|
|
Post by marksenior on Dec 17, 2013 15:31:02 GMT
I think Mark is probably right to be sceptical of these constituency polls as a means of predicting the outcome of the next election in those constituencies. In any case, constituency polls do not generally have a great record of accuracy. However, from the UKIP point of view, they do show the direction of travel and political trends, even if they show an inaccurate starting point. No contest Trident. I agree all you say. What got to me was the "No you are totally wrong". And the absolutist assertions without evidence, by Mark. Most of us on this site know our way about polls and polling and theory of weighting. It is imprecise stuff but useful for trends of direction. These constituency based polls provide a focus and however crude the result I would wager they are not too far out as to the relative positioning and a possible if not probable result if there had been a poll on that day. My original short overview was not in its totality 'totally wrong' and I have a right to resent that. Mark is too certain and too dogmatic. I am not a tribal Kipper puffing our chances and denigrating the LibDems. No evidence ? I have referred you to an article on ukpollingreport and elsewhere to research papers by Curtice and Sparrow and Martin Boon , what further evidence do you need ? |
|
|
|
Post by greenchristian on Dec 17, 2013 16:56:37 GMT
Even accepting - and I do - that constituency polls are prone to error, I think having these is good, not just from the point of view of having more information about these specific constituencies, but also because in the long run it might help polling companies get better at constituency polls. I think having a number of these in the short run up to May 2015 could be very interesting for analysing data, even if they are not all that accurate for that election Indeed. It might ultimately lead to a method of predicting election results that performs better than uniform swing. If we had relatively widespread (and high quality) constituency polling then it might be possible to construct a model along the lines that Nate Silver uses for US Presidential elections, and we could have prediction models that tell us which seats are likely to change hands, rather than a model that just gives us an idea of how many. |
|
|
|
Post by erlend on Dec 17, 2013 18:26:52 GMT
It is possible for something to be agreed to exist without understanding the cause. Gravity is generally accepted to exist. The cause remains elusive. I will stick with 'The earth sucks' until they detect gravitons.
|
|
Deleted
Deleted Member
Posts: 0
|
Post by Deleted on Dec 17, 2013 19:25:59 GMT
Even accepting - and I do - that constituency polls are prone to error, I think having these is good, not just from the point of view of having more information about these specific constituencies, but also because in the long run it might help polling companies get better at constituency polls. I think having a number of these in the short run up to May 2015 could be very interesting for analysing data, even if they are not all that accurate for that election Indeed. It might ultimately lead to a method of predicting election results that performs better than uniform swing. If we had relatively widespread (and high quality) constituency polling then it might be possible to construct a model along the lines that Nate Silver uses for US Presidential elections, and we could have prediction models that tell us which seats are likely to change hands, rather than a model that just gives us an idea of how many. Exactly. |
|
|
|
Post by gwynthegriff on Dec 17, 2013 22:02:47 GMT
Obviously there is a margin of error on these polls, but the Crewe one is not as unlikely, given the local election results. These would be the local election results where Labour failed to win a single seat outside the town of Crewe. [I should expand on that point, not least because somebody will doubtless point out that Labour won every seat within the town of Crewe! Throughout the Dunwoody years the Labour Party normally dominated Crewe (with small Ratepayer, then Conservative, then Lib Dem enclaves) but found it difficult to win seats elsewhere. They could sometimes pick up a couple of seats in Nantwich and post-1990 also won in Haslington, Shavington and - quite remarkably - in Willaston! Those post-1990 gains all went, but perhaps more significant is their collapse in Nantwich which became a Conservative v Nantwich Independent battle. On balance, and notwithstanding local election results on some very low turnouts, it is going to be a big challenge for Labour to shift an incumbent who is well-funded.] |
|
Sibboleth
Labour
'Sit on my finger, sing in my ear, O littleblood.'
Posts: 16,044
|
Post by Sibboleth on Dec 18, 2013 0:50:56 GMT
Labour don't need to win outside Crewe to win the seat. As you know perfectly well...
|
|
|
|
Post by erlend on Dec 18, 2013 8:27:47 GMT
I think it boils down to neck and neck if they don't. 1983 was I think (?) an example where that probably occurred.
|
|
The Bishop
Labour
Down With Factionalism!
Posts: 39,015
Member is Online
|
Post by The Bishop on Dec 18, 2013 11:18:19 GMT
Labour don't need to win outside Crewe to win the seat. As you know perfectly well... Plus there must be a good chance that a decent chunk of the "Nantwich Independent" vote will go Labour at a GE. This won't be an easy pick-up for us, the Survation findings notwithstanding. But it is definitely doable. |
|
|
|
Post by East Anglian Lefty on Dec 18, 2013 13:57:51 GMT
2011 local election results, using average vote method to deal with patchy contestation:
Con: 11535.5
Lab: 11365.5
LD: 1311.5 (only stood in 5 out of 13 wards)
Independents: 3054 (only stood in 4 out of 13 wards)
I'd also note that Gwyn's comments about Nantwich aren't entirely accurate - in the North and West ward, Labour ran the Tories a pretty close second and except for the leading Independent it was basically a three-way marginal. In the South and Stapeley ward the Tories were the clear victors, but that's got areas outside Nantwich and Labour were still comfortably ahead of the Independents.
|
|
|
|
Post by independentukip on Dec 18, 2013 21:04:16 GMT
|
|
Deleted
Deleted Member
Posts: 0
|
Post by Deleted on Dec 18, 2013 21:10:13 GMT
It is possible for something to be agreed to exist without understanding the cause. Gravity is generally accepted to exist. The cause remains elusive. I will stick with 'The earth sucks' until they detect gravitons. Can we turn this into a discussion relating to the quantum mechanics behind string theory? Anyone ...  |
|
|
|
Post by carlton43 on Dec 18, 2013 23:17:43 GMT
My first thought is 'Don't show it to Mark or he will have a hissy fit'! |
|
|
|
Post by East Anglian Lefty on Dec 19, 2013 10:49:48 GMT
Con 34, Lab 38, LD 10, UKIP 11
It does look like there's been a genuine uptick in the Tory numbers over the past ten days. I'm not sure what's caused it, because whilst immigration has been in the news a lot recently, the beginning of the uptick preceded it. One wonders if there's something about YouGov's sample that produces better results for the Tories around Christmas. If that's the case, they may have slightly overegged Cameron's Christmas bounce last year.
|
|
|
|
Post by marksenior on Dec 19, 2013 13:22:00 GMT
My first thought is 'Don't show it to Mark or he will have a hissy fit'! No I won't have a hissy fit . Patrick's " well argued piece " concentrates on the phenomenum of false recall so as to discredit past vote weighting but completely ignores the pro Labour bias in telephone polling which has been in evidence for 25 years and subject to a number of ways to correct for this by pollsters over the last 20 years . Ashcroft , Rentoul , Wells and ICM's track record over 20 years and myself versus Survation with no record of a successful polling record , not a difficult choice . |
|
Deleted
Deleted Member
Posts: 0
|
Post by Deleted on Dec 19, 2013 15:33:33 GMT
At the end of the day, pollsters showing smaller swings are almost always more accurate in any GE polling.
Survation's arguments may be correct - but the fudge factors do help to model the GE squeeze.
The real question is, do you try to show what the nation is thinking at a given time or try to push the data towards what people would be thinking in 2 years time come the real election - and that is a more philospohical one. Survation are clearly in the former camp, as UKIP getting 20% in a GE is totally laughable.
|
|
johnr
Labour & Co-operative
Posts: 1,944 
|
Post by johnr on Dec 19, 2013 15:59:13 GMT
completely ignores the pro Labour bias in telephone polling which has been in evidence for 25 years Yet again - your evidence for this? I have pointed out reasons why this may be in fact the opposite. |
|
Deleted
Deleted Member
Posts: 0
|
Post by Deleted on Dec 19, 2013 16:29:02 GMT
At the end of the day, pollsters showing smaller swings are almost always more accurate in any GE polling. Survation's arguments may be correct - but the fudge factors do help to model the GE squeeze. The real question is, do you try to show what the nation is thinking at a given time or try to push the data towards what people would be thinking in 2 years time come the real election - and that is a more philospohical one. Survation are clearly in the former camp, as UKIP getting 20% in a GE is totally laughable. Opinion polling is supposed to be about taking a snapshot, not gazing into a crystal ball. So you favour no redistribution at all - even if evidence has shown that those umming and aahing often go back to their last port of call? |
|
|
|
Post by greenchristian on Dec 19, 2013 16:35:56 GMT
It is possible for something to be agreed to exist without understanding the cause. Gravity is generally accepted to exist. The cause remains elusive. I will stick with 'The earth sucks' until they detect gravitons. Can we turn this into a discussion relating to the quantum mechanics behind string theory? Anyone ... Sadly, I've forgotten almost everything I know about string theory. Though I do recall that at one point it looked like the most promising way to produce a Grand Unified Theory that manages to align quantum mechanics with relativity, and come up with a common explanation for the four fundamental forces (strong nuclear, weak nuclear, electromagnetic & gravitational). |
|
|
|
Post by marksenior on Dec 19, 2013 16:58:19 GMT
completely ignores the pro Labour bias in telephone polling which has been in evidence for 25 years Yet again - your evidence for this? I have pointed out reasons why this may be in fact the opposite. I have given references to the evidence in at least 2 earlier posts . I suggest you read through the thread and follow them up instead of asking for it again . If your assertion that it is in fact the opposite which is the correct one why on earth did the pollsters change their methodology post 1992 . Some of my references also discuss false recall as to how people do not recall correctly how they have previously voted . |
|
|
|
Post by marksenior on Dec 19, 2013 17:05:23 GMT
At the end of the day, pollsters showing smaller swings are almost always more accurate in any GE polling. Survation's arguments may be correct - but the fudge factors do help to model the GE squeeze. The real question is, do you try to show what the nation is thinking at a given time or try to push the data towards what people would be thinking in 2 years time come the real election - and that is a more philospohical one. Survation are clearly in the former camp, as UKIP getting 20% in a GE is totally laughable. Opinion polling is supposed to be about taking a snapshot, not gazing into a crystal ball. Your comment about taking a snapshot is correct but the adjustments or "fudge factors" are not concerned with making a more accurate forecast but with ensuring that the sample you are polling is fully representative . The Comres online and telephone polls have the same methodology but consistently very different results , the difference must therefore be due differing sample groups . |
|
