YouGov

[manchesterman]

Some interesting and unexpected 3-way+ marginals I noted based on their seat share percentages...

Basildon/Billericay: Ref 30, Lab 28, Con 26
S.Basil/E.Thurrock: L32, C 30, R 28
Bognor Regis & Lit: L 31, C 28, R 27
S.holland & deepings: C 32, R 31, L 29
havant: C 29, R 29, L 27
IoW East: L28, C26, R23
N.Abbot: C 29, LD 25, R 23
Tamworth: C32, L32, R 26
Montgomery: C 28, L 25, R 21
Bridgwater: C 30, L28, R 23
Torridge: C 30, LD 24, R 23

4-way
Waveney; C 31, R 23, L 21, G 21

[batman]

Jun 19, 2024 16:51:36 GMT johng said:

I'll copy my post from the Ipsos thread:

I think people are obsessing a bit too much on individual seats.

An MRP is just a much enhanced universal swing projection. It is less accurate in Wales and Scotland due to nationalist parties with a small sample sizes, it finds it hard to pin down new parties like ReformUK, and is pretty much useless in projecting seats where there a specific circumstances (a strong independent, a disgraced MP, a particularly popular candidates, a very strong ground game etc. etc.).

Absolutely agree with you John. Modelling seats is a waste of time unless it is predicated on an understanding of its specific politics. A particularly extreme example is totally omitting Jason Zadrozny in Ashfield, but there are plenty of examples of where the pollster has simply not taken into account the ways in which some seats are clearly different from each other over & above the raw figures. It is sort of moderately interesting or even amusing when an MRP extrapolates that, for example, Rishi Sunak is behind Labour in his constituency, but unless it's a properly-conducted constituency poll it's nearly worthless. Some of the results are, even in the present extraordinary circumstances, quite laughable. To give just one example, the idea that the Tories would lose more than half their seats & yet hold Finchley & Golders Green bears no relation to the political situation in that constituency whatsoever; and that would have been the case even if Mike Freer had stood for re-election, which he is not. I pledge myself not believe a single MRP prediction which is not supported by a constituency poll, however rare such things are. And of course I am mostly disbelieving results which are highly favourable to my own party.

[Adam in Stroud]

Jun 19, 2024 16:11:34 GMT binky said:

This model has the Tories losing Huntingdon, Maidenhead, Uxbridge, and Witney. I buy Labour will win Johnson's old seat, but color me very skeptical they'll take Major's. It's also a stretch to imagine the Liberals winning May and Cameron's old stomping grounds...

I hadn't spotted that. Wasn't that the largest Conservative majority in the Commons when Major held it?

[batman]

Jun 19, 2024 21:26:38 GMT Adam in Stroud said:

Jun 19, 2024 16:11:34 GMT binky said:

This model has the Tories losing Huntingdon, Maidenhead, Uxbridge, and Witney. I buy Labour will win Johnson's old seat, but color me very skeptical they'll take Major's. It's also a stretch to imagine the Liberals winning May and Cameron's old stomping grounds...

I hadn't spotted that. Wasn't that the largest Conservative majority in the Commons when Major held it?

yes it was the largest Tory majority both in 1992 on its larger old boundaries and in 1997 on its smaller revised ones.

[polupolu]

From the YouGov website for 18th Jun: (change vs 13th Jun)
Con 20% +2
Lab 36% -1
Lib 14% =
SNP 3% = (27% in sub-sample -7 so probably wat too small)
PC 1% =
Ref 18% -1
Grn 7% =
Oth 2% =
yougov.co.uk/topics/politics/trackers/voting-intention

[polupolu]

Plug those numbers into the FT model:
Con 84
Lab 429
Lib 92
SNP 21
Ref 0
Grn 1
PC 3

In Electoral Calculus UNS model with the scottish sub-sample numbers:
Con 145
Lab 400
Lib 64
Ref 1
Grn 1
SNP 16
PC 3

Electoral Calculus Non-UNS with Scot sub-sample:
Con 108
Lab 429
Lib 64
Ref 4
Grn 2
SNP 18
PC 4

[gibbon]

Those who claim that it is bad for democracy for there to be a very large Labour majority ignore the fact that in 2019 they campaigned for a very large Consevative majority. The electoral system is the same as in 2019, first past the post, and the main changes are that spending limits have increased and voter ID is now required. Both of these were introduced by the Conservatives.

[graham]

In vote share terms 36% would be a pretty poor result for Labour - well below the 41% Corbyn polled in 2017 and only 3% higher than 2019.

[cogload]

Should the slow decline in polling numbers worry Labour? It is one polling form and it is noise but the trend is clear.

[woollyliberal]

Jun 20, 2024 14:25:18 GMT graham said:

In vote share terms 36% would be a pretty poor result for Labour - well below the 41% Corbyn polled in 2017 and only 3% higher than 2019.

Vote share is wonderful, but not as good as seat share. Several but not all MRP models show the Lib Dems beating their all time record number of MPs on a lower vote share than 2019. Would I rather have 11.4% of the votes and 67 MPs, or 11.7% of the vote and 11 MPs? Let me think...

[binky]

The Labour/LibDem #s aren't too far off the Goodwin poll, although the Tory/Reform #s conform with the other stuff we've seen.

[johnloony]

Jun 20, 2024 14:56:57 GMT cogload said:

Should the slow decline in polling numbers worry Labour? It is one polling form and it is noise but the trend is clear.

It should. If I was Labour I would be worried about opinion polls saying 37% or 35% or whatever, regardless of how other parties split.

[The Bishop]

Jun 20, 2024 14:56:57 GMT cogload said:

Should the slow decline in polling numbers worry Labour? It is one polling form and it is noise but the trend is clear.

No serious pollster other than YouGov has put Labour under 39% since the Truss era. By their own admission, their new methodology in "regular" polling has taken a deliberate decision to heavily downweight the Labour score. And their MRP still has the party at 39%.

[Sibboleth]

Jun 20, 2024 14:56:57 GMT cogload said:

Should the slow decline in polling numbers worry Labour? It is one polling form and it is noise but the trend is clear.

The new methodology that they suddenly switched to halfway through the campaign seems to be putting a very heavy downwards push on it.

[The Bishop]

Indeed, in fact I'm pretty sure that all the "methodology changes" by pollsters since this campaign started have only worked in one direction.

If the ones who haven't done this (basically the "nowcasters") actually turn out to be the most accurate, then some questions will have to be asked in the industry.

[mattbewilson]

What is up with people trying to make out pollsters are bubkus atm

[batman]

Well there are these changes of methodology and one sometimes wonders why they feel they're necessary. Normally pollsters change their methodology after election results not in the middle of a campaign.

[greenchristian]

Jun 20, 2024 14:24:31 GMT gibbon said:

Those who claim that it is bad for democracy for there to be a very large Labour majority ignore the fact that in 2019 they campaigned for a very large Consevative majority. The electoral system is the same as in 2019, first past the post, and the main changes are that spending limits have increased and voter ID is now required. Both of these were introduced by the Conservatives.

Not everybody who is worried about the consequences of a very large Labour majority supported (let alone campaigned for) a Conservative majority in 2019. I have certainly expressed concerns that a massive Labour majority could be bad for democracy and good governance, and there's no way I'm the only person on the left who feels that way.

[doktorb🏳️‍🌈🏳️‍⚧️]

Jun 19, 2024 16:55:07 GMT michaelarden said:

Jun 19, 2024 16:51:36 GMT johng said:

I'll copy my post from the Ipsos thread:

I think people are obsessing a bit too much on individual seats.

An MRP is just a much enhanced universal swing projection. It is less accurate in Wales and Scotland due to nationalist parties with a small sample sizes, it finds it hard to pin down new parties like ReformUK, and is pretty much useless in projecting seats where there a specific circumstances (a strong independent, a disgraced MP, a particularly popular candidates, a very strong ground game etc. etc.).

Exactly - there is no way Galloway is going to get 7% in Rochdale and come 5th for example.

There's something broken about how many players don't include any of the prominent 'outsiders', be it Galloway, Corbyn, even Farage. There will have to be lots of reviews into the way these things work after polling day.

[No Offence Alan]

Jun 20, 2024 20:50:52 GMT doktorb🏳️‍🌈🏳️‍⚧️ said:

Jun 19, 2024 16:55:07 GMT michaelarden said:

Exactly - there is no way Galloway is going to get 7% in Rochdale and come 5th for example.

There's something broken about how many players don't include any of the prominent 'outsiders', be it Galloway, Corbyn, even Farage. There will have to be lots of reviews into the way these things work after polling day.

Not really, on a national scale, these localised candidates won't register. And at the next election, with the possible exception of the Ashfield Independents, they will have disappeared.