We have two new GB polls today, plus YouGov polls for London & Wales.

Firstly, the weekly YouGov poll for the Times has topline figures of CON 42%(-1), LAB 39%(+3), LDEM 7%(-2), UKIP 4%(nc). Fieldwork was Tuesday and Wednesday and changes are from the YouGov/Sunday Times poll at the weekend. The trend continues to be towards Labour and, given YouGov tend to show the most favourable figures for Labour, that’s now heading into hung Parliament territory. Tabs are here. YouGov also have a new election model on their site here, providing a seat estimate – currently that is also showing a hung parliament, with the Conservatives on 317 seats.

Secondly we have this week’s Panelbase poll. Topline figures there are CON 44%(-4), LAB 36%(+3), LDEM 7%(nc), UKIP 5%(+1). Fieldwork was conducted between Friday and today, and changes are from their poll conducted at the start of last week. A sharp narrowing of the Tory lead here, and Panelbase now weight their voting intention figures to the age profile of 2015 voters, not the whole adult population, so they are using a method that we’d expect to show a big Tory lead. When they changed their method last week it increased the Tory lead by seven points, so without the change they’d presumably have been showing a very close race indeed.

YouGov’s London poll for Phil Cowley at Queen Mary University London has topline figures of CON 33%, LAB 50%, LDEM 11%, UKIP 3%. In 2015 Labour had a nine point lead in London, so this would reflect a swing of four points from Conservative to Labour, and would likely produce several Labour gains. That’s better for Labour than even the most favourable GB polls, but isn’t necessarily unfeasible. Given the different demographics in the capital the swing in London is increasingly divorced from the rest of Britain – there was also a sizable shift towards towards Labour in London in 2015, despite there being relatively little movement in England as a whole. It is also younger than the rest of England, and more anti-Brexit than the rest of England. Tabs are here. (For what it’s worth, there are actually two YouGov/QMUL London polls today – there was a YouGov London poll coming out of field at the time of the Manchester bombing, which given the timing was held back to release both together today. Not that there has been any real movement between them… the Conservatives are down one compared to last week. That does, at least mean we can be confident that the big shift towards Labour in London happened around the time of the manifestos, rather than in the last week).

Just out, there is also new YouGov poll of Wales, which has topline figures of CON 35%(+1), LAB 46%(+2), LDEM 5%(-1), Plaid 8%(-1), UKIP 5%(nc). I’ll update with Roger Scully’s commentary and the tables when they appear.

Finally, I have a longer piece over on the YouGov website about the differences between the polls, implied turnout figures, what is likely to happen at the election.

710 Responses to “Latest Panelbase poll and YouGov polls for Britain, London and Wales”

1 2 3 15
  1. Nice, just read your piece – will be very interesting to see which model works, but it seems like there is no way to know which is most likely with pollster going in both directions…

  2. Gripping contest still.

  3. Panelbase method change seems really odd to me. If they are weighting everything to the turnout at the last election then they will automatically be wrong if the turnout distribution changes between elections. What is the evidence behind this decision? Doesn’t their own polling data show there is a change in attitude amongst younger people?

  4. AW do you know if there are any further polls out tonight?

  5. Whatever your politics, it can’t be denied that this is turning into the most exciting election for a while!

  6. So Panelbase looked at their results, didn’t like them and changed their methodology to change result?

    Very definition of herd mentality?

  7. I’m wondering about SLABs relationship with English labour. It seems that theu are encouraging anti SNP tactical voting which means that they are helping to install a Tory govt in Westminster. If this is what they are doing should the English party cut them adrift and disown them?

  8. @Steven

    Presumably there has been a management decision to go with a transparent methodology, as it may be perceived as a safe choice, and they can keep on saying “if we’d stuck to our other methodology the difference would have been 6% the other way” or whatever it would have been. Seems a little like the herd mentality after Kantar also did it recently (or am I mistaken?), although they timed it to perfectly coincide with YouGov really sticking their flag in the ground over their methodology…

  9. @ Lazlo, BBZ (fpt)

    I thought that the most worrying sign for TM wasn’t Crick’s question itself, but the way it is being reported. Yes this is from the Guardian, but it isn’t an isolated example and it rings true. Yesterday we had Laura Kuenssberg (of all people) getting tetchy with TM unwillingness to move beyond slogans, and Evan Davies praising Amber Rudd by mocking TM. Turning round a struggling campaign with the press against you is a lot harder.

    This is the comment from the Guardian’s live feed that caught my eye.


    May’s Q&A – Summary

    Journalists covering Theresa May’s campaign are getting frustrated at her refusal to answer questions and the highlight of the Q&A came when Channel 4 News’ Michael Crick delivered a mini-diatribe on their behalf. Addressing May, he asked her:

    ‘Isn’t the reason that you are doing so badly that whenever people ask you about policy, all they get are cliches and platitudes, and we’ve seen the same today. People think there is nothing there. ….’

  10. It is a most interesting election from a polling point of view. And if larger turnout is a factor then I would expect subsequent elections to have polling with many more turnout based questions.

    I think turnout depends in part upon expectation of result (c.f. the 2001 and 2005 elections). As people sense that this election can result in change it will cause more people to vote – on both sides of the political divide. This may add to the volatility of the polls over the next week.

  11. Philotes – no, not that I know of

  12. This is definitely a change election

  13. The polls which matter start from this Saturday. They will be the ones held to account.
    Volatile VI can explain April and May snapshots.
    But it is getting to the business end of polling with exam on June 8.

  14. YouGov Wales tables already out:


    though there may be more details to come

  15. I strongly dislike pollsters changing their methods mid-election at a time when the polls are changing dramatically. If you are allowed to do that you can just pick your method to get whatever result you want. It’s more a case of “who does the managing director at Panelbase think will do best” rather than an actual evidence based prediction.

    I’d have far more repsect for a pollster who sticks with their method, get’s it wildly wrong and then changes their method in the light of evidence, rather than one who gets the “right” result by fudging their result to be closer to the other pollsters.

  16. Steven/CoventryFefe – most of the pollsters are now using variations on this approach. ICM, ComRes, ORB also use similar methods – weighting towards the age profile of 2015 voters. Kantar use a method that takes into account age, but uses people’s self-reported likelihood to vote too, so is presumably more flexible.

  17. I think there is a growing sense of frustration from Tories at TM’s unwillingness to commit to anything when she so easily could: the Plymouth Herald interview embodied everything the public loathe about politicians. It would be sad to see her lose her majority merely because she wasn’t prepared to open up and commit to something.

  18. There should be an indication of potential change in turnout from the likelihood to vote questions. Even if inaccurate as an indicator of turnout in absolute terms, one would expect a change in turnout to be reflected in the pattern of answers compared with those for 2015.

    Although I suppose this could be skewed if there has been a change in the demographic make up of the polling panels.

  19. Thanks Anthony

    So are they just varying the factor by which that age profile correction works (based on expected turnout), or do any show disimilar raw data too?

  20. Reported by Guardian, May continues to go missing:

    “BREAKING Theresa May has refused to appear on @BBCRadio4’s Woman’s Hour tomorrow morning. She is sending Justine Greening.”

    This seems insane. It’s just going to feed the “No-show Theresa” meme. The lack of confidence in her must surely be profound.

  21. @PAULMCH

    The most exciting since 2015, certainly!

  22. Reading AW’s longer piece, a lot seems to ride on the gap between younger age-group turnout and older age-group turnout. But there does appear to be an expectation that only the younger part of this equation is an unknown.

    The Con manifesto seems to have gone down quite badly with some of its traditional supporters in the older demographic. What if they don’t turnout? If the young do vote and the old don’t, what are the chances of an even bigger shock?

  23. Spent 10mins playing around with YouGov’s by seat predictor and it looks like their model is assuming a much more even CON/LAB split of UKIP vote in seats where UKIP are not standing a candidate than the “loyalty” section of VI versus 2015 vote would suggest (e.g. Bristol East, North Norfolk, Wrexham, etc…)

    They are showing UKIP as 0% chance of winning in seats they have no candidate so I’m guessing they have simply zeroed UKIP out and using local swing data for change from 2015 to 2017 – risky in my opinion, but anything else is subjective.

    Having said that the strong (80%) UKIP-CON move that we saw in the past has diminished significantly so it is possible that at a local level LAB are gaining more like a 50/50 split of the ex-UKIP voter and its my model that is “wrong” by assuming the ex-UKIP voter is going to split out with a CON/LAB/Abstain split along a national swing in line with the “loyalty” section in the polls.

    How you treat UKIP in the pulled candidate seats where they received a large number of votes in 2015 can make quite a difference to predictions on seat numbers – especially now some people are considering hung parliament territory!

  24. ExileinYorks

    To me the question was the financing of the NHS (true, Crick introduced it in the context of not answering). It was not a tricky one – basically if you say that you are not funding the NHS if the economy wobbles, you lose a million plus (the plus means any positive integer) votes. That’s it.


    I still think that the Tories had a decent manifesto, but very poorly sold . I’m really not sure if May really understands what is in it.

    It became redundant in the campaign, but the thrust of both manifestos (not the emphasised one) were centrist.

    The Tories didn’t put costing, and they can’t, because it’s either money tree (so Labour cannot be attacked) or more severe cuts (losing votes). So Labour can promise anything (and they continued today), and Conservatives can’t name any source of funding, because then they have to name funding for the other spending, and the cuts of funding.

  25. Have just been looking in detail at the YouGov model. Can someone confirm to me (and apologies, as I am sure this as been already covered in detail on the board), that the constituency by constituency poll is not that at all, but a pooled national poll to maintain credible sample size superimposed back into the constituencies based upon an untested (in UK elections) methodology?

  26. Anthony,
    Thanks for that. Am I right in thinking ICM, ORB and ComRes have been doing that for a while though? I still think it’s a bit dodgy but at least they have stayed consistent. It could be argued that since it’s only 2 years since the last election turnout is less likely to have changed. Kantar’s approach seems more reasonable as it at least allows for the possibility of a larger turnout amongst young people. I still don’t like them changing method during the election period.
    I’m sticking with YouGov, which conveniently conforms to my own political preferences as well. ;)

  27. @ Rich

    “Yep forgot about that, 3 newt species with palmates.
    So, I have never had problems with small garden ponds and tend to go for the following; (but need to know size of yours)”

    Thanks. It’s about 10000 litres, 20′ by 10′. We have one big lily established, just bought another one. It’s a wildlife pond so plenty of oxygenators and marginal plants. The blanketweed killed 2 lilies unfortunately.

    I’ll print the list off, I don’t we have any water snails unless we inherited some with the plants but I’ll check. We’re getting electricity run down to it, I was hoping to just get away with lighting but maybe not.

  28. @ Robin

    If true, given that this is the very programme hosted by Emma Barnett where Crobyn tripped up on his costings, the accusation will probably quickly turn into TM is scared to turn up.

  29. Ive been calling a hung parliment with prime minster Corbyn for a few weeks now but this does worry that Labour will stick up votes in safe city and student seats, Could get very close in the popular vote say a couple of points but Torys still get a majority

    This will lead to mass scale protests in London. I can see Labour in the next election campaign for PR if that happened

  30. Steven – yes. ComRes did it first, way back before the referendum. ICM did it just after the referendum. Not sure when ORB adopted it.

  31. @CambridgeRachel

    Given the electoral dominance of the SNP, the only realistic prospect for SLab, SCon and SLD at this election is to play the anti-SNP card.

    The only alternative strategy in Scotland would be to urge a vote for SNP to keep out the Tories – not one that helps Labour in Scotland or Westminster.

  32. I’ve had a vision

    The scene: Strasbourg in 3 weeks time.Jeremy Corbyn and Emily Thornberry meets Chancellor Merkel and Mr Macron for negotiations.

    JC: Mrs Merton guten tag (JC snaps heels together) such a pleasure to meet you. Can I offer you this Battenburg made by my own fair hand. And Monsieur Macron (JC dons stripey jumper and takes out large baguette) I am so pleased to meet you also and test out my French in which I am effluent.
    JC: Well as you know I have won a snap election against the wicked witch and have gone to the Queen to form a government. I also mentioned while I was there that I was nationalising her but that need not concern you here.
    JC. Now let’s sit down with a nice piece of Battenburg and a cuppa and thrash out a deal that suits us both eh? You will already understand my negotiating position of course which is that ANY deal is better than no deal so there is no need for any nastiness between us unlike with that old witch.
    JC: So then what can you do for us ? as you know I will not walk away as ANY deal is better than no deal so what can you give us to cheer up the British people as Wimbledon was a complete horlicks?
    JC: Sorry how much? (adjusts hearing device), Really? (takes back baguette)
    JC: Well I suppose any deal is better than no deal eh? It’s our fault for invading you both – I knew that would come back to bite us
    JC: Emily – get me the air force on the phone – we need a lift home
    JC :What do you mean you have disbanded them?

  33. @LMZDEE,

    You can be sure of one thing, if TM scrapes home there will be angry protests, if JC scrapes in, no protests. That in a way helps show the difference between left and right. I hope this isn’t seen as partisan, as frankly it’s inevitable…

  34. One week to go. That BBC exit poll will determine whether I stay up all night or just go to bed. I’m staying with my 90-100 majority in which case I’ll be staying up but the polls say I’m wrong.

    So I will either have egg on my face or I will be collecting the TOH prediction award!

  35. If Lab get in I do think it puts a question mark over Brexit. I wouldn’t be surprised if we never leave. Just a gut feel.

    Anybody also think this way?

  36. @ Laszlo

    What I can’t understand is who could possibly think it was a good idea to go for a long campaign when the strategy was “We are not telling you details on anything, just trust Theresa”.

  37. “This will lead to mass scale protests in London”

    Don’t they protest whenever things don’t go their way?

  38. HertsAndy – it’s not really a poll at all – it is definitely NOT a collection of constituency polls. It’s an exercise in data modelling, specifically Multilevel Regression and Poststratification (or MRP… or MisterP, if you like giving things stupid sounding names). I don’t really understand it, hence why I haven’t sought to explain it – Chris Hanretty has done a lot of it in this country, and there are a couple of posts on his page here http://constituencyopinion.org.uk/posts/ explaining what it is.


    Yes and no.

    What they do is by using certain variables (demography, political inclination, turnout, etc, etc) they group the constituencies. Then by running the monte calro chain they have a more reliable distribution model of voters’ behaviour, and they basically use this distribution model of the group of constituencies for the particular constituency to predict the outcome.

    It’s methodologically sound, but it doesn’t mean that it’s correct. Three cautions: 1) if the respondent data is correct; 2) if the stratification of constituencies and respondents is correct; 3) if the Monte Carlo improves the distribution.

    Having said that, I’m sure that YouGov tested it no end, and I was (not much credibility) impressed already with their 2015 version.


    Why even call an election? It was completely unnecessary.

  41. @Anthony,

    Are you allowed to say personally what you think will happen results wise? Or do you just get represented by the relevant polls?

  42. Posted this on the old thread by mistake first, so will put it here instead! I’ve been reading this blog for a while and thought I’d get involved! I’m just interested as there seem to be so many conflicting accounts of what is going to happen. I’m a 28 year old Tory and it has to be said, the polls look very bad. However, I do have some sympathy wiring the argument some have made here about “shy Tories” especially among the young. I know quite a few other people my age who vote Tory and UKIP and none of them are “out” – as in they only like to talk about it with those like me who agree and never on Facebook, etc. I myself am very vocal and don’t particularly care what people think of me but I’m an exception among my young right wing peers! I don’t know if this is affecting the polls at all but just wondered what you all think.

    Also, how come Theresa May is in Middlesbrough today with the polls so close? It has a 12,000 majority. Shouldn’t they be focussing on shoring up marginals? Does anyone else know about where the Tory campaign is most intense now? Has it shifted back to ultra marginals?

  43. @ HERTSANDY – somewhere in between!! It seems they are trying to use local data as much as possible but given the small sample sizes that would produce they then look at the demographics of each seat and use a ‘national swing’ of that demographic to help “fix” the small sample issue. Given the amount of data they have there is no way for other modellers to fully check the results but their model will be a better predictor of the outcome if voters with similar demographics vote the same way at a national level. EG:

    1/ a young urban female remain voter in a London constituency voting LAB is similar to a young urban female remain voter in every other constituency in UK (and same likelihood to vote)
    2/ an old rural male leave voter in S.East constituency voting CON is similar to an old rural male leave voter in every other constituency in UK (and same likelihood to vote).

    Sorry to create/pick stereotypes but as you can see in the polls those two demographics do seem to fit a stereotype assumption!

    I think I just about understand what they are trying to do. Whether or not that is the best way to go about it or not is a different matter. It feels like quite a leap of faith to me to assume specific demographics have similar VI across the nation but they have a lot of data so you have to assume they feel some confidence in making that assumption.

  44. Anthony Wells

    Thanks for Hanretty’s link.

  45. @RICH

    If you read his piece, he states what his original projection was…

  46. ExileinYorks – I don’t think they could’ve gone for a much shorter campaign. The statutory minimum length between dissolution and polling day is longer now to begin with, and they only gave themselves a week and a half or so of wash-up between the announcement and dissolution, which wasn’t much for short notice.

  47. @ Mike88

    Middlesbrough South & East Cleveland is still a realistic Cons target seat.

    Not far away are Hartlepool, Bishop Aukland and Darlington which, just a few short weeks ago, also looked in range – but not anymore even on the polls more favourable to Cons.

  48. @Trevor

    “Spent 10mins playing around with YouGov’s by seat predictor and it looks like their model is assuming a much more even CON/LAB split of UKIP vote in seats where UKIP are not standing a candidate ”

    Interesting observation. I can certainly see that, unless explicitly built into the model, this might lead to incorrect modelling of such seats. My unevidenced but (I think) not entirely unreasonable opinion is that the 3% who are doggedly sticking to UKIP are more likely to be blue UKIP, compared to those who have already deserted the sinking ship.

    But maybe I’m wrong in thinking that.

    I guess an alternative is that, although the model looks like that, that is because the demographics of those seats are such that there are very few hardcore UKIPpers left.

  49. Thanks Anthony and @Laszlo

    I have now increased my scepticism on the YouGov results significantly, particularly when looking at individual constituencies.

    I don’t know if you saw Jim Messina’s tweet yesterday: ‘spent the day laughing ….’

  50. @Exile

    Eddie Mair just confirmed that May has run away from the Woman’s Hour interview.

1 2 3 15