Options
politicalbetting.com » Blog Archive » There’s no clear picture in England where 532 of the 650 We

At the end of last year I started monitoring England only data from the main pollsters partly in response to the very different political environment that’s developed in Scotland.
0
This discussion has been closed.
Comments
We certainly live in interesting times - with so many variables in play, I don't envy the punter.
It's been said by yourself time and again the main reason the Tory votes turn into seats less efficiently than Labour's is the spread of the votes. The large majorities in many seats with higher turnout. That could change this time. While the SNP could take upto 40 seats off Labour, UKIP could struggle to win 5.
If the UKIP rise is universally spread it's irrelevant. If it's much higher (but not enough) in safe seats while being squeezed in marginals then that could change the bias in the system.
Unless you no longer believe that the spread of votes matters for looking into the bias or you seriously believe that the UKIP rise is a Universally equal one, the 11.4% mattered for last time not necessarily this time.
Kellner's argument for improved Tory efficiency is the incumbency bonus of a lot of the first term Tory MPs and the SNP in Scotland. The British Election Study also suggest Labour will do disproportionately better in Lab/Lib marginals and seats where Labour are third behind the Lib Dems.
More than anything we need regional numbers for the midlands, that's where the election will be decided.
I agree with your argument. I think that UKIP will win very few seats on their voteshare, certainly in single figures and possibly only Clacton. What they will achieve will be a lot of second places, particularly in coastal seats and in the Tory shires. Whether they will build on this for the subsequent election or whether they will implode due to infighting is still uncertain.
Looking at my County of Leics and Rutland: Labour hold the 3 Leicester seats, the Tories hold all 7 seats in Leics and Rutland. The LDs were in second place in Harborough and in Leics South. I expect none of these 10 seats to change hands, the only possibility is Loughborough and I am fairly sure Nicky Morgan will hold on. I am anticipating a Leics result rather like 2010 Scotland. UKIP will come second in at least one city seat (Leicester West) and is likely to come second in many of the shire seats (Melton and Rutland, Charnwood, Leics South, NW Leics).
Is Leics representative? I think so. It is both literally and metaphorically middle England, with age spectrum and income levels very typical of England as a whole. When I look at the wider East Midlands I see a fairly similar picture, with a few Labour gains but not enough to make up for losses in Scotland.
I am not a poll watching obsessive, and regard polls rather like long term weather forecasts, simply indicative rather than diagnostic. Since 1992 there has been a lot of weighting based on past voting, and I am not convinced that this is accurate in the new environmend of LD collapse and UKIP rise.
Weighting is a method for adjusting data to attempt to compensate for having an unrepresentative sample. I think pollsters would be better off concentrating on improving sampling techniqes rather than using secondary voodoo adjustments to attempt to fix voodoo polls. My experience of this potential for bias comes from my experience in epidemiology and surveys in public health medicine.
The country still seems to be substantially uninterested in the election and I very rarely hear it discussed socially except when I raise it myself. When I do raise it, I hear a palpable lack of enthusiasm for the various alternatives and a weariness for the continuing austerity. This impression of a lot of undecideds matches the polls, where the don't knows make up a fair hefty bloc particularly in swing groups of voters such as women.
I forecast a low turnout, some Labour gains in England, but not enough to form a stable government (though the value bet is on Lab majority).
http://www.dailymail.co.uk/news/article-2967223/Emma-Thompson-won-t-pay-penny-tax-evil-b-s-prison-says-husband-Greg-Wise-HSBC-row.html
Though any polling company trying to introduce their new, improved (more expensive), yet unproven sampling technique probably struggles to sell it when the current system is not yet proven to be broken (though it might be after the election)
I must say I hadn't treally thought about it but if they're typical the Green vote's screwed..
https://www.youtube.com/watch?v=gvN6OwqFMuU
(and what a great theme tune)
https://www.dropbox.com/s/83sxf8225s7h3mm/2015 General Election Possibilities Latest.xlsx?dl=0
https://www.dropbox.com/s/4djbtovu64zto8g/Gen Election Calculator Notes.docx?dl=0
It suggests that the decline of the LibDems mean Con requires a much lower lead over Lab than 11.4% in England to win an overall majority.
Even if UKIP draws votes from Con (say 18% of 2010 Con voters) and Lab (say 6% of 2010 Lab voters) uniformly across Eng and Wales, then a Con lead of around 8% in England would give them 326 seats.
Should (as I expect) the switchers from both Con and Lab to UKIP be lower in Con/Lab marginals, then even an England lead of 7% would be enough for an overall majority; while assuming Lab lose 20 seats to SNP. then a bare 2.5% leave in England (0.1% over all of the UK) would leave Con with most seats.
As I said on my previous post, I may have got something horribly wrong, but I would appreciate it if someone could have a play with my model and let me know.
I shall be away now for a couple of days so won't be able to see any replies until then. But it would be great if someone could.
Thanks Mike.
That's a formidable piece of work. Will get back to you when I have had more time to study it.
"I shall be away now for a couple of days so won't be able to see any replies until then. But it would be great if someone could."
Crosby!!
"Looking at my County of Leics and Rutland: ....."
............................................................................
The Grand Duchy of Rutland is not an appendage of "Leics"
Take one hundred lines :
Ed Miliband Will Never Be Prime Minister.
So along with transport, housing, adult education the NHS and social services have joined numerous areas being devolved to Manchester.
I'd suggest this adds a significant layer of complexity to the English votes for English MPs given Manchester is to have a similar amount of devolution compared to Wales, more so than London.
There are over 20 MPs in Greater Manchester, how would they be treated in the future?
http://www.bmj.com/content/338/bmj.b2393
Though this too is susceptible to bias depending on both random and non-random sampling errors, as well as deciding which variables to incorporate in the analysis.
Political polls also have the problem of self selection. We can opt out of voting, but not out of heart attacks!
Somethings can be forecast accurately such as the time that the sun will come up on May 7, and we can reasonably expect May 7th to be warmer than today, but the range of possibilities is quite wide. Pollsters rarely give detailed statistical confidence intervals, but bias is a much bigger issue than statistical confidence.
In the absence of statistical certainty we should assume that there is everything to play for, and play to the whistle.
Would one of the weekend threads be a good time to go over it?
Rutland is undeniably a distinct entity, but for electoral purposes is part of a contituency that incorporates a lot of East Leics so is only half the population and area of the Parliamentary seat. Something that Mr Duncan often seems not to notice!
I am sure that things were much better arranged before the scandalous abandonment of tradition that was the 1832 Reform Act.
From punters perspective I'm worried that 2015 demographics are unlike anything previous. I don't trust online polling & have them in my div 1 category as opposed to PL. I'm also now jittery about phone polling. For instance I never answer my landline. Loads peeps I know don't have one or do like me 'cos always cold callers. So how do you do demographic sampling on mobiles? You can't.
Anyhows can someone do me favour & list which pollsters do online & which do phoning?
We can make a working assumption that Labour won't perform better in England than in 2005, when the parties were level-pegging. Labour won't be winning back seats like Rochester & Strood or St. Alban's.
"For instance I never answer my landline. Loads peeps I know don't have one or do like me 'cos always cold callers. So how do you do demographic sampling on mobiles? You can't."
Someone posted on here that in a room of 23 the chances of two having the same birthday are 50/50. It sounded so improbable I looked it up on WIKI and discovered the even more improbable fact that with 70 people the chances are 99.9%. I then read the way the maths worked and couldn't figure it.
The people who make a living out of polling have obviously thought through these imponderables-their careers depend on it- so all the posters pointing out why various pollsters have missed or forgtten something are likely to be wide of the mark.
Also this:
http://www.telegraph.co.uk/finance/comment/jeremy-warner/11432919/Dramatic-political-change-is-in-prospect-and-markets-dont-yet-know-it.html
If it starts to look like Labour will scrape in (or better) then sometime in April the markets will start to react to that. It's not clear to me that this is a self-correcting thing. If there's a market crash of some sort as the prospect of Ed being charge causes a serious wobbly it's not for sure that the the electorate will pin the blame on Ed! Market wobblies tend to be blamed on the current government. In fact an Ed driven market meltdown may be vicious circle of nastiness from Dave's p.o.v.
Red Liberals who are minded to vote Labour in disgust at their former party whether Labour are in contention or not. Could easily be worth 2%.
Labour supporters who are no longer willing to vote tactically for the Lib Dems on the basis that they are the same as the Tories. Difficult to quantify but one of the reasons that the Lib Dems are polling poorly on constituency specific questions (if not as poorly as the national question).
UKIP reducing pointless majorities in very safe seats. This should, all things being equal, reduce the Tory share but if they replicate their 2010 share then they are getting compensating votes where it will do more good.
A large part of Labour's efficiency at the last election came from Scotland. This time it looks likely to be a drag on their efficiency in that their share of the seats will be less than their share of the vote.
The 2010 Labour vote was depressed by Brown and his staggering incompetence. This resulted in very low turnout in comparatively safe Labour seats. It would be possible for the Labour vote to increase by hundreds of thousands without gaining a single seat as a result. In fairness Ed is trying to offset this problem by being crap.
There are counterarguments. If Labour regain a seat like Broxtowe because of a differential gain from the Lib Dem collapse this has the possibility of improving their efficiency. It does look as if the collapse of the Lib Dems will gift Labour some seats. I just do not think this will offset the issues identified above. The Labour vote will be more efficient. The differential turnout between safe Labour and safe Tory seats makes this inevitable. But will it be as efficient as it was in 2010? I don't think so.
I am not technically proficient to comment on the mechanics of Mike Greene's model but if he is suggesting that a lead of 8% would replicate what a lead of 11.4% achieved the last time that would not surprise me at all.
" It's not clear to me that this is a self-correcting thing. If there's a market crash of some sort as the prospect of Ed being charge causes a serious wobbly it's not for sure that the the electorate will pin the blame on Ed! Market wobblies"
The most prolonged bull markets are always under a Labour government. Having said that my stock broker called to suggest I sell United Utilities on the basis Labour look likely winners
"The chance of 2 people in a group having the same birthday is obviously much higher than 2 people in the same group having the same day (and year) of birth"
Yes but there are still 366 possible birthdays
If (as is not impossible) the Conservatives were to win 6 out of 6 Cornish seats on c35% of the vote, that would be thanks to UKIP.
I'm on Nick @ 1-2, perhaps I should have taken some Anna 7-2 when it was available but I don't really fancy topping up on Nick at 2-5 or taking Anna 2-1.
I'd give Nick a 75% chance of recapture or so.
1) the rise of the SNP
2) the collapse of the LD vote
3) the rise of UKIP in England (though Wales too is interesting kipper territory)
4) individual voter registration and its impact on youth and BME voters
5) the lack of enthusiasm for any party leader or programme
These compound each other to make this the most interesting and uncertain election in a long time. As such the predictibility is low, and likelihood of surprise results high. The value bets are not the favourites as much as usual. Very little is "nailed on".
Personally, I'd prefer England & Wales to be polled entirely separately from England. A really good sub-sample for the Conservatives in Scotland can make Labour's position in England look better than it is.
The established pollsters that have been tested at a General Election before all have the Tories ahead in England, only the two new pollsters that haven't been tested before have Labour ahead in England.
As an example, would you expect the UKIP vote to be larger in leafy Oxfordshire (Witney) or in urban South Yorkshire (Doncaster North)?
It looks likely that the two main parties will have a combined seat total little unchanged from 2010, but on a lower combined share of the vote - thus they could both see an improvement in vote efficiency.
1. UKIP is picking up a fairly even spread of protest/NOTA votes everywhere. In a few seats which we know are Kipper targets they are doing better, and they are doing less well in London and Scotland, but it's unsafe to assume they have the sort of pattern that we expect from the LibDems (strong in citadels, crushed elsewhere).
2. The Tory first-term incumbency boost seems not to be happening consistently. Anecdotally there seems to be an anti-incumbency vote in some places. More solidly, every constituency poll shows the marginals doing much the same as everyone else.
3. The comments that people aren't switched onto the election are I suspect untypical of marginals. I'm literally unable to go shopping (without a rosette and just minding my own business) without people stopping me to ask anxiously about the election; people stop their cars to call out; I've rarely encountered such intense feeling on the doorstep (certainly not in 2005 or 2010), and "not sure" voters are largely apathetic-looking people who I suspect mostly won't vote. That doesn't mean voters are necessarily enthusiastic - far more are saying "We must stop party/candidate X at all costs" rather than "I really like what party Y is saying". But they're really engaged and dug in - I can't shift many Tories, and I doubt if Anna is shifting many Labour voters.
FWIW, my guess is that the average of the current polls - which have been broadly stesdy for 4 months - is roughly what we'll get in English marginals, i.e. a swing of 4% or so in England. We are running out of time for black swans.
The big variables for the Tories are the extent to which LAB inclined voters switch to LD in yellow defences to keep the blues out and what the 20% of 2010 LDs yet to make up their minds do.
The Ashcroft marginals polling of CON-LD battlegrounds has found an increasing propensity of LAB voters to switch.
There are two non Westminster elections which demonstrate the readiness of LAB voters to switch to the LDs - the successful Mayoral defences in Bedford and Watford. Unlike the Tories the yellows have successfully hung on to both the elected mayors they had 5 years ago.
These are highly relevant because the LD MP defences will be run like Mayoral campaigns with all the focus on the individual incumbent.
I'll post some of the detailed data
Edited 8.29
As well as the headline figure being hard to discern, the Greens and UKIP will also lead to some perhaps counter-intuitive results
http://www.dailymail.co.uk/news/article-2967589/An-unspeakable-obscenity-Minister-s-rant-Ed-riddle-scene-cut-BBC-Inside-Commons-documentary.html
Ed's been off the news - apart from suggesting no second jobs for MPs and that resonates.
Straw comes across as a greedy tosser but Rifkind comes across as an arrogant, greedy tosser and that's a lot worse. If he'd gone quietly, it would have helped but being arrogant ...
The Greens and Ukip have had bad publicity - Nat being gormless and the Ukip woman saying gormless things on camera.
I'm surprised the corpse of the LDs hasn't stirred.
,
"There's Green Party 'mind blank' day to factor in as well. I think all the publicity might actually push them up a point or two in the polls later this week."
I refer you to my and OKC's earlier posts. Anecdodes notwithstanding I think they're in trouble
London and the North (especially NW) above average, West Mids about average, other regions below par.
"Anna Soubry isn't my favourite politician, but I like her style here."
You might even agree with her comment but threatening to sue???
But by how much and where will the votes be is the crucial question.
What's happened to the old stiff upper lip. Public school BOYS doing kiss and tells now. Hewitt's got a lot to answer for.
"Last night, watched the BBC news at ten. The reporter helpfully told anyone considering going to ISIS that there was a bus straight to the border, the fare was eminently affordable, and passport checks were somewhere between minimal and non-existent. It was practically a 'how to' guide."
It was a toss up between there and Cannes but I didn't fancy the 18 hour bus ride
I suspect Mike is right for Watford and Bedford, perhaps for the wider SE and London, but I would not be so sure elsewhere.
More seats is probable, but Labour has a crucial efficiency advantage in Wales and for all their woes will have some more Scottish seats than the Cons too.
I was rather hoping somebody might have had time to look at the model I posted a couple of days ago which I have put together.
https://www.dropbox.com/s/83sxf8225s7h3mm/2015 General Election Possibilities Latest.xlsx?dl=0
https://www.dropbox.com/s/4djbtovu64zto8g/Gen Election Calculator Notes.docx?dl=0
It suggests that the decline of the LibDems mean Con requires a much lower lead over Lab than 11.4% in England to win an overall majority.
Even if UKIP draws votes from Con (say 18% of 2010 Con voters) and Lab (say 6% of 2010 Lab voters) uniformly across Eng and Wales, then a Con lead of around 8% in England would give them 326 seats.
Should (as I expect) the switchers from both Con and Lab to UKIP be lower in Con/Lab marginals, then even an England lead of 7% would be enough for an overall majority; while assuming Lab lose 20 seats to SNP. then a bare 2.5% leave in England (0.1% over all of the UK) would leave Con with most seats.
As I said on my previous post, I may have got something horribly wrong, but I would appreciate it if someone could have a play with my model and let me know.
I shall be away now for a couple of days so won't be able to see any replies until then. But it would be great if someone could.
I prefer the latest Professor John Curtice model which I have featured here before. To win majority, even with current Scottish situation, LAB needs 5% lead. The Tories need 7-10% dependent on national voting performance of LDs.
The big variables for the Tories are the extent to which LAB inclined voters switch to LD in yellow defences to keep the blues out and what the 20% of 2010 LDs yet to make up their minds do.
The Ashcroft marginals polling of CON-LD battlegrounds has found an increasing propensity of LAB voters to switch.
There are two non Westminster elections which demonstrate the readiness of LAB voters to switch to the LDs - the successful Mayoral defences in Bedford and Watford. Unlike the Tories the yellows have successfully hung on to both the elected mayors they had 5 years ago.
These are highly relevant because the LD MP defences will be run like Mayoral campaigns with all the focus on the individual incumbent.
I'll post some of the detailed data
Edited 8.29
WiFi on the train. Whatever next?
Mike, of course I am not comparing my skills to those of Prof Curtice. But my model does allow you to put in your own assumptions as to how vote changes since 2010 vary in different types of seat, including Lab/Con marginals and LibDem held. It then translates that to national vote shares and seats. It is also transparent so I hope it might be of some use.
Mr. Lilburne, I favour not broadcasting to millions a how-to guide on the ease of reaching ISIS. I also favour not broadcasting the recipe for Molotov Cocktails.
That is worrying, and if it's true it means that over 200,000 British Muslims know someone with that view. And "a fifth of those polled said they thought Western liberal society could never be compatible with Islam." - that's around half a million.
I think whichever party finishes ahead, the difference in overall vote share will be relatively small.
Mr. CD13, I saw that as well.
"But 27% of the 1,000 Muslims polled by ComRes said they had some sympathy for the motives behind the Paris attacks."
http://www.bbc.co.uk/news/uk-31293196
No model can ever work perfectly and to discuss why, for example, Mike S prefers Prof Curtice's to mine would be very helpful. I am sure I would learn a lot and be able to improve my model (and my betting returns!)
Relevant to vote efficiency debate I suppose, and also indicates they should win those seats if the national polling rises again
However, the aggregate of the English sub samples will give you a reasonable sample particularly over a week, say, due to its size.
Purists will not agree though.
Exactly!
Also, as I've perhaps mentioned, I'm not overflowing with wealth to lump on a huge sum.