The big danger for Labour is piling up even more votes than usual in their safe seats, while flatlining everywhere else.
But YouGov suggests the exact opposite: Tories piling up most of their extra votes in safe Labour seats where they're too far behind to win, and in safe Tory seats where extra votes are just wasted on bigger majorities.
ICM has the Tories 5% ahead in Labour held marginals
While also showing a hefty swing to Labour in Tory-held marginals.
That is not inconsistent with the YouGov model, which even with only a 3% national lead still shows Tories gaining some seats from Labour, while Labour gains some seats from the Tories
ICM also still had the Tories ahead in Tory held marginal, if Yougov is correct then virtually all the other pollsters are wrong but unlike ICM and Comres Yougov does not use a 2015 turnout model
.....but ICM showed a swing to Labour in Tory marginals nonetheless, which indicates Labour gaining some seats from the Tories, even if they lose the popular vote by 12%.
The big difference between ICM and YouGov is that ICM thinks the national gap between the two parties is much larger than YouGov thinks. And they may well be right about that. But what I'm talking about here is the idea that the Conservatives are underperforming in marginal seats, as compared to their national voteshares - something which the YouGov model says is the case, and which the ICM poll also doesn't disprove, since it's showing a swing to Labour in a big batch of marginals, despite the very same data showing a national swing to the Tories.
Subsamples.
Indeed, but it was HYUFD who started using an ICM subsample to disprove the YouGov model!
Whereas YouGov's model has Rudd losing her seat, Hanretty has her doubling her majority to 18 points.
Wm Hill have the odds of her losing Hastings and Rye at 1/50. I guess YouGov might be a little out on that one. Also the Lab candidate there is Captain Uninspiring himself (he was on an election debate on BBC SE on Tuesday)..
There is a deeply troubling tendency towards misogyny by some on the Left. We have seen some unpleasant examples of it here in the past. It is so counter-intuitive, but there we are.
I can honestly say I can't recall any examples from posters of the Right.
I'm not sure that SeanT has been noted for his gallantry. You get a few nasty comments on all parts of the spectrum in my experience.
The big danger for Labour is piling up even more votes than usual in their safe seats, while flatlining everywhere else.
But YouGov suggests the exact opposite: Tories piling up most of their extra votes in safe Labour seats where they're too far behind to win, and in safe Tory seats where extra votes are just wasted on bigger majorities.
ICM has the Tories 5% ahead in Labour held marginals
And according to my analysis that represents a 7% swing to the Tories in those seats (cf. 2% swing across GB).
We have to be careful with these small subsamples but the three ICM polls since Manchester have shown similar figures for that subset.
Too late to edit, I meant since the Con manifesto launch.
I'm not entirely sure there are words sufficient to express my contempt for the leadership of the Tory party, from John Major to David Cameron and George Osborne, right down to Theresa May.
The constant frustration of the public's rightful scepticism over the EU, may now result in a quasi-Marxist government of the UK.
Well done, chaps. No, really. Well DONE.
BRAVO.
Plenty of 'Tories' cheering it too
Must say though, the prospect of PM Corbyn gives me some idea of how @AlastairMeeks and @Scott_P must feel about the EU ref. I simply can't believe I might live in a country that could vote this prick and his lunatic posse in as our PM, CoftheE etc but it might happen IF SOME of the polls are accurate
Let us mark them. Any PB-er who votes for Corbyn, is, in essence, engaging in treachery. No matter how well-meant.
eh? a fair selection of PB righties (along with most of the commenters on order-order) did vote for corbyn.
You don't have a problem with the presumptive home secretary approving of the greatest mass murderer in history? Especially given she's in perfect alignment with Corbyn on pretty much everything.
I'm not entirely sure there are words sufficient to express my contempt for the leadership of the Tory party, from John Major to David Cameron and George Osborne, right down to Theresa May.
The constant frustration of the public's rightful scepticism over the EU, may now result in a quasi-Marxist government of the UK.
Well done, chaps. No, really. Well DONE.
BRAVO.
It was chavs in Hartlepool that voted for Brexit. Dave told them not to. All those who willingly boarded the leavers' ship of fools now bear the shame and culpability for the incoming Corbyn government.
If it weren't for Breggzit we'd now be entering the second year of a boringly competent posh boy regime and wondering exactly when #GO4PM would happen.
There is a deeply troubling tendency towards misogyny by some on the Left. We have seen some unpleasant examples of it here in the past. It is so counter-intuitive, but there we are.
I can honestly say I can't recall any examples from posters of the Right.
I'm not sure that SeanT has been noted for his gallantry. You get a few nasty comments on all parts of the spectrum in my experience.
@SeanT is an equal opportunity offensive w*nker (when he wants to be)
Perhaps it's the tendency on the left to prioritise group rights over individual rights? Anyone who then breaks out from their assigned category is therefore somehow immoral and fair game for any bigotry
I'm not entirely sure there are words sufficient to express my contempt for the leadership of the Tory party, from John Major to David Cameron and George Osborne, right down to Theresa May.
The constant frustration of the public's rightful scepticism over the EU, may now result in a quasi-Marxist government of the UK.
Well done, chaps. No, really. Well DONE.
BRAVO.
It was chavs in Hartlepool that voted for Brexit. Dave told them not to. All those who willingly boarded the leavers' ship of fools now bear the shame and culpability for the incoming Corbyn government.
If it weren't for Breggzit we'd now be entering the second year of a boringly competent posh boy regime and wondering exactly when #GO4PM would happen.
Thank you. As someone who has been consistent in my anti-Corbyn position on this site, I was rather shocked by the response of some PBers. I had always maintained I wouldn't vote Labour, so I don't know why they were so surprised.
@jonny83 Yes, I did - via YT. I felt sorry for Twyman!
No problem.I am anti-Corbyn but will be voting Labour. It is a democratic election.
I must say for election that feels more leader focused than any before it, it astounds me just how disliked both May and Corbyn are by most people I know and seemingly here (very unrepresentative samples I imagine of course!).
Is there anyone on PB actually supportive of May or Corbyn (or Farron for that matter)? I mean not just voting against the other, or for their party to win / not be crushed, but actually in favour of either of them being PM? It amazes me we are in this situation.
Absolutely. I think it'll be terrific if Corbyn makes it - a huge sea change in British politics to adult, redistributive politics delivered without artifice, instead of the ineffective jostling into decline of recent years. (I know him moderately well so am of course biased.)
But of course you're absolutely entitled to take the opposite view, Apoc, and I'm sorry that you have evidently been hassled over it (I've only been looking at the threads intermittently so hadn't seen it myself).
Just for completeness, and just for fun (and because I've been working on this for a while and have been looking to shoe-horn it into a conversation somewhere):
ICM - Lab marginal seats, majority of <15% (over ANY second-placed party; average Lab lead over Con in these seats was 10 points in 2015.)
19-21 May: Con lead of 3 (swing to Con of 6%) 24-26 May: Con lead of 9 (swing to Con of 9%) 26-29 May: Con lead of 5 (swing to Con of 7%)
I take Nick's point about ICM's demographic turnout filter, but that applies to their headline numbers too, so we can at least compare the above with the GB swing (4, 4, and 2 to Con respectively in the above polls).
Why on earth does Newton Dunn feel the need to publicly comment on an anonymous Twitterer with 45 followers? Oh yes, I remember, he works for the same newspaper that has just run an entire front page on an offhand tweet by the New Statesman's political editor.
I know Twitter can be a vile place but ignoring the low-follower anonynumpties is surely Twitter 101.
Why on earth does Newton Dunn feel the need to publicly comment on an anonymous Twitterer with 45 followers? Oh yes, I remember, he works for the same newspaper that has just run an entire front page on an offhand tweet by the New Statesman's political editor.
I know Twitter can be a vile place but ignoring the low-follower anonynumpties is surely Twitter 101.
Why on earth does Newton Dunn feel the need to publicly comment on an anonymous Twitterer with 45 followers? Oh yes, I remember, he works for the same newspaper that has just run an entire front page on an offhand tweet by the New Statesman's political editor.
I know Twitter can be a vile place but ignoring the low-follower anonynumpties is surely Twitter 101.
I don’t see why he shouldn’t comment, if he wants to.
Tweets that are racist probably come from anonynumpties with very few followers -- that doesn’t mean they shouldn’t be prosecuted, however.
Whatever, the comment about Amber Rudd is very unpleasant, and we all hoping for a kinder, fairer politics.
Absolutely. Jezza gives me the first manifesto I can 100% agree with.
Just curious, do you agree with Diane Abbott's 2008 comment that Mao "on balance did more good than harm"?
After all, he only killed 50 million or so.
I do think the Tories have a real problem here. They decided to save their main attacks on Corbyn and Co's past (and near present tbh) for the last few weeks of this campaign and the truth is far too lurid for anyone to believe unless it's really common knowledge (and beyond the high towers of PB and the like it seems it really hasn't been well known). Given this I think a lot of people are just convinced these stories are just Tory attacks. Just like the "Tories for Corbyn" thing I think it shows the folly of trying to keep a weak opponent in place by playing games.
Mao Zedong was a mass-murdering scumbag and so were Winston Churchill and King Leopold. Diane Abbott doesn't half talk shit sometimes. Perhaps it was at Cambridge that she learnt the attitude that she can gob off on stuff she knows little about?
People simply aren't interested in what Corbyn said in the 1980s. They're not that stupid. They are worried about the near future and rightly so. It's great that so many, including people in all age groups, are rejecting the "F*** you, I'm all right, Jack" theory of security and are showing an understanding that security will be nothing if it's not collective. The fabric of society is worth defending against the motherf*ckers of Wonga, the City and offshore private-equity firms who want to tear it apart with their moneygrabbing fangs bared and who are Tory supporters to a man. Similarly the reason the Tories hate Corbyn has nothing to do with what he said in the 1980s. It's precisely because he's a socialist and socialism is on the up again.
Notably the actual quote was 2008 and to be honest if May start applauding Leopold I would definitely count that as a black mark against her. I do appreciate the point that these can seem like trivialities to some but they are an indication of character from my view.
Obviously I'm no fan of May myself and I do think there are some real long term policy screw ups that screw my generation in particular. In particular the policy of building no houses where needed due nimbyism and the religion of house price rises as pensions is a big bug bear of mine and really needs a solution. I just don't think Corbyn is the answer.
Whereas YouGov's model has Rudd losing her seat, Hanretty has her doubling her majority to 18 points.
Wm Hill have the odds of her losing Hastings and Rye at 1/50. I guess YouGov might be a little out on that one. Also the Lab candidate there is Captain Uninspiring himself (he was on an election debate on BBC SE on Tuesday)..
You'd think it was even less likely after her creditable performance in tonight's debate.
Why on earth does Newton Dunn feel the need to publicly comment on an anonymous Twitterer with 45 followers? Oh yes, I remember, he works for the same newspaper that has just run an entire front page on an offhand tweet by the New Statesman's political editor.
I know Twitter can be a vile place but ignoring the low-follower anonynumpties is surely Twitter 101.
I don’t see why he shouldn’t comment, if he wants to.
Tweets that are racist probably come from anonynumpties with very few followers -- that doesn’t mean they shouldn’t be prosecuted, however.
Whatever, the comment about Amber Rudd is very unpleasant, and we all hoping for a kinder, fairer politics.
Of course we all hope for a kinder fairer politics, but that wan't politics. It was just some air-head after a bit of publicity. Why give it?
Whereas YouGov's model has Rudd losing her seat, Hanretty has her doubling her majority to 18 points.
Wm Hill have the odds of her losing Hastings and Rye at 1/50. I guess YouGov might be a little out on that one. Also the Lab candidate there is Captain Uninspiring himself (he was on an election debate on BBC SE on Tuesday)..
You'd think it was even less likely after her creditable performance in tonight's debate.
Amber will keep her seat. The Tory campaign has been a shambles bit not *that* much of a shambles.
Rudd has done very well, even more so following the bereavement. I am not sure the rest of the Tory front bench have done well this election campaign and I suspect a clear out is on the cards come this Summer if they get back in. May herself is a wounded leader even with a healthy win as I can see no credit she can claim for this car crash of an election
She's clearly under a lot of pressure. Who wouldn't be, with "I've totally f***ed up, and I'm getting dumped in the dustbin of history very soon" written all over them? Is she cracking up? She's laughing at inappropriate times.
Whereas YouGov's model has Rudd losing her seat, Hanretty has her doubling her majority to 18 points.
Wm Hill have the odds of her losing Hastings and Rye at 1/50. I guess YouGov might be a little out on that one. Also the Lab candidate there is Captain Uninspiring himself (he was on an election debate on BBC SE on Tuesday)..
You'd think it was even less likely after her creditable performance in tonight's debate.
Amber will keep her seat. The Tory campaign has been a shambles bit not *that* much of a shambles.
Would be a turn-up for the books, that's for sure!
Clive Lewis is 3 points ahead of his Tory challenger in Norwich South according to Hanretty's model, and has a 2/1 shot of losing it. YouGov? Lewis has a 32-point cushion!
Why on earth does Newton Dunn feel the need to publicly comment on an anonymous Twitterer with 45 followers? Oh yes, I remember, he works for the same newspaper that has just run an entire front page on an offhand tweet by the New Statesman's political editor.
I know Twitter can be a vile place but ignoring the low-follower anonynumpties is surely Twitter 101.
I don’t see why he shouldn’t comment, if he wants to.
Tweets that are racist probably come from anonynumpties with very few followers -- that doesn’t mean they shouldn’t be prosecuted, however.
Whatever, the comment about Amber Rudd is very unpleasant, and we all hoping for a kinder, fairer politics.
Of course we all hope for a kinder fairer politics, but that wan't politics. It was just some air-head after a bit of publicity. Why give it?
I don’t think it is unreasonable to call out bad behaviour myself, tbh.
Pb.com itself veered in some rather uncomfortable directions this evening.
Let’s all hope kindness prevails in what’s left of this election.
She's clearly under a lot of pressure. Who wouldn't be, with "I've totally f***ed up, and I'm getting dumped in the dustbin of history very soon" written all over them? Is she cracking up? She's laughing at inappropriate times.
Why on earth does Newton Dunn feel the need to publicly comment on an anonymous Twitterer with 45 followers? Oh yes, I remember, he works for the same newspaper that has just run an entire front page on an offhand tweet by the New Statesman's political editor.
I know Twitter can be a vile place but ignoring the low-follower anonynumpties is surely Twitter 101.
I don’t see why he shouldn’t comment, if he wants to.
Tweets that are racist probably come from anonynumpties with very few followers -- that doesn’t mean they shouldn’t be prosecuted, however.
Whatever, the comment about Amber Rudd is very unpleasant, and we all hoping for a kinder, fairer politics.
Of course we all hope for a kinder fairer politics, but that wan't politics. It was just some air-head after a bit of publicity. Why give it?
I don’t think it is unreasonable to call out bad behaviour myself, tbh.
Pb.com itself veered in some rather uncomfortable directions this evening.
Let’s all hope kindness prevails in what’s left of this election.
Little hope of kindness. Perhaps just aim for maturity.
She's clearly under a lot of pressure. Who wouldn't be, with "I've totally f***ed up, and I'm getting dumped in the dustbin of history very soon" written all over them? Is she cracking up? She's laughing at inappropriate times.
2m15s: "Will you resign if you lose seats?" She answers with a lot of spew, completely ignoring the question.
She needs to be asked again. And again.
What politician in their right mind would answer yes to that question?
She's clearly under a lot of pressure. Who wouldn't be, with "I've totally f***ed up, and I'm getting dumped in the dustbin of history very soon" written all over them? Is she cracking up? She's laughing at inappropriate times.
2m15s: "Will you resign if you lose seats?" She answers with a lot of spew, completely ignoring the question.
She needs to be asked again. And again.
What politician in their right mind would answer yes to that question?
She doesn't have to ignore it. She could say "I trust the electorate, and so I don't believe we are going to lose seats".
The thing that strikes me is the huge breadth of outcomes, all considered broadly equally likely.
Size of Con Maj:
NOM: 5.3 1 to 24: 9.4 25 to 49: 8.6 50 to 74: 7.8 75 to 99: 7.8 100 to 124: 6.8 125 to 150: 11 151 to 174: 12
The implication is that nobody really has a clue - for such massively different outcomes to be all considered roughly equally likely - ie up to the 100 to 124 band. And even above that is still considered perfectly plausible.
Someone asked earlier about Hanretty showing 1 "Other" in his forecast. It may have been answered already - it must be the Speaker, as the Con 2015 baseline is 330.
Whereas YouGov's model has Rudd losing her seat, Hanretty has her doubling her majority to 18 points.
Wm Hill have the odds of her losing Hastings and Rye at 1/50. I guess YouGov might be a little out on that one. Also the Lab candidate there is Captain Uninspiring himself (he was on an election debate on BBC SE on Tuesday)..
You'd think it was even less likely after her creditable performance in tonight's debate.
Amber will keep her seat. The Tory campaign has been a shambles bit not *that* much of a shambles.
Would be a turn-up for the books, that's for sure!
Clive Lewis is 3 points ahead of his Tory challenger in Norwich South according to Hanretty's model, and has a 2/1 shot of losing it. YouGov? Lewis has a 32-point cushion!
It would be interesting to seem some real details of the Yougov model.
The podcast describes the model as a “massive, complicated and difficult model accounting for 100s of factors”.
It sounds extraordinary to introduce such a complicated model without extensive prior testing and calibration against previous elections.
Mike L - bettors must be confused by the variance of models. YouGov has certainly thrown things. Either Clive Lewis holds on by c. 1,500 (Henretty) or c. 15,000 (YouGov) at the moment!
Whereas YouGov's model has Rudd losing her seat, Hanretty has her doubling her majority to 18 points.
Wm Hill have the odds of her losing Hastings and Rye at 1/50. I guess YouGov might be a little out on that one. Also the Lab candidate there is Captain Uninspiring himself (he was on an election debate on BBC SE on Tuesday)..
You'd think it was even less likely after her creditable performance in tonight's debate.
Amber will keep her seat. The Tory campaign has been a shambles bit not *that* much of a shambles.
Would be a turn-up for the books, that's for sure!
Clive Lewis is 3 points ahead of his Tory challenger in Norwich South according to Hanretty's model, and has a 2/1 shot of losing it. YouGov? Lewis has a 32-point cushion!
It would be interesting to seem some real details of the Yougov model.
The podcast describes the model as a “massive, complicated and difficult model accounting for 100s of factors”.
It sounds extraordinary to introduce such a complicated model without extensive prior testing and calibration against previous elections.
They say they have tested it against earlier electoral events, but not UK GEs (I think?).
Mike L - bettors must be confused by the variance of models. YouGov has certainly thrown things. Either Clive Lewis holds on by c. 1,500 (Henretty) or c. 15,000 (YouGov) at the moment!
Whereas YouGov's model has Rudd losing her seat, Hanretty has her doubling her majority to 18 points.
Wm Hill have the odds of her losing Hastings and Rye at 1/50. I guess YouGov might be a little out on that one. Also the Lab candidate there is Captain Uninspiring himself (he was on an election debate on BBC SE on Tuesday)..
You'd think it was even less likely after her creditable performance in tonight's debate.
Amber will keep her seat. The Tory campaign has been a shambles bit not *that* much of a shambles.
Would be a turn-up for the books, that's for sure!
Clive Lewis is 3 points ahead of his Tory challenger in Norwich South according to Hanretty's model, and has a 2/1 shot of losing it. YouGov? Lewis has a 32-point cushion!
It would be interesting to seem some real details of the Yougov model.
The podcast describes the model as a “massive, complicated and difficult model accounting for 100s of factors”.
It sounds extraordinary to introduce such a complicated model without extensive prior testing and calibration against previous elections.
They say they have tested it against earlier electoral events, but not UK GEs (I think?).
The thing that strikes me is the huge breadth of outcomes, all considered broadly equally likely.
Size of Con Maj:
NOM: 5.3 1 to 24: 9.4 25 to 49: 8.6 50 to 74: 7.8 75 to 99: 7.8 100 to 124: 6.8 125 to 150: 11 151 to 174: 12
The implication is that nobody really has a clue - for such massively different outcomes to be all considered roughly equally likely - ie up to the 100 to 124 band. And even above that is still considered perfectly plausible.
Thanks Mike for the kind words about my Tory% analysis on the other thread.
One thing I should have also noted was that for the Tories to get over 400 seats would require them to win seats where they were below 30% in 2015.
She's clearly under a lot of pressure. Who wouldn't be, with "I've totally f***ed up, and I'm getting dumped in the dustbin of history very soon" written all over them? Is she cracking up? She's laughing at inappropriate times.
"Theresa May is to abandon her strategy of attacking Jeremy Corbyn as she urges voters to join her on a “great national mission” to deliver a successful Brexit."
(...)
"In an email to voters on Wednesday, Mrs May also said she was “excited about the future” - a marked departure from the “project fear” tactic that has characterised much of the campaign so far."
"It will be seen as an admission that the Tories’ largely negative campaign strategy, targeting Jeremy Corbyn’s “weak leadership”, has failed sufficiently to inspire voters."
Mike L - bettors must be confused by the variance of models. YouGov has certainly thrown things. Either Clive Lewis holds on by c. 1,500 (Henretty) or c. 15,000 (YouGov) at the moment!
Whereas YouGov's model has Rudd losing her seat, Hanretty has her doubling her majority to 18 points.
Wm Hill have the odds of her losing Hastings and Rye at 1/50. I guess YouGov might be a little out on that one. Also the Lab candidate there is Captain Uninspiring himself (he was on an election debate on BBC SE on Tuesday)..
You'd think it was even less likely after her creditable performance in tonight's debate.
Amber will keep her seat. The Tory campaign has been a shambles bit not *that* much of a shambles.
Would be a turn-up for the books, that's for sure!
Clive Lewis is 3 points ahead of his Tory challenger in Norwich South according to Hanretty's model, and has a 2/1 shot of losing it. YouGov? Lewis has a 32-point cushion!
It would be interesting to seem some real details of the Yougov model.
The podcast describes the model as a “massive, complicated and difficult model accounting for 100s of factors”.
It sounds extraordinary to introduce such a complicated model without extensive prior testing and calibration against previous elections.
They say they have tested it against earlier electoral events, but not UK GEs (I think?).
I think, given the size of the 95 per cent CLs, the model is not very constraining.
It would be interesting to see the posterior probability distributions, not just the CLs.
My guess is that the posteriors are very flat, and so the model has little predictive power.
The thing that strikes me is the huge breadth of outcomes, all considered broadly equally likely.
Size of Con Maj:
NOM: 5.3 1 to 24: 9.4 25 to 49: 8.6 50 to 74: 7.8 75 to 99: 7.8 100 to 124: 6.8 125 to 150: 11 151 to 174: 12
The implication is that nobody really has a clue - for such massively different outcomes to be all considered roughly equally likely - ie up to the 100 to 124 band. And even above that is still considered perfectly plausible.
Thanks Mike for the kind words about my Tory% analysis on the other thread.
One thing I should have also noted was that for the Tories to get over 400 seats would require them to win seats where they were below 30% in 2015.
Thanks for that further piece of quality info.
Suggests it would be a massive stretch for Con to get over 400 - especially given recent direction of polls.
The thing that strikes me is the huge breadth of outcomes, all considered broadly equally likely.
Size of Con Maj:
NOM: 5.3 1 to 24: 9.4 25 to 49: 8.6 50 to 74: 7.8 75 to 99: 7.8 100 to 124: 6.8 125 to 150: 11 151 to 174: 12
The implication is that nobody really has a clue - for such massively different outcomes to be all considered roughly equally likely - ie up to the 100 to 124 band. And even above that is still considered perfectly plausible.
Thanks Mike for the kind words about my Tory% analysis on the other thread.
One thing I should have also noted was that for the Tories to get over 400 seats would require them to win seats where they were below 30% in 2015.
Scotland a special case of course, but in 2015 the Tories were below 30% in three of the six seats Baxter currently predicts them to gain (22.8% in Aberdeen South).
The thing that strikes me is the huge breadth of outcomes, all considered broadly equally likely.
Size of Con Maj:
NOM: 5.3 1 to 24: 9.4 25 to 49: 8.6 50 to 74: 7.8 75 to 99: 7.8 100 to 124: 6.8 125 to 150: 11 151 to 174: 12
The implication is that nobody really has a clue - for such massively different outcomes to be all considered roughly equally likely - ie up to the 100 to 124 band. And even above that is still considered perfectly plausible.
Thanks Mike for the kind words about my Tory% analysis on the other thread.
One thing I should have also noted was that for the Tories to get over 400 seats would require them to win seats where they were below 30% in 2015.
Scotland a special case of course, but in 2015 the Tories were below 30% in three of the six seats Baxter currently predicts them to gain (22.8% in Aberdeen South).
Is there a place we can play around with the Yougov model on their site? I can only click on their constituencies and see the probability predictions.
The thing that strikes me is the huge breadth of outcomes, all considered broadly equally likely.
Size of Con Maj:
NOM: 5.3 1 to 24: 9.4 25 to 49: 8.6 50 to 74: 7.8 75 to 99: 7.8 100 to 124: 6.8 125 to 150: 11 151 to 174: 12
The implication is that nobody really has a clue - for such massively different outcomes to be all considered roughly equally likely - ie up to the 100 to 124 band. And even above that is still considered perfectly plausible.
Thanks Mike for the kind words about my Tory% analysis on the other thread.
One thing I should have also noted was that for the Tories to get over 400 seats would require them to win seats where they were below 30% in 2015.
Scotland a special case of course, but in 2015 the Tories were below 30% in three of the six seats Baxter currently predicts them to gain (22.8% in Aberdeen South).
Fair point - but as you say Scotland is special due to an "extra" main party - and a very big one at that.
Even if Con were to get 400, Scotland would only be making a very small contribution - they would have to do the job principally in England & Wales.
The thing that strikes me is the huge breadth of outcomes, all considered broadly equally likely.
Size of Con Maj:
NOM: 5.3 1 to 24: 9.4 25 to 49: 8.6 50 to 74: 7.8 75 to 99: 7.8 100 to 124: 6.8 125 to 150: 11 151 to 174: 12
The implication is that nobody really has a clue - for such massively different outcomes to be all considered roughly equally likely - ie up to the 100 to 124 band. And even above that is still considered perfectly plausible.
Thanks Mike for the kind words about my Tory% analysis on the other thread.
One thing I should have also noted was that for the Tories to get over 400 seats would require them to win seats where they were below 30% in 2015.
Scotland a special case of course, but in 2015 the Tories were below 30% in three of the six seats Baxter currently predicts them to gain (22.8% in Aberdeen South).
Is there a place we can play around with the Yougov model on their site? I can only click on their constituencies and see the probability predictions.
How do you want to play around with it? I don't think you can change their assumptions.
The thing that strikes me is the huge breadth of outcomes, all considered broadly equally likely.
Size of Con Maj:
NOM: 5.3 1 to 24: 9.4 25 to 49: 8.6 50 to 74: 7.8 75 to 99: 7.8 100 to 124: 6.8 125 to 150: 11 151 to 174: 12
The implication is that nobody really has a clue - for such massively different outcomes to be all considered roughly equally likely - ie up to the 100 to 124 band. And even above that is still considered perfectly plausible.
Thanks Mike for the kind words about my Tory% analysis on the other thread.
One thing I should have also noted was that for the Tories to get over 400 seats would require them to win seats where they were below 30% in 2015.
Scotland a special case of course, but in 2015 the Tories were below 30% in three of the six seats Baxter currently predicts them to gain (22.8% in Aberdeen South).
That's why I don't trust Baxter. I tend to the view that by and large the election in each seat will be Tory vs Non-Tory, and that Tory gains will come from growing the Tory vote, and that the remaining non Tory vote mix is likely to adjust in ways not helpful to the Tories, even in Scotland.
The thing that strikes me is the huge breadth of outcomes, all considered broadly equally likely.
Size of Con Maj:
NOM: 5.3 1 to 24: 9.4 25 to 49: 8.6 50 to 74: 7.8 75 to 99: 7.8 100 to 124: 6.8 125 to 150: 11 151 to 174: 12
The implication is that nobody really has a clue - for such massively different outcomes to be all considered roughly equally likely - ie up to the 100 to 124 band. And even above that is still considered perfectly plausible.
Thanks Mike for the kind words about my Tory% analysis on the other thread.
One thing I should have also noted was that for the Tories to get over 400 seats would require them to win seats where they were below 30% in 2015.
Scotland a special case of course, but in 2015 the Tories were below 30% in three of the six seats Baxter currently predicts them to gain (22.8% in Aberdeen South).
Fair point - but as you say Scotland is special due to an "extra" main party - and a very big one at that.
Even if Con were to get 400, Scotland would only be making a very small contribution - they would have to do the job principally in England & Wales.
There is an extra main party, but in Scottish seats where the Tories have relative strength, there is often significant anti-Tory switching where Lab or the Lib Dems get squeezed
The thing that strikes me is the huge breadth of outcomes, all considered broadly equally likely.
Size of Con Maj:
NOM: 5.3 1 to 24: 9.4 25 to 49: 8.6 50 to 74: 7.8 75 to 99: 7.8 100 to 124: 6.8 125 to 150: 11 151 to 174: 12
The implication is that nobody really has a clue - for such massively different outcomes to be all considered roughly equally likely - ie up to the 100 to 124 band. And even above that is still considered perfectly plausible.
Thanks Mike for the kind words about my Tory% analysis on the other thread.
One thing I should have also noted was that for the Tories to get over 400 seats would require them to win seats where they were below 30% in 2015.
Scotland a special case of course, but in 2015 the Tories were below 30% in three of the six seats Baxter currently predicts them to gain (22.8% in Aberdeen South).
Is there a place we can play around with the Yougov model on their site? I can only click on their constituencies and see the probability predictions.
How do you want to play around with it? I don't think you can change their assumptions.
oh, yougov had a turnout filter you could change in EuRef.
The thing that strikes me is the huge breadth of outcomes, all considered broadly equally likely.
Size of Con Maj:
NOM: 5.3 1 to 24: 9.4 25 to 49: 8.6 50 to 74: 7.8 75 to 99: 7.8 100 to 124: 6.8 125 to 150: 11 151 to 174: 12
The implication is that nobody really has a clue - for such massively different outcomes to be all considered roughly equally likely - ie up to the 100 to 124 band. And even above that is still considered perfectly plausible.
Thanks Mike for the kind words about my Tory% analysis on the other thread.
One thing I should have also noted was that for the Tories to get over 400 seats would require them to win seats where they were below 30% in 2015.
Scotland a special case of course, but in 2015 the Tories were below 30% in three of the six seats Baxter currently predicts them to gain (22.8% in Aberdeen South).
Is there a place we can play around with the Yougov model on their site? I can only click on their constituencies and see the probability predictions.
How do you want to play around with it? I don't think you can change their assumptions.
oh, yougov had a turnout filter you could change in EuRef.
Suspect their model is too complex to be able to recalculate in real time on a website. Although you could pre-compute a bunch of different scenarios.
My assumption is it's the ex-lab, 2015 ukip, then brexit vote (that was blue a month ago) - which is boosting lab.
Any other theories?
I can't see why anyone fitting that description would vote for Corbyn. I am one of them, maybe not representative though. I live in a very Ukip area and have never heard anyone say one good thing about Corbyn, he is a laughing stock. In fact the only time two of my mates have ever brought up politics was to say 'wtf is that new labour weirdo all about'
My assumption is it's the ex-lab, 2015 ukip, then brexit vote (that was blue a month ago) - which is boosting lab.
Any other theories?
I can't see why anyone fitting that description would vote for Corbyn. I am one of them, maybe not representative though. I live in a very Ukip area and have never heard anyone say one good thing about Corbyn, he is a laughing stock. In fact the only time two of my mates have ever brought up politics was to say 'wtf is that new labour weirdo all about'
He also states that 20% of UKIP 2015 voters are undecided at the moment. Will be interesting to know which way they break, although they only represent ~2.5% of votes up for grabs
Rudd has done very well, even more so following the bereavement. I am not sure the rest of the Tory front bench have done well this election campaign and I suspect a clear out is on the cards come this Summer if they get back in. May herself is a wounded leader even with a healthy win as I can see no credit she can claim for this car crash of an election
If the tories do win -
Ruth Davidson needs a seat,she's the tories only hope for the next election unless the new batch brings riches.
Any euphoria (if the Tories do win) on election night, will soon be replaced by a muttering campaign against TM as the pressures of BREXIT kicks in, and as I remember from 1990 - they are not afraid to stick the knife in-the list of assassins is endless - ranging from Johnson to Hammond in her cabinet and a whole lot in between ranging from sceptics (Redwood, Rees Mogg) through to Clarke and the Tory elder statesmen from the Lords (which I would imagine will soon include Cameron) with a few nudges from Osborne et al. I genuinely cannot see May staying in very long........( please remind me in 2020 when I am proved wrong)
How come there are two YG 50,000-respondent polls in Wikipedia?
I think it's a rolling 7 day, with 7,000 per day.
You therefore need to be careful with ELBOW - to avoid double counting you only want one poll every 7 days - ie include the first one, then ignore the next six and then take the one exactly a week after the first one.
Sounds bloody complicated, with just a week to go!
Anyhow, including all the polls in Wiki, with fieldwork ending after the 29th inclusive, I get:
Con 42.8 Lab 36.5 LD 8.3 UKIP 4.2
Tory lead 6.3 (-2.1)
Sorry if complicated.
I was just making the point that if you include every YouGov "50,000" poll then you are massively doublecounting - as each poll is 6/7 the same as the previous one..
OK so just including ICM, Kantar, SurveyMonkey and tonight's "normal YG", I get:
Mike L - bettors must be confused by the variance of models. YouGov has certainly thrown things. Either Clive Lewis holds on by c. 1,500 (Henretty) or c. 15,000 (YouGov) at the moment!
It would be interesting to seem some real details of the Yougov model.
The podcast describes the model as a “massive, complicated and difficult model accounting for 100s of factors”.
It sounds extraordinary to introduce such a complicated model
They say they have tested it against earlier electoral events, but not UK GEs (I think?).
(1/2) Yes, against Brexit and in the US.
I am one of the 49,000. On the upside, in the poll I get a screen with the candidates, named and with their party descriptions in ballot paper order for my constituency. That must add a level of accuracy, particularly for seats where known minor party candidates are in contention, that the "which party will you vote for" polls do not have.
The "100s of factors" will be socio-economic, based on their exceptionally detailed data on Yougov panellists. They know what I buy, what I watch and listen to, what I read, and what I think on a whole host of issues. Yougov specialises in cross-correlating all of this data: much of it is on their website and 'for fun', you can click on a celebrity and see, for (as a made up) example, "people who like JK Rowling also like Ed Miliband, Cornflakes, and Radio 4".
The potential downside, however, is this. Their model only has 75 people in my constituency, which they use as a 'base' to get some idea on trends, but it then finds people who are 'similar' to me (and all the other demographics they need to match the constituency population, data probably derived from the census) living in nearby constituencies. They then use the voting preferences and trends of such people to 'boost' the sample for my local area.
In areas where there is a swathe of seats with similar characteristics, for example the shires or Liverpool or other big cities, fine. People switching from Labour to Tory or vice versa in these seats probably do share similar characteristics. Where the model must be weak, however, is in picking up the effect of ultra-local candidates or the actual campaign in unusual seats. This must be particularly relevant for the minor parties - for example in Norman Lamb's case he only gets elected because the type of people who vote for him in his Norfolk seat won't be thinking of backing the LibDems in large numbers in the other Norfolk seats. If the habits of the rest of Norfolk were projected into his seat, he would be nowhere in contention.
(2/2) I don't see how Yougov can resolve this issue without relying over-heavily on the responses of the local 75 people? For example they have the Independent leading in East Devon seat - that can't be because similar demographics in the rest of Devon are leaning Indy, and must be because a randomly high proportion of the East Devon 75 are saying they will vote for her. At the same time, the demographics of surrounding seats is clearly having an undue impact - for example they have Kensington as potentially marginal for the Tories, which can only be because they have 'imported' people from surrounding areas like Hammersmith who are not as Tory as residents of Kensington. Contrarywise, the demographics of Hastings are very different from the surrounding rural Kent seats (many more people on benefits, for example) and Yougov appears to be over-projecting non-Tory voters into that seat.
These two potential errors work in opposite directions - reliance on the 75 increases the randomness and the apparent chances of outsiders (East Devon) - reliance on voters from the surrounding area 'smooths' the politics of an area and reduces the chances of candidates known only in their seat. I really don't see any way that Yougov can resolve this except by trying to 'guess' a crude balance between two different sorts of potential error?
Finally, one element is missing altogether from their model. British GE's are fascinating because the political history and local campaigns have a big local impact. Parties can retain a few voters in a seat because of some freak by-election or issue years back into the past, or the reputation of one of the candidates. YouGov is probably excellent on the demographics but politics is more than that - and local political history will be missing from its model and cannot be derived accurately from a sample of just 75 people.
These are in addition to the well-discussed problems of YouGov's panel being both self-selected and of people who enjoy (or are willing, for 50p a time) to fill in lots of surveys. This is unlikely to reflect the population as a whole, however hard they try to weight the different demographics of their panel.
Even at this late stage of the GE campaign, one still stumbles across some staggering unlikely predictions. A case in point is the West Wales constituency of Preseli Pembrokeshire, just about as safe a Tory seat as one is likely to find in the entire principality, as reflected by the betting odds, where the most "generous" price available on the Tories is the 1/25 available from Betfair Sportsbook, while those nice folk at Bet365 will give you odds of 18/1 against Labour capturing this seat from the Blue Team. The outcome therefore is absolutely cut and dried or at least so one would have thought ....... but wait ....... along comes Chris Hanretty's Election Forecast team who decides that this is the third most likely seat Labour is capable of capturing in the entire nation and equally the third most likely seat the Tories are likely to lose, in both instances making this a 76% likelihood, or a better than 3 in 4 chance of actually happening. Is there something unusual about this seat which I've missed or has it somehow slipped through the most elementary of filters presumably applied by Election Forecast to result in such a seemingly near impossible result. I thought it was pushing things somewhat to imagine that either the Tories or Labour could conceivably win Brighton Pavilion, but this seems to take things onto an entirely new level.
If I recall, in the 2015 election aftermath, there was a huge amount of criticism of YouGov, not just because of the inaccuracy of their polling, but because the fact that they were producing daily polls was creating a narrative that was driving out the results of other pollsters. This was compounded then by the belief that because of this narrative it was resulting in "herding" with the other pollsters reluctant to publish polls which seemed to be so obviously against the narrative that had been built up.
And yet, only a couple of years later, they have been allowed to do it again. The other polls are barely allowed to get a word in edgeways because once again YouGov are driving the agenda with their highly regular polling creating a narrative of rapidly converging gaps and very close polls. Just think what this campaign would be looking like if we didn't have YouGov leading the headlines every couple of days. The closest poll would be about a 6-7 point Tory lead, and they would be looking like outliers with a mid point lead of about 9-10% and a likely healthy Tory majority.
If YouGov screw up again, after the influence they have once again had on the campaign, then they should be banned from political polling in future.
@alex. Well you'd hope no newspaper would hire them in the future to do polls if they have messed this one up. Agree that they [polls] probably play too much of a role these days, despite how much we love them on here.
Interesting podcast - after listening to Prof Curtis’ illuminations on his recent YouGov poll, I've decided not to place any bets on a hung parliament.
@alex. Well you'd hope no newspaper would hire them in the future to do polls if they have messed this one up. Agree that they [polls] probably play too much of a role these days, despite how much we love them on here.
I doubt it. Newspapers love them because they are creating stories. They don't care whether they are actually right or not. Imagine how boring and mundane the election would look without them. In fact it will be even better for the newspapers if they are wildly wrong. Then they get a whole load of stories to write in the aftermath as well.
@alex. Well you'd hope no newspaper would hire them in the future to do polls if they have messed this one up. Agree that they [polls] probably play too much of a role these days, despite how much we love them on here.
I doubt it. Newspapers love them because they are creating stories. They don't care whether they are actually right or not. Imagine how boring and mundane the election would look without them. In fact it will be even better for the newspapers if they are wildly wrong. Then they get a whole load of stories to write in the aftermath as well.
Yeah, that's a good point
If they are seriously out, I wonder if there will be talk about banning them during the campaign. Not sure I'd want that, but wonder if it will be discussed.
If I recall, in the 2015 election aftermath, there was a huge amount of criticism of YouGov, not just because of the inaccuracy of their polling, but because the fact that they were producing daily polls was creating a narrative that was driving out the results of other pollsters. This was compounded then by the belief that because of this narrative it was resulting in "herding" with the other pollsters reluctant to publish polls which seemed to be so obviously against the narrative that had been built up.
And yet, only a couple of years later, they have been allowed to do it again. The other polls are barely allowed to get a word in edgeways because once again YouGov are driving the agenda with their highly regular polling creating a narrative of rapidly converging gaps and very close polls. Just think what this campaign would be looking like if we didn't have YouGov leading the headlines every couple of days. The closest poll would be about a 6-7 point Tory lead, and they would be looking like outliers with a mid point lead of about 9-10% and a likely healthy Tory majority.
If YouGov screw up again, after the influence they have once again had on the campaign, then they should be banned from political polling in future.
They are nevertheless selling a product in demand, from both Tories eager to frighten in the waverers and as forlorn hope for all the non-Tories. They even provided us with a full day's entertainment yesterday.
If you want an 'it's close' poll, who else do you call?
O/T - the scariness of Labour is embodies in the leading tri-umverate of Corbyn, McDonnell and Abbott. But don't forget Corbyn's press man is an actual bonafide North Korean admiring Stalinist.
@alex. Well you'd hope no newspaper would hire them in the future to do polls if they have messed this one up. Agree that they [polls] probably play too much of a role these days, despite how much we love them on here.
I doubt it. Newspapers love them because they are creating stories. Imagine how boring and mundane the election would look without them.
Yeah, that's a good point
If they are seriously out, I wonder if there will be talk about banning them during the campaign. Not sure I'd want that, but wonder if it will be discussed.
Of course, that is a big if.
I'm opposed to banning things in principle. If there was found to be a deliberate attempt to subvert or influence the election result that would be a different matter - but I think the pollsters are genuinely trying to get the result right and the different outcomes we are seeing largely trace to different assumptions on turn out - all of which you can make a reasonable case for. The nerves of punters, candidates or party machines are neither here nor there - with all due respect to them all!
Didn't catch the debate last night - was it the shouty mess it's being portrayed as in some of the press?
@alex. Well you'd hope no newspaper would hire them in the future to do polls if they have messed this one up. Agree that they [polls] probably play too much of a role these days, despite how much we love them on here.
I doubt it. Newspapers love them because they are creating stories. Imagine how boring and mundane the election would look without them.
Yeah, that's a good point
If they are seriously out, I wonder if there will be talk about banning them during the campaign. Not sure I'd want that, but wonder if it will be discussed.
Of course, that is a big if.
I'm opposed to banning things in principle. If there was found to be a deliberate attempt to subvert or influence the election result that would be a different matter - but I think the pollsters are genuinely trying to get the result right and the different outcomes we are seeing largely trace to different assumptions on turn out - all of which you can make a reasonable case for. The nerves of punters, candidates or party machines are neither here nor there - with all due respect to them all!
Didn't catch the debate last night - was it the shouty mess it's being portrayed as in some of the press?
People still get punished for accidental death They might not be meaning to influence it, but I suspect they are! (Of course, caveat that they might be right... titters)
Debate was alright, although I missed the middle bit. Definitely shouty, and probably not at all useful in debating anything.
@alex. Well you'd hope no newspaper would hire them in the future to do polls if they have messed this one up. Agree that they [polls] probably play too much of a role these days, despite how much we love them on here.
I doubt it. Newspapers love them because they are creating stories. Imagine how boring and mundane the election would look without them.
Yeah, that's a good point
If they are seriously out, I wonder if there will be talk about banning them during the campaign. Not sure I'd want that, but wonder if it will be discussed.
Of course, that is a big if.
I'm opposed to banning things in principle. If there was found to be a deliberate attempt to subvert or influence the election result that would be a different matter - but I think the pollsters are genuinely trying to get the result right and the different outcomes we are seeing largely trace to different assumptions on turn out - all of which you can make a reasonable case for. The nerves of punters, candidates or party machines are neither here nor there - with all due respect to them all!
It is the ubiquity of YouGov that is the issue, not necessarily the fact that there are polls. Look at how other pollsters are almost being ignored - you would be forgiven for not knowing there was a Kantar poll yesterday showing a 10% lead. Who cares when YouGov are available with their latest daily offering, with added excitement of Labour within touching distance of a majority?
Labour need to be careful with the whooping/jeering crowds - thoughts return of Kinnock's Sheffield rally. Youthful exuberance tends not to go down too well with the humbug-sucking blue-rinse brigade.
Labour need to be careful with the whooping/jeering crowds - thoughts return of Kinnock's Sheffield rally. Youthful exuberance tends not to go down too well with the humbug-sucking blue-rinse brigade.
@alex. Well you'd hope no newspaper would hire them in the future to do polls if they have messed this one up. Agree that they [polls] probably play too much of a role these days, despite how much we love them on here.
You would think that by now, aided by massively powerful computers and with polling having become a major element of all western democratic elections over several decades, that the ever-evolving methodologies would have reached such a stage of development that only the smallest refinements would now be considered, let alone thought necessary for implementation.
Yet it would seem that we are nowhere near this stage, with the major UK pollsters, possibly the most experienced in the world, fighting tooth and claw over who has the most reliable methodology in place with just one week left before polling day. This is amply demonstrated by arguably the two pre-eminent firms, YouGov and ICM who are currently around 8% apart as regards what they claim is the Tories' lead over Labour. As things currently stand, I can only imagine that within the losing team, heads will roll big time after the result is known. Of course should their predictions converge over the final few days, which I fully expect will be the case and they finish up within 3% - 4% or possibly closer still to correctly forecasting the outcome, then of course both firms will claim to have been right all long and the irritating thing for us PBers and other commentators in general is that none of us will never know who was telling the truth!
Labour need to be careful with the whooping/jeering crowds - thoughts return of Kinnock's Sheffield rally. Youthful exuberance tends not to go down too well with the humbug-sucking blue-rinse brigade.
Labour need to be careful with the whooping/jeering crowds - thoughts return of Kinnock's Sheffield rally. Youthful exuberance tends not to go down too well with the humbug-sucking blue-rinse brigade.
True, but what can they do?
The "blue-rinse brigade" you mean? - What they can and will do is to vote in enormous numbers as they did post-Sheffield.
@alex. Well you'd hope no newspaper would hire them in the future to do polls if they have messed this one up. Agree that they [polls] probably play too much of a role these days, despite how much we love them on here.
You would think that by now, aided by massively powerful computers and with polling having become a major element of all western democratic elections over several decades, that the ever-evolving methodologies would have reached such a stage of development that only the smallest refinements would now be considered, let alone thought necessary for implementation.
Yet it would seem that we are nowhere near this stage, with the major UK pollsters, possibly the most experienced in the world, fighting tooth and claw over who has the most reliable methodology in place with just one week left before polling day. This is amply demonstrated by arguably the two pre-eminent firms, YouGov and ICM who are currently around 8% apart as regards what they claim is the Tories' lead over Labour. As things currently stand, I can only imagine that within the losing team, heads will roll big time after the result is known. Of course should their predictions converge over the final few days, which I fully expect will be the case and they finish up within 3% - 4% or possibly closer still to correctly forecasting the outcome, then of course both firms will claim to have been right all long and the irritating thing for us PBers and other commentators in general is that none of us will never know who was telling the truth!
Yeah, I wonder what "tweaks" the various companies will make in the next coming days. Panelbase have already tweaked their methodologies mid-campaign, which seems pretty crappy. I wonder if we will also see some movement back to the Conservatives due to the perceived tightening of the race, although I'm not sure the wider public follow this stuff that much.
Labour need to be careful with the whooping/jeering crowds - thoughts return of Kinnock's Sheffield rally. Youthful exuberance tends not to go down too well with the humbug-sucking blue-rinse brigade.
Labour need to be careful with the whooping/jeering crowds - thoughts return of Kinnock's Sheffield rally. Youthful exuberance tends not to go down too well with the humbug-sucking blue-rinse brigade.
True, but what can they do?
The "blue-rinse brigade" you mean? - What they can and will do is to vote in enormous numbers as they did post-Sheffield.
Labour need to be careful with the whooping/jeering crowds - thoughts return of Kinnock's Sheffield rally. Youthful exuberance tends not to go down too well with the humbug-sucking blue-rinse brigade.
Labour need to be careful with the whooping/jeering crowds - thoughts return of Kinnock's Sheffield rally. Youthful exuberance tends not to go down too well with the humbug-sucking blue-rinse brigade.
True, but what can they do?
The "blue-rinse brigade" you mean? - What they can and will do is to vote in enormous numbers as they did post-Sheffield.
Afternoon all, I hope that upper lips are stiff once more after that appauling display of bed wetting yesterday. Seems like not many people watched the debates even here on PB, and the general view in the papers seems to be that Jezza looked shagged out and the rest would not look out of place in a sixth form common room
Who put something in BigJohnFreeOwls's coffee ? More ramping posts from him today that the rest of the last month!
@alex. Well you'd hope no newspaper would hire them in the future to do polls if they have messed this one up. Agree that they [polls] probably play too much of a role these days, despite how much we love them on here.
I doubt it. Newspapers love them because they are creating stories. Imagine how boring and mundane the election would look without them.
Yeah, that's a good point
If they are seriously out, I wonder if there will be talk about banning them during the campaign. Not sure I'd want that, but wonder if it will be discussed.
Of course, that is a big if.
I'm opposed to banning things in principle. If there was found to be a deliberate attempt to subvert or influence the election result that would be a different matter - but I think the pollsters are genuinely trying to get the result right and the different outcomes we are seeing largely trace to different assumptions on turn out - all of which you can make a reasonable case for. The nerves of punters, candidates or party machines are neither here nor there - with all due respect to them all!
It is the ubiquity of YouGov that is the issue, not necessarily the fact that there are polls. Look at how other pollsters are almost being ignored - you would be forgiven for not knowing there was a Kantar poll yesterday showing a 10% lead. Who cares when YouGov are available with their latest daily offering, with added excitement of Labour within touching distance of a majority?
Exactly, yougov must be loving all the publicity. The polls and pollsters are the news, not the politicians. Boring 'tories are still 15% ahead' stories day after day don't cut it in this media obsessed age.
In the end this is all manufactured, if you assume 16-24 year olds turn out at 110% (students of course somehow voting twice) and that 65+ all decide to go to the seaside for the day then Mr. Corbyn is prime minister and you have your headline.
@alex. Well you'd hope no newspaper would hire them in the future to do polls if they have messed this one up. Agree that they [polls] probably play too much of a role these days, despite how much we love them on here.
You would think that by now, aided by massively powerful computers and with polling having become a major element of all western democratic elections over several decades, that the ever-evolving methodologies would have reached such a stage of development that only the smallest refinements would now be considered, let alone thought necessary for implementation.
Yet it would seem that we are nowhere near this stage, with the major UK pollsters, possibly the most experienced in the world, fighting tooth and claw over who has the most reliable methodology in place with just one week left before polling day. This is amply demonstrated by arguably the two pre-eminent firms, YouGov and ICM who are currently around 8% apart as regards what they claim is the Tories' lead over Labour. As things currently stand, I can only imagine that within the losing team, heads will roll big time after the result is known. Of course should their predictions converge over the final few days, which I fully expect will be the case and they finish up within 3% - 4% or possibly closer still to correctly forecasting the outcome, then of course both firms will claim to have been right all long and the irritating thing for us PBers and other commentators in general is that none of us will never know who was telling the truth!
Yeah, I wonder what "tweaks" the various companies will make in the next coming days. Panelbase have already tweaked their methodologies mid-campaign, which seems pretty crappy. I wonder if we will also see some movement back to the Conservatives due to the perceived tightening of the race, although I'm not sure the wider public follow this stuff that much.
There is still the question of why UK polls are so much worse than in the US or France? OK the capricious semi-random nature of our voting system with 650 micro-level results each potentially swung by a single vote magnifies any error in the polls, making reports of "mistakes" that are actually within margin of error more likely. This is why Yougov is trying to use its much larger reach to probe down toward constituency level. But the fact remains that on vote share the French polls were almost spot on and the error in the US election (also with a result distorted by the system, if at larger state level) was much less than is typical here in the UK? Why are we such uniquely difficult buggers to poll accurately?
Seems like not many people watched the debates even here on PB, and the general view in the papers seems to be that Jezza looked shagged out and the rest would not look out of place in a sixth form common room
Theresa May's judgment has seldom looked sounder than when she decided to stay the heck away from this awful, bent, babyish custard-pie fight.
@alex. Well you'd hope no newspaper would hire them in the future to do polls if they have messed this one up. Agree that they [polls] probably play too much of a role these days, despite how much we love them on here.
I doubt it. Newspapers love them because they are creating stories. Imagine how boring and mundane the election would look without them.
Yeah, that's a good point
If they are seriously out, I wonder if there will be talk about banning them during the campaign. Not sure I'd want that, but wonder if it will be discussed.
Of course, that is a big if.
I'm opposed to banning things in principle. If there was found to be a deliberate attempt to subvert or influence the election result that would be a different matter - but I think the pollsters are genuinely trying to get the result right and the different outcomes we are seeing largely trace to different assumptions on turn out - all of which you can make a reasonable case for. The nerves of punters, candidates or party machines are neither here nor there - with all due respect to them all!
It is the ubiquity of YouGov that is the issue, not necessarily the fact that there are polls. Look at how other pollsters are almost being ignored - you would be forgiven for not knowing there was a Kantar poll yesterday showing a 10% lead. Who cares when YouGov are available with their latest daily offering, with added excitement of Labour within touching distance of a majority?
Exactly, yougov must be loving all the publicity. The polls and pollsters are the news, not the politicians. Boring 'tories are still 15% ahead' stories day after day don't cut it in this media obsessed age.
They won't enjoy the publicity on June 9th if the Tories are 15% ahead....
@alex. Well you'd hope no newspaper would hire them in the future to do polls if they have messed this one up. Agree that they [polls] probably play too much of a role these days, despite how much we love them on here.
I doubt it. Newspapers love them because they are creating stories. Imagine how boring and mundane the election would look without them.
Yeah, that's a good point
If they are seriously out, I wonder if there will be talk about banning them during the campaign. Not sure I'd want that, but wonder if it will be discussed.
Of course, that is a big if.
I'm opposed to banning things in principle. If there was found to be a deliberate attempt to subvert or influence the election result that would be a different matter - but I think the pollsters are genuinely trying to get the result right and the different outcomes we are seeing largely trace to different assumptions on turn out - all of which you can make a reasonable case for. The nerves of punters, candidates or party machines are neither here nor there - with all due respect to them all!
It is the ubiquity of YouGov that is the issue, not necessarily the fact that there are polls. Look at how other pollsters are almost being ignored - you would be forgiven for not knowing there was a Kantar poll yesterday showing a 10% lead. Who cares when YouGov are available with their latest daily offering, with added excitement of Labour within touching distance of a majority?
There is money at stake for all the pollsters. YouGov is gambling that its attempt at a deep-analysis model will pay off. If they are judged "right", they will be able to dine out on it for years (and their shares would represent a great 'buy' right now!). If they fail, they'll suffer accordingly - but maybe the commercial upside of being uniquely right is the greater? Hence their gamble is rational from a business perspective?
@alex. Well you'd hope no newspaper would hire them in the future to do polls if they have messed this one up. Agree that they [polls] probably play too much of a role these days, despite how much we love them on here.
I doubt it. Newspapers love them because they are creating stories. Imagine how boring and mundane the election would look without them.
Yeah, that's a good point
If they are seriously out, I wonder if there will be talk about banning them during the campaign. Not sure I'd want that, but wonder if it will be discussed.
Of course, that is a big if.
I'm opposed to banning things in principle. If there was found to be a deliberate attempt to subvert or influence the election result that would be a different matter - but I think the pollsters are genuinely trying to get the result right and the different outcomes we are seeing largely trace to different assumptions on turn out - all of which you can make a reasonable case for. The nerves of punters, candidates or party machines are neither here nor there - with all due respect to them all!
It is the ubiquity of YouGov that is the issue, not necessarily the fact that there are polls. Look at how other pollsters are almost being ignored - you would be forgiven for not knowing there was a Kantar poll yesterday showing a 10% lead. Who cares when YouGov are available with their latest daily offering, with added excitement of Labour within touching distance of a majority?
Exactly, yougov must be loving all the publicity. The polls and pollsters are the news, not the politicians. Boring 'tories are still 15% ahead' stories day after day don't cut it in this media obsessed age.
They won't enjoy the publicity on June 9th if the Tories are 15% ahead....
I'm sure they'll conveniently have a couple of unpublished polls from the day of the election, state "late swing" and move on. Or a couple of experimental models that conveniently showed the outcome of the election that they'll then be able to correct for going forward.
Seems like not many people watched the debates even here on PB, and the general view in the papers seems to be that Jezza looked shagged out and the rest would not look out of place in a sixth form common room
Theresa May's judgment has seldom looked sounder than when she decided to stay the heck away from this awful, bent, babyish custard-pie fight.
I think the Daily Mail wrote that without having seen it.
Anyone who watched it or isn't the Daily Mail would accept it was very interesting and informative and that Mrs May made a serious error of judgement by not showing up. The broadcast media this morning are talking about nothing else.
Seems like not many people watched the debates even here on PB, and the general view in the papers seems to be that Jezza looked shagged out and the rest would not look out of place in a sixth form common room
Theresa May's judgment has seldom looked sounder than when she decided to stay the heck away from this awful, bent, babyish custard-pie fight.
@alex. Well you'd hope no newspaper would hire them in the future to do polls if they have messed this one up. Agree that they [polls] probably play too much of a role these days, despite how much we love them on here.
I doubt it. Newspapers love them because they are creating stories. Imagine how boring and mundane the election would look without them.
Yeah, that's a good point
If they are seriously out, I wonder if there will be talk about banning them during the campaign. Not sure I'd want that, but wonder if it will be discussed.
Of course, that is a big if.
I'm opposed to banning things in principle. If there was found to be a deliberate attempt to subvert or influence the election result that would be a different matter - but I think the pollsters are genuinely trying to get the result right and the different outcomes we are seeing largely trace to different assumptions on turn out - all of which you can make a reasonable case for. The nerves of punters, candidates or party machines are neither here nor there - with all due respect to them all!
It is the ubiquity of YouGov that is the issue, not necessarily the fact that there are polls. Look at how other pollsters are almost being ignored - you would be forgiven for not knowing there was a Kantar poll yesterday showing a 10% lead. Who cares when YouGov are available with their latest daily offering, with added excitement of Labour within touching distance of a majority?
There is money at stake for all the pollsters. YouGov is gambling that its attempt at a deep-analysis model will pay off. If they are judged "right", they will be able to dine out on it for years (and their shares would represent a great 'buy' right now!). If they fail, they'll suffer accordingly - but maybe the commercial upside of being uniquely right is the greater? Hence their gamble is rational from a business perspective?
They are going to have to get very lucky though. On their sample size they are going to be averaging something like 70 people per constituency, which gives a margin of error over over 10% making the results in danger of being worthless except for good fortune. Additionally a sample of 70 is hideously easy to game, with only a very small handful of activists with the right profile being required to massively swing the result in a small group of continuencies.
Seems like not many people watched the debates even here on PB, and the general view in the papers seems to be that Jezza looked shagged out and the rest would not look out of place in a sixth form common room
Theresa May's judgment has seldom looked sounder than when she decided to stay the heck away from this awful, bent, babyish custard-pie fight.
Anyone who watched it or isn't the Daily Mail would accept it was very interesting and informative and that Mrs May made a serious error of judgement by not showing up. The broadcast media this morning are talking about nothing else.
Big surprise. So the rightleaning DM says it was the right move, and the leftie broadcast media say it was a screw up.... and the news here is ?
I'm not entirely sure there are words sufficient to express my contempt for the leadership of the Tory party, from John Major to David Cameron and George Osborne, right down to Theresa May.
The constant frustration of the public's rightful scepticism over the EU, may now result in a quasi-Marxist government of the UK.
Well done, chaps. No, really. Well DONE.
BRAVO.
I'm sanguine.
They think Brexit will be a clusterf**k, they ain't seen nothin yet.
But you are right, the Conservatives are not an innocent party to this.
I think the Queen would be sanguine as well. After all, there has never been a new monarch under a Labour government.
YouGov is not doing these polls for free, is it? Surely it is doing what any business does - delivering on orders placed by customers. It's the media that commissions this stuff. And the Times has definitely got good value for its money.
@alex. Well you'd hope no newspaper would hire them in the future to do polls if they have messed this one up. Agree that they [polls] probably play too much of a role these days, despite how much we love them on here.
I doubt it. Newspapers love them because they are creating stories. Imagine how boring and mundane the election would look without them.
Yeah, that's a good point
If they are seriously out, I wonder if there will be talk about banning them during the campaign. Not sure I'd want that, but wonder if it will be discussed.
Of course, that is a big if.
I'm opposed to banning things in principle. If there was found to be a deliberate attempt to subvert or influence the election result that would be a different matter - but I think the pollsters are genuinely trying to get the result right and the different outcomes we are seeing largely trace to different assumptions on turn out - all of which you can make a reasonable case for. The nerves of punters, candidates or party machines are neither here nor there - with all due respect to them all!
It is the ubiquity of YouGov that is the issue, not necessarily the fact that there are polls. Look at how other pollsters are almost being ignored - you would be forgiven for not knowing there was a Kantar poll yesterday showing a 10% lead. Who cares when YouGov are available with their latest daily offering, with added excitement of Labour within touching distance of a majority?
Exactly, yougov must be loving all the publicity. The polls and pollsters are the news, not the politicians. Boring 'tories are still 15% ahead' stories day after day don't cut it in this media obsessed age.
They won't enjoy the publicity on June 9th if the Tories are 15% ahead....
I'm sure they'll conveniently have a couple of unpublished polls from the day of the election, state "late swing" and move on. Or a couple of experimental models that conveniently showed the outcome of the election that they'll then be able to correct for going forward.
and isn't 'pollsters being wrong' a story people expect? And if the result comes out in between YG and ICM then they'll all be wrong!
By being so prominently different YG is risking some damage to its brand, but I still think the potential upside is greater. Although my judgement remains that their model is flawed in principle, however good their work in practice.
Seems like not many people watched the debates even here on PB, and the general view in the papers seems to be that Jezza looked shagged out and the rest would not look out of place in a sixth form common room
Theresa May's judgment has seldom looked sounder than when she decided to stay the heck away from this awful, bent, babyish custard-pie fight.
Morning all. I'm beginning to wonder if the polls with massive Tory leads were ever that accurate.
True they have not had a good campaign and May has not shown any leadership qualities or answered detailed questions (she should have been at the debate last night) and Corbyn has had a spirited campaign but he does not have very convincing answers either and has no experience of government or negotiating with the EU. Yet the election is on a knife edge if you believe YouGov and media coverage.
On QT on Friday I think May should debate Corbyn directly.
YouGov is not doing these polls for free, is it? Surely it is doing what any business does - delivering on orders placed by customers. It's the media that commissions this stuff. And the Times has definitely got good value for its money.
On reflection - maybe that's wrong. As others have said, it's great brand building for YG and name recognition alone will get them plenty of commissions. If it is about brand building there would definitely be a temptation to develop ways of looking at things that create noise.
By being so prominently different YG is risking some damage to its brand, but I still think the potential upside is greater. Although my judgement remains that their model is flawed in principle, however good their work in practice.
I think its more fundamental than that. As someone said here yesterday, we are all used to lying online, friending people we don't like, liking posts we actually hate, why should opinion polls be any different. Its a way to send a free, no risk, message to the party of your choice, and lying by mouse-click is so much less socially awkward, and so easy, compared to looking an interviewer in the eye and telling them a porkie.
If you were pissed off blue-rinser, you might well tell a few opinion pollsters you were going to vote Labour/LD to give the Tories a kick up the arse, but on the day you are not going to be voting for Corbyn.
YouGov is not doing these polls for free, is it? Surely it is doing what any business does - delivering on orders placed by customers. It's the media that commissions this stuff. And the Times has definitely got good value for its money.
On reflection - maybe that's wrong. As others have said, it's great brand building for YG and name recognition alone will get them plenty of commissions. If it is about brand building there would definitely be a temptation to develop ways of looking at things that create noise.
Having a poll with the Tories three points ahead is one thing. Translating that into seats using some methodology which has been ridiculed by many on here is another.
@alex. Well you'd hope no newspaper would hire them in the future to do polls if they have messed this one up. Agree that they [polls] probably play too much of a role these days, despite how much we love them on here.
I doubt it. Newspapers love them because they are creating stories. Imagine how boring and mundane the election would look without them.
Yeah, that's a good point
If they are seriously out, I wonder if there will be talk about banning them during the campaign. Not sure I'd want that, but wonder if it will be discussed.
Of course, that is a big if.
It is the ubiquity of YouGov that is the issue, not necessarily the fact that there are polls. Look at how other pollsters are almost being ignored - you would be forgiven for not knowing there was a Kantar poll yesterday showing a 10% lead. Who cares when YouGov are available with their latest daily offering, with added excitement of Labour within touching distance of a majority?
There is money at stake for all the pollsters. YouGov is gambling that its attempt at a deep-analysis model will pay off. If they are judged "right", they will be able to dine out on it for years (and their shares would represent a great 'buy' right now!). If they fail, they'll suffer accordingly - but maybe the commercial upside of being uniquely right is the greater? Hence their gamble is rational from a business perspective?
They are going to have to get very lucky though. On their sample size they are going to be averaging something like 70 people per constituency, which gives a margin of error over over 10% making the results in danger of being worthless except for good fortune. Additionally a sample of 70 is hideously easy to game, with only a very small handful of activists with the right profile being required to massively swing the result in a small group of continuencies.
Except that - if they can make their national 49,000 panel representative - the Labour/Tory differential share is the one indicator that statistics suggest they should be able to get closer than anyone else. Take out all the noise from the minor party targets and other unusual seats, and it is the respective shares for the two big parties that decides the election.
I woke up early and posted a way-too-long post about YouGov down below. In summary I think that their panel is probably biased in ways they will struggle to compensate for, and their model whilst clever cannot hope with any constituency-level 'political history'.
You've also got to remember re: political polling that the newspapers who commission them don't have anything at stake if the polling is wrong. All they want is polling that is "interesting". Whereas this is very different to the pollsters core business which is, polling for businesses with the commissioners looking to take actual money determining decisions upon them.
Where they have probably done badly is in the area of private polling for financial organisations though...
Yet the election is on a knife edge if you believe YouGov and media coverage.
Neither of whom have the slightest interest in a boring election with a massive lead on one side. No one reads boring polls, and no one watches TV coverage of foregone conclusions... that fact that almost no one watched the debates should tell you all you need to know of what the country things, they think its all over.
Looking at the Yougov tables it is interesting to see how the 18-24 vote has been almost entirely hoovered up by Labour. Greens showing 0. Labour showing 69 weighted, 50 unweighted.
Also likelihood to vote amongst the 18-24 group is about the same as the 25-64 groups, though the 65+ old timers show slightly higher.
I have seen on here many people post that the young voters don't turn out on the day. However, (anecdotally) from my contacts in this age group there is huge voter engagement. Labour has three killer policies for all those in the age bracket which will break the traditional low turnout model:
For those not working: housing benefit restored for 18-21s, For those who are working: an immediate massive pay rise for those on National Minimum wage aged less than 25, especially under 21s, For students: no debt, plus grants
From a betting perspective, I'm nailing my colours to labour holds in marginals with lots of young voters, such as Nottingham South, plus one gain in Plymouth Sutton. There is still value out there, though not as good as 2 weeks ago.
Comments
I can honestly say I can't recall any examples from posters of the Right.
I'm not sure that SeanT has been noted for his gallantry. You get a few nasty comments on all parts of the spectrum in my experience.
I want them named and shamed.
If it weren't for Breggzit we'd now be entering the second year of a boringly competent posh boy regime and wondering exactly when #GO4PM would happen.
Perhaps it's the tendency on the left to prioritise group rights over individual rights? Anyone who then breaks out from their assigned category is therefore somehow immoral and fair game for any bigotry
But of course you're absolutely entitled to take the opposite view, Apoc, and I'm sorry that you have evidently been hassled over it (I've only been looking at the threads intermittently so hadn't seen it myself).
ICM - Lab marginal seats, majority of <15% (over ANY second-placed party; average Lab lead over Con in these seats was 10 points in 2015.)
19-21 May: Con lead of 3 (swing to Con of 6%)
24-26 May: Con lead of 9 (swing to Con of 9%)
26-29 May: Con lead of 5 (swing to Con of 7%)
I take Nick's point about ICM's demographic turnout filter, but that applies to their headline numbers too, so we can at least compare the above with the GB swing (4, 4, and 2 to Con respectively in the above polls).
I know Twitter can be a vile place but ignoring the low-follower anonynumpties is surely Twitter 101.
But I'l still buy you a drink if we turn up at the same PB event. Have you ever actually attended one, to meet these scary traitors in person?
Tweets that are racist probably come from anonynumpties with very few followers -- that doesn’t mean they shouldn’t be prosecuted, however.
Whatever, the comment about Amber Rudd is very unpleasant, and we all hoping for a kinder, fairer politics.
Obviously I'm no fan of May myself and I do think there are some real long term policy screw ups that screw my generation in particular. In particular the policy of building no houses where needed due nimbyism and the religion of house price rises as pensions is a big bug bear of mine and really needs a solution. I just don't think Corbyn is the answer.
http://brightcove04.brightcove.com/35/4221396001/201705/2076/4221396001_5454414414001_5454399287001.mp4
2m15s: "Will you resign if you lose seats?" She answers with a lot of spew, completely ignoring the question.
She needs to be asked again. And again.
Clive Lewis is 3 points ahead of his Tory challenger in Norwich South according to Hanretty's model, and has a 2/1 shot of losing it. YouGov? Lewis has a 32-point cushion!
Pb.com itself veered in some rather uncomfortable directions this evening.
Let’s all hope kindness prevails in what’s left of this election.
Night all.
Size of Con Maj:
NOM: 5.3
1 to 24: 9.4
25 to 49: 8.6
50 to 74: 7.8
75 to 99: 7.8
100 to 124: 6.8
125 to 150: 11
151 to 174: 12
The implication is that nobody really has a clue - for such massively different outcomes to be all considered roughly equally likely - ie up to the 100 to 124 band. And even above that is still considered perfectly plausible.
The podcast describes the model as a “massive, complicated and difficult model accounting for 100s of factors”.
It sounds extraordinary to introduce such a complicated model without extensive prior testing and calibration against previous elections.
One thing I should have also noted was that for the Tories to get over 400 seats would require them to win seats where they were below 30% in 2015.
Meanwhile, by Gordon Rayner of the Daily Telegraph:
"Theresa May to abandon strategy of attacking Jeremy Corbyn".
"Theresa May is to abandon her strategy of attacking Jeremy Corbyn as she urges voters to join her on a “great national mission” to deliver a successful Brexit."
(...)
"In an email to voters on Wednesday, Mrs May also said she was “excited about the future” - a marked departure from the “project fear” tactic that has characterised much of the campaign so far."
"It will be seen as an admission that the Tories’ largely negative campaign strategy, targeting Jeremy Corbyn’s “weak leadership”, has failed sufficiently to inspire voters."
It would be interesting to see the posterior probability distributions, not just the CLs.
My guess is that the posteriors are very flat, and so the model has little predictive power.
Suggests it would be a massive stretch for Con to get over 400 - especially given recent direction of polls.
Even if Con were to get 400, Scotland would only be making a very small contribution - they would have to do the job principally in England & Wales.
e.g. Angus SNP 54 Con 29 Lab 8.8 LD 2.7
^ Red kippers comin' home to lab, after briefly going blue.
The tories have misunderstood kippers. They're anti-status quo voters.
Ruth Davidson needs a seat,she's the tories only hope for the next election unless the new batch brings riches.
Con 43.50
Lab 35.75
LD 8.00
UKIP 4.25
Tory lead 7.75 (-0.69)
Yes, against Brexit and in the US.
I am one of the 49,000. On the upside, in the poll I get a screen with the candidates, named and with their party descriptions in ballot paper order for my constituency. That must add a level of accuracy, particularly for seats where known minor party candidates are in contention, that the "which party will you vote for" polls do not have.
The "100s of factors" will be socio-economic, based on their exceptionally detailed data on Yougov panellists. They know what I buy, what I watch and listen to, what I read, and what I think on a whole host of issues. Yougov specialises in cross-correlating all of this data: much of it is on their website and 'for fun', you can click on a celebrity and see, for (as a made up) example, "people who like JK Rowling also like Ed Miliband, Cornflakes, and Radio 4".
The potential downside, however, is this. Their model only has 75 people in my constituency, which they use as a 'base' to get some idea on trends, but it then finds people who are 'similar' to me (and all the other demographics they need to match the constituency population, data probably derived from the census) living in nearby constituencies. They then use the voting preferences and trends of such people to 'boost' the sample for my local area.
In areas where there is a swathe of seats with similar characteristics, for example the shires or Liverpool or other big cities, fine. People switching from Labour to Tory or vice versa in these seats probably do share similar characteristics. Where the model must be weak, however, is in picking up the effect of ultra-local candidates or the actual campaign in unusual seats. This must be particularly relevant for the minor parties - for example in Norman Lamb's case he only gets elected because the type of people who vote for him in his Norfolk seat won't be thinking of backing the LibDems in large numbers in the other Norfolk seats. If the habits of the rest of Norfolk were projected into his seat, he would be nowhere in contention.
I don't see how Yougov can resolve this issue without relying over-heavily on the responses of the local 75 people? For example they have the Independent leading in East Devon seat - that can't be because similar demographics in the rest of Devon are leaning Indy, and must be because a randomly high proportion of the East Devon 75 are saying they will vote for her. At the same time, the demographics of surrounding seats is clearly having an undue impact - for example they have Kensington as potentially marginal for the Tories, which can only be because they have 'imported' people from surrounding areas like Hammersmith who are not as Tory as residents of Kensington. Contrarywise, the demographics of Hastings are very different from the surrounding rural Kent seats (many more people on benefits, for example) and Yougov appears to be over-projecting non-Tory voters into that seat.
These two potential errors work in opposite directions - reliance on the 75 increases the randomness and the apparent chances of outsiders (East Devon) - reliance on voters from the surrounding area 'smooths' the politics of an area and reduces the chances of candidates known only in their seat. I really don't see any way that Yougov can resolve this except by trying to 'guess' a crude balance between two different sorts of potential error?
Finally, one element is missing altogether from their model. British GE's are fascinating because the political history and local campaigns have a big local impact. Parties can retain a few voters in a seat because of some freak by-election or issue years back into the past, or the reputation of one of the candidates. YouGov is probably excellent on the demographics but politics is more than that - and local political history will be missing from its model and cannot be derived accurately from a sample of just 75 people.
These are in addition to the well-discussed problems of YouGov's panel being both self-selected and of people who enjoy (or are willing, for 50p a time) to fill in lots of surveys. This is unlikely to reflect the population as a whole, however hard they try to weight the different demographics of their panel.
A case in point is the West Wales constituency of Preseli Pembrokeshire, just about as safe a Tory seat as one is likely to find in the entire principality, as reflected by the betting odds, where the most "generous" price available on the Tories is the 1/25 available from Betfair Sportsbook, while those nice folk at Bet365 will give you odds of 18/1 against Labour capturing this seat from the Blue Team.
The outcome therefore is absolutely cut and dried or at least so one would have thought ....... but wait ....... along comes Chris Hanretty's Election Forecast team who decides that this is the third most likely seat Labour is capable of capturing in the entire nation and equally the third most likely seat the Tories are likely to lose, in both instances making this a 76% likelihood, or a better than 3 in 4 chance of actually happening. Is there something unusual about this seat which I've missed or has it somehow slipped through the most elementary of filters presumably applied by Election Forecast to result in such a seemingly near impossible result.
I thought it was pushing things somewhat to imagine that either the Tories or Labour could conceivably win Brighton Pavilion, but this seems to take things onto an entirely new level.
And yet, only a couple of years later, they have been allowed to do it again. The other polls are barely allowed to get a word in edgeways because once again YouGov are driving the agenda with their highly regular polling creating a narrative of rapidly converging gaps and very close polls. Just think what this campaign would be looking like if we didn't have YouGov leading the headlines every couple of days. The closest poll would be about a 6-7 point Tory lead, and they would be looking like outliers with a mid point lead of about 9-10% and a likely healthy Tory majority.
If YouGov screw up again, after the influence they have once again had on the campaign, then they should be banned from political polling in future.
Interesting podcast - after listening to Prof Curtis’ illuminations on his recent YouGov poll, I've decided not to place any bets on a hung parliament.
If they are seriously out, I wonder if there will be talk about banning them during the campaign. Not sure I'd want that, but wonder if it will be discussed.
Of course, that is a big if.
If you want an 'it's close' poll, who else do you call?
Didn't catch the debate last night - was it the shouty mess it's being portrayed as in some of the press?
Debate was alright, although I missed the middle bit. Definitely shouty, and probably not at all useful in debating anything.
Yet it would seem that we are nowhere near this stage, with the major UK pollsters, possibly the most experienced in the world, fighting tooth and claw over who has the most reliable methodology in place with just one week left before polling day. This is amply demonstrated by arguably the two pre-eminent firms, YouGov and ICM who are currently around 8% apart as regards what they claim is the Tories' lead over Labour. As things currently stand, I can only imagine that within the losing team, heads will roll big time after the result is known. Of course should their predictions converge over the final few days, which I fully expect will be the case and they finish up within 3% - 4% or possibly closer still to correctly forecasting the outcome, then of course both firms will claim to have been right all long and the irritating thing for us PBers and other commentators in general is that none of us will never know who was telling the truth!
Have fun!
Who put something in BigJohnFreeOwls's coffee ? More ramping posts from him today that the rest of the last month!
In the end this is all manufactured, if you assume 16-24 year olds turn out at 110% (students of course somehow voting twice) and that 65+ all decide to go to the seaside for the day then Mr. Corbyn is prime minister and you have your headline.
http://www.dailymail.co.uk/debate/article-4560540/An-audience-balanced-gorilla-unicycle.html
Anyone who watched it or isn't the Daily Mail would accept it was very interesting and informative and that Mrs May made a serious error of judgement by not showing up. The broadcast media this morning are talking about nothing else.
One compensation for the passing of time is seeing the pernicious influence of our trashy tabloids slowly withering away.
By being so prominently different YG is risking some damage to its brand, but I still think the potential upside is greater. Although my judgement remains that their model is flawed in principle, however good their work in practice.
Labour majority nailed on then, Roger?
True they have not had a good campaign and May has not shown any leadership qualities or answered detailed questions (she should have been at the debate last night) and Corbyn has had a spirited campaign but he does not have very convincing answers either and has no experience of government or negotiating with the EU. Yet the election is on a knife edge if you believe YouGov and media coverage.
On QT on Friday I think May should debate Corbyn directly.
If you were pissed off blue-rinser, you might well tell a few opinion pollsters you were going to vote Labour/LD to give the Tories a kick up the arse, but on the day you are not going to be voting for Corbyn.
I woke up early and posted a way-too-long post about YouGov down below. In summary I think that their panel is probably biased in ways they will struggle to compensate for, and their model whilst clever cannot hope with any constituency-level 'political history'.
Where they have probably done badly is in the area of private polling for financial organisations though...
Also likelihood to vote amongst the 18-24 group is about the same as the 25-64 groups, though the 65+ old timers show slightly higher.
I have seen on here many people post that the young voters don't turn out on the day. However, (anecdotally) from my contacts in this age group there is huge voter engagement.
Labour has three killer policies for all those in the age bracket which will break the traditional low turnout model:
For those not working: housing benefit restored for 18-21s,
For those who are working: an immediate massive pay rise for those on National Minimum wage aged less than 25, especially under 21s,
For students: no debt, plus grants
From a betting perspective, I'm nailing my colours to labour holds in marginals with lots of young voters, such as Nottingham South, plus one gain in Plymouth Sutton. There is still value out there, though not as good as 2 weeks ago.