J. L. Partners

[manchesterman]

I stopped looking after O & S!

Nonsense.

[carolus]

You'd think they might have been able to invest in a better map...

At least you can see the shapes:

I've expressed my wider scepticism of MRPs before, but this one really does suggest it's pure guesswork. 4.5k sample doesn't exactly fill one with confidence either.

Wonder if JL would like to put some money down on some of those seats?

[YL]

In the southern half of England it doesn't actually look that weird as these things go. It's in the North (and down into parts of the East Midlands and Wales) where it seems really out there; Red Tatton is the standout bizarre prediction for me but there are several others as already mentioned. Only Westmorland & Lonsdale and Penrith & the Border stop Labour sweeping the NW as well as the NE...

[Andrew_S]

The Sunday Times also seems to believe that 352 seats = a 14 seat majority.

[batman]

Feb 27, 2022 18:20:05 GMT Andrew_S said:

The Sunday Times also seems to believe that 352 seats = a 14 seat majority.

what a bunch of nitwits. it's hardly rocket science for them to work out otherwise

[batman]

Feb 27, 2022 12:29:28 GMT andrewp said:

Feb 27, 2022 8:55:36 GMT batman said:

Well if Labour were to win 352 seats that would be an overall majority of 54, not 14, officially. And if we assume that Sinn Fein have 7 seats  the effective majority would be 61. Either the number of Labour seats or the majority figure is wrong.

Tim Shipman reports on the poll in the Sunday times today and repeats the 352 and majority of 14 mistake.

It also says Boris Johnson has a majority of 5000 in Uxbridge

It also says that Labour would gain Wyre and Preston North and Berwick Upon Tweed but that Esher and Walton would be a Conservative hold. A very odd combination.

you are certainly right in your last sentence. Labour will definitely not win  Wyre & N Preston. At least Ribble Valley contains Bamber Bridge, but Labour would be easily outvoted elsewhere.

[jamie]

Some of the specific constituency predictions look suspect, but a 13% Labour lead will produce Labour wins in constituencies that haven’t voted Labour in a long time/ever, and the swing will be bigger in some places than others.

[Andrew_S]

Feb 27, 2022 19:31:23 GMT jamie said:

Some of the specific constituency predictions look suspect, but a 13% Labour lead will produce Labour wins in constituencies that haven’t voted Labour in a long time/ever, and the swing will be bigger in some places than others.

Although probably not Ben Wallace's seat which has a majority of 16,800 or 32%.

[jamie]

Feb 27, 2022 22:29:41 GMT Andrew_S said:

Feb 27, 2022 19:31:23 GMT jamie said:

Some of the specific constituency predictions look suspect, but a 13% Labour lead will produce Labour wins in constituencies that haven’t voted Labour in a long time/ever, and the swing will be bigger in some places than others.

Although probably not Ben Wallace's seat which has a majority of 16,800 or 32%.

Probably not, but the poll assumes a 25% shift in margin from 2019 and thats obviously not likely (or in some cases, borderline not possible) in Labour's safest seats, so there would be a greater swing in the Conservatives safer seats. This poll is one of the best for Labour in years and I don't think many of us are expecting it to actually happen, but a proper Labour landslide would result in some before the event unexplainable Labour gains as happened in 1997 (or to some extent happened for the Conservatives in 2019). 

[clyde1998]

Feb 27, 2022 22:37:56 GMT jamie said:

Feb 27, 2022 22:29:41 GMT Andrew_S said:

Although probably not Ben Wallace's seat which has a majority of 16,800 or 32%.

Probably not, but the poll assumes a 25% shift in margin from 2019 and thats obviously not likely (or in some cases, borderline not possible) in Labour's safest seats, so there would be a greater swing in the Conservatives safer seats. This poll is one of the best for Labour in years and I don't think many of us are expecting it to actually happen, but a proper Labour landslide would result in some before the event unexplainable Labour gains as happened in 1997 (or to some extent happened for the Conservatives in 2019). 

I believe the poll was conducted at the height of the 'partygate' scandal, so the VI numbers are probably the best case scenario for Labour. Also, as I've said with other MRP analyses, it's very hard to accurately judge tactically voting - I haven't seen any specific numbers, but it's likely there are a lot of seats where reasonable tactical voting between Labour, Lib Dem and Green voters could be enough to change the outcome of the constituency (as seen in 1997 as well).

[Andrew_S]

Feb 27, 2022 22:42:26 GMT clyde1998 said:

Feb 27, 2022 22:37:56 GMT jamie said:

Probably not, but the poll assumes a 25% shift in margin from 2019 and thats obviously not likely (or in some cases, borderline not possible) in Labour's safest seats, so there would be a greater swing in the Conservatives safer seats. This poll is one of the best for Labour in years and I don't think many of us are expecting it to actually happen, but a proper Labour landslide would result in some before the event unexplainable Labour gains as happened in 1997 (or to some extent happened for the Conservatives in 2019). 

I believe the poll was conducted at the height of the 'partygate' scandal, so the VI numbers are probably the best case scenario for Labour. Also, as I've said with other MRP analyses, it's very hard to accurately judge tactically voting - I haven't seen any specific numbers, but it's likely there are a lot of seats where reasonable tactical voting between Labour, Lib Dem and Green voters could be enough to change the outcome of the constituency (as seen in 1997 as well).

If they're using a computer database to do the research, as I assume they are, I don't know why it can't be more up-to-date, ie. a short time between conducting it and publishing the results. I could understand a delay if it was being done using one of the old-fashioned methods.

[clyde1998]

Feb 27, 2022 23:43:13 GMT Andrew_S said:

Feb 27, 2022 22:42:26 GMT clyde1998 said:

I believe the poll was conducted at the height of the 'partygate' scandal, so the VI numbers are probably the best case scenario for Labour. Also, as I've said with other MRP analyses, it's very hard to accurately judge tactically voting - I haven't seen any specific numbers, but it's likely there are a lot of seats where reasonable tactical voting between Labour, Lib Dem and Green voters could be enough to change the outcome of the constituency (as seen in 1997 as well).

If they're using a computer database to do the research, as I assume they are, I don't know why it can't be more up-to-date, ie. a short time between conducting it and publishing the results. I could understand a delay if it was being done using one of the old-fashioned methods.

You would expect they'd have some fairly modern program to 'project' each constituency. YouGov were able to have daily updates in the previous two general election campaigns, so it's certainly technically possible. Perhaps it was down to the Sunday Times deciding on the timing to publish the results, rather than any technical restrictions on the pollsters part.

[YL]

Tables and the MRP figures for each constituency are available on the JL Partners website.

Unlike some recent MRPs this doesn't show the same level of collapse of the Lib Dem vote towards Labour in LD/Con battlegrounds; it does show Labour moving into second in Bath but the Lib Dems are still some way ahead. I suspect this one is more accurate in that respect even if several pinches of salt are needed for other aspects of it.

The tables for the poll show regional subsamples, so I suspect region has been used as a variable in the regression and is responsible for the very strong Labour results it shows in e.g. the NW and East Midlands. (There's no way they can have actually modelled individual constituency effects in the regression with this sample size.) The sample size in each region is not very large so there's a substantial margin of error there.

[andrewp]

Mar 1, 2022 11:40:42 GMT YL said:

Tables and the MRP figures for each constituency are available on the JL Partners website.

Unlike some recent MRPs this doesn't show the same level of collapse of the Lib Dem vote towards Labour in LD/Con battlegrounds; it does show Labour moving into second in Bath but the Lib Dems are still some way ahead. I suspect this one is more accurate in that respect even if several pinches of salt are needed for other aspects of it.

The tables for the poll show regional subsamples, so I suspect region has been used as a variable in the regression and is responsible for the very strong Labour results it shows in e.g. the NW and East Midlands. (There's no way they can have actually modelled individual constituency effects in the regression with this sample size.) The sample size in each region is not very large so there's a substantial margin of error there.

Interesting. As usual one can point to individual constituencies and I would be prepared to bet quite a lot that some of those won’t happen. Labour winning Altrincham by c.10000.

They still have Labour far too high in some Con/ LD contests. In my constituency in Taunton, Labour on 23%. The Tories holding with 37%. If the Tories drop to 37% in Taunton, the Lib Dems will win by at least 5000 votes.

[East Anglian Lefty]

Mar 1, 2022 11:40:42 GMT YL said:

Tables and the MRP figures for each constituency are available on the JL Partners website.

Unlike some recent MRPs this doesn't show the same level of collapse of the Lib Dem vote towards Labour in LD/Con battlegrounds; it does show Labour moving into second in Bath but the Lib Dems are still some way ahead. I suspect this one is more accurate in that respect even if several pinches of salt are needed for other aspects of it.

The tables for the poll show regional subsamples, so I suspect region has been used as a variable in the regression and is responsible for the very strong Labour results it shows in e.g. the NW and East Midlands. (There's no way they can have actually modelled individual constituency effects in the regression with this sample size.) The sample size in each region is not very large so there's a substantial margin of error there.

And if you're working at a regional level, you're basically asking for junk figures. I'm not sure what the likelihood of Eddisbury falling whilst Stoke South stays Tory is, but you'd need a lot of decimal points to be in the right order of magnitude.