The YouGov/Sunday Times poll this morning showed a four point Labour lead, interpreted in some quarters of the commentariat as showing an advance for Labour after the Paxman interviews. As ever, it was only one poll. Now we have a second post-Paxman poll, a ComRes telephone poll for the Daily Mail & ITV, and this one shows the complete opposite – CON 36%(+1), LAB 32%(-3), LDEM 9%(+1), UKIP 12%(+2), GRN 5%(-2) (tabs).

If the four point Labour lead in this morning’s YouGov poll equalled their best this year, this poll is the biggest Conservative lead ComRes have delivered since 2010. Where YouGov showed Miliband’s ratings improving, ComRes show Cameron widening his lead as best Prime Minister.

There is no great mystery here, I expect we’re just seeing normal sample variation. I said this morning we needed to wait for some more polling to have any idea whether the Paxman interviews had really had any effect, whether there was a consistent trend. With two polls now showing movement in opposition directions there certainly isn’t yet. It could still be that the rest of the week’s polls show a similar movement to YouGov and the ComRes poll was just a blip… or that the rest of the week’s polls show a similar movement to ComRes and the YouGov poll was a blip. I’ve a sneaky suspicion though that we’ve just happened to see two outliers in opposite directions, and we’re going to see lots of polls showing no clear movement. Time will tell.

442 Responses to “ComRes/Mail/ITV – CON 36, LAB 32, LD 9, UKIP 12, GRN 5”

1 5 6 7 8 9
  1. James

    Umm, put a comma after England only, to read what I meant. Alternatively, drop the ‘only’.

  2. So what Govt do people on here actually think we will get? Non partisan honest view?

  3. Rich
    A Lab govt, – most seats and votes.



    Current polling would probably lead to a Labour-led government.

    The question is whether there would be much of a move in the polls from now until May.

    I looked at my calendar. I’ll be sick to death of politics by May 7th, so goodness knows what the average voter will think.

  5. I notice that commentators elsewhere are drawing attention to the fact that the two polls that have shown the Conservatives ahead since the YouGov Sunday Times have been telephone surveys.

    I’m not sure if the implication is that they are (a) less reliable (b) have a pro-Conservative bias (c) simply that as the methodology is different they may throw out different results.

    I know AW treats ComRes phone and online polls differently because of differing results, but I’m not clear if there is any pattern to the way in which these vary. Anyone know?

    Likewise, Anthony mentioned that the Ashcroft phone poll follow a Populus methodology with ICM adjustments, but do they differ in any predictable way?

    Again I’m sure the house experts may have the information to hand.

  6. @ Rich

    A mess

    Which will probably result in Ed leading a highly unstable minority. For a while.

  7. Rich. I think Con will win most seats but not enough to form a majority or a coalition with the LibDems losing a stack of seats. I can only really see there being a Lab coalition with LibDems boosted by informal support from the SNP.

  8. @ Aberdeenangus,

    1) A major shift in the two outlying pollsters- 3 consecutive Labour leads from Ashcroft, 3 consecutive Tory leads from Populus- would suggest something had changed in the underlying position, as would 5 consecutive leads for either party from YouGov. ComRes, ICM, Opinium, and Ipsos Mori are too hard to benchmark and TNS are just wacky, although a week in which all five of those came back with a lead for a single party might also suggest that party had taken the lead.

    It’s also worth looking at the size of the leads with the pollsters we can benchmark, given what we know of the house effects. A week with Ashcroft +1, Populus -4 -5, YouGov -4 0 -3 -2 0 would suggest to me Labour had nudged ahead a bit even though it doesn’t meet any of the above criteria, because everyone would be more Laboury than their average Labour lead for the year.

    2) As far as Ukip is concerned, I’m satisfied to rely on the YouGov rolling average, which does show a gradual decline.

  9. Rich, I think most votes/most seats is a toss up (I lean to Labour on both but I can’t say such a view is entirely free of partisanship). At this point I would make Labour strong favourites to form a government simply because their path is much clearer – on current numbers, with C&S from SNP and Lib Dems they are comfortably home, whereas the Tories have no configuration which gives them adequate numbers. Things can change, of course.

  10. @Little Red

    I know by now that the conviction of STV proponents are well nigh impossible to shake, so I won’t try. But I also know where your coming from it’s just that I’m not coming from there. I’d far prefer local reprentatives to be as local as possible, and elect the Lords by St Lague to hold the disproportionate Commons to account. STV tries to square the circle and we end up with the worst of all worlds.

    Of course you’re right that smaller parties can win votes under STV, but its particular attraction to the Liberals and LDs was that it benefitted them a lot and smaller parties only a little. If they get into another Coalition deal don’t expect them to push very hard for it – they’d risk losing more seats where they were strongest without much gain everywhere else.

  11. “I think most votes/most seats is a toss up (I lean to Labour on both but I can’t say such a view is entirely free of partisanship)”

    Though I did say less than an hour ago that I make the Tories slight favourites to win a plurality on vote share… so I don’t even agree with myself!

  12. @AberdeenAngus

    Your second question is easier to answer. Some weeks ago Anthony suggested there had been a Ukip decline since Jan 1 and there was a lot of going and froing about what counted as good evidence for this.

    @Catmanjeff offered his CUSUM data and there were some straightforward comparisons of January vs February means. My preferred approach was to fit a linear regression line to the VI scores since Jan 1 and determine whether the best- fitting line had a slope that differed from zero. After a couple of weeks of extra data I was able to show that this was the case using YouGov data alone but that the downward using sll polls was not reliable.

    I have just repeated the exercise and I can confirm that both are now reliable (all polls at the 1% level and YouGov only at the 0.1% level). The respective slopes are drops of 0.21% and 0.36% per month (somewhat lower now than the figure when I first did the snalysis).

    These are markedly more stringent criteria than MoEs normally used for polls. So, there is clear evidence of a Ukip VI decline over this period (though this might have been flattening out most recently).

    In relation to you first question, I have tried to persuade UKPR contributors to agree conventions on this. The point is that you have to agree a time period over which the VIs are to be compared. (If there is no prior agreement on this, then two calculations could give different answers).

    One of my suggestions was that there is a case for settling on the batch of polls covered by any one of Anthony’s Polling Averages. (The current batch shows no reliable difference.) You could decide instead to base the decision on all polls reported over the last fortnight … Or the last month. I, too, would welcome other views on this.

  13. @Rich

    “So what Govt do people on here actually think we will get? Non partisan honest view?”

    I tend to look at the polls, and they say a Lab-led minority with C&S from SNP.

  14. I would agree with Statgeek.

  15. @ Assiduosity,

    There’s no reason on the face of it to dismiss phone polls as unreliable, but ComRes tends to be highly volatile and Ashcroft tends to be both highly volatile and Con-leaning. (This isn’t necessarily bias, it just means Ashcroft is outside the main poll pack. Maybe all the other polls are too Lab-leaning.)

    To be honest, I think the main implication is that the commentators you mention are Labour-supporting and they didn’t like the poll results, but there are good reasons not to assume a Tory lead with those firms means there will be a Tory lead in the polling average.

    Re. the two ComReses, ComRes Online has (recently) consistently higher Labour scores, and therefore consistently higher Labour leads, than ComRes Phone. They also have higher Ukip and lower Lib Dems. We don’t really know why.

  16. @Rich

    I’m going to go for a Conservative/Lib Dem coalition.

    The polls tell me to agree with Mr Nameless & Statgeek, but then I look at how much of current Labour VI is made up of 2010 non voters and demographics who don’t typically turn out, and the general lack of enthusiasm of the Labour VI for their leaders, and policies to appeal to their base (in fact many policies seem against the values of their core vote) and I just can’t see these people making the effort to turn out to vote.

    Labour will lose on turnout.

  17. Whoops! That was the 2010 version. There’s a new one I think.

  18. @RICH

    I’d go along with what seems to be the majority view on here. Very close between the two main parties, but Labour to form Government with support of other parties, something the Conservatives would seem unable to do.

  19. @mike n

    Conservatives lose some to UKIP, but they gain some Lib Dems + incumbency bonus

    “So what Govt do people on here actually think we will get? Non partisan honest view?”

    Tory/LD/DUP coalition or C&S

    I think it’s more likely based on my judgement of how the undecided/wavering voters will break, and I think the Tories have stronger “fundamentals” which will be effective in winning votes.

    That’s my honest view, I’m not very partisan, indeed I voted for Labour in 2010 and I’d vote for any of Lab/Con/Lib depending on policy rather than ideology.

  20. @Spearmint

    Thanks very much for this. My understanding then is that whilst Ashcroft phone polls are Con leaning, there’s no similar pattern among the ComRes phone polls?

    To be fair, amidst the obviously partisan statements, the figure I saw quoted that caught my eye was:

    ‘With the four most recent telephone polls from each company, the Conservatives lead 35.3% to 33.7%.’

    If true, this seemed at variance with the average of all polls to a degree that was worth noting (even taking into account longer averaging periods etc).

    Your point is very well made of course, even if the telephone polls do tend towards a higher Conservative VI it could well be because they are correct and the online polls are wrong.

    Also, I should be clear, my use of the word ‘bias’ wasn’t intended to imply any pre-meditated attempt to rig the figures rather that the methodology consistently favoured Conservative VI for reasons not understood.

    However, if they do consistently deliver different results on a party basis useful knowledge when viewing new result releases.

  21. @Spearmint

    As far as Ukip is concerned, I’m satisfied to rely on the YouGov rolling average, which does show a gradual decline.

    So, how steep does such a decline have to be before you decide it falls beyond the MoE boundary? Regression answers this question (yes – outside MoE). But as you peer at a rolling average graph how do you distinguish between a drop that comes from a single rogue poll and a reliable decline?

    Also, on Con/Lab differences your 5-in-a-row YouGov test would certainly pick up a big difference. But if the VIs are close (though still reliably different) you could go for moths without getting a 5-in-a-row sequence. For example, a test on the Polls that Anthony including in his averages on Feb 20 shows that Labour were then reliably higher than the Tories. But none of your criteris would have picked this up.

    I don’t think there is any escape from relying on a statistical test over an agreed period of time.

  22. “I’m going to go for a Conservative/Lib Dem coalition.”

    Good luck with that…

    Yes, Mikey sorry, misread what you are saying…the swing is still 2.75% from con to lab on the ashcroft poll in england only…

    the snp surge has damaged the lib dem/con coalition chances…the snp are primed to take abt. 5 lib dem seats in scotland which labour would have struggled to take….

    the arithmetic is tight for con/lib dem…

    let’s say the tories get 295 seats. the lib dems would need to get 28 seats for both of them to hit 323 which is needed for a working majority (remember take speaker and 5 sinn fein from 650 and halve then add 1, you get 323)

    BUT the world in which the tories get 295, presupposes about 15 gains by the tories from the liberals. If the lib dems lose 15 to the tories and 8 to the snp, they would have to limit their losses to labour to 6. A world in which the liberals lose 15 to the tories but only 8 to labour is not the one painted by any polling data.

    Similarly, a world in which the tories gain 15 seats from the lib dems suggests a collapse in the lib dem votes in their seats where they face tory challenges. Such a collapse would probably be accompanied by a collapse in con/lab marginals which would benefit labour…

    it’s difficult to see how the tories can limit their losses to labour to 23 while gaining 15 from the lib dems…this what the tories on 295 implies.

    unsurprisingly, the constituency betting isn’t buying this at all.

  23. Oberta


    The more the merrier on here.

  24. As the various polling companies disagree with each other, albeit within a fairly narrow range, has anyone kept data about their by-election accuracy for the last year or two?
    Surely that would help us to decide which companies to place most credence with.

    P.S. Whatever happened to Angus something (Prune?)

  25. @james peel

    The Tories can also win seats from Labour where the Lib Dems don’t feature much. There are Tory/Labour marginals where the 3rd party is UKIP, for example.

  26. which seats are the tories going to win from labour?

    maybe one or two freak results…but which ones?

    Southampton Itchen, Dudley North? Grimsby?

  27. I agree with Ashcroft’s findings for England, as they are much more in line with the national poll findings for Scotland, Wales and London.

    But if I was Labour l would be very worried about the Midland crossbreak that shows the UKIP to Tory swing, with some votes still hanging on from Labour.

    I also think Ashcroft is showing Green strong in precisely the area where they were strongest in the European election, London, Southeast and Southwest, and that Labour are well up in those regions too with the Tory hurting most because of UKIP residual strength left over from the European election.

    I really doubt that Green is on 8% in Scotland and would give half that back to SNP, which is where it will go anyway based on Panelbase data.

    I think the Comres poll is a bunch of rubbish as it was before the EU election, but it does help smooth out the over exubernace of other pollsters.

    Three poll avererage of Populos, Comres and Lord A gives Con a bounce of about 2% taken from UKIP down 1.5% and Green down .5%, with Labour down .3%, LD up .3% and nationalists up .2%.

    I am used to seeing a Con-Green movement in Canada, but acknowledge in the UK the flows could also be going back and forth between LD, Labour and the nationalists as well, depending on the region and constituency.

    It will be fascinating to see where things are at one week into the campaign next Monday, after the seven way debate on April 2nd, but will the Easter Weekend be a distraction to the seven way debate?

    I think it is a mistake for Clegg to stay out of the April 16th debate, as coverage, good or bad, is what it is all about in an election.

    And I am going to be interested to see if Fargae openly agrees with Miliband’s critcisms of Cameron, becauee if he does so, then he has learnt to soften his image and we all know the angry right-wing Tory vote that will not go to Labour.

    The goal of Farage in this campaign, surely, is to do maximum damage to Cameron to prevent him from continuing as PM. A shrill or overly harsh Farage will scare some voters off.

    Strange alliance that, but I would be looking for the classic gang up on the PM in the seven way, and the inter-oppositional nastiness to come out in the five way April 16th.

    And I think Clegg not being there, will lose him the chance of being seen as the “reasonable man”. The “reasonable” “plodding man” in this election may turn out to be Milliband.

    In my opinion all he has to do in this election is to be seen as “thoughtful”, caring and thinking long term.

    And Cameron’s posh public school boy privileged view that says “you are a fool” to Miliband and “I know what I am doing” , will come across and has started to come across as a “bully” making fun of a middle class working kid.

    Ashcroft has tried to say that more nicely, than my blunt acerbic tone.

  28. @ Rich

    Con/LD. I think Labour are going to completely collapse, despite what the polls say and UKIP will do better than expected. Just a gut feeling.

  29. Well, we call it Ashcroft poll, but we actually don’t know did it for him.

    I wouldn’t dismiss any of the polls. Lots of noise in the data, but there you are.

    A bit off:

    By the way 5% significance level is about the data and not the hypothesis – the probability of the data with the given hypothesis and not hypothesis with the given data (so the only thing you can say whether the data supports your hypothesis or not). Even stats textbooks get it wrong sometimes.

  30. @james peel

    Well there are quite a few Tory-Labour marginal seats here

    I should have said defend from and win from Labour. Most of these are predicted to fall to Labour. If the Tories are boosted from their current position in the polls, as I believe they will be, they should be able to 1) defend most of their seats from Labour 2) Pick up the odd seat in a constituency destabilised by small parties 3) Gain around 10 from the Lib Dems.

    So I don’t think 300 seats is that far-fetched. Of course it it’s impossible given where they’re polling now, but I don’t think the polls will stay like this

  31. @Unicorn You are talking of drops of .21 and .36 a month. I accept that the slope is statistically significant. Do you think it is politically significant?

  32. Rich,

    All of the forecasters except Stephen Fisher are saying that Lab plus left-leaning nationalists is going to be an absolute majority. So I think that is where the polling evidence is pointing right now: minority Labour government with informal support from minor parties.

    The interesting question would be exactly which combinations of minor parties would be required to get a majority a) overall and b) if the SNP abstain.

  33. Someone asked everyone to give a prediction.

    Based on last 10 polls, and assuming a small amount of change towards incumbent government, I think Labour, reliant on SNP.

  34. HAL

    I think one of the most interesting questions (perhaps the most interesting) is whether Labour will need the SNP or can have a choice of partner to get votes through.

  35. Postage Included,

    One might even hypothesise that PR in general will become more popular over time on the right and less popular on the left, since Con-UKIP coalitions could currently be the natural PR coalitions rather than Lab-LD coalitions. On the other hand, the rise of the Greens might complicate that; however, thus far Green success has been overwhelmingly at the expense of Labour/LDs, not the Tories/UKIP, and the same goes for the SNP.

  36. Charles

    Going from being predicted to take 3-4 seats to taking 1 seat isn’t very significant IMO.

    Where it might prove significant is where that vote is going. There will be strong negative correlations with the other parties (the votes have to add up to 100% still) removing half the purple balls in a pot will increase the VI of the other parties, the question is are the balls getting painted another colour in significant numbers (and will further balls continue to be removed/repainted as the election progresses.)

  37. “assuming a small amount of change towards incumbent government, I think Labour, reliant on SNP.”

    This is becoming a consensus view. labour minority shows the shortest odds on any outcome of a government, though it’s still quite long.

    The assumption is that the minority labour government will be propped up by the snp (roughly 40+ seats), the green, and the smaller nationalist parties (sdlp, pc) (abt. 7 seats)

    Assuming snp on 40, labour need to get 323-47, 276 seats…But the number they fall short of 276 will probably be made up from snp gains. so if snp get 45, the labour number of seats required for this grouping to work is 271 (323-45-7=271)

  38. Rich

    You are conducting a Wisdom Index poll!

    I’ll agree with the wise folk on here. Currently it looks like a Miliband led government, but Prof Howard makes the sensible proviso that he might have a choice of which party(ies), from different parts of the UK, he can persuade to support him at various times.

    The “Progressive Alliance” – SNP/PC/Eng Greens + SDLP (maybe – unless they just take the Lab whip as formerly)

    Whether Miliband wants to create a stable governing arrangement, or rely on vote by vote arrangements – time will tell )as the wisest person of all says all the time :-) )

  39. Hello everyone from Bournemouth East; wet here.
    14% London lead for Labour reported by M.Smithson.
    How does that compare with 2010.

  40. Oldnat

    Technically it’s a voodoo wisdom poll, it being self selecting and unweighted!

  41. @james peel

    I just had a quick play with the May2015 calculator. It doesn’t take much to get the Conservatives on 295 seats.

    CON: 37%
    LAB: 34%
    LDEM: 9%
    UKIP: 10%
    GRN: 5%
    SNP: 43% in Scotland

    I think you’ll agree that’s NOT very far from the last batch of polls were saying. That puts the Tories on 295 seats. Lib Dems on 23. DUP on 8.

    If you put CON on 38%, you’re on 306 seats.

  42. @Alan

    So like bookmakers’ odds then.

  43. @ Omni

    Entirely plausible.

    As are about two dozen other outcomes.

  44. Alan

    Good point.

  45. @CL45

    Lab were 2% ahead in London in 2010.

    However, treat ComRes with extreme caution always. No other pollster has put Labour 14% ahead in London recently. YouGov’s 10% is more likely. Especially as YG have extensive recent experience of pilling in London.

  46. the tories won’t get 37%!

    Anyway, i am sick of these games. you, omnishambles, are relying on the polling average picking up 3% for the tories. I am not talking about individual polls, I am talking about the average…the average for both main parties is still under 34%

  47. @little red rock
    “As are about two dozen other outcomes.”

    Of course! I’m just defending the 295 prediction as very plausible because some people act like it would break the laws of physics. As I’ve said before, a high Labour score reaching ~300 would only take a few% swing to them, it too is entirely plausible.

    @james peel

    Well I’m assuming either the polling average will increase or, as we’ve seen in the European elections, the Conservative result to overshoot the polling average by a couple of %

  48. Little Red Rock

    Which is why I don’t consider bookmakers’ odds as data. They may or may not reflect the data but there’s no necessity for them to do so.

    Bookmakers’ odds reflect where they think the money will go, largely that is going to be based on where money is going now.

  49. Yesterday I pointed out that ComRes had not shown a Labour lead in their phone polls in 2015. It should be noted also that Ashcroft has only shown about 2 Labour leads in his phone polls this year. As a result, using ComRes and Ashcroft phone polls to show evidence of swingback or of a Tory lead is a bit doubtful. Not wrong, but doubtful.

  50. James Peel

    “Assuming the SNP get 40 seats”
    I think you are assuming a lot here James.
    A lot of people in Scotland are sick tired of hearing theSNP/Salmond/ Stugeon
    continually banging on about breaking up the UK.
    You may be more accurate if you assume an SNP total of 30 seats max.

1 5 6 7 8 9