Focaldata

[The Bishop]

Nov 1, 2019 22:21:18 GMT liverpoolliberal said:

Has anyone collated the Focaldata MRP poll seat by seat? The Best for Britian site is pure garbage and I just want to see what they say for each seat.

That's not following the party line

[liverpoolliberal]

Nov 2, 2019 10:26:00 GMT The Bishop said:

Nov 1, 2019 22:21:18 GMT liverpoolliberal said:

Has anyone collated the Focaldata MRP poll seat by seat? The Best for Britian site is pure garbage and I just want to see what they say for each seat.

That's not following the party line

I meant in terms of user-friendlyness, but I would humbly point out that the website currently advocates voting for Labour in Hampstead & Kilburn and Battersea... even though their projections show us winning in both! Amazing that the righteous wing of twitter haven't bothered commenting on those seats 

[iain]

I’m not at all convinced by this data. There are a few weird looking Labour holds - for example Grimsby, Blackpool South and Darlington. The data for Labour in the Welsh valleys is abysmal.

There is no data given for East Devon, Eastbourne, Runnymede & Weybridge, Beaconsfield, South West Hertfordshire, Leicester East, Broxtowe, Eddisbury

There is an exact draw between Labour and the Tories in Heywood & Middleton.


The site gives the following seat changes:

Conservative gain from Labour (61):
Blyth Valley, Bishop Auckland, Sedgefield, Stockton South, Workington, Barrow & Furness, Lancaster & Fleetwood, Chorley, Burnley, Bolton North East, Bury North, Bury South, Stalybridge & Hyde, Wirral West, Weaver Vale, Warrington South, Crewe & Nantwich, Scunthorpe, Bradford South, Keighley, Colne Valley, Dewsbury, Wakefield, Penistone & Stocksbridge, Don Valley, Rother Valley, High Peak, Derby North, Bolsover, Ashfield, Bassetlaw, Lincoln, Newcastle-under-Lyme, Stoke North, Stoke Central, Wolverhampton South West, Dudley North, Warwick & Leamington, Peterborough, Ipswich, Bedford, Reading East, Portsmouth South, Canterbury, Stroud, Bristol North West, Plymouth Sutton & Devonport, Enfield Southgate, Kensington, Croydon South, Eltham, Newport West, Cardiff North, Bridgend, Gower, Clwyd South, Wrexham, Alyn & Deeside, Delyn, Vale of Clwyd, Ynys Mon

Liberal Democrat gain from Conservative (6):
St Ives, Cheltenham, Winchester, Lewes, Richmond Park, St Albans

Liberal Democrat gain from Labour (5):
Battersea, Hampstead & Kilburn, Cambridge, Sheffield Hallam, Leeds North West

Liberal Democrat gain from Plaid Cymru (1):
Ceredigion

[Tony Otim]

Nov 2, 2019 20:44:23 GMT iain said:

I’m not at all convinced by this data. There are a few weird looking Labour holds - for example Grimsby, Blackpool South and Darlington. The data for Labour in the Welsh valleys is abysmal.

There is no data given for East Devon, Eastbourne, Runnymede & Weybridge, Beaconsfield, South West Hertfordshire, Leicester East, Broxtowe, Eddisbury

There is an exact draw between Labour and the Tories in Heywood & Middleton.


The site gives the following seat changes:

Labour gain from Conservative (61):
Blyth Valley, Bishop Auckland, Sedgefield, Stockton South, Workington, Barrow & Furness, Lancaster & Fleetwood, Chorley, Burnley, Bolton North East, Bury North, Bury South, Stalybridge & Hyde, Wirral West, Weaver Vale, Warrington South, Crewe & Nantwich, Scunthorpe, Bradford South, Keighley, Colne Valley, Dewsbury, Wakefield, Penistone & Stocksbridge, Don Valley, Rother Valley, High Peak, Derby North, Bolsover, Ashfield, Bassetlaw, Lincoln, Newcastle-under-Lyme, Stoke North, Stoke Central, Wolverhampton South West, Dudley North, Warwick & Leamington, Peterborough, Ipswich, Bedford, Reading East, Portsmouth South, Canterbury, Stroud, Bristol North West, Plymouth Sutton & Devonport, Enfield Southgate, Kensington, Croydon South, Eltham, Newport West, Cardiff North, Bridgend, Gower, Clwyd South, Wrexham, Alyn & Deeside, Delyn, Vale of Clwyd, Ynys Mon

Liberal Democrat gain from Conservative (6):
St Ives, Cheltenham, Winchester, Lewes, Richmond Park, St Albans

Liberal Democrat gain from Labour (5):
Battersea, Hampstead & Kilburn, Cambridge, Sheffield Hallam, Leeds North West

Liberal Democrat gain from Plaid Cymru (1):
Ceredigion

Surely 61 Con gain from lab rather than vice-versa

[afleitch]

Nov 2, 2019 20:44:23 GMT iain said:

I’m not at all convinced by this data. There are a few weird looking Labour holds - for example Grimsby, Blackpool South and Darlington. The data for Labour in the Welsh valleys is abysmal.

There is no data given for East Devon, Eastbourne, Runnymede & Weybridge, Beaconsfield, South West Hertfordshire, Leicester East, Broxtowe, Eddisbury

There is an exact draw between Labour and the Tories in Heywood & Middleton.


The site gives the following seat changes:

Conservative gain from Labour (61):
Blyth Valley, Bishop Auckland, Sedgefield, Stockton South, Workington, Barrow & Furness, Lancaster & Fleetwood, Chorley, Burnley, Bolton North East, Bury North, Bury South, Stalybridge & Hyde, Wirral West, Weaver Vale, Warrington South, Crewe & Nantwich, Scunthorpe, Bradford South, Keighley, Colne Valley, Dewsbury, Wakefield, Penistone & Stocksbridge, Don Valley, Rother Valley, High Peak, Derby North, Bolsover, Ashfield, Bassetlaw, Lincoln, Newcastle-under-Lyme, Stoke North, Stoke Central, Wolverhampton South West, Dudley North, Warwick & Leamington, Peterborough, Ipswich, Bedford, Reading East, Portsmouth South, Canterbury, Stroud, Bristol North West, Plymouth Sutton & Devonport, Enfield Southgate, Kensington, Croydon South, Eltham, Newport West, Cardiff North, Bridgend, Gower, Clwyd South, Wrexham, Alyn & Deeside, Delyn, Vale of Clwyd, Ynys Mon

Liberal Democrat gain from Conservative (6):
St Ives, Cheltenham, Winchester, Lewes, Richmond Park, St Albans

Liberal Democrat gain from Labour (5):
Battersea, Hampstead & Kilburn, Cambridge, Sheffield Hallam, Leeds North West

Liberal Democrat gain from Plaid Cymru (1):
Ceredigion

Tallying that up (ignoring the missing data) that would put the Tories on 4 seats in Scotland, the Lib Dems on 3 (down 1) and Labour on zero. With the SNP on 52 that would work out. But it means the Lib Dems losing a seat in Scotland and Labour being gutted in Edinburgh South. Shame they've not released the Scottish data.

[Andrew_S]

The LDs gaining Battersea isn't credible in my opinion. Putney, possibly, but not Battersea.

[Tony Otim]

Nov 2, 2019 21:00:43 GMT afleitch said:

Nov 2, 2019 20:44:23 GMT iain said:

I’m not at all convinced by this data. There are a few weird looking Labour holds - for example Grimsby, Blackpool South and Darlington. The data for Labour in the Welsh valleys is abysmal.

There is no data given for East Devon, Eastbourne, Runnymede & Weybridge, Beaconsfield, South West Hertfordshire, Leicester East, Broxtowe, Eddisbury

There is an exact draw between Labour and the Tories in Heywood & Middleton.


The site gives the following seat changes:

Conservative gain from Labour (61):
Blyth Valley, Bishop Auckland, Sedgefield, Stockton South, Workington, Barrow & Furness, Lancaster & Fleetwood, Chorley, Burnley, Bolton North East, Bury North, Bury South, Stalybridge & Hyde, Wirral West, Weaver Vale, Warrington South, Crewe & Nantwich, Scunthorpe, Bradford South, Keighley, Colne Valley, Dewsbury, Wakefield, Penistone & Stocksbridge, Don Valley, Rother Valley, High Peak, Derby North, Bolsover, Ashfield, Bassetlaw, Lincoln, Newcastle-under-Lyme, Stoke North, Stoke Central, Wolverhampton South West, Dudley North, Warwick & Leamington, Peterborough, Ipswich, Bedford, Reading East, Portsmouth South, Canterbury, Stroud, Bristol North West, Plymouth Sutton & Devonport, Enfield Southgate, Kensington, Croydon South, Eltham, Newport West, Cardiff North, Bridgend, Gower, Clwyd South, Wrexham, Alyn & Deeside, Delyn, Vale of Clwyd, Ynys Mon

Liberal Democrat gain from Conservative (6):
St Ives, Cheltenham, Winchester, Lewes, Richmond Park, St Albans

Liberal Democrat gain from Labour (5):
Battersea, Hampstead & Kilburn, Cambridge, Sheffield Hallam, Leeds North West

Liberal Democrat gain from Plaid Cymru (1):
Ceredigion

Tallying that up (ignoring the missing data) that would put the Tories on 4 seats in Scotland, the Lib Dems on 3 (down 1) and Labour on zero. With the SNP on 52 that would work out. But it means the Lib Dems losing a seat in Scotland and Labour being gutted in Edinburgh South. Shame they've not released the Scottish data.

So, basically it's complete bobbins...

[hullenedge]

Chris Hanretty's spreadsheet of the MRP data (England & Wales bar a few seats):-

Correlations of % change for MRP & Euro19:-

[jamie]

They've updated their model with new data.

[iain]

Nov 27, 2019 14:36:43 GMT jamie said:

They've updated their model with new data.

It looks even barmier than before

[jamie]

My quick scan of a few constituencies suggest they think Labour are doing very well (well, relatively well) with remain voters while the Tories are surging with leavers everywhere. Labour doing well in London and the Tories in the North East.

[afleitch]

Seat data:

gallery.mailchimp.com/a706d4e03dc82629db3f7acf9/files/ae6c2fb1-48aa-4c6f-8a3d-b5f8fd958956/Best_for_Britain_Recommendation_Seats.pdf

Scotland only seems to have Tory 2017 seats and some marginals. SNP gain Ochil (Stirling not polled) only. SNP and Green vote bigger than LD's in Edinburgh W and Duns E. And Labour holding Edinburgh S and surprisingly Glasgow N E.

[jacoblamsden]

Conservatives 10% ahead in Ilford North and 19% ahead in St Ives apparently!

[curiousliberal]

Nov 27, 2019 15:22:31 GMT jacoblamsden said:

Conservatives 10% ahead in Ilford North and 19% ahead in St Ives apparently!

Sounds nuts and I don't buy it, but Andrew George may well have spoken out against #revoke with his constituents as well as his principles in mind.

[cogload]

Yeah and the SNP only 4% behind in Orkney and Shetland.

[curiousliberal]

Not all MRP is made equal. The model essentially works by weighting factors likely to decide someone's vote (which presumably vary to some extent in each electoral cycle). This means that even with good data, getting the weights wrong (which, from experience of dabbling in machine learning and a fair bit of reading on the subject, seems very easy to do) is entirely possible and can render an MRP model useless.

[iain]

Tories up by 30% in Esher but losing Wimbledon to us ... hmmm ...

[Chris from Brum]

So, to sum up, it's probably pish?

[Vibe]

Am I blind? Where is Broxtowe?

[curiousliberal]

Nov 27, 2019 15:33:12 GMT Chris from Brum said:

So, to sum up, it's probably pish?

It's probably pish, but that isn't proof of the pudding. I'd be shocked if we didn't do better in Wimbledon (and probably significantly so) than Esher and Walton. Wimbledon returned someone other than a Conservative as late as 2001; to find the last time E&W voted anything but Conservative, you have to go back to 1906.