysis that seats like the Rhondda should be prime UKIP targets is, in the technical term, ignorant Anglo-centric bollocks. UKIP themselves think that their target seats here are the border seats….especially the most porous borders in the north east


    4-4-2 worked well for the GNR in Edwardian days and had improved pulling power over their famous Victorian 4-2-2s, but before George V popped his clogs they were obscelescent and being replaced with 4-6-2s.


    Actually, 4-4-2s stopped being built just before WW1, barely into George V’s reign, although one lasted in service until 1958.

  29. @Thomas Robinson


    According to oxforddictionaries.com a spoof is “A humorous imitation of something…” A “Hoax or trick” or “a trick played on someone as a joke”

    Not sure you have something in mind that you think I might be imitating. ‘Hoax’ or ‘trick’ suggests a degree of subterfuge and I was perfectly upfront about everything that underpins the calculations. There are any numbers of contributors to this site who are perfectly capable of checking the accuracy of my figures, and you can easily do the same if you choose to do so.

    Do I think the SNP VI trend is going to follow that curve inexorably? No – I made that perfectly clear in the comment itself. Has anyone else done a better job of capturing what is happening to the SNP VI trend? I don’t think so. I’d welcome your advice if you have something better to offer.

  30. On the debates.

    To keep everyone happy why don’t EM DC and NC hold debates in England with all the main parties in England in one debate, then the 3 of them hold a debate in Wales Scotland and NI along side the main parties in each of the 3 countries!

    After all NC EM and DC want to be PM of all the UK and not just England so they collectively should travel the UK with the debates.

  31. @Unicorn

    “The equation ‘thinks’ that today’s SNP VI should be precisely 39.7%”

    I get different data.

    25-poll weighted calcs for all UK regions, and the UK total taken from the sum of the regions. In addition, the samples for Labour and the SNP in Scotland have been ‘visualised’.

    All weighted MAD calcs have 325 samples. I prefer to total the regions for arriving at the UK values, rather than take the weighted UK data. Hopefully it drills down past the UK rounding (one can only hope).

    http://www.statgeek.co.uk/2014/09/here/ (Scroll to second block of images)

    Allan Christie,
    “On QT the SNP are taking a pasting and no one there to defend their position. The same would go for the debates if the SNP were not on them.”
    “That’s a good point. Miliband was going to be under pressure to go after SNP votes in the debate regardless of whether Sturgeon was actually there (in fact, moreso if she wasn’t!) and so perhaps it’s not surprising that the debate on the debates is going the way that it debatably is going”

    Yes it appears to be going wobblydoo. Some people are not happy standing next to the Greens and SNP. Can’t think why?

  33. LURKER

    I agree that building stopped well before grouping although some were built after Edward VII shuffled off, but a few remained in service through to BR days, and many were in mainline service throughout the 1920s.

    BP’s original post just had me thinking I’d logged on to UK Public Railways blog by accident. :<}

  34. It looks like all the NI parties and George Galloway want an invite,

    Who’da thunk it?

  35. Those putting their faith in the mano-a-mano Cameron vs Miliband debate should of course be aware that it seems likely to be the one on SKY and thus only watched by a quarter of the population (I think that was about the amount in 2010).

    In fact, I await Ofcom’s rule on whether SKY constitutes a major broadcaster. Perhaps it should only be allowed a shorter debate?

  36. Lurker,

    Do Sinn Fein really want an invite? “Vote for us and we won’t show up”. I suppose at least they’re honest about it.

    NI is one where I can see the logic in having a regional debate with SF, DUP, UUP, SDLP and Alliance included.

  37. Etienne,

    If the offer of work I received from Sky is anything to go by (broadcasting all counts online) then I imagine they will stream their debate online free.

  38. @Mr Nameless



  39. Looking at Question Time, is there some parliamentary convention that no MP is allowed to discuss a health system other than the NHS and the US system?

  40. @ Statgeek

    “I get different data”

    Clearly there are any number of different ways of compiling the figures. For this ‘spoof’ – as @ TR dubs it – I stuck with YouGov tallies alone. One reason for this is that other polls come along at irregular intervals and use markedly different method. I felt that the intrusion of these numbers would disguise any trends that might otherwise emerge.

    Nothing but good can come of offering a different perspective on these issues. In what ways do you think my approach might be introducing systematic biases? Where do *you* think the SNP curve is heading?

    Note that the projection of 39.7% today was based on an equation based on input no later than November, and so it doesn’t enjoy the benefit of taking account of recent polling data. Given that, I doubt whether we’ll find it is too far adrift of tonight’s YouGov figure when it turns up.


    I thought it was a joint Sky/Channel 4 debate.

  42. Case in point:

    Member of the public: we should look seriously at the Western European systems.

    Diane Abbott: the American system isn’t good!


    In any case, you can get Sky News on Freeview channel 132.

  44. As for Galloway, his party only has three candidates selected and one of them is him. They’ve as much right to be on the debates as Justice For Men and Boys, the Pirate Party and Yorkshire First.

  45. Is Populus the new TNS/BMRB?

    If so, they’ve now taken on the mantle of being the most accurate, reliable and trusted of all the pollsters. TNS disappointed me greatly with their last poll, caddishly reverting to the mean after their two excellent 7% Labour leads, but it’s good to see that Populus are now striking out as the pollster of choice. The Gold Standard, if you like; giants amongst a motley collection of permanently inaccurate polling pygmies!



    It is funny how the posters on a certain other site no longer see ICM as the Gold Standard.

  47. At this point it looks like almost equal seat counts are being predicted and nowhere near enough even to form a wobbly minority government. Probably not much point celebrating if your party happens to be largest by 1 or 2 seats.

    I wonder how many parties will be in the next government? I suppose the old two-party system had to end sometime.

    NI is one where I can see the logic in having a regional debate with SF, DUP, UUP, SDLP and Alliance included.

    In 2010 I seem to recall being able to watch the national NI, Wales & Scotland debates via the full set of BBC1 variations available on Sky. I have a vaguer recollection of watching an STV one but neither access to nor recollection of publicity for UTV or ITV Wales debates.

  49. KEITHP

    My memory might be playing tricks, but most of the academic predictions up to 2014 showed the Tories being ahead on seats as their polling improves due to swingback. The longer this polling improvement doesn’t occur, the more the models will show Labour taking the most seats.

    Therefore, the seat predictive models are moving in favour to Labour. If the polls stay as they are (a big if) they should move Labour to being ahead on seats as we get nearer to polling day.

  50. The proposal does seem to be for a joint Sky/C4 show, which I hadn’t realised – pretty sure the 2010 wasn’t, so I withdraw my comment!

1 5 6 7 8 9