For poll watchers there’s a ritual every night at 10.30pm. We go into Twitter an wait for the Tweet from the Sun politics team giving the headline numbers from the latest YouGov daily poll findings. Fieldwork for the surveys starts early evening the previous day and continues to the following afternoon.
Comments
The point about how YG has dominated the polling agenda is so good. If we took YG out would we view the election differently? Would we definitely say its neck & neck? Not sure tbh.
Here's an interesting fact, if you aggregate the all the polling for the last 7 days then remove the YouGovs from the dataset, the tory vote share goes from 33.55% to 33.55%. Absolutely zero change.
The labour vote share goes from 33.57% to 32.91%
So, I guess you can ask if YouGov are very slightly over reporting labour, but given that their polling is pretty much exactly in line with the non YouGov polling, I really don't get the YouGov are busted meme. There is no basis for it in the numbers.
We're at the cool heads and look at the numbers phase, emotion isn't the punters friend right now
One has to feel a little sympathy for yougov and their sponsors. They devised a system that would cut out the noise and simply report on the trend. They put in place frequency to ensure that yougov was the one who would show trending as it happened. THE place to go to spot the trend.
And then there was no trend to see.
We are also told there are as many as 10 million undecideds. That looks quite plausible, from my experience of door-knocking.
And whilst he has narrowed the gap some, it is clear that Ed Miliband is a considerable way behind David Cameron as the preferred Prime Minister. A common thread amongst the undecideds I have spoken to is their dislike of Ed.
So come Thursday, the message the voters will get from the polls is that there is still a very real risk of Prime Minister Ed Miliband - supported by the SNP....unless you vote to prevent it. The normal rule of thumb is that the undecideds break 2 to 1 for the incumbent. I suspect that could be a little higher. Maybe 70:30. In which case incumbent Tories (and LibDems) could get a very late-breaking net boost of several million votes over Labour. And the polls will not have picked it up.
Somebody's right, and somebody's wrong. And I don't mean the fickle electorate.
As against this phone pollsters reach the parts internet pollsters cannot reach but this is tempered by the possibility that people don't want to show their darker side.
Do UKIP fare badly in phone polls?
There's so much phone selling and scams these days that I imagine only saints don't hang up on them.
Do the Greens do better in phone polls?
(I wonder what the phone pollsters opening line is? It'd have to be good. )
Marquee Marks posted a really good comment. Ive met lots of people still undecideds quite a lot of them saying they'll probably vote Cameron so a reluctant tory vote. Lots of them mention not liking Miliband & not wanting the SNP.
Regardless of the genuine strong support for another Referendum (often apparently lost completely to media commentators who seem to prefer "the opinion of potentially biased audiences" to actual polling showing 48% support for another within 5 years and 60-odd percent within 10 years) the main benefit isn't this. The main benefit is that it offers Sturgeon and the SNP a "free ride" in the coverage.
While handling them well, there were genuinely difficult questions for Nicola in the Question Time. But coverage on BBC bulletins has only mentioned the Referendum question. Nothing on an NHS employee in the audience who was fired (albeit a management post) or any of other potential banana skins thrown at the FM.
It seems bizarre but the overwhelming bias of BBC Scotland in their attempts to sabotage the SNP campaign they have ended up helping the SNP by completely misunderstanding the relevant issues. They've created a narrative to discredit the SNP which, in the public's mind, isn't at all discrediting to start with.
The fieldwork for 6/7 April is when YG implemented their methodological change. That's when the regular Tory leads ceased and the regular Labour ones started.
Weighting the sample back to Jan/Feb respondents and assuming a one point Lab lead may well measure change, but possibly from an inaccurate starting point. The phone polls had a 0.71% Tory lead at that time.
A one point Tory lead now is really a +2 for the Tories from whatever the true base position was in February.
YouGov's methology changed at data point number 77 and took 5 days to fully impact upon the moving average.
"Even if the share it the land were 34/34 with daily polling we should get "rogues" throwing up 5% leads."
I don't see the logic in that. If the data on your panel is accurate enough then I would think the chances of a rogue poll would be severely reduced.
"We are also told there are as many as 10 million undecideds. That looks quite plausible, from my experience of door-knocking."
Wouldn't you interpret someone telling you they were undecided as suggesting they didn't want to hurt your feelings?
That's true, but it can also drive a nervous desire to cluster round the mean. If noone's really sure what the election result is going to be, and they don't have full confidence in their methodology, then they might not want to be the one pollster who's miles out.
Better for everyone to risk getting it similarly wrong than take the business risk of being the lone wolf.
A major consequence for me of the dead hand of Yougov is that it has been impossible for the Tories to build any kind of narrative of momentum in their favour. So very good polls for them like the ICM before last and Ipsos MORI yesterday are swiftly trumped in the way Mike describes indicating that there is no change.
This does not mean that Yougov is wrong. There was another excellent thread a few days ago which showed if you compared like with like none of the pollsters were picking up any movement outside the MoE. The illusion of movement came when the results of one polling company were compared with others.
I think it is fair to conclude that Yougov are right in this in that, like most campaigns, there is very little evidence that this one is changing anything (outside Scotland). There has also been no evidence of swing back since the start of the year which is why the initially very favourable Fisher type models have progressively become less so for the Tories.
So the question that arises is which polling company is right? Is it ICM, whose methodology indicates that the Tories have a modest lead of 3-4% or Yougov who indicate it is a dead heat? The short answer is that we will not know until after 10 on Thursday when the exit poll comes out. If it is ICM the Tories will have a modest plurality although it is beyond me how they could form a government. If it is Yougov then Labour will have the most seats and a Labour/SNP co-operation will have a majority.
For any poll with weighting or sampling the normal rules of random sampling (chi squared etc) are actually an over-simplification. If the thing you weight to exactly corresponds to the thing you're trying to find out, sampling error should be zero.
For example, say you're taking a poll of Argentians and Brits on who owns the Falklands. Say for the sake of argument that the Argentinians unanimously say Argentina, and the Brits unanimously say Britain. Say you call Brits and Argentinians in exact proportion to their populations. Since nobody ever changes their mind, and you have exact correspndance to your weighting variable (nationality) you will get exactly the same result every time.
What YouGov are doing is weighting to declared voting intention in Jan/Feb (which was in turn weighted/sampled to be representative of the population). This varies randomly a little bit each time because even without the overall vote shares changing, some Jan/Feb Labs will go Con and vice versa, so the weighting variable (Jan/Feb VI) isn't an exact proxy for the thing you're measuring (May VI). But it's pretty close, so it doesn't change much from day to day unless those people's voting intention changes.
There are two things which are striking about this elections polling and both are to do with the online/phone divide. Firstly, the phone polls are producing consistently better scores for Con and the online ones favouring Lab; secondly, the online polls are incredibly static while the phone polls have shown quite a bit of movement. Are the two related through some internal methodology? I don't know but I'm suspicious of any results that come back with metronomic regularity in a field that should be inherently changing and uncertain.
But the point is an important one. Does the fact that YouGov and Populus have reported suspiciously consistent results mean that we should also be sceptical about their small Lab leads? Or are the two features independent? Or are the two right? Alternatively, do the swings reported by ICM, Mori and Ashcroft give them more credibility or less, and if more, does that mean we can put more store in their Con leads or is that too a separate matter?
Ultimately it probably comes down to gut instinct. We can't know until it's too late and there are arguments on both sides though my own take is that the phone companies are producing more believable results: I just don't think opinion is that static throughout an election campaign.
http://fivethirtyeight.com/features/the-polls-might-be-skewed-against-democrats-or-republicans/
In brief - don't count on it, and if they are, they may not be biased in the direction you'd prefer...
http://blogs.spectator.co.uk/steerpike/2015/04/question-time-will-ed-miliband-take-his-lectern-with-him/
Has Call me Dave really done that many more appearances than Miliband in the campaign?
B & H Independent @BrightonIndy 11h11 hours ago
Supposing Green activists had burned @UKIP adverts: http://bit.ly/1EUvtLe ? Non-story? Or central to free democracy?
Lord Ashcroft@LordAshcroft·16s16 seconds ago
Well @Nigel_Farage just said on Sky that my South Thanet poll involved "some voodoo"! That's a first. Thanks Nigel.....
If you bear in mind that the VI of the sample was (as expected for a TV audience) more favourable to Labour (a 5% lead over the Tories) then Cameron did even better than the headline 6% lead over Miliband (and a 14% lead on the forced question of who would be the better PM).
See tables 3 & 9: http://www.icmunlimited.com/data/media/pdf/Flash-Poll-30th-April.pdf
http://www.theguardian.com/politics/2015/apr/30/unite-leader-len-mccluskey-backs-dismissed-mayor-lutfur-rahman
http://blogs.spectator.co.uk/coffeehouse/2015/04/the-truth-about-labour-and-overspending/
My point was that a Second Referendum is not a handicap for the SNP. Leading with that argument their opponents cannot win, firstly because they're doing their campaign so badly that the SNP is just batting them off but more importantly, even if they did trap the FM or another SNP senior into saying "it might happen in 2016 Manifesto", there is too much actual, real support for this in Scotland for it to hurt the SNP.
The attacks made on them are "no win" for Labour/Tories/Libs.
And while no-one individual knows enough people for their numbers be important, I know a lot of my “mature” friends take the same attitude.
Populus today should tell us whether there's been real momentum to the Tories - if they show their first Tory lead of the year then it's safe to say there has been.
I've seen studies saying the opposite:
http://www.pollingreport.com/incumbent.htm
I dare say it varies from country to country and era to era and upon the 'national mood' at the time.
Who's been full of hot air? Who's been massaging their internals?
Time to vote off ......the weakest link.......
https://yougov.co.uk/news/2015/04/19/tories-are-losing-both-air-war-and-ground-war/
1.Do phone polls also contact mobile phone users?Younger and unattached people are less likely to have landlines.
2.ICM and Ashcroft reallocate don`t knows to party voted in 2010-50% in ICM and 100% in Ashcroft.With Ashcroft`s own polls and piece in the Independent suggesting don`t knows were swinging more to Labour,the wisdom of this approach is debatable.
3.MORI was an outlier for various reasons as pointed out before.
4.Despite the above,large Tory leads in phone polls suggests a smaller national Tory lead nationwide is probably correct.
I would not be surprised if the SNP were aware of this and are using it themselves. They could say it would never be ruled out, not suffer in the polls and close the issue down. By refusing to make such a statement, they allow the issue to stay front and centre, keep all criticism of them as a party and government focused on this and avoid any difficult coverage for their campaign.
I think the phone polls reach a different part of the public, let's call it the general population who aren't bothered about politics normally but will interest themselves every election. This as appears to move towards the Tories.
Now if you are on here you are probably in the first section. I know that most of the time my anorak status on politics is not the norm and even though yougov is probably balanced, it may have a representation of groups, but inherently with more politically interested and aware types.
I know it has been successful in the euros and London materials but I think on that the more political part of the population vote in the euros, and London is different to the rest of the country with regards to political climate attracting the more engaged.
I just cannot believe that the series of yougov results is so stable. You would expect one in twenty outliers so we should be surprised fairly frequently by yougov and we aren't. The only mitigation I can think of is that, their change of methodology could have had the wrong effect.
It's impossible to comment on YouGov's sampling technique without knowing how big its panel is and what the participation rate is. As others have noticed, the lack of range in its results is suspicious.
If he had done so earlier,could have slowed the mass hysteria north of the border.
http://ukpollingreport.co.uk/blog/archives/9281
If UKIP draw more support from Conservatives than Labour, that could skew their numbers.
---
ICM and Ipsos have the lowest UKIP vote shares.
http://ukpollingreport.co.uk/blog/archives/9350
What surprised me was the long memory of the QT audience last night, and the evident anger of people. They remembered Labour's spending and borrowing (even if the message has been reinforced by the Tories) and it will take until the next election for many to forgive and forget.
This has been highlighted in the polling by the way ("who is to blame for the cuts"), and one reason why a Labour victory has been impossible this year.
. I fully expect the Tories to win the national vote share. The national polls point to Ed (In aggregate) being ahead on seats or it being too close to call. The likely distribution from the models (UKElect, FIsher, Hanretty) points to a Tory lead on votes and seats. BUT Here is the thing... Con + LD + DUP must equal 323. In order for Ed to NOT become PM he musty abstain the Conservative Queen speech. SNP, PC, SDLP, Green voting it down is even more sure than the Labour party. No doubt if it really is Labour 260, Tories 285 he will give it serious consideration. But then he will put forward a Labour Queen speech and DARE the SNP to vote it down. There will be some Labour rebels who will be expelled, Danczuk, Mann, perhaps one or two others.
The SNP will vote it through.. And thus as Rod Crosby has pointed towards in his constitutional musings Ed is PM. I expect him to remain in post till Holyrood 2016. Past that the SNP may well pull the plug.
My book is well enough balanced that the final betting consequences of either Ed or Dave (Or other) should be fine for me. But this is my prediction.
Not sure why you're ignoring the point at hand. Labour and the other opposition are highlighting the wrong issue.
Really quite scary Miliband denying Labour overspending. We can't let these clowns near the levers of power again! I don't really buy the current narrative from the tories about how they have got everything going great again, but they are at least not living in complete la-la land like Miliband seems to be.
What a fool. People aren't going to let him be in charge ...are they? Please...?
Danczuk perhaps but I doubt it.
What was striking in 2008-2009 was that no Labour MPs defected even though some of them looked, and as it turned out were, certain to be defeated.
But what it might have done was reassured a few English voters. Labour's mistake was not to have given up on Scotland a while ago. It should have focused entirely on England and Wales.
Only 966 comments on the last thread - PB lightweights !! ....
"It seems bizarre but the overwhelming bias of BBC Scotland in their attempts to sabotage the SNP campaign they have ended up helping the SNP by completely misunderstanding the relevant issues. They've created a narrative to discredit the SNP which, in the public's mind, isn't at all discrediting to start with."
"Danczuk perhaps but I doubt it."
The only comfort in the Tories winning the election would be that Danczuk won't be sitting on the government benches.
"Oh, you're not voting for us? There must be something wrong with you."
Whilst the YG polls show a lot of steadiness, there is a lot of unpredictable daily churn inside them, especially since the change in methodology. This is most noticeable in the 2010 LD VI (the Cons and LAB ones are more stable) which veers widely from large losses to LAB to large retention by LD and smaller losses to LAB.
Thus are the 2010LD voters still in a state of flux as suggested by YG or are they more stable?
Then Ed has to try to form a government.... its as simple as that. Ed can't turn it down.
Thats when it gets difficult for him....for all his bleating, the SNP will hold great influence over him, and can time their effective veto at any and all times.
Given that any minority government will be so weak, becoming the government in that situation is truely a posioned chalice.
I do not like Dave much, and find much of the Conservative agenda scary. I would not want to see a Conservative government with a working majority.
I would quite like to see a continuity coalition, but the LDs look to be losing too many seats for that to be tenable.
It is all going to be quite a mess in a weeks time.
On the thread- I don't think YG has changed the narrative. Disregard Yougov and Populus if you wish as many do.
I genuinely thought that YouGov would show a 2% Tory lead- one that I expect on the 7th May. A 2% Tory lead would ensure that all the pollsters are within the MOE, just about.