YouGov’s first poll of the year is out tonight, with topline figures of CON 31%, LAB 34%, LDEM 7%, UKIP 14%, GRN 8%.

YouGov have made a couple of methodological changes to start the election year. The first and most interesting is to include UKIP in the main prompt for voting intention. Prompting is something I’ve written about here many times before, most recently here. It’s tricky because it really can make a difference, yet there is no real way of knowing when it is appropriate and when it isn’t. There are instances when pollsters have overestimated the level of support for minor parties because they prompted when they probably shouldn’t have (YouGov & UKIP in the 2004 European election, and the Greens in the 2007 Scottish election), but go back to the 1980s and polls that failed to prompt for the Liberals & SDP tended to underestimate their support. It’s clear from history that you can both get it wrong by prompting when you shouldn’t, and get it wrong by failing to prompt when you should.

I’ve written before about the difficulties of making the judgement call on this. There is no obvious way of drawing the line – whenever I write about it in the comments section people make helpful comments saying “why not if the party is in third place” or “if the party is over x%” or whatever… but all these are utterly arbitrary – perfectly reasonable in themselves, but no help in getting it right. It’s not even clear what we should be looking for – it it a certain level of support, or a level of public awareness and familiarity, or a level of media coverage?

In the event however the difficult decision pretty much made itself. It’s something YouGov have been quietly testing on and off over the years, and it seems to be making less of a difference. Testing at the end of last year showed about the same level of support for UKIP in prompted polls as in unprompted polls. Presumably UKIP are established enough in the public mind for prompting not to make a difference, at which point the decision became a simple one. Note that the fairly low UKIP score in today’s YouGov poll – 14% – is not a result of the change, in our testing last year we were showing UKIP at around 17% when prompted, at at time when they were around 16%-17% in unprompted polls.

With YouGov shifting over it means three companies (YouGov, ComRes and Survation) now include UKIP in the main prompt, Populus, Opinium, Ipsos MORI, ICM and Lord Ashcroft polls do not. When ComRes made the switch they too seemed to show very little difference in levels of UKIP support (though their online polls seem to have made some changes to turnout weighting at the same time) but just because prompting doesn’t make much difference to YouGov polls, it does not follow that it won’t make a difference anywhere else – in particular, the impact of prompting may be very different in an internet survey with two pages to click through than in a telephone survey with a human interviewer. What is the correct approach for one sort of polling will not necessarily be the correct approach for another company’s polls.

The other change in YouGov’s methods is much smaller, a tidying up of the sample spec to try and reduce some of the oversampling and reducing the amount of weighting needed. The overall quota targets are the same and the weighting remains exactly the same so there should be no difference at all in the published voting intention figures. The only difference anyone might notice is in crossbreaks: YouGov have started to include ethnicity in the sample quotas for London, which may have an impact in the London crossbreak.

UPDATE: I somehow managed to miss the first Populus poll of this year this morning – figures there were CON 34%, LAB 36%, LDEM 9%, UKIP 12%. Tabs are here.

214 Responses to “YouGov/Sun – CON 31, LAB 34, LD 7, UKIP 14, GRN 8”

1 2 3 5
  1. First

  2. 5,2,4,3 seems yougov are finding a pattern.

  3. Listening to some of today’s politicking and premature electioneering, I was reminded of that old dictum about campaigning in poetry and governing in prose. Sadly, it looks as if most of our campaigning is conducted in prose too now, and pretty badly written stuff as well. Martin Kettle, slowly returning to the fold now after a lengthy dalliance with the Coalition, wrote a good piece about this recently, talking about the need for politicians, particularly progressive ones, to engage people’s emotions as much their rational side. I think there is space for emotion and feeling in our political conversation and I yearn for it; an appeal to the sense of justice that lies within most of us, a generation of excitement about what could be, something that goes beyond sterile debates about financial instruments, tax and public spending. Bush once talked, ham fistedly as usual, about the vision thing, but he was on to something, probably recognising that his political persona couldn’t or wouldn’t generate the elusive elixir.

    Politics shouldn’t be about who is most effective in scaring the electorate about the other lot, or pandering to and generating resentments and grievances, even hatreds sometimes. This negativity isn’t the preserve of either the right or left, both are guilty, but I agree with Kettle that the emotional call to arms is usually most effectively deployed by the political left. As a man of the left myself, I sincerely hope Miliband eschews the yahboo and bickering and sticks to what I thought he did quite effectively in parts of his speech today; quietly outlining why he thinks he can make Britain a fairer and more decent place to live in that it is today. It’s the high road to Downing Street in some ways, and he can’t afford to totally leave his street-fighting skills at home, but while it might be the more precarious route, I think he might be pleasantly surprised by the number of voters who may follow him on his whimsical jaunt. Brits don’t generally do spite and bile, and Miliband, a deeply flawed political leader in many ways, might just do decency better than all the others.

  4. But how does you gov explain being so out of synch with other polls re libs v greens? Sorry but these figures are so different to other polling that it doesn’t add up. What are you gov doing in terms of lib and green questioning that results in greens being ahead or at least level yet most other polls show greens with less than half the lib dem support. Seems rather odd to me.

  5. Could it be they’re prompting green but not lib dem to get this result?

  6. Three poll now for 2015, and seemingly continuing contradictions. Labour ahead in all of them, but the Opinium poll wasn’t great, while Populus and now YG look better, although nothing that can be distinguished from MoE.

    I’d tentatively suggest that this provides the first indication of the festive break meaning no real change in polling numbers, and it will be interesting to see if there is any traction gained by either side in the first few days of politics this week.

  7. Does UKIP angling for Labour votes – by all the ‘liblabcon westminster bubble’ propaganda just make the Greens more acceptable as a protest rather than UKIP itself?

  8. @Ashley re. Lib Dems & Greens –

    Opinium had them level on 6 / 6 in their last poll before Christmas
    Mori had them level on 9 / 9;
    TNS had the Greens 2 points ahead on 7 / 5
    Ashcroft frequently shows them neck and neck.

    It’s true that other pollsters show the Lib Dems far ahead of the Greens, but it’s far from YouGov showing the rogue results you imply, and some of those seem to be due to aggressively weighting to past vote. Bear in mind the two parties’ votes in 2010 were more than 20 points different.

  9. In relation to the methodological issue of prompting, I have a number of questions about the status of Lord Ashcroft’s ‘considering your own constituency’ prompt. In this case it seems to have been accepted universally that the constituency prompted question provides the best VI measure to use. But I can’t find any evidence at all in support of doing this.

    Anthony (or anyone else), can you point to any evidence that real voting patterns are more accurately predicted by Constituency Voting Intention (CVI) figures than by Standard Voting Intention (SVI) measures?

    As far as I can tell, Lord Ashcroft only started using his CVI question (Question 3 in his questionnaire) a couple of years into the present election cycle and it seems that the measure has never been validated against actual voting behaviour. At one point I thought I might be able to check the relative predictive values of CVI and SVI by looking at the data for the six polls he has done just before by elections. However, in these cases it turns out that he doesn’t use either of his standard VI questions. Instead, he has a special prompt reminding the respondent of the circumstances of the by election and proceeds from there.

    So I haven’t been able to turn up any evidence at all that we should rely upon CVI rather than SVI data.

  10. On another tack, I have recently been trying to look at the possible effects on election outcomes of future changes in Ukip VI.

    Some of the influences turn out to be rather complex and unexpected.

    As before, I start with the current Electionforecast Nowcast database and I then apply VI changes to the various parties to model possible developments between now and the election. (For background and details see my comment uploaded at 3.54 pm on Jan 1st, appearing on p.2 of Anthony’s Dec 31 post.)

    For analyses that assume a *rise* in Ukip VI, using figures from Anthony’s November 7 churn analysis I assume that 50% of the VI comes from the Tories (and so reduces their projected VI by that amount). Using the same figure, I assume that 19% of the VI increase comes from Labour and 15% from the LibDems.

    In modelling any Ukip VI *falls* over the coming four months I make a slightly different set of assumptions. That is, I don’t assume that current Ukip support will automatically head straight back to the party it came from. For these analyses I use the results of a fairly recent poll published by Lord Ashcroft:

    In this poll, respondents were told to imagine they were using a transferable vote system and were then asked which party would receive their second vote. (3rd vote, too, but I have ignored that.)

    For respondents whose first choice was Ukip, 51% of the 2nd votes went to the Tories, 27% to Labour, 8% to the LibDems, 9% to the Greens (and the last 5% to other parties or ‘won’t vote’). I have assumed that if current supporters drop away, then their support will shift elsewhere according to the same pattern. In other words, a 1% Ukip VI drop is linked with w 0.51% Tory rise, a boost of 0.27% for Labour and so on.

    Using these figures I have tracked seat projections ranging from Ukip VIs 1% above my standard (Continuing Trends) projection down to 6% *below* this figure.

    Over this range the following ten seats are in play for Ukip:

    Boston and Skegness
    ++++++Continuing Trends minus 6%++++++++
    South Thanet
    Great Yarmouth
    South Basildon and East Thurrock
    South West Norfolk
    Louth and Horncastle
    ———-Continuing Trends———-
    Rochester and Strood
    St Austell and Newquay

    The calculations suggest that Ukip will hold Clapton and gain the next two seats in the list even if their VI falls away quite sharply before the election. The next five seats are expected gains under Continuing Trends assumptions, and only if they do a little better than that will then hold Rochester and Strood and gain St Austell & N.

    All such gains come from the Tories and so – as expected – they look set to lose a few seats if the Ukip VI holds up.

    With analyses of this kind it is interesting to track the effects on other parties that are less closely involved in the action.

    Under the (perhaps unlikely) assumption that Ukip support were to rise to the top of the range, the calculations suggest that the following five seats would pass from the Tories to Labour: Blackpool & Cleveleys; Brigg and Goole. Halesowen and Rowley Regis; Stevenage; and Hove. (This happens because the Tories are penalised more than Labour and so lose control of some of the more marginal seats).

    In the (arguably) more likely event that UKIP support *falls* there is some exchange of seats in the opposite direction, but interesting Labour suffers very little net damage.

    Specifically, with the biggest Ukip VI drops I looked at Calder valley, Norwich North and Worcester pass from Labour to the Tories. However, as part of the same process Labour GAINS three seats from other parties (namely Arfon, Bermondsey and Old Southwark and Bristol North West.)

    It is worth looking at each of these in more detail as they provide some indication how complicated multi-party elections can be.

    Bermondsey and Old Southwark (Simon Hughes’ seat) is teetering on the cusp according to my Continuing Trends analyses. (A personal vote or a Swingback in support of the LibDems could remove this seat from risk, but the calculations are what they are.). On these figures the seat stays with the LDs if current trends continue, but if Ukip support drops away then Labour benefits from this more than the LDs do, and the seat therefore changes hands.

    Bristol North West is down as a Green gain if current trends continue (favouring the Greens and further punishing Labour). However, the present analysis shows that this is by no means a free ride for the Greens. If Ukip support drops, the released votes flow to Labour as opposed to the Greens in a ratio 3:1, and the seat stays in present hands.

    Even more remote is the Ukip impact on Arfon. The Electionforecast Nowcast has this down as a very marginal Labour gain. Continuing Trends put it back with Plaid Cymru (as a result of the assumption that Labour is on a steadily declining trajectory). However, if UKIP fades the votes go preferentially to Labour and the seat is back in Labour’s hands again.

    Summary: The calculations show remote effects of the *type* I was expecting. But I, for one, wouldn’t have had a clue where these effects would have manifested themselves. If Swingback emerges in other party VIs then the battleground would shift to a different set of marginals. But I would expect the same general phenomena to reappear. Personally, I had expected Labour’s seat tallies to be much more vulnerable in the face of Ukip VI drops. The fact that the counts hold up is due in part to the way in which departing Ukip support is redistributed. It may also be that there are not that many seats with Labour margins that are small enough to be affected by the swings in question. Another factor that might assist Labour is that in 2010 the marginal seats had a lower-than-average Ukip presence. Given that there is at least some evidence of proportional growth (in addition to an intercept change), there is reason to believe that in Labour/Tory battleground seats there will be fewer Ukip supporters available to convert to other causes.

  11. On Newsnight , Electionforcast. co. uk estimating most likely outcome to be Lab 286, Con 280 withe UKIP on 3 and Greens on 1

  12. Due to the strong end to the year, it seems inevitable that pollster to pollster Labour’s lead will fall when the next poll comes out, the 5 and 7 for ICM and Opinium will be unlikely to be repeated.

    Could result in some favourable headlines for the Tories, just as the first inklings of a policy or two are being heard.

  13. @Unicorn
    I have similar doubts. In particular, I wonder how many voters, after an election campaign conducted nationally, will retain their “thinking about your constituency” VI. Is the question in fact prompting a more byelectionish response that will be forgotten during a general election?

    It’s an interesting experiment, and we’ll soon have some data on how effective it is. Interpreting that data is going to be a minefield.

  14. Unicorn

    Thanks for that post. That’s given me a greater insight into what might happen in E&W than I’ve seen for a long time.

    If we assume that UNS is now a dead duck, I wonder if some kind of UGS (Unified Group Swing) might be developed – where groups of seats which behave in a similar fashion could be identified.

  15. The figure to keep checking is the Tory VI: will it climb?

  16. Only three polls so far in 2015, so it’s a bit early to determine whether the trend that established itself over the two weeks before Christmas, showing an increasing Labour lead and VI, is maintaining itself. Opinium says not, and shows a sharp fall in both Labour’s VI and lead, but Populus shows a Labour lead of 2% as opposed to a tie pre-Christmas. Tonight’s YouGov is broadly in line with its last four or five polls of 2014. Pick the bones out of that, if you will, but what we can say is that there is no hard evidence that the sands are shifting in any appreciable way. Opinium was first out of the blocks and set some pulses racing, but today’s Populus and YouGov are suggesting things are broadly as they were.

    Fascinating stuff and now that the campaigning wraps are coming off, let’s see how things move from hereon in.

    It feels like game on now and I hear the gathering noise of the hustings approaching. Time to get those old campaigning boots out of the attic again and I’m getting as excited as a little kid on Christmas Eve!


  17. Bristol West not North West surely? And even then it’s a huge stretch to imagine a Green gain.

  18. Crossbat,

    It’s all change now. I’ve got a stack of “Students for Oliver Coppard” posters sat here on top of student petition clipboards and I’m monitoring traffic for “Clegg”, “@olivercoppard” and “Sheffield Hallam” on Tweetdeck.

    Of course none of that replaces early-morning trudging up to Crookes or Ecclesall to bang on doors. Something traditional and endlessly rewarding about that.

  19. @Unicorn

    Thanks – fascinating analysis.

  20. @R&D

    “The figure to keep checking is the Tory VI: will it climb?”

    Up in Opinium, down in YouGov and Populus, but you’re absolutely right; the Tory VI, and how it now behaves, holds the key. I would think they need to see a steady rise across all the pollsters pretty soon now if they’re going to be in a winning position by May. It could be why they’ve gone early on the campaign front, with all guns blazing.

    Big Mo needs to start rolling for them now, methinks.

  21. Ashcroft’s marginal polling is to be released on Wednesday fieldwork is being done now. I am sure you will be really happy to hear that it inclues … Scottish Marginals incuding my neighbouring seat Dundee West and Danny Alexanders’s Inverness seat.

  22. @Peter Bell

    Current betting market predictions
    Con 281 to 285
    Lab 286 to 288*
    LD 27* to 28
    SNP 25 to 27
    UKIP 6* to 7
    Green 1

    * shows that the figure marked could be slightly above/below (as appropriate) this end of the range, as odds are unbalanced

  23. COUPER2802
    Ashcroft’s marginal polling is to be released on Wednesday fieldwork is being done now. I am sure you will be really happy to hear that it includes … Scottish Marginals including my neighbouring seat Dundee West and Danny Alexander’s Inverness seat

    Should be interesting.

  24. @Ashley

    Unless I’m mistaken YouGov prompt for Lib Dem but not for Greens, so actually the opposite of what you’re suggesting.

  25. @PostageInc, @Unicorn

    I would have more confidence in the “thinking about your constituency….(etc)” question if it were the first question asked rather than the second. Asking it as the second is with a very slightly different wording is almost inviting a reappraisal of the first answer.

    i.e. The secondary nature of the question means that it strays a little down the road towards prompting along the lines of “But remember that this is (say) a LD/Con marginal and your MP is X. Are you sure about that last answer?”

  26. @Hannah

    Having just read Unicorn’s post I was about to type exactly what you put. You win. (Don’t the ladies ALWAYS win?).

  27. @Phil H

    I think I said something similar re the question order a couple of years ago and Mr Wells had a smart answer. Unfortunately I’ve forgotten what that answer was! Perhaps he will remind us.

  28. @ Phil Haines

    I agree entirely.

    Imagine you are polling respondent and you have just answered Q2 (..which party would you vote for?) saying Party A or Party B. You are then faced with a new question carrying presuppositions that you might have missed something in the answer you have just given, and you might therefore like to reconsider your answer. (Linguistic philosophers call these presuppositions “implicatures”). This is a leading question carrying the suggestion that you first answer may be imperfect, and urging you to give a more carefully considered response.

    Well, this may all be fine. (As i said above, I can’t find any evidence for or against highlighting such responses.). But it does raise the issue of whether it is the open question (Q2) or the leading question (Q3) which is the closer to simulating the circumstances of the polling booth.

    In favour of CVI the real voter will cast his/her vote within the constituency itself and will be required to do this using a ballot paper listing all the candidate names (as well as their respective party affiliations). They may also have received recent constituency-relevant leaflets and/or canvassing visits. Against CVI, however, most of the campaign material will have features national leaders and nationwide issues rather than local material. The tone of the debate will have been set by speeches reported by the national media, by leaderships debates (if they go ahead), by manifesto material, by (again national) party political broadcasts. Given all of this, it seems to me that the ‘polling booth mindset’ could be much closer to that encapsulated by Question 2 rather than one in which the voter is primarily considering their own constituency.

    The problem is that it makes an enormous difference to what we make of Ashcroft’s polling evidence. Quite a high proportion of his polls have one party ahead on SVI and the reverse on CVI.

    Despite the doubt we both obviously share about placing trust in this measure there seems to be a complete consensus among the professionals that it is the right thing to do.

    Specifically, CVI figures are routinely used by Lord Ashcroft himself when he drafts his his own posts summarising each batch of results. The same figures are then also used by Anthony when he posts constituency polling threads on this site. Similarly, CVI snd not SVI tallies are used by Electionforecast in their exercises benchmarking the performance of their model. (Given this, it seems a safe assumption that the same figures are also the ones they use to update their database). Moreover, CVI figures are used by Wikipedia to post UK constituency poll results on:,_2010-5

    Where is the justification for taking this course of action?

  29. @ Postageincluded

    Apologies. I meant to acknowledge that you had made the same point as well.

  30. And now, just to remind us what we’re looking at, I bring you… the December 2014 churn report! It was a short month for polling but one with some interesting developments. We’ll soon see whether they’ve been sustained into the new year.

    The State of the Parties

    The Conservatives remain becalmed on 32%. After collapsing like a chancellor’s commitment to fiscal consolidation in an election year, Labour briefly dipped down to parity with the Tories before staging an abrupt recovery at the end of December. Meanwhile, in the land of the submerging parties the Lib Dems continued to decline, allowing the Greens to overtake them as the formal fourth party by mid-month. The SNP and Ukip appear to have stabilised on their new plateaus, which will give the SNP all of Scotland and Ukip all of Clacton if they can sustain their positions until May.

    The Conservatives

    The Autumn Statement didn’t cost the Tories another 4% of the vote share, so I guess as George Osborne interventions go it wasn’t a disaster, but it doesn’t appear to have helped them at all either. They remain firmly stuck on 32%, and so far no one has produced a lever big enough to dislodge them. Those giddy August heights of 34% seem like a distant memory now.

    Not much to note about the churn. There was a mid-December dip in Con -> Ukip switching which promptly reversed, possibly from getting a bit more media coverage around the Autumn Statement. Con -> Lab switching has overtaken Lab -> Con switching again but that’s probably just noise.


    Ed Miliband has clearly been a good boy this year, because after a ghastly November Father Christmas brought him a modest recovery at the end of December. Has the Labour Party finally got its act together? This is Labour we’re talking about so the safe money is on “No”, or possibly on “Lol no”, but thanks to the good offices of Messrs Cameron, Osborne and Clegg it may not matter. The racist candidates in Ukip also deserve a shout-out here, as it appears the shine may be going off Nigel’s pint glass.

    Whatever the cause, the Red team will be relieved to see that trendline finally moving in the right direction, although an average of 34% isn’t something we’d have been celebrating six months ago.

    The recovery seems to be due to a modest increase in retention and LD -> Lab switching, and a corresponding fall in Lab -> Ukip switching. Perhaps working class people call Chinese food “Chinese food” after all?

    The Liberal Democrats

    The Lib Dems are astoundingly resilient… in their capacity to continue losing voters long after most of us thought they’d been pared down to an irreducible core. How low can they go? 6%? 5%? Only Clement Davies’ ghost knows for sure!

    The latest exodus is to Labour and the Greens. LD -> Tory and LD -> SNP churn is steady, LD -> Ukip churn is down.


    Ukip VI doesn’t move much without imminent elections to drive the churn, and with no new by-elections they’ve stalled on 15.5%. This is simultaneously much better than anyone expected them to do at this stage of the election cycle and not good enough to win more than a handful of seats.

    Oh well, if the goal is to prevent Cameron from holding an EU referendum and securing an “In” vote, they’re still firmly on course.

    Normally Ukip churn from all the parties moves in tandem, but in December Lab/LD -> Ukip fell while Tory -> Ukip churn spiked. Were voters scared off by the Conservative cuts agenda, with Redkippers moving back home to keep the Tories out and former Tories defecting to Ukip? Have we discovered the “chinky” gap between left and rightwing voters? Is it just noise? Who knows, but it’s a bit odd.

    The Greens

    The Greens continue to shoot upward like a little sapling with no plateau in sight. Whether or not the tree will bear fruit anywhere but Brighton Pavilion remains to be seen, but they can be very, very happy with this trend.

    Interestingly, their December gains seem to be almost 100% from the Lib Dems, with Labour contributions leveling off in November and the Tories back in May.

    The Nationalists

    The SNP survey the Scottish countryside from the commanding heights of their massive, massive poll lead and laugh. They’ve leveled off, but this is probably because everyone in Scotland is already voting for them.

    (Plaid sit in a corner and cry, most likely, but we’ll never know because no one is polling Wales. Sorry Plaid.)

    I hereby dub that huge spike in Lab -> SNP churn in mid-December the “Murphy Spike”.

    Contributions from Labour and the Lib Dems appear to have stabilised, but who knows… Certainly the Murphy-based recovery predicted by certain commentators on the Labour right has yet to make itself known.

    Don’t Knows and Not Voting

    All is stable and boring. After teasing us with a brief moment of crossover in the autumn, Labour DKs have returned to their usual position below the Tories, but they’re still relatively high by Labour standards. Lib Dems dipped a bit but seem to be rising again; it was probably just noise.

  31. Okay, churn report is out whenever Anthony releases it from auto-mod. He’s probably gone to bed like a sensible person, so check back tomorrow morning if you’re interested.

  32. @Unicorn, Phil H

    On reflection, I think AW pointed out that after being asked CVI, a respondent wouldn’t then be able to make a sensible answer to the SVI – in effect you’d have to imagine what you would have thought if you hadn’t been reminded about your constuency – a difficult hypothetical question.

    I suppose just asking CVI might be best. But testing that would mean polling twice. Exactly how rich is Ashcroft?

  33. @ Unicorn,

    Fascinating analysis. (And rather reassuring for Labour on the Ukip front.)

    I’m starting to wonder if UNS will give roughly the right seat totals even if the predictions for each individual seat are wildly off, just because there’s so much churn flowing around in all directions!

  34. @ Hannah

    “Bristol West not North West surely? And even then it’s a huge stretch to imagine a Green gain.”

    Yes – you’re right. It is Bristol West that flips back and forth between the Greens and Labour. I got this right in most posts the other day about Green projections, but managed to get a bit confused today.

    The reason for this is that Bristol North West is is *also* down as a marginal seat that flips back and forth in this analysis (this time a Labour/Torymarginal). At this time of night the names were too similar for me to get all the details right in my earlier summary.

    Sorry about the glitches…

  35. @Spearmint

    That wouldn’t surprise me, but as the Queen of Churn your word carries more authority. Looking forward to your piece, as always.

  36. Spearmint

    As always, your churn report is both informative and (more importantly) entertaining.

  37. @ Peter Bell

    “On Newsnight , Electionforcast. co. uk estimating most likely outcome to be Lab 286, Con 280 withe UKIP on 3 and Greens on 1”

    But bear in mind that the *forecasting* component of the Electionforecast model includes a good dose of Swingback (or regression-to-mean, as they call it). For example they predict that the LD and Ukip VIs will cross back over again, taking the former to 13.9% and the latter to 10.7%.

    It might just happen, but there is no hint of it in the data with just four months to go. Until we see some positive evidence for regression to the mean I am inclined to treat the forecasting part of their model with a health dose of scepticism.

  38. …How many votes will UKIP take off the Tories….?…..that is surely what will determine the next election… guess is enough to deny the Tories a victory as predicted by most polls……but also enough for Miliband to form a coalition with the SNP……yes UKIP will also hit Labour but nothing like the Tories……the key to all this is Nigel Farage…..he will be on the TV non-stop for a month……and he is very good on TV…..this will make a huge difference……and he has some momentum and more credibility now with 2 MPs. The Tories need to attack him….but are afraid of upsetting UKIP voters by attacking their policies….which are by and large…the policies most Tory MPs and voters agree with.

  39. @ Unicorn


    And in making the cross-back comment I had intended to refer back to the poll at the top of the thread – which as the the Ukip VI at double the level of that for the LibDems…

  40. @Spearmint

    Thanks for the churn report, it explains things quite nicely as always, looks like that small 2%ish Labour recovery towards the end of Dec was about half Lib Dems who moved Green moving back to Labour, and half UKIP moving back to Labour. I guess that was the impact of the ‘back to the 1930’s’ scare means some people started to see a difference between Labour and the Tories again.

    Also good to see that the Yougov London crossbreaks now include ethnicity – hopefully it reduces some of the volatility and gives us more accurate polls.

  41. @Guymonde

    Thanks for that link & I agree the questions posed are strange in that they set scenarios but don’t permit respondents to give the answer Both.

    An example given here but all follow the same pattern:

    Thinking ahead to 2015, which, if either, of these would be better for Britain?
    – Give head teachers and hospital managers more freedom to decide local priorities, and set fewer national rules on running schools and hospitals
    – Means test age-related benefits, so that better-off pensioners no longer receive free bus passes and the winter fuel allowance
    – Neither
    – Don’t Know

    In this particular scenario, there is no relationship between the policies & I can imagine many people might choose to both give HTs & hospital managers freedom to decide whilst also thinking means testing OAPs would be a good idea.

    Very strange indeed.

  42. @Anthony

    On crossbreaks in general, I noticed when Yougov did their big analysis of the young vote they seemed to weight that by work status, which is not something I recall seeing on the normal crossbreaks for the daily Yougov’s. (and did not weight by newspaper readership) I assume that is done because like ethnicity in London, work status amongst the young also results in very different VI. – see tables here:

    Is that something that is done on the daily polls but is just not visible?

    Could polls become more precise by using different weightings for different sub-samples, or would that just make the whole thing too hard to do/ not worth the effort?

    If I think of the most volatile crossbreaks, regionally it tended to be London (hopefully fixed now), and age-wise it seems to be that 18-24 age bucket. (where you don’t have past VI to be able to politically weight?)

  43. WRT – “regression to mean” based projections.

    Like other people on here – i really cant see this lib dem recovery happening. Their ratings have got even worse over the past year. Their goose is well and truly cooked.

    Its all fascinating stuff and I think their are way too many varibles for any projection models to be relied on.

    has any research been done on how voters will behave in marginal seats as opposed to the rest of the country?

    I suspect that the likes of greens and UKIP will do well in seats with big lab or tory majorities. I.e i live in l Rachael reeves constituency – its safe labour. I would be far more likely to vote green or other left than if i lived down the road in greg mullholland’s Lib dem seat – where labour would be the main challenger and a vote for green or other left would help him keep his seat.

  44. With previously ignored parties gaining traction, it’ll be pretty interesting to see the constituency results in safe seats, because they are probably more susceptible for minority party growth. In the safe seats, people can vote for who they really want, rather than tactically, which means we can expect a share more in line with what you’d expect from PR rather than the more bipartisan result you’d expect from a marginal. Such a result can be built upon in further elections. Would it therefore be fair to say that safe seats are actually better for outsiders like UKIP and the Greens in the long run?

  45. Reading the comments, Reggie seems to think the same. It’s pretty fascinating, really, how ‘safe’ seats could paradoxically be a breeding ground for their own demise.

  46. “…How many votes will UKIP take off the Tories….?…..that is surely what will determine the next election… guess is enough to deny the Tories a victory as predicted by most polls……but also enough for Miliband to form a coalition with the SNP……yes UKIP will also hit Labour but nothing like the Tories……the key to all this is Nigel Farage…..he will be on the TV non-stop for a month……and he is very good on TV…..this will make a huge difference……and he has some momentum and more credibility now with 2 MPs. The Tories need to attack him….but are afraid of upsetting UKIP voters by attacking their policies….which are by and large…the policies most Tory MPs and voters agree with.”

    Bloody hell! Wot’s with all the dots?

    If it is supposed to add to ease of reading think again.

  47. @Spearmint

    Thank you very kindly for the churn analysis.

  48. ALEC
    ” no real change in polling numbers”

    From tied or close 31/31 in mid-December?

    You have the Bluekilppers on 0.51 and the Redkippers on 0.27, but in terms of movement back to Mummy, is there not evidence of a qualitative difference which should be factored in – with a last in first out element evident Redkipper and Lab Green movers in the possibly now consistent and upward moving Lab 34-36. The upward and downward movement 12 to 17/18 s UKP VI and its clear linkage to movement in the Lab and Con VI seems to me the one to watch, not so much the lead of one over the other.

  49. Yougov Scotland in numeric descending order.







  50. Unicorn

    Bear in mind electionforecast model does take into account of the uncertainty in the levels of swing back looking at their seats based on 90% (two tailed) confidence intervals. No swingback is certainly within the scope of their predictions.

    Lab 246-331
    Con 238-319
    Lib 17-35
    SNP 22-44
    UKIP 1-6
    Greens 1-2

    I’d be pretty confident all of the results will lie within this range.

    I’d say the underlying prediction is one of great uncertainty in the numbers of seats, although with a more certain prediction of an OM for either party being very unlikely.

    There’s two aspects to the model, one is predicting swing back (as a low amount of drift with a pretty high level of diffusion as “events” have historically outweighed the historical “swingback”) based on data, albeit with a small amount of real data leading to large amounts of sampling error on these terms.

    As more elections pass, one might expect to get more sophisticated “swing back” models. It’s plausible that a party like UKIP rising from a very low level will suffer less swingback than a major party, but until the data shows it trying to fudge a model to account for “gut” is an unscientific way to go about creating a model.

    I’m fairly supportive of anyone attempting a model that doesn’t rely on “gut feeling (which probably incorporates a large level of personal prejudice)” and instead relying purely on data (using experience to create the models and correcting them where they fall short instead of adding a fudge factor based on “gut feeling” from election to election)

    The other side of the model is to create a “better than UNS” model for predicting constituencies. Again there will be large levels of uncertainty due to the few election data points but I suspect their model (with a good dose of Ashcroft polling) will better take into account where things deviate from UNS models.

    One area of criticism might be their predictions are too vague, although probably less vague than the range of “guts” out there if you compare odds (a pseudo measure of cumulative current gut feeling) of a hung parliament being much longer than their predicted likelihood of an OM.

    I’d rather have a model which has a large level of uncertainty (but at least reports that uncertainty) than try and pick the best gut from a forest of guts with wildly varying outcomes.

1 2 3 5