|
Post by batman on Feb 12, 2022 20:09:53 GMT
so that puts them at variance with all other pollsters. Either they're right & everyone else is wrong, or the reverse.
|
|
Sibboleth
Labour
'Sit on my finger, sing in my ear, O littleblood.'
Posts: 15,261
|
Post by Sibboleth on Feb 12, 2022 20:14:22 GMT
They're using an entirely new methodology that eliminates the usual variable turnout dynamic in favour of fixed assumptions about voter behaviour projected from the last election. The next stage in the shift of polling away from being an actual survey and towards a projection model that incorporates survey data. I'm dubious, but we shall see.
|
|
|
Post by woollyliberal on Feb 12, 2022 20:18:21 GMT
so that puts them at variance with all other pollsters. Either they're right & everyone else is wrong, or the reverse. Well yes, because the other pollsters still use the traditional methodology for dealing with don't knows. I think what's going on here is a reaction to the last 6 months or so. Lots of voters have left the Tories but many of them have moved to Don't Know. Traditionally, polls ignore Don't Knows as they tend to split in the same proportions as those who Do Know. Since the Don't Knows are now full of ex-Tories, the old assumption probably doesn't hold. When forced to vote, these ex-Tory Don't Knows will have to choose. I guess that this new methodology tries to predict how that will go.
|
|
|
Opinium
Feb 12, 2022 20:19:38 GMT
via mobile
Post by andrew111 on Feb 12, 2022 20:19:38 GMT
They're using an entirely new methodology that eliminates the usual variable turnout dynamic in favour of fixed assumptions about voter behaviour projected from the last election. The next stage in the shift of polling away from being an actual survey and towards a projection model that incorporates survey data. I'm dubious, but we shall see. Isn't that essentially what MRP is doing?
|
|
|
Opinium
Feb 12, 2022 20:21:25 GMT
via mobile
Post by andrew111 on Feb 12, 2022 20:21:25 GMT
so that puts them at variance with all other pollsters. Either they're right & everyone else is wrong, or the reverse. Well yes, because the other pollsters still use the traditional methodology for dealing with don't knows. I think what's going on here is a reaction to the last 6 months or so. Lots of voters have left the Tories but many of them have moved to Don't Know. Traditionally, polls ignore Don't Knows as they tend to split in the same proportions as those who Do Know. Since the Don't Knows are now full of ex-Tories, the old assumption probably doesn't hold. When forced to vote, these ex-Tory Don't Knows will have to choose. I guess that this new methodology tries to predict how that will go. And it predicts more will go Lib Dem than go Labour, I guess
|
|
|
Post by woollyliberal on Feb 12, 2022 20:22:48 GMT
Well yes, because the other pollsters still use the traditional methodology for dealing with don't knows. I think what's going on here is a reaction to the last 6 months or so. Lots of voters have left the Tories but many of them have moved to Don't Know. Traditionally, polls ignore Don't Knows as they tend to split in the same proportions as those who Do Know. Since the Don't Knows are now full of ex-Tories, the old assumption probably doesn't hold. When forced to vote, these ex-Tory Don't Knows will have to choose. I guess that this new methodology tries to predict how that will go. And it predicts more will go Lib Dem than go Labour, I guess And some will return to the Tories. Some will go to the other parties. Some will stay at home.
|
|
Sibboleth
Labour
'Sit on my finger, sing in my ear, O littleblood.'
Posts: 15,261
|
Post by Sibboleth on Feb 12, 2022 20:25:26 GMT
I think what's going on here is a reaction to the last 6 months or so. Lots of voters have left the Tories but many of them have moved to Don't Know. Traditionally, polls ignore Don't Knows as they tend to split in the same proportions as those who Do Know. Since the Don't Knows are now full of ex-Tories, the old assumption probably doesn't hold. When forced to vote, these ex-Tory Don't Knows will have to choose. I guess that this new methodology tries to predict how that will go. The problem is that no one is forced to vote. I've read the theoretical justifications and I'm really not convinced that this is a good idea: it might work out (and it should produce much more stable polls than we're used to), but there's a risk that they might just be ICMing themselves.
|
|
|
Post by woollyliberal on Feb 12, 2022 20:29:57 GMT
I think what's going on here is a reaction to the last 6 months or so. Lots of voters have left the Tories but many of them have moved to Don't Know. Traditionally, polls ignore Don't Knows as they tend to split in the same proportions as those who Do Know. Since the Don't Knows are now full of ex-Tories, the old assumption probably doesn't hold. When forced to vote, these ex-Tory Don't Knows will have to choose. I guess that this new methodology tries to predict how that will go. The problem is that no one is forced to vote. I've read the theoretical justifications and I'm really not convinced that this is a good idea: it might work out (and it should produce much more stable polls than we're used to), but there's a risk that they might just be ICMing themselves. OK, badly worded. When forced to decide by a vote. Tory voters are religious when it comes to turning out on polling day. They currently can't back Boris, but they can't bring themselves to back anyone else. Come polling day they will.
|
|
|
Post by John Chanin on Feb 12, 2022 20:36:53 GMT
I have long held the view that it is deeply misleading to exclude don’t knows from polls. Changes in this figure are often more illuminating than changes in the percentages supporting different parties, and always provide a context to it, particularly when the polling companies know the previous affiliation of don’t knows, which they do with online panel polls.
|
|
|
Post by Andrew_S on Feb 12, 2022 20:38:18 GMT
3% lead for Labour with Opinium. Excellent news for Johnson. It's dangerous to just look at one poll. I'm not sure this will shift the average very much.
|
|
clyde1998
SNP
Green (E&W) member; SNP supporter
Posts: 1,765
|
Post by clyde1998 on Feb 12, 2022 20:38:56 GMT
The methodology changes appear to be: - Introduction of weighting by political attention.
- Introduction of weighting by ethnicity.
- Introduction of weighting by education by age.
- Replacing turnout weighting with demographic weights of voters at the previous two general elections (ie. the target weights will be the same as those who voted in the previous general elections, not the whole adult population).
- Weights will only be applied to those giving a voting intention (ie. weights are applied once undecided voters are excluded).
All seem reasonable, but the latter two are at odds with traditional polling methodology. www.opinium.com/resource-center/uk-voting-intention-27th-january-2022-2/
|
|
|
Post by mattbewilson on Feb 12, 2022 20:52:07 GMT
Raw figures are Con 23 Lab 29. Last time round it was Con 28 Lab 33
|
|
|
Post by andrew111 on Feb 12, 2022 20:58:09 GMT
3% lead for Labour with Opinium. Excellent news for Johnson. It's dangerous to just look at one poll. I'm not sure this will shift the average very much. This poll is worse than the previous Opinium for Johnson than the last one isn't it? Only the methodology is better
|
|
|
Post by carolus on Feb 12, 2022 21:16:18 GMT
Opinium's magic number box now gives different magic numbers.
|
|
|
Post by redtony on Feb 12, 2022 21:34:17 GMT
What about the Labour dontknows must be more than a few Left Labour voters who now say they do not know.They have only one party to go to and thatis the Greens
|
|
|
Opinium
Feb 12, 2022 22:22:18 GMT
via mobile
Post by mattbewilson on Feb 12, 2022 22:22:18 GMT
What about the Labour dontknows must be more than a few Left Labour voters who now say they do not know.They have only one party to go to and thatis the Greens 11%
|
|
jamie
Top Poster
Posts: 6,839
|
Post by jamie on Feb 12, 2022 22:39:17 GMT
They're using an entirely new methodology that eliminates the usual variable turnout dynamic in favour of fixed assumptions about voter behaviour projected from the last election. The next stage in the shift of polling away from being an actual survey and towards a projection model that incorporates survey data. I'm dubious, but we shall see. I'm naturally sceptical of pollsters attempting to model, but its not like demographic turnout modelling even works out in practice eg; its disastrous performance in 2017. Lets hope the other pollsters don't return to it.
|
|
The Bishop
Labour
Down With Factionalism!
Posts: 36,474
|
Post by The Bishop on Feb 13, 2022 10:42:39 GMT
Well if this "new methodology" has done anything, its got people talking about Opinium's surveys. Cynics might even suspect that was a significant motivation
|
|
|
Post by manchesterman on Feb 13, 2022 14:22:15 GMT
To try to get a wider reading of current political trends, Ive taken the polling figures over the last 6 months or so, this covers pre-Partygate times from Sept, OCt and early Nov (when Tories had a consistent but narrow lead) and the polling since then which has varied between narrow to significant Labour leads. So it takes account of the lack of trust in a Johnson government currently felt, with an idea of how things may change if politics returns to something approaching "normal" in a post-Johnson world.
The figures are Lab 37.4%, Con 35.6%, LD 9.7%, SNP 4.7%, Green 6.1%, Reform 3.6%, Oth 3.3%
Putting those stats into Electoral Calculus gives: Lab 291 (+88) Con 272 (-93) LD 6 (-5) SNP 56 (+8) PC 5 (+1) Ind 1 (+1) Devon E!
|
|
CatholicLeft
Labour
2032 posts until I was "accidentally" deleted.
Posts: 6,232
|
Post by CatholicLeft on Feb 13, 2022 16:17:10 GMT
|
|