14 weeks to go

Week four of the year we had the regular YouGov, Ashcroft and Populus polls, the first ComRes telephone poll of the year and the first 2015 GB poll from Survation – the first in a regular series for the Daily Mirror.

YouGov/S Times (23/1/15) – CON 32%, LAB 32%, LDEM 7%, UKIP 15%, GRN 7%
Survation/Mirror (25/1/15) – CON 31%, LAB 30%, LDEM 7%, UKIP 23%, GRN 3%
Populus (25/1/15) – CON 34%, LAB 35%, LDEM 9%, UKIP 13%, GRN 6%
Ashcroft (25/1/15) – CON 32%, LAB 32%, LDEM 6%, UKIP 15%, GRN 9%
ComRes/Indy (25/1/15) – CON 31%, LAB 30%, LDEM 8%, UKIP 17%, GRN 7%
YouGov/Sun (26/1/15) – CON 34%, LAB 33%, LDEM 6%, UKIP 15%, GRN 7%
YouGov/Sun (27/1/15) – CON 34%, LAB 33%, LDEM 7%, UKIP 14%, GRN 7%
YouGov/Sun (28/1/15) – CON 33%, LAB 33%, LDEM 6%, UKIP 16%, GRN 7%
YouGov/Sun (29/1/15) – CON 34%, LAB 34%, LDEM 6%, UKIP 14%, GRN 7%
Populus (29/1/15) – CON 34%, LAB 35%, LDEM 10%, UKIP 14%, GRN 4%

The polls this week continued to show an extremely tight race – every single poll had the two main parties within one point of each other, and unlike last week there were slightly more polls with the Tories ahead than with Labour ahead. The UKPR average though still has figures of CON 32%(nc), LAB 33%(nc), LDEM 8%(nc), UKIP 15%(nc), GRN 6%(-1), as Opinium and ICM polls from last week are still contributing towards the average. For anyone interested in the differences between some of the polls from different companies, I explored them in this post earlier this week.

Welsh polls

There were also two Welsh voting intention polls out this week, the regular YouGov/ITV/University of Cardiff poll and an ICM poll for the BBC. Westminster voting intention figures for the two polls were:

ICM/BBC – CON 21%, LAB 38%, LDEM 7%, Plaid 12%, UKIP 13%, GRN 6%
YouGov/ITV – CON 23%, LAB 37%, LDEM 6%, Plaid 10%, UKIP 16%, GRN 8%

Week four

  • At the beginning of the week there was a lot of froth about UKIP’s NHS policy and the Green party’s policies on membership of extremist groups and a citizen’s income. It’s unlikely that either will make much difference for the simple fact that most people have no idea at all about what their policies are on such issues. For UKIP, the majority of people think they have at least a fairly good idea of what sort of approach they would take on immigration and Europe, but on other subjects people draw a blank. For the Green party 54% think they’ve got some idea what the Green party would do on the environment, but on everything else at least three quarters know nothing. It doesn’t necessarily stop people backing them, as broad perceptions of a party’s values, principle and competence are far more important than specific policies anyway. I suspect that maybe even more the case for parties who have no realistic chance of getting a majority and putting said policies into action.
  • As we passed the 100 days to go mark both Labour and the Conservatives put out new policies, Labour on the NHS, the Conservatives on welfare benefits. The Conservatives headline pledge to reduce the benefit cap to £23,000 was supported by 61% to 25% (including amongst Labour voters), even though people didn’t think it made people look for work. The idea of stopping housing benefit for young people was more divisive – 42% supported the idea, 40% opposed it.
  • The NHS is generally a rock solid issue for Labour anyway – last week they had a thirteen point lead over the Tories on which party people thought would handle the issue the best. Welfare benefits is actually much more contested ground, in the same poll 28% of people thought Labour would handle the issue the best, 28% of people thought the Conservatives would handle the issue the best.


The latest forecasts from Election Forecast, May 2015 and Elections Etc are below. All are still predicting a hung Parliament, though Election Forecast and May2015 have the Conservatives catching up with Labour after a week of close polls.

Elections Etc – Hung Parliament, CON 282(-1), LAB 280(+2), LD 24(+1), SNP 40(-1), UKIP 3(nc)
Election Forecast – Hung Parliament, CON 283(+5), LAB 285(-1), LD 27(-1), SNP 32(-2), UKIP 2(-1)
May 2015 – Hung Parliament, CON 280(+11), LAB 280(-9), LD 24(-3), SNP 38(nc), UKIP 5(+1)

299 Responses to “14 weeks to go”

1 2 3 4 5 6
  1. @ Lurker

    Looking at my Yougov 7 day average chart, it looks as though both Lab and Con VI has edged up over the last week, but remains at a historically low level.

    There may be something in that. In my earlier summary of this week’s trends what I didn’t say (because the evidence wasn’t quite there) was that if just one of the Labour VIs had shifted from below trend to above trend, I would then have been reporting Labour as polling above trend.

    The fact is that they are hovering near the confidence boundary but – unlike the Tories – didn’t pass the barrier this week. It is quite possible that the next batch of polls will see the Labour VI rising above trend as well (in a statistical, rather than numerical sense, that is..)

  2. Unicorn

    Sounds like your quadratic formula is “over fitting” the data.

    I assume you are doing a minimising the squares of the residuals approach to fitting the curve to the data.

    If the model is of the form y = a_0 + a_1 *x +a_2*x^2, I’d try adding a term
    lambda*(a_0+a_1+a_2) to the sum of the squared residuals and minimise the total expression.

    Increasing lambda will have the effect of damping down the data a bit and create something less extreme, while still plausibly fitting the data.

    Ideally you’d keep a separate set of data (about 20% of the data, which is not used to fit the values of a_0, a_1 and a_2) and optimise the value of lamba to fit this backtest.

    I’m not sure at all about quadratic models for long term forecasting as the inevitable will lead to a “blowing up of the prediction, eventually either sending the prediction over 100% or below 0%”. The main reason for disliking them is very small changes now can massively affect the quadratic term and so leading to an unstable prediction, a small blip in the data and in order to best fit, distorts the trendline, magnifying the the relevance of a recent blip in terms of the final prediction.

    You might have more success in trying to fit an AR(1) model. If that is somewhat successful but doesn’t contain enough “momentum”, try an
    ARMA(1 1) model, which should allow for momentum in the short term, with a gentle long term push to stabilise the long term predictor.

    I suspect when trying to model the SNP VI, you’ll want to fit to the post Indyref date as it seems the change in VI isn’t due to “a random walk” but a sea change which will be contained in the variables. (If you were really feeling up for it you could create a 2-part function whose variables depend if they are before of after t_0).

  3. While people earnestly discuss the effect of policies on voters, they might consider the voter described in this tweet –

    “Sat behind lady pensioner on bus as she moaned about benefits, saying they should all be scrapped. Used her bus pass for the journey”

  4. @ Alan

    Lots of food for thought there!

    I rather doubt whether any autoregressive model would do any better. The problem is that there is only a tiny amount of post-sea-change data that can be used to provide any initial fit. (In relation to your final comment, I am not really interested in 2-part functions because I am less concerned with describing the past than trying to predict where the VI function might go next.)

    In the absence of better evidence, reluctantly I think ‘hunchwork’ might be the best bet.

  5. Had to smile. Where else would one read:

    ‘You might have more success in trying to fit an AR(1) model. If that is somewhat successful but doesn’t contain enough “momentum”, try an
    ARMA(1 1) model, which should allow for momentum in the short term, with a gentle long term push to stabilise the long term predictor.’

    Followed by:

    “Sat behind lady pensioner on bus as she moaned about benefits, saying they should all be scrapped. Used her bus pass for the journey”

  6. couper

    “it looks like another Tory government with or without the LibDems.”

    A very strange analysis of current polling.


    “both Lab and Con VI has edged up over the last week, but remains at a historically low level.”

    How does that work??

  7. @Couper

    I’d be fairly surprised if we got any sort of Conservative-led government next time. I’m wondering who would else be willing to support them if they (Cons) had say 295-300 seats, and also enough seats on offer to help them to a combined majority, and also who the Conservatives could cut a workable deal with. There’s few options on offer.

    LD support in the opinion poll list for the week seems to be hovering just above 7%. I think we can safely predict the UKPR will soon drop from 8% to 7%. If you are a large party, you’ll need to get over 300 seats to get a majority in the current and foreseeable situation.

    It seems like the events just after the Feb 1974: Heath trying and failing to find coalition partners for the above reasons.

  8. Rosie

    “it looks like another Tory government with or without the LibDems.”
    A very strange analysis of current polling”

    There seems to be an assumption by many on here and other places.
    Dan Hodges is the perfect example that the polls are not really close, They are all saying it is like 92 mainly because of Ed Milliband.

    This I believe is like parts of the police used to act in the 80s on gut reaction.

  9. It (still) looks the same.

    There will either be a minority government or a coalition of Lab/LD/SNP/SDLP/PC.

    And I (still) cant see it being a minority government.

    LD will not agree to support a CON minority.
    And I doubt Labour will have a shot at it because they will have lost to CON both in number of seats and votes.

    So it will be that coalition.

    The SNP will go along with it because the alternative is to let CON form some sort of government again. And I can’t see the SNP letting that happen.

    So it will be that coalition.

  10. Chris Huhne in the Guardian(featured in a pb thread) argues that a minority government is scotched by the Fixed Term Parliament Act.

  11. Also re the LD position, it seems remarkable that so many predictions now and in the past claim they will suddenly rise from 7-8% to 10-14% when there has been little or no sign of that happening, now or even since they dropped like a stone in 2010.

  12. Colin

    Wouldn’t be Scotched by a C&S arrangement with the SNP?

  13. OLDNAT

    Why would such an arrangement be an exception to the risks Huhne envisages.?

  14. Colin

    Sigh. Capital “S”.

  15. OLDNAT

    Ah yes-:-)

    Certainly more secure than one with PC . They would probably Welsh on the deal.

  16. Seems to moi highly unlikely that Cons will do better than 2010, in votes or seats: they will almost certainly do a fair bit worse.

    The Lib Dems will be much reduced also.

    Given this overwhelmingly likely outcome, on what basis could the Tories govern alone or the Lib Dems prop up a party that has lost support when it has lost support itself for – err – well, propping them up in the first place.

    Whilst I still thank a Labour OM is possible none of us can be certain of anything. But, to me, Labour as the lead party in government seems far more likely than the Cons in that position.

    Even if all the above is bolloc*s I really think that to say:

    ““it looks like another Tory government with or without the LibDems.”

    comes over more like anti Labour bitterness than as the result of keen, analytical logic which takes current polling into account.

  17. PAUL

    @”comes over more like anti Labour bitterness than as the result of keen, analytical logic which takes current polling into account.”

    You mean keen analytical logic like this :-

    “Seems to moi highly unlikely that Cons will do better than 2010, in votes or seats: they will almost certainly do a fair bit worse.”

    -a keen analytical logic which is , at once , “almost certain” and also “overwhelmingly likely ”


  18. NorthumbrianScot – “However it’s 99% likely that Independence won’t be a policy in the SNP manifesto (partly for the Constitutional reason that SNP policy is that the Scottish Parliament should make such decisions, partly for the practical “too soon” reason.)
    Therefore the concerns mainly boil down to the fact that they might support policies you don’t like.”

    LOL, like most Scots you’ve deliberately misunderstood what I was trying to say. Your argument boils down to “OK they told fibs about independence, but that’s off the table so what’s the problem?”

    Our argument is that if they told dangerous fibs about independence they are likely to be telling dangerous fibs in other areas too.

    Here’s an analogy with Greece. It’s unlikely that Greek politicians in the ’90’s knew that the euro would cause severe hardship but thought lets go ahead anyway and who cares about the consequences because there is a political benefit to us. There were theories that it might not work but no-one knew for sure because the euro was an unprecedented experiment. They were fools but they didn’t have bad intentions.

    Now [i]everyone[/i] knows that you shouldn’t be in a currency you can’t control – yet the SNP were wilfully urging their supporters on. Their attitude was, who cares if Scots are being urged off a cliff, it’s a price worth paying. Combine that with the fairy-tales about the oil price to trick people into voting Yes, and you get a sort of wickedness we haven’t seen in a British party before.

    You arn’t experiencing the consequences of that wickedness because the No people pulled you back from the brink – and I guess like the Millennium bug you’ve rationalised that because the crisis was prevented, it would have all worked out OK anyway without anyone’s effort.

    But at some point the penny will drop about just what risks they wanted you to take. And people will start to struggle to maintain the “They fooled me once, I’m determined to let them fool me again, just to defy London”, stance they’ve adopted. Think about this – the English and London and even the supposedly evil Tories were more concerned about the fate about vulnerable Scots than the SNP were.

    From where we’re sitting it’s just too dangerous to have a party like the SNP in government. Yes they can run a regional outfit under strict limitations, but the untrammelled power that comes with Westminster? No.

    If Ed Miliband had any sense he’d rule out a coalition with the SNP – it’s a measure of how poor he is at reading the public mood that he has not.

  19. colin

    It is an analysis based on polling and the history of governments managing to improve their polling figures in a subsequent general election. I have added to that the new idea of a five year parliament and a GE which is not necessarily at the government’s preferred time.

    If the evidence pointed the other way then so would I.

    If you wish to offer your analysis I’d like to read it and compare.

    I did also say: “Even if all the above is bolloc*s I really think that to say: [etc etc”, which hardly makes me mister dogmatic – and I was responding to a post that was.

  20. @R&D

    I agree.

    For me the key stat is that the Tories have struggled to get to 35+ at any point in this parliament. A lot of people are expecting the Tories to get there eventually, but there is no current data to support this. In addition, how can the Tories get closer to or emulate the percentage they reached after 13 years if Labour governments? At present the only logical answer is they won’t. Bear in mind too that the Tories have not won an OM since 1992.

    So, the best outcome for the Tories is to get to 34ish and hope Labour do roughly the same as last time. That would given the Tories the chance of winning a plurality of seats. But as Labour seat loses are only like to go to the SNP (and not the Tories), and Lab will most likely make gains from the LDs and the Tories, the Tories will not be able to govern.

  21. Candy

    The “L” word is normally caught by automod – for very good reason.

  22. PAUL

    Thanks-that clarifies things :-)

  23. @Candy
    “From where we’re sitting it’s just too dangerous to have a party like the SNP in government. Yes they can run a regional outfit under strict limitations, but the untrammelled power that comes with Westminster? No.”

    Hmm…so you’re saying that the SNP shouldn’t be around in a UK minority government situation because they will have untrammelled power?

    They’ve been successful at running Scotland, first as a minority, then as a majority. It seems to me that they have invaluable experience of the skills necessary to deal with an inconclusive election outcome.

  24. @Oldnat

    Are you saying that the SNP’s policies on currency and oil were instead because they were just too daft to understand the consequences?

    Okaaay. But it’s dangerous to have daft people in govt too.

  25. Candy

    You really aren’t worth communicating with if you are incapable of being civil.

  26. CANDY
    @”They were fools but they didn’t have bad intentions.”

    They probably did actually :-


  27. I’m with @Colin. I think the SNP knew full well much of their policy prognosis was nonsense, designed for a specific purpose to win a single referendum vote.

    In this, they placed themselves right alongside people who claimed to be the greenest government ever, had no plans to raise VAT, who pledged not to raise tuition fees, and who said we had 24 hours to save the NHS from privatisation, among other things.

    Quite why the SNP should be barred from high office and all the other parties not, I’m a bit non plussed.

  28. ALEC

    I was responding to Candy’s view on Greece.

    Don’t drag me into a discussion about SNP intentions please :-)

  29. The fashionistas of the Media are reporting that we are back to the 1992 scenario and given all their wisdom and wit who am I to argue with them? this received wisdom has the Conservatives quietly confident and purring at the prospect of being returned to power.

    It would be a foolish man to discount that as a real possibility but to recap the famous ’92 victory snatched from the jaws of Kinnock – as I remember the % lead on the night was very large – say 7% or thereabouts yet Labour still won 271 seats and did better in seats than the popular vote math might have implied. The conservative lead then was partly because the opinion polls under recorded the shy Conservatives and since then polling data has made considerable strides to take account of that feature.

    Thus, the 7% lead in ’92 was not a swing-back pure and simple. Secondly, at the present moment it seems highly unlikely that the % lead over Labour is likely to be anything like 7%. If the conservatives lone put on 3% that would not necessarily much reduce Labour’s polling number as more likely dent the UKIP vote. Similarly the Green or even the SNP poll numbers might similarly move to Labour were that process to get underway.

    It seems most likely this election most likely will reproduce the model of ’74 rather than ’92 since we can take the neck and neck situation in the polls more accurately to reflect the reality of voting intention. There’s even an outside chance there might be a greater earthquake alike Spain. All that said I’m not Cassandra and I have no very special powers to foresee the future. But I’m already mightily weary of the turgid campaign that has hardly yet begun.

    Oh bliss was that world when snap elections could be sprung and we only had to endure 3 weeks of this! For me – and I love US elections – this has the Somme-like qualities of their current electoral process where the last man standing is crowned victor but there is remarkably little of the victory left to savour once the battle is over.

  30. There’s an Opinium poll Mike Smithson tweeted earlier today of first time voters alongside what appears to be a national VI poll:

    Firstly, first time voters:
    Lab 41
    Con 26
    Green 19
    LD 6
    Ukip 3
    Others 5.

    Now the VI poll:
    Lab 33
    Con 29
    Ukip 19
    LD 6
    Green 6
    Others 7.

  31. R&D

    By historically low, I am comparing to the pre-1990s situation.

  32. The Daily Mail says it is allover for Labour this week due to they have not ruled out a coalition with the SNP.
    Rich opened the thread with the same sentiment.

    Is there any polling as yet to confirm this view ?

  33. The punters on Betfair seem to think the Tory party will “win”, in so much as there will be a winner. Its ok keeping on about 35%, and the Tory inability to hit it, but there are 14 weeks left to do that. Labour are not convincing any but the faithful, that they can run an economy. The well known football, “OUR NHS”, is not the vote catcher they had hoped, mainly because the principality where they run health is a disaster. As I said the other day, people are beginning to feel some improvement in their circumstance and will be less likely to flirt with the lunatic fringe. In the area’s where Ed Balls insists things are still on the breadline, most voters are Labour anyway and the public popularity of Miliband is a Tory gift that keeps giving.

  34. Alec
    The SNP aren’t barred. In fact I commented here in the aftermath of the last general election that Labour should have tries to form a wide coalition involving the then SNP.
    But they do pose particular difficulties for potential partners because they can’t under any circumstances want a UK government to succeed. They want the UK to break up.
    Anyone working with them has to face the charge they are working with an organisation that wants to break up the country which must be a bit of an albatross.
    Labour minority (in the event of a hung parliament) seems much stronger than most permutations.

  35. @Alec

    And how’s the stuff about tuition fees working out for the LibDems and the stuff about VAT for teh conservatives? And on the richter scale they weren’t even a fraction as devastating as the stuff the SNP were proposing.

    Voters are bloody-minded about punishing stuff like that.

    People wanted to know why there was so much hostility to a coalition with the SNP and I offered an explanation, which obviously they didn’t want to hear!

    But there’s a reason why that “Vote Labour get SNP” poster is having an effect in England. You can date when Labour started to lose it’s tiny lead to the moment Sturgeon pronounced about voting on English laws.

  36. @RAF
    Yep! and far more old buggers like me want a Tory led government.
    Whats new? There have never been many 17 year old Tories.
    The main difference is, people like me actually vote.

  37. @RAF – I thought that about the Opinium poll. In fact, it’s their poll from before Christmas. The headline vote (29/33) is actually the ‘all adults’ vote, which when added to the first voters bit where Labour lead by 15%, presumably brings us back to the 28/33 split that was reported on Jan 16th.

    At least that’s what I think it means.

  38. An SNP/Lab coalition would be difficult to sustain and probably not in either parties interest in the medium term imo.

    I confiedently predictit will not happen and will be rules out by both parties well in advance (In fact I thought I have seen such statements already but may be mistaken)

    A Lab minority giving certain concessions to the SNP under a C&S arrangement can be sold to both sets of supporters in my view.

    In fact I suspect that some on the left of the Labour spectrum might secretly like the idea just as perhaps some moderate Cons allegedly like the LDs hving an influence.

  39. candy

    “Yes they can run a regional outfit under strict limitations, but the untrammelled power that comes with Westminster? No.”

    Untrammelled? Oh dear.


    Glad that is clear: and your own analysis, to counteract my own view, [based on the views of people wot know and care a lot more than I do about all this stuff} as to why you think it is not “overwhelmingly” unlikely that the Cons will improve on 2010 is what ferzackerley?

    It is very easy to knock other people with brief comments but, if you have any ideas on the above, or the likelihood of the |Lib Dems [1] doing well enough in seats and [2] having the will of the entire party to prop up the Tories again, please share them and don’t just criticise me.

  40. Doing an Ashcroft…the May 2010 to Nov 2011 data is updated and you can have a sneak preview of the London crossbreak:

    http://www.statgeek.co.uk/2014/09/here/ (1st image)

    It quite reasonably spells out one reason why the Conservatives will struggle to get an overall majority in 2015. They were ahead in London in 2010 and aren’t ahead now. They are perhaps 12 or so points down on 2010, while Labour are up 5 or so.

    I’ll mull over how best to bring the rest to you. Don’t expect them too quickly. I’ve got a new spreadsheet to populate with fancy tricks and I have to learn em all over again. :))

  41. @Unicorn

    “I wondered if you had given this a try.”

    I wouldn’t know where to start, and every prediction is based on today’s snapshot, rather than the election once every 4-5 years. It’s fine saying “here’s what the parliament might look like”, but unfortunately the media and politicians pervert this information to suit their own ends.

    Will another UNS prediction tool be of use to anyone over an above the ones that exist?

  42. @Alec
    Thanks. You must be right as while Opinium tweeted the first time voters numbers earlier, they haven’t published any new VI polls since 16 January (the next one being due tonight for tomorrow’s Observer).

  43. PAUL

    My reasons why “Cons will /must/etc etc get less than 36%” is not cast in stone ?:-

    The last YouGov Poll + MOE + effects of the Campaign interviews =
    a possible 36%

  44. @Jim Jam
    “A Lab minority giving certain concessions to the SNP under a C&S arrangement can be sold to both sets of supporters in my view.”

    Yes – I don’t think anyone has or will be suggesting a coalition.

  45. RAF @Jim Jam

    And on Radio Ulster today, SDLP guy [1] saying that they were minded to back a Lab Government at Westminster, but would use their influence to extract benefits for Northern Ireland.

    Someone else [1] expressing concern that Miliband’s visit was to woo DUP not them. Of course, DUP would also use their influence to extract benefits for Northern Ireland.

    [1] Didn’t catch their names – no disrespect intended!

  46. TOH
    Lots of good micro-biology going on down there, Howard. There should be steam arising in this weather if you wish to try some empirical research with a stick.

  47. @UNICORN

    Thanks. Apparently it was my misunderstanding that they would be out this week. They should throw some light on the Scottish situation.

  48. Colin

    Hopeful, rather than analytical. MOE works both ways – and so will the prospective interviews…………….

    Another thing I would add is that, right or wrong, the larger part of VI does not anticipate further cuts announced by the Tories with much enthusiasm and just about every other party will be pointing that out.

    Far from underestimating the Tories I am impressed with how close they have remained to their 2010 VI: but I don’t expect them to equal it – and nor do the overwhelming majority on expert commentators……………

  49. Some great analysis, as ever, on this site.

    But there is also too much “IMHO” on here and not enough comment based on the facts-as-we-have-them (ie. the polls).

    HINT: The clue is in the name of this website: UKPollingReport. It really makes sense to try to analyse things based on the polls we have.

    So what do the polls actually tell us?

    Polls show neither main party is very likely to get a OM. (LAB are still close. But historically – and recent poll movements support it – the gap closes in favour of the governing party as the GE approaches. So LAB are currently drifting, as the bookies would say).

    That means – as things stand but they can change – that the GE is likely to lead to a minority government or a coalition.

    Any other conclusion is not based on the polls we have.

    Speculation, if any at the moment, should be on who might or might not deal with who.

    But even there the maths are pretty limited.

    So, for example, the polls currently tell us that LD will not get enough seats to form a coalition with CON. The two comined will not have enough seats.

    By contrast, the polls do tell us that SNP + LD + LAB would have enough seats.

    And those are the facts. Not opinions.

  50. @ Statgeek

    Thought-provoking charts as usual. I was surprised to see the unexpected (though presumably random) drop in Scottish crossbreak numbers (with the associated MoE implications). Awkward to get increased scatter just when we trying to work out what is going on.

    On the value of further variants of UNS model .. probably not much. It is just that all models could do with improving their VI-reallocation algorithms and I am not sure they make as much use as they could of regional crossbreak information.

1 2 3 4 5 6