ComRes have released their first voting intention poll since the election, and have topline figures of CON 41%, LAB 29%, LDEM 8%, UKIP 10%, GRN 5%. Full details are here.

The ComRes poll also had the first attempt at a methodology change to address the failings of the polls at the 2015 general election – though as ComRes make clear in their explanation this is not ComRes’s final word on the topic, they are continuing their internal review and may make extra changes too.

As with all the pollsters who use political weighting, the initial change is to move from past vote weighting using the 2010 election to past vote weighting using the 2015 election, something that would have been done anyway. The second change is a new model of turnout weighting. This is based on the theory that a cause of the error was people overestimating their likelihood to vote in an uneven way – that is, we all know people overestimate their vote, but ComRes suggest they overestimate it unevenly, that people in some social groups (who happened to support Labour this time) overestimated their likelihood to vote more than other groups, thus skewing the polls.

In the past almost all the pollsters accounted for likelihood to vote using a straightforward system of asking people to rate their likelihood to vote on a scale of 0 to 10, and then either filtering out those people who gave a low score, weighting people according to how likely they said they were to vote, or a combination of the two. ComRes’s new method still filters out people who say they are less than 5/10 likely to vote, but after that bases likelihood to vote weighting on demographics, based upon patterns of turnout at the general election, specifically that there tends to be lower turnout in areas of social deprivation and in areas with a high proportion of social classes DE and low proportions of ABs.

The mechanics of this aren’t completely clear yet (I’ve asked ComRes for some more details which I’ll update later), but essentially it looks as if younger and more working class respondents are assumed to be less likely to vote than they claim they are and weighted downwards accordingly. It means, in effect, that the final headline voting intention figures are made up of 41% AB, 31% C1s, 19% C2, and just 9% DEs, so the effective sample once it’s modelled for the sort of people who actually turn out to vote is far more middle class than the pre-election samples that got it wrong.

The impact of the change is, as you might expect, to produce significantly more Conservative figures. In this particular poll it increased the Conservative lead from eight points to twelve points. In ComRes’s final pre-election poll it would have changed the result from a one point Tory lead to a five point Tory lead, significantly nearer what actually happened.

UPDATE: ComRes have got back to me with some more details of their turnout model. In my original version of this post I’d assumed ComRes were still weighting people according to their 0-10 score, but were adjusting this score based on demographics too. In fact ComRes are now only using the 0-10 score to filter out people who say they are less than 5/10 likely, otherwise the turnout weights are all based on demographics.


401 Responses to “ComRes/Daily Mail – CON 41, LAB 29, LD 8, UKIP 10, GRN 5”

1 2 3 9
  1. Anthony,

    What about indications that in Scotland the type of people who say they will vote Labour but don’t vote is too high while people in the same demographic came out in numbers for the SNP.

    ComRes don’t like the term “lazy Labour” but is something different happening in Scotland?

    Peter.

  2. I refer the reader to the post I made on June 4th here: http://ukpollingreport.co.uk/blog/archives/9433#comment-1020015

  3. Thanks AW.

    I had a look at the data earlier, and am wondering whether ComRes might be double counting by factoring in BOTH the results of a turnout question and the results of historic evidence on turnout into their model.

    Table 1/2 provides some evidence of this. Here’s how.

    If you take the weighted figures of how people say they voted in 2015, you get the following responses after discounting the DKs:
    Con 37.3%
    Lab 31.1%
    LD 8.1%
    UKIP 12.8%
    SNP 4.9%
    Other 5.8%.

    So that’s a difference between Con and Lab of 6.2%, 0.3% short of the actual 6.5%.

    Now take just those same people who voted in 2015 and the results for how they say they would vote now (the totals in the columns in Table 1/2 apart from those who did not vote). That means you’re excluding others who didn’t vote but who say they will now, so you’ll be picking up any issues about reluctance to turn out or electoral registration. I think you’ve then got a pretty good alternative and better way of gauging current voting intention without putting in place a turnout adjustment, based on the simple expedient of assuming that those who didn’t vote a month ago won’t now.

    That gives a split of current voting intention from 2015 voters of:
    Con 260 39.9%
    Lab 200 30.7%
    LD 56 8.6%
    UKIP 73 11.2%
    SNP 4.9%
    Green 4.6%
    Other 0% (since not given, in practice you would have to scale down all the above differences by about 1/100th)

    So that’s a difference between Con and Lab of 9.2%, which if you add back the 0.3% difference in the original weighting would get to a 9.5% Conservative lead.

    That’s very different though to the Con lead that ComRes have published of 12.0%, which is why I think their methodology may be overstating the Con lead. I would expect some minimal difference, but nothing of that scale.

    Mind, after a month in which leading figures in Labour have been running around disowning what seems to be just about every popular policy that the party fought the 2015 election on, I wouldn’t regard such a big Conservative lead as in any way implausible.

  4. @AW

    By the way, if you’re reading this, one thing I thought really worked in the last YouGov poll was the questions on government priorities i.e. “which of the following do you think should be the government’s main priorities”. It would be good to see these developed into a regular tracker, although you would need to drop the “thinking about the next five years” preamble. Obviously issues could be added or dropped over time. I found the choice of questions and their extended form really informative when looking at the cross breaks.

  5. I calculated YouGov’s figures for the Labour leadership with Don’t Knows removed:

    ALL RESPONDENTS
    Burnham: 50%
    Cooper: 25%
    Kendall: 19.4%
    Creagh: 5.6%

    CON VOTERS:
    Burnham: 42%
    Kendall: 29%
    Cooper: 22.6%
    Creagh: 6.4%

    LAB VOTERS:
    Burnham: 59.1%
    Cooper: 30.6%
    Kendall: 12.2%
    Creagh: 2.1%

    LIB DEM VOTERS:
    Burnham: 44.2%
    Kendall: 27.9%
    Cooper: 20.9%
    Creagh: 7%

    UKIP VOTERS:
    Burnham: 47%
    Kendall: 25%
    Cooper: 21.8%
    Creagh: 6.2%

    Caveats: Small sample size due to stripping out about 60% of respondents who don’t know. It was also published before Corbyn announced his candidacy.

  6. “ComRes don’t like the term “lazy Labour” but is something different happening in Scotland?”

    ———-

    Always good to start a thread with summat about Scotland. One might expect summat different to happen in Scotland ‘cos as Scots peeps are apt to point out, “Scotland’s different”. Not escaping the cuts though…

    “Scotland will face a £176.8m cut in public spending this year as a result of a savings plan announced by the UK Chancellor.”

    From da Beeb. I suppose proportionally, that’s not too bad out of 4.5Bn in cuts. Wonder which region of the UK will take the heaviest hit?

  7. I continue to think public opinion of future leaders is a suspect measure – both of who might win the contest and of how they would be received in practice. This is because the people voting in the contest are a tiny, engaged sub-set of the electorate and the wider electorate knows very little about politicians outside the PM, Leader of the Oppo and the Chancellor. At this stage for the public Andy Burnham is that northern guy who likes the NHS, Yvette Cooper is Ed Balls’ wife, Liz Kendall and Mary Creagh are virtually unknown. Once they become leader whoever wins will break out from these typecasts and be portrayed differently. It is for the people voting in the contest to predict how this might play out, not base their decisions on current perceptions.

    As for Jeremy Corbyn’s entry I expect Yvette Cooper and Liz Kendall will be very keen that he gets on the ballot in order to take some votes off Burnham.

  8. A key question in all this turn out stuff, is how much the differential turnout may have changed for this election, and how much impact voter registration changes had on this.

  9. Jack Sheldon

    As for Jeremy Corbyn’s entry I expect Yvette Cooper and Liz Kendall will be very keen that he gets on the ballot in order to take some votes off Burnham.”

    It won’t make any difference. That’s the beauty of AV!

  10. @Norbold

    You make a good point.

  11. Pollsters are flailing about at the moment. Less than a month after the General Election is no time to produce an opinion poll. It means nothing – even less than nothing now because of the awful failings of the polls before the GE.
    Pollsters are in trouble – and it will take a lot to regain the trust of the people, let alone the parties and newspapers who paid for the pre-election drivel that was served up.

  12. Good evening all from East Renfrewshire. Wee bit windy but quite mild. In fact its rather pleasant.

    AW..
    “The impact of the change is, as you might expect, to produce significantly more Conservative figures. In this particular poll it increased the Conservative lead from eight points to twelve points. In ComRes’s final pre-election poll it would have changed the result from a one point Tory lead to a five point Tory lead, significantly nearer what actually happened”
    ……..

    You were skeptical of past national polls being a bit too “Laboury”
    ..

    “So what can we conclude? Well, looking at the figures by-election polls do seem to produce figures that are a bit too Laboury, but I’d be wary of assuming that the same pattern necessarily holds in national polls (especially given Survation use completely different methods for their constituency polling). At the European elections the polls also seemed to be a bit Laboury… but the pollsters who produced figures for that election included those pollsters that tend to produce the more Laboury figures anyway, and didn’t include any telephone pollsters. It would be arrogant of me to rule out the possibility that the old problems of pro-Labour bias may return, but for the time being consider me unconvinced by the argument”

    http://ukpollingreport.co.uk/blog/archives/9146

    Anyway its a very good poll for the Tories and a bit of a disaster for Labour and UKIP. It looks as if the blue kippers have gone back to Cameron and the Red kippers have stuck with UKIP.

    That’s the uncomplicated scenario!!

  13. @Jasper22

    I’m inclined to agree… but there are contracts to fulfill and there is always a thirst for information, even questionable information.

  14. Well a poll to comment on at last.

    Who knows if it is any good?

    Back to the tea leaves!

  15. JASPER

    “Pollsters are in trouble – and it will take a lot to regain the trust of the people, let alone the parties and newspapers who paid for the pre-election drivel that was served up”
    ______

    Don’t agree with you. It’s not the pollsters fault if white working class males who said they would vote Labour when asked decided to go to the pub instead of the polling booth resulting in skewed polling.

    Also, the political parties were carrying out their own canvassing (thousands each day) so they must had known the polls were a bit out!!

  16. Boost for the winners at the expense of the losers.

    Quelle surprise.

    In 2010 Labour’s prolonged Leadership Election arguably allowed a narrative to take hold about over-spending with little challenge; and, ultimately was a factor in the 2015 GE.

    This time around that ‘risk’ is not there and a period of reflection that may have a short term impact on notional support is necessary.

    In short a pretty meaningless poll which doesn’t add to Labour’s issues.

  17. @AC

    I don’t know if canvassing would have helped much. I know there were anecdotal examples of seats looking better for CON/worse for LAB than polls suggested but there were also plenty of cases where the rumour ‘on the ground’ was that things were looking good for a party and it turned out to be total rubbish. This isn’t particularly surprising because, especially during the campaign period, canvassing tends to focus on narrowed down groups of people who were previously supporters or undecideds.

  18. @Anthony Wells
    Although this comment is entirely anecdotal I am going to make it anyway. As I think you know, I have darkened the door of polling stations in Buckinghamshire for many years. Never, have I seen so many voters, nor so many very elderly voters struggling on zimmer frames with great determination in order to vote. I am assured by other Tory activists in Buckingham and North and South MK constituencies, that it was the same across the county. For those who see no humanity in Cameron, look at it this way, keep the old buggers going with the NHS, to vote Tory.

  19. Allan Christie
    I quite agree, White van man, Bob the Builder, and Barry the roofer,
    were well taking the p.ss wasn’t they. Camerons a toff, but Miliband, do leave it out !

  20. JACK SHELDON

    I absolutely agree with you in some cases political parties were getting more favorable fro themselves vibes in some seats and in the end they ended up losing the seats on election night.

    However surely if they are canvessing right across the country then they should had seen that their own polling was out of step (either + or -) from what the national polls were predicting.

    The SNP (even though the polls were great for them) said on election night that they knew the polls were underestimating their support.

    The Tories Jackson Carlaw said that the election felt like 92, he and his party must had known that the polls were underestimating the Tory vote.

    To me it looks like Labour were too lazy to do their own homework and relied too much on the polls as did the Lib/Dems but the Lib/Dems did think their incumbency factor would help them out even though in Scotland the Ascroft polls showed they were facing wipe out as did the ITN poll of the South West, it too showed Lib/Dem wipe out.

  21. The ComRes interim revision to their polling methodology is a bit of a fraud, in my view.

    Tweaking the result to take account of the likelihood of respondents fulfilling their voting prediction, according to their social status/circumstances, is surely just the same as tweaking according to their voting intention.

    In other words, ComRes might just as well have said, we are going to add 4%, or whatever figure they judge appropriate, to the Tory vote.

    In this country, the relationship between a political party and certain ‘social groups’ is pretty constant. I don’t think the ‘lazy/non-voting’ social groups ‘happened to support Labour this time’. They probably always do.

    All ComRes are saying is that the pollsters have consistently under-estimated Tory support. This, I recall was the line of the prescient ‘NumberCruncher’ pre-election post, which predicted a Tory victory.

  22. @Jack Sheldon

    I agree that canvassing isn’t that good a guide. While it’s possible to get a view of how your own support is holding up (or not), it’s hard to tell what’s happening on the other side.

    So, from my own experience, while it was clear that support for Labour had weakened a bit in 2015 compared to a year earlier, which could have been gleaned from the polls, it wasn’t at all clear though whether many of the “Againsts” and “Doubtfuls” were supporters of the Con/UKIP/Diehard LD or the Cant Stand the Lot of You parties, so it was hard to pick up any Conservative revival. There was furthermore an apparent firming up of support in the last week or so, which might have given grounds for optimism but which in practice I think was just people making up their minds one way or another, with the ones on your side being prepared to come off the fence.

    I think that Conservative canvassers will therefore have been in a much better position to pick up the revival in their fortunes. That said, it still wasn’t quite enough for them to retain this seat, which they had gained in 2010.

  23. You wonder if Corbyn has been put up to stand, in order to suddenly make Burnham and Cooper seem centrist and having ‘wide appeal’ after all

  24. Oh dear

    I think they’ve gone too far the other way now, I don’t think the Tory lead is this big. No doubt in 2020 we will see labour outperform the polls.

  25. Have any pollsters yet done any polling to identify reasons for voting / not voting yet?

    Could they please add the following questions:

    To Labour-leaning respondents who didn’t bother voting on the day:

    “Did the popular perception given by the opinion polls of a near-impossibility for the Conservative Party to win an overall majority make you feel it was less important that you cast your vote?”

    To Conservative-leaning respondents who did bother voting on the day:

    “Did the popular perception given by the opinion polls of a near-impossibility for the Conservative Party to win an overall majority make you feel it was very important that you cast your vote?”

    Not all of the electorate are wise to the dangers of self-fulfilling, or perhaps in this case self-negating, prophecies…..and people fight hardest when they feel their backs to the wall.

  26. (Having said that, I realise that those might be construed as leading questions for post-hoc rationalisation. I don’t know if there is a better way of teasing it out of people.)

  27. Good early evening, everyone, on a beautiful day here by the sea.
    Anthony; thanks, as ever, for this work on the polls.

    I think the Cons will be hopeful for 2020 due to swing back.

    David Aaronovitch’s article this morning is excellent about Liz Kendall and the left-wing-deniers.

    Lib Dem numbers?

  28. Aaranovitch isnt a communist any more then chris -more proof that thatcher and blair have shifted the political sprectrum to the right.

    In my youth pro europe ,anti inequality pro social.justice politicians were the labour right -crosland,hattersley ,smith now miliband,who shares their views ,methods and values is a dangerous left winger.

    Only scotland keeps the faith.

    Reckon red ed is going to keep old one nation dave and young liz,andy and yvette up to the mark on that score if his speech today is anything to go by.

    Scottish crossbreak looks hopeful for labour in comres (ok I made that bit up )

  29. @Millie

    If you genuinely want prescience, try RodCrosby’s post on politicalbetting.com. He nailed it. A full year before. http://www2.politicalbetting.com/index.php/archives/2014/05/27/guest-slot-rod-crosby-the-bell-tolls-for-labour-and-miliband/

  30. “Your humble and most obedient servant” -think young andy is taking this big tent and canvassing mularchy a bit far ,that mrs blair she was no forelock tugger -leftie burnham you are having a laugh.

  31. 07052015.
    I agree with you; things shifted to the right, gradually, from 1955, imo. That is why Hugh Gaitskell tried to ditch the old clause four of the Constitution, written by the Webbs in 1918, who saw the Soviet Union as the future.

  32. Martyn (fpt)

    Double AAARGH!!

    From a quick look at the ComRes tables they have assumed that a “British” model can be equally applied to E&W and Scotland.

    Applying their “age and social grade” (ie yoiung and working class) won’t vote concept, they manage to downweight Scotland more heavily than England – yet turnout in Scotland was 71% as against 66% in England.

    Whether ComRes are capable of casting a glance northwards and wondering why their model doesn’t work is something we might await with interest (but little hope :-( )

  33. Tiny subsample but Tories moving ahead of Labour in Scotland?

    Also interesting that the South West is now UKIP’s worst area outside Scotland or London – might back up the idea of “blue UKIP” returning to Con

  34. Oldnat

    Really? Unbelievable. They had an open goal: surely Scotland has diverged from E&W to an extent sufficient for differential polling based on country to make sense. It wouldn’t cost that much more to run two defacto polls simultaneously, you just need a few more respondents on each side of the border.

  35. mikeinsdevon

    “Tiny subsample but Tories moving ahead of Labour in Scotland?”

    Unlikely, I think – much more down to ComRes assuming English weightings can be simply applied in Scotland.

    If you look at the “What party do you think of yourself as?” question, the responses from that wee sample were –

    SNP 26% : Lab 21% : Con 14% : LD 6% : Oth 1% : None/Refused 20%

    The VI figures (DK excluded) were –

    SNP 47% : Lab 16% : Con 22% : LD 9% : Oth 2%

    It would take a considerable stretch to see that kind of VI – yet.

  36. Martyn

    Well a lot more north of the border! :-)

    I suggested some time back that it made infinitely more sense for pollsters to poll E&W as regularly as they can attract the funding to do – but to save up the resources for sampling those in Scotland and produce less frequent Scottish polls than for E&W, but polls that would be more accurate for both political systems.

  37. I am bothered that pollsters are arguing there is a repeatable effect here which can reliably be transferred to the next election. Obviously, that is what pollsters set out to do, but it might be the stay home effect was governed by the circumstances of the campaign. Maybe that just pushes the target a little further down the road, and sets the task of identifying those factors, but it might mean there is no pattern which can be found in these results and applied in a simple way to the next.

    Strikes me, what really happened in this election was that the liberal vote collapsed. So that means the question might be, how do disenchanted liberals react? Given the principle attraction of the liberals has been for a long time the protest vote, would these people continue to vote that way? If so, then the fact of the opinion polls calling a tied result might be critical to these people, might be enough to change their minds at the last minute.

    Being a floating, though liberal inclined, voter, myself, I can see how this might work. Sometimes I have voted for my preferred candidate, which would be liberal, despite the knowledge they will not win. Other times I have voted tactically, to influence the the result if it is not one I like.Personally I felt the best possible outcome would be a hung parliament, and that was exactly what was forecast. This time round I did not regard a vote for liberal as useful in furthering this goal, since they were plain and simple conservative light.

    So I suggest some considerations. A growing number of voters disenchanted with all the possible winners, who want a hung result. Did anyone poll to ask people if they wanted this?

    The self fulfiling prophecy (or self unfulfilling prophecy) of an opinion poll, feeding back to people trying to engineer a hung parliament.

    The enormous instability in the system created by first past the post, which is designed to amplify a small change into a big result.

    This might be a system inherently impossible to predict by polling, because it is the minority flip voters who are critical to the result, not the majority main supporters. Whatever the result of a true poll giving accurate and preise intentions at that instant in time, its publication might immediately invalidate it because of the effect on those same voters in changing their minds.

    I dont think this phenomenon should be any surprise to pollsters, though a dificult thing to admit your entire industry is self- cancelling.

    great vested interest for parties in biasing poll results to cause the outcome you want. if I was one of the two main parties, I would seek to get low poll figures in the hope floating voters will be more inclined to vote for me. Not too low, of course, dont want to scare them off completely, but any way to bias the pollsters results a little would be great.

  38. ROLANDGATINOISE
    Allan Christie
    I quite agree, White van man, Bob the Builder, and Barry the roofer,
    were well taking the p.ss wasn’t they. Camerons a toff, but Miliband, do leave it out
    ______

    It appears so I’m afraid. One thing I will give the Tories credit for and that’s getting the over 90’s (their core voters) out to the polling stations to mark that wee X in the box.

  39. ALLAN CHRISTIE
    We are now expanding into the over 80’s, especially if they play nudist leapfrog.

  40. I agree with you about LAB relying too much on polls. In my neck of the woods the probably rogue Ashcroft poll that put them in the lead in Finchley & Golders Green seemed to trigger activists being poured into what was always an unlikely gain at this election. As a result they seemed to put less resource into neighbouring Hendon and ended up losing both – Hendon by a majority 35x bigger than in 2010! Easy to say this with hindsight though.

  41. I can still foresee a potential issue with this ComRes methodology. It’s based on demographic turnout at the last election, so just like the party voted for last time, could be prone to projecting historic assumptions about things that have changed.

    @Roly makes the point about older people voting strongly, which isn’t uncommon, but may also be a direct result of Tory policies deliberately aimed at older voters. If, for the sake of argument, something happens in the next 5 years that makes older voters more disenchanted and less like to vote, then basing polling methodology on past turnout by demographic might still produce errors.

  42. Repairing polls with a sledgehammer…

  43. re the Comres methodology:

    How do they decide which social group (A, B, C1 etc) people belong to? I have looked at this in the past and can’t even decide which group I belong to myself.

    Do they judge people on their occupation, or lack of it? I can see problems with that. For instance, a student from an middle class family might be stocking shelves in a supermarket in the holidays.

  44. Peter Cairns

    “What about indications that in Scotland the type of people who say they will vote Labour but don’t vote is too high while people in the same demographic came out in numbers for the SNP.”

    I think the effect they picked up is real i.e. there are lots of people who say they vote (cos they know they *ought* to) but who never do and in my experience this effect is skewed towards lower SES but as your Scottish point suggests Comres are currently over weighting the effect.

    I would suggest this is because there is a specific problem with chunks of the native working class population that used to be Labour voters who have stopped but haven’t found a new home yet except in Scotland.

  45. “Repairing polls with a sledgehammer…”
    ——–

    If they’ve repaired them…

  46. “@Roly makes the point about older people voting strongly, which isn’t uncommon, but may also be a direct result of Tory policies deliberately aimed at older voters.”

    ———–

    Or fear of Labour policies, e.g. Mansion Tax. Or fear of the Scottish jig thing.

    But it isn’t necessarily that they are more motivated, just that more are living longer, and stay healthier for longer, so more are less constrained than before.

    Or maybe they just like my storage being taxed…

  47. “For instance, a student from an middle class family might be stocking shelves in a supermarket in the holidays.”

    ——–

    Could do with a few more of them doing the shelf thing,’cos students keep cleaning out my local Tesco Expres s round the corner.

  48. @carfrew

    Except it isn’t really £4.5 billion in cuts. More like £2 billion, with the rest of the “savings” filled up with asset sales (mostly the Royal Mail stake). So the Scottish £177 million is more like the proportionate share of the £2 billion actual cuts.

  49. I find all the discussions we are having quite depressing really. Isn’t it the case that polling simply doesn’t work or at very is least is incapable of being accurate enough? How can we have any idea whether ComRes’s change will work? Presumably we wait until the next election and find out after the poll has taken place – what is the point of that?

  50. Martyn

    Thanks.

    As your link points out, most observers took the view that the Labour performance in actual elections looked good for the Tories throughout the Parliament. This was underlined by the good economic numbers as the election approached. No doubt this explained the large amounts of money that were placed on the Blues.

    It seems that everyone is very unimpressed with the ComRes ‘solution’ and I note that it is an interim observation. Frankly, if the ComRes comments are representative of the quality of work undertaken by the pollsters, I am not at all surprised they got it wrong.

    I would describe their effort as a simplistic analysis leading to an excessively complicated solution.

    OldNat’s comments sum this up well.

    I was talking to a former Conservative MP yesterday, and he maintained that the pollsters will only get it right when they interview face-to-face.

1 2 3 9