|
|
Post by bjornhattan on Nov 3, 2023 19:11:08 GMT
I’m sceptical that the new Blaydon and Consett constituency was less than 54%, but it has Gateshead and Whickham at less than 55% Leave as well (Gateshead borough voted 57% Leave). Birtley and Felling must be very Leave voting for all this to do up. There's also Lamesley and Windy Nook & Whitehills - both will have been among the stronger wards for Leave in the borough. Assuming the five Blaydon wards voted ~55% Leave, the seven southern and eastern wards will have had to have been 62-63% Leave for the figures to add up - and that doesn't seem plausible given (for example) the South Tyneside result. I actually think the mistake may be at the other end for that Blaydon and Consett figure. County Durham is probably a difficult area to model generally - it's huge and has a massive range of areas which will vary from some of the most strongly Remain in the country (Neville's Cross) to potentially over 70% Leave (Horden). I note Hanretty has City at 46% Leave, North Durham at 58% Leave, and Easington at 66% Leave (!) all of which strike me as a couple of percentage points higher than I'd expect - so perhaps he has Consett and the surrounding areas as being too strongly Remain voting which has fed through to the constituency. |
|
|
|
Post by manchesterman on Nov 3, 2023 20:52:59 GMT
On Times Radio this morning, Prof John Curtice said that the 'official' BBC/ITN/Sky News estimates of notional 2019 results for new boundaries will be coming out shortly after the Tamworth and Mid Bedfordshire byelections. Question: How shortly is shortly? It seems he was using the same time frame as Dorries' "immediately"  |
|
|
|
Post by greatkingrat on Nov 4, 2023 19:30:47 GMT
Technically we haven't got any new boundaries yet!
|
|
Harry Hayfield
Green
Cavalier Gentleman (as in 17th century Cavalier)
Posts: 2,922 
|
Post by Harry Hayfield on Nov 11, 2023 19:50:34 GMT
On Times Radio this morning, Prof John Curtice said that the 'official' BBC/ITN/Sky News estimates of notional 2019 results for new boundaries will be coming out shortly after the Tamworth and Mid Bedfordshire byelections. Twenty nine days later and twenty one days after Tamworth and Mid Bedfordshire, is or is not "shortly" because there is still no sign of them! |
|
Harry Hayfield
Green
Cavalier Gentleman (as in 17th century Cavalier)
Posts: 2,922
|
Post by Harry Hayfield on Nov 14, 2023 14:56:44 GMT
On Times Radio this morning, Prof John Curtice said that the 'official' BBC/ITN/Sky News estimates of notional 2019 results for new boundaries will be coming out shortly after the Tamworth and Mid Bedfordshire byelections. Twenty nine days later and twenty one days after Tamworth and Mid Bedfordshire, is or is not "shortly" because there is still no sign of them! We have an answer (of sorts) from the BBC: "John Curtice forwarded your email about the 2019 notionals: the estimates of what the election results would have been in 2019, had the new boundaries been in place then. We are planning to publish them soon but I’m afraid I can’t give you a firm date. They have been jointly commissioned by the BBC, ITV News, Sky News and the Press Association which means we all need to agree on when to release them. It’s really just a case of everybody being ready to go. This has taken longer than anticipated, largely because it’s been such a busy period for news organisations, but I do hope it won’t be much longer now" Which was sent to me by Peter Barnes of BBC News |
|
|
|
Post by Pete Whitehead on Nov 14, 2023 15:05:16 GMT
Twenty nine days later and twenty one days after Tamworth and Mid Bedfordshire, is or is not "shortly" because there is still no sign of them! We have an answer (of sorts) from the BBC: "John Curtice forwarded your email about the 2019 notionals: the estimates of what the election results would have been in 2019, had the new boundaries been in place then. We are planning to publish them soon but I’m afraid I can’t give you a firm date. They have been jointly commissioned by the BBC, ITV News, Sky News and the Press Association which means we all need to agree on when to release them. It’s really just a case of everybody being ready to go. This has taken longer than anticipated, largely because it’s been such a busy period for news organisations, but I do hope it won’t be much longer now" Which was sent to me by Peter Barnes of BBC News I contacted the BBC about something to do with elections/notional results back in the 90s (possibly touting my own skills IIRC) and I'm pretty sure Peter Barnes was the person who replied to me then |
|
nodealbrexiteer
Forum Regular
non aligned favour no deal brexit!
Posts: 4,456
|
Post by nodealbrexiteer on Nov 14, 2023 15:19:24 GMT
We have an answer (of sorts) from the BBC: "John Curtice forwarded your email about the 2019 notionals: the estimates of what the election results would have been in 2019, had the new boundaries been in place then. We are planning to publish them soon but I’m afraid I can’t give you a firm date. They have been jointly commissioned by the BBC, ITV News, Sky News and the Press Association which means we all need to agree on when to release them. It’s really just a case of everybody being ready to go. This has taken longer than anticipated, largely because it’s been such a busy period for news organisations, but I do hope it won’t be much longer now" Which was sent to me by Peter Barnes of BBC News I contacted the BBC about something to do with elections/notional results back in the 90s (possibly touting my own skills IIRC) and I'm pretty sure Peter Barnes was the person who replied to me then it's always worth doing something like that,if you don't ask you don't get etc |
|
nodealbrexiteer
Forum Regular
non aligned favour no deal brexit!
Posts: 4,456
|
Post by nodealbrexiteer on Nov 14, 2023 15:20:22 GMT
Twenty nine days later and twenty one days after Tamworth and Mid Bedfordshire, is or is not "shortly" because there is still no sign of them! We have an answer (of sorts) from the BBC: "John Curtice forwarded your email about the 2019 notionals: the estimates of what the election results would have been in 2019, had the new boundaries been in place then. We are planning to publish them soon but I’m afraid I can’t give you a firm date. They have been jointly commissioned by the BBC, ITV News, Sky News and the Press Association which means we all need to agree on when to release them. It’s really just a case of everybody being ready to go. This has taken longer than anticipated, largely because it’s been such a busy period for news organisations, but I do hope it won’t be much longer now" Which was sent to me by Peter Barnes of BBC News thanks for posting this |
|
|
|
Post by bjornhattan on Jan 14, 2024 16:49:43 GMT
I have now updated my model and notional estimates for the final proposals. These are still based purely on demographic data but now use updated figures from the 2021 Census and analysis is done at a more granular level (Output Areas rather than LSOAs which my old model used). The caveat to this is that these updated estimates are England only - Scottish census data is not yet available and while Welsh census data is accessible, the need to incorporate variables like Welsh speaking complicates the process there.
Full tables are available here - I've given estimated results for 2015, 2017, and 2019 as well as a summary tab showing who won each seat in each election and giving basic totals. As always, it's worth stressing that my model is unable to pick up anomalous sources of local strength which a more traditional notional calculation based on local elections would detect. However, I did cross-check all figures with the Ben Walker model (number 2 above) and in general errors were small, indicating reasonable levels of agreement with a more conventional approach. Using Prediction Competition scoring rules (adding up the absolute difference between my estimates and the Walker notionals), the median discrepancy was 1.74% and in nine tenths of seats our figures differed by less than 6%. In general my model tends to overestimate differences within seats, so the biggest differences were in seats like Croydon West (which essentially combines the strongest areas for Labour in three existing seats) or Ossett and Denby Dale (similarly strongly Conservative areas in three existing seats). There were eight constituencies where my model gave a different winner to Ben's, and most of them are interesting: - Beckenham and Penge (me: Con maj 9, BW: Lab maj 823) - this is a highly marginal seat with large boundary changes so a small difference is to be expected.
- Bedford (me: Lab maj 317, BW: Con maj 194) - here the changes are extremely small with a few small new build estates switching either way. The estate coming in looks to be slightly more Labour friendly than the estates going out (it is more ethnically diverse and more social housing), which is why my model increases Labour's majority. However, it is in a rural (and strongly Conservative) ward in Ben Walker's ward model, and I think he treats split wards as homogeneous - so in his calculations the change helps the Conservatives.
- Cheltenham (me: LD maj 203, BW: Con maj 1,224) - the Lib Dems often do well in relatively working class areas within middle class seats where they win tactical votes from people who would otherwise probably favour Labour. This causes a major ecological fallacy in my model - which generally assumes well educated middle class areas are the strongest for the Lib Dems because that's the trend at constituency level. For example, the only change in Cheltenham is the removal of Springbank ward (a relatively deprived area with a fair amount of social housing), where my model has the Lib Dems on about 20%; in local elections they win by huge margins and took 77% of the vote least time. Removing this obviously hurts the Lib Dems and certainly doesn't flip the seat their way - my model is categorically wrong here.
- Chorley (me: Lab maj 17,055, BW: Oth maj 31,316) - in the modelling stage I ignore the Speaker, but for these estimates I've classed him as Labour. In practice our results are the same (the figures are quite different as I classed the independent as , whereas Ben has classed both Mark Smith and Lindsay Hoyle as "other"). This has a slightly more meaningful impact on South Ribble next door - about 3,000 electors in the Charnock Richard area are moved, and in normal circumstances many of these (probably 1,000-1,200) would vote Conservative; the majority in the new South Ribble is likely to be closer to 20% rather than the 15.7% my estimates suggest.
- Cramlington and Killingworth (me: Con maj 582, BW: Lab maj 1,797) - quite difficult to account for the discrepancy here; most wards are similar in the two models but mine has a far stronger Conservative vote in Killingworth and parts of the Seaton Valley. There are new builds here such as at Holystone, but my model may be overestimating how Conservative these areas are.
- Dagenham and Rainham (me: Con maj 474, BW: Lab maj 992) - the boundary changes here remove one area and add two more. As mentioned above, my model tends to emphasise differences within seats. Therefore Labour do extremely well in (most of) Dagenham whereas the Conservatives run up the numbers in Rainham - of course this is somewhat accurate but probably not to the extent my model suggests. The ward being removed, Chadwell Heath, illustrates this well - Ben has this as 40% Conservative (perhaps picking up on historic local Tory strength here) whereas I have it as more like 32% Conservative. This accounts for most of the difference between the two results; I also have slightly stronger Conservative performances than Ben in the areas added in Valence ward and near Elm Park station.
- Kensington and Bayswater (me: Con maj 220, BW: Lab maj 1,137) - intuitively this seems like a fairly obvious Labour win but my model seems to have them as too weak in the Westminster part (perhaps assuming Karen Buck build up her majority in the Church Street and Harrow Road sections of her seat). It estimates that Labour only won the Bayswater and Lancaster Gate wards by about 500 votes, whereas Ben has them ahead there by about 1,800 which seems more realistic - they need to win by around 750 votes to overturn the Conservative majority in Kensington (this being bolstered a little by some extra Tory territory around Hans Town).
- Walsall and Bloxwich (me: Lab maj 1,442, BW: Con maj 4,181) - this is probably the constituency with the starkest difference between my model and Ben's, with a difference of over 13% (I have a Labour lead of 48-45, he has a Conservative lead of 51-42). While our Walsall South figures are generally quite similar, it's in the former Walsall North wards that the differences really pile up. This is epitomised by the Birchills Leamore ward: I have it as Conservative by 60 votes, compared with Ben having it as Conservative by 1,800. Unusually, this is one seat where we actually know the real result by ward, and generally speaking these are somewhere between the two models - for instance Birchills Leamore actually voted Conservative by just under 1,000 votes. However, for the new seat as a whole, it's still not possible to conclude with certainty who won - a ward (Paddock) is split: two thirds of it are in this seat and a third are in Aldridge-Brownhills. If you assume Paddock voted entirely uniformly, then the Conservatives won Walsall and Bloxwich by 100 votes. But you would only need a very slight difference to put Labour ahead across the seat as a whole - western Paddock (which is younger, more working class, and more Muslim) would only have to be 3% more Labour than eastern Paddock for the constituency as a whole to go Labour.
I look forward to next week, when the Rallings and Thrasher figures are due to be published by the broadcasters. It will be interesting to see how their figures vary from those already in this thread...
|
|
Harry Hayfield
Green
Cavalier Gentleman (as in 17th century Cavalier)
Posts: 2,922
|
Post by Harry Hayfield on Jan 15, 2024 10:39:55 GMT
I look forward to next week, when the Rallings and Thrasher figures are due to be published by the broadcasters. It will be interesting to see how their figures vary from those already in this thread...
Next week? I was told they would be published tomorrow (January 16th) |
|
|
|
Post by Pete Whitehead on Jan 15, 2024 11:06:48 GMT
I have now updated my model and notional estimates for the final proposals. These are still based purely on demographic data but now use updated figures from the 2021 Census and analysis is done at a more granular level (Output Areas rather than LSOAs which my old model used). The caveat to this is that these updated estimates are England only - Scottish census data is not yet available and while Welsh census data is accessible, the need to incorporate variables like Welsh speaking complicates the process there.
Full tables are available here - I've given estimated results for 2015, 2017, and 2019 as well as a summary tab showing who won each seat in each election and giving basic totals. As always, it's worth stressing that my model is unable to pick up anomalous sources of local strength which a more traditional notional calculation based on local elections would detect. However, I did cross-check all figures with the Ben Walker model (number 2 above) and in general errors were small, indicating reasonable levels of agreement with a more conventional approach. Using Prediction Competition scoring rules (adding up the absolute difference between my estimates and the Walker notionals), the median discrepancy was 1.74% and in nine tenths of seats our figures differed by less than 6%. In general my model tends to overestimate differences within seats, so the biggest differences were in seats like Croydon West (which essentially combines the strongest areas for Labour in three existing seats) or Ossett and Denby Dale (similarly strongly Conservative areas in three existing seats). There were eight constituencies where my model gave a different winner to Ben's, and most of them are interesting: - Beckenham and Penge (me: Con maj 9, BW: Lab maj 823) - this is a highly marginal seat with large boundary changes so a small difference is to be expected.
- Bedford (me: Lab maj 317, BW: Con maj 194) - here the changes are extremely small with a few small new build estates switching either way. The estate coming in looks to be slightly more Labour friendly than the estates going out (it is more ethnically diverse and more social housing), which is why my model increases Labour's majority. However, it is in a rural (and strongly Conservative) ward in Ben Walker's ward model, and I think he treats split wards as homogeneous - so in his calculations the change helps the Conservatives.
- Cheltenham (me: LD maj 203, BW: Con maj 1,224) - the Lib Dems often do well in relatively working class areas within middle class seats where they win tactical votes from people who would otherwise probably favour Labour. This causes a major ecological fallacy in my model - which generally assumes well educated middle class areas are the strongest for the Lib Dems because that's the trend at constituency level. For example, the only change in Cheltenham is the removal of Springbank ward (a relatively deprived area with a fair amount of social housing), where my model has the Lib Dems on about 20%; in local elections they win by huge margins and took 77% of the vote least time. Removing this obviously hurts the Lib Dems and certainly doesn't flip the seat their way - my model is categorically wrong here.
- Chorley (me: Lab maj 17,055, BW: Oth maj 31,316) - in the modelling stage I ignore the Speaker, but for these estimates I've classed him as Labour. In practice our results are the same (the figures are quite different as I classed the independent as , whereas Ben has classed both Mark Smith and Lindsay Hoyle as "other"). This has a slightly more meaningful impact on South Ribble next door - about 3,000 electors in the Charnock Richard area are moved, and in normal circumstances many of these (probably 1,000-1,200) would vote Conservative; the majority in the new South Ribble is likely to be closer to 20% rather than the 15.7% my estimates suggest.
- Cramlington and Killingworth (me: Con maj 582, BW: Lab maj 1,797) - quite difficult to account for the discrepancy here; most wards are similar in the two models but mine has a far stronger Conservative vote in Killingworth and parts of the Seaton Valley. There are new builds here such as at Holystone, but my model may be overestimating how Conservative these areas are.
- Dagenham and Rainham (me: Con maj 474, BW: Lab maj 992) - the boundary changes here remove one area and add two more. As mentioned above, my model tends to emphasise differences within seats. Therefore Labour do extremely well in (most of) Dagenham whereas the Conservatives run up the numbers in Rainham - of course this is somewhat accurate but probably not to the extent my model suggests. The ward being removed, Chadwell Heath, illustrates this well - Ben has this as 40% Conservative (perhaps picking up on historic local Tory strength here) whereas I have it as more like 32% Conservative. This accounts for most of the difference between the two results; I also have slightly stronger Conservative performances than Ben in the areas added in Valence ward and near Elm Park station.
- Kensington and Bayswater (me: Con maj 220, BW: Lab maj 1,137) - intuitively this seems like a fairly obvious Labour win but my model seems to have them as too weak in the Westminster part (perhaps assuming Karen Buck build up her majority in the Church Street and Harrow Road sections of her seat). It estimates that Labour only won the Bayswater and Lancaster Gate wards by about 500 votes, whereas Ben has them ahead there by about 1,800 which seems more realistic - they need to win by around 750 votes to overturn the Conservative majority in Kensington (this being bolstered a little by some extra Tory territory around Hans Town).
- Walsall and Bloxwich (me: Lab maj 1,442, BW: Con maj 4,181) - this is probably the constituency with the starkest difference between my model and Ben's, with a difference of over 13% (I have a Labour lead of 48-45, he has a Conservative lead of 51-42). While our Walsall South figures are generally quite similar, it's in the former Walsall North wards that the differences really pile up. This is epitomised by the Birchills Leamore ward: I have it as Conservative by 60 votes, compared with Ben having it as Conservative by 1,800. Unusually, this is one seat where we actually know the real result by ward, and generally speaking these are somewhere between the two models - for instance Birchills Leamore actually voted Conservative by just under 1,000 votes. However, for the new seat as a whole, it's still not possible to conclude with certainty who won - a ward (Paddock) is split: two thirds of it are in this seat and a third are in Aldridge-Brownhills. If you assume Paddock voted entirely uniformly, then the Conservatives won Walsall and Bloxwich by 100 votes. But you would only need a very slight difference to put Labour ahead across the seat as a whole - western Paddock (which is younger, more working class, and more Muslim) would only have to be 3% more Labour than eastern Paddock for the constituency as a whole to go Labour.
I look forward to next week, when the Rallings and Thrasher figures are due to be published by the broadcasters. It will be interesting to see how their figures vary from those already in this thread...
My notionals had a similar Conservaticve lead to yours in Kensington & Bayswater although my methodology is more like Ben Walker's than yours. I'm not sure that an 1800 Labour lead in Basyweater and Lancaster Gate is that realistic although I agree that intuitively it feels like the boundary changes should make this a Labour seat. I wonder if Davıd Boothroyd has any insights on that particular area. The Walsall situation shows up the flaws in both methods. Again my own notionals (had I not had the actual results to work with) would show a larger Conservative lead in Walsall & Bloxwich because local elections would tend to show a stronger Conservative vote in the Asian dominated wards of Walsall South (Palfrey, St Matthews etc) than in the Darlaston wards (which we know was not the case). Your own methodology addresses that but in some cases perhaps overcompensates (as in your Cheltenham example) |
|
wysall
Forum Regular
Posts: 326
|
Post by wysall on Jan 15, 2024 16:46:21 GMT
I have now updated my model and notional estimates for the final proposals. These are still based purely on demographic data but now use updated figures from the 2021 Census and analysis is done at a more granular level (Output Areas rather than LSOAs which my old model used). The caveat to this is that these updated estimates are England only - Scottish census data is not yet available and while Welsh census data is accessible, the need to incorporate variables like Welsh speaking complicates the process there.
Full tables are available here - I've given estimated results for 2015, 2017, and 2019 as well as a summary tab showing who won each seat in each election and giving basic totals. As always, it's worth stressing that my model is unable to pick up anomalous sources of local strength which a more traditional notional calculation based on local elections would detect. However, I did cross-check all figures with the Ben Walker model (number 2 above) and in general errors were small, indicating reasonable levels of agreement with a more conventional approach. Using Prediction Competition scoring rules (adding up the absolute difference between my estimates and the Walker notionals), the median discrepancy was 1.74% and in nine tenths of seats our figures differed by less than 6%. In general my model tends to overestimate differences within seats, so the biggest differences were in seats like Croydon West (which essentially combines the strongest areas for Labour in three existing seats) or Ossett and Denby Dale (similarly strongly Conservative areas in three existing seats). There were eight constituencies where my model gave a different winner to Ben's, and most of them are interesting: - Beckenham and Penge (me: Con maj 9, BW: Lab maj 823) - this is a highly marginal seat with large boundary changes so a small difference is to be expected.
- Bedford (me: Lab maj 317, BW: Con maj 194) - here the changes are extremely small with a few small new build estates switching either way. The estate coming in looks to be slightly more Labour friendly than the estates going out (it is more ethnically diverse and more social housing), which is why my model increases Labour's majority. However, it is in a rural (and strongly Conservative) ward in Ben Walker's ward model, and I think he treats split wards as homogeneous - so in his calculations the change helps the Conservatives.
- Cheltenham (me: LD maj 203, BW: Con maj 1,224) - the Lib Dems often do well in relatively working class areas within middle class seats where they win tactical votes from people who would otherwise probably favour Labour. This causes a major ecological fallacy in my model - which generally assumes well educated middle class areas are the strongest for the Lib Dems because that's the trend at constituency level. For example, the only change in Cheltenham is the removal of Springbank ward (a relatively deprived area with a fair amount of social housing), where my model has the Lib Dems on about 20%; in local elections they win by huge margins and took 77% of the vote least time. Removing this obviously hurts the Lib Dems and certainly doesn't flip the seat their way - my model is categorically wrong here.
- Chorley (me: Lab maj 17,055, BW: Oth maj 31,316) - in the modelling stage I ignore the Speaker, but for these estimates I've classed him as Labour. In practice our results are the same (the figures are quite different as I classed the independent as , whereas Ben has classed both Mark Smith and Lindsay Hoyle as "other"). This has a slightly more meaningful impact on South Ribble next door - about 3,000 electors in the Charnock Richard area are moved, and in normal circumstances many of these (probably 1,000-1,200) would vote Conservative; the majority in the new South Ribble is likely to be closer to 20% rather than the 15.7% my estimates suggest.
- Cramlington and Killingworth (me: Con maj 582, BW: Lab maj 1,797) - quite difficult to account for the discrepancy here; most wards are similar in the two models but mine has a far stronger Conservative vote in Killingworth and parts of the Seaton Valley. There are new builds here such as at Holystone, but my model may be overestimating how Conservative these areas are.
- Dagenham and Rainham (me: Con maj 474, BW: Lab maj 992) - the boundary changes here remove one area and add two more. As mentioned above, my model tends to emphasise differences within seats. Therefore Labour do extremely well in (most of) Dagenham whereas the Conservatives run up the numbers in Rainham - of course this is somewhat accurate but probably not to the extent my model suggests. The ward being removed, Chadwell Heath, illustrates this well - Ben has this as 40% Conservative (perhaps picking up on historic local Tory strength here) whereas I have it as more like 32% Conservative. This accounts for most of the difference between the two results; I also have slightly stronger Conservative performances than Ben in the areas added in Valence ward and near Elm Park station.
- Kensington and Bayswater (me: Con maj 220, BW: Lab maj 1,137) - intuitively this seems like a fairly obvious Labour win but my model seems to have them as too weak in the Westminster part (perhaps assuming Karen Buck build up her majority in the Church Street and Harrow Road sections of her seat). It estimates that Labour only won the Bayswater and Lancaster Gate wards by about 500 votes, whereas Ben has them ahead there by about 1,800 which seems more realistic - they need to win by around 750 votes to overturn the Conservative majority in Kensington (this being bolstered a little by some extra Tory territory around Hans Town).
- Walsall and Bloxwich (me: Lab maj 1,442, BW: Con maj 4,181) - this is probably the constituency with the starkest difference between my model and Ben's, with a difference of over 13% (I have a Labour lead of 48-45, he has a Conservative lead of 51-42). While our Walsall South figures are generally quite similar, it's in the former Walsall North wards that the differences really pile up. This is epitomised by the Birchills Leamore ward: I have it as Conservative by 60 votes, compared with Ben having it as Conservative by 1,800. Unusually, this is one seat where we actually know the real result by ward, and generally speaking these are somewhere between the two models - for instance Birchills Leamore actually voted Conservative by just under 1,000 votes. However, for the new seat as a whole, it's still not possible to conclude with certainty who won - a ward (Paddock) is split: two thirds of it are in this seat and a third are in Aldridge-Brownhills. If you assume Paddock voted entirely uniformly, then the Conservatives won Walsall and Bloxwich by 100 votes. But you would only need a very slight difference to put Labour ahead across the seat as a whole - western Paddock (which is younger, more working class, and more Muslim) would only have to be 3% more Labour than eastern Paddock for the constituency as a whole to go Labour.
I look forward to next week, when the Rallings and Thrasher figures are due to be published by the broadcasters. It will be interesting to see how their figures vary from those already in this thread...
Sometimes you can over-account for local factors. Shard End was 52.7% Labour with the Conservatives at 37.1% in 2017 according to the released ward breakdowns; according to Ben Walker's ward results, its successors voted over 70% Labour in 2019. His also give a Conservative majority of 6% in Coseley East when the released ward breakdowns have it at 35%. Also the Holystone newbuilds are solidly Labour. |
|
YL
Non-Aligned
Either Labour leaning or Lib Dem leaning but not sure which
Posts: 4,915
|
Post by YL on Jan 15, 2024 17:35:50 GMT
As always, it's worth stressing that my model is unable to pick up anomalous sources of local strength which a more traditional notional calculation based on local elections would detect. However, I did cross-check all figures with the Ben Walker model (number 2 above) and in general errors were small, indicating reasonable levels of agreement with a more conventional approach. Using Prediction Competition scoring rules (adding up the absolute difference between my estimates and the Walker notionals), the median discrepancy was 1.74% and in nine tenths of seats our figures differed by less than 6%. In general my model tends to overestimate differences within seats, so the biggest differences were in seats like Croydon West (which essentially combines the strongest areas for Labour in three existing seats) or Ossett and Denby Dale (similarly strongly Conservative areas in three existing seats). On the other hand I have looked at a few cases where I think Ben's model may underestimate differences within seats. One, which due to limited boundary changes, isn't very important, is Sheffield South East. Here Electoral Calculus thinks that the Tories carried Beighton and Mosborough, whereas Ben's model has Labour carrying all five wards. Of course we don't know the truth, but my suspicion would be that Labour carried Darnall by so much that given the size of the majority in the constituency as a whole it's very unlikely that they carried all the other wards, and Beighton and Mosborough would be the two most likely to have gone Tory. Beighton and Mosborough are of course two of many wards around the country which have a strong Lib Dem vote locally which doesn't carry over to General Elections. On Cheltenham: I'm sure I remember seeing a comment somewhere that the Lib Dems had done less well than usual in the more working class parts of the town, which wouldn't be surprising given the general context of the election. I can't remember whether Springbank was specifically mentioned, but it's at least plausible that the strong Lib Dem vote there carried across less well to the General Election than it usually does (though perhaps not to the extent that your model thinks). More generally, the existence of places like Beighton and Mosborough and Cleckheaton and Darton East and Knottingley suggests that even in constituencies where the Lib Dems do quite well nationally the patterns in local elections may not tell us so much. (E.g., on the other side of Sheffield, how well do they do in Stannington in General Elections compared with the other four Hallam wards?) |
|
|
|
Post by Pete Whitehead on Jan 15, 2024 17:50:39 GMT
As always, it's worth stressing that my model is unable to pick up anomalous sources of local strength which a more traditional notional calculation based on local elections would detect. However, I did cross-check all figures with the Ben Walker model (number 2 above) and in general errors were small, indicating reasonable levels of agreement with a more conventional approach. Using Prediction Competition scoring rules (adding up the absolute difference between my estimates and the Walker notionals), the median discrepancy was 1.74% and in nine tenths of seats our figures differed by less than 6%. In general my model tends to overestimate differences within seats, so the biggest differences were in seats like Croydon West (which essentially combines the strongest areas for Labour in three existing seats) or Ossett and Denby Dale (similarly strongly Conservative areas in three existing seats). On the other hand I have looked at a few cases where I think Ben's model may underestimate differences within seats. One, which due to limited boundary changes, isn't very important, is Sheffield South East. Here Electoral Calculus thinks that the Tories carried Beighton and Mosborough, whereas Ben's model has Labour carrying all five wards. Of course we don't know the truth, but my suspicion would be that Labour carried Darnall by so much that given the size of the majority in the constituency as a whole it's very unlikely that they carried all the other wards, and Beighton and Mosborough would be the two most likely to have gone Tory. Beighton and Mosborough are of course two of many wards around the country which have a strong Lib Dem vote locally which doesn't carry over to General Elections. On Cheltenham: I'm sure I remember seeing a comment somewhere that the Lib Dems had done less well than usual in the more working class parts of the town, which wouldn't be surprising given the general context of the election. I can't remember whether Springbank was specifically mentioned, but it's at least plausible that the strong Lib Dem vote there carried across less well to the General Election than it usually does (though perhaps not to the extent that your model thinks). More generally, the existence of places like Beighton and Mosborough and Cleckheaton and Darton East and Knottingley suggests that even in constituencies where the Lib Dems do quite well nationally the patterns in local elections may not tell us so much. (E.g., on the other side of Sheffield, how well do they do in Stannington in General Elections compared with the other four Hallam wards?) I think that was iain in response to some of the notionals I posted (not sure if it was 2019 or an earlier election). Certainly there was a partial inversion of the traditional voting patterns in Cheltenham (where the working class, council estate wards have usually been the strongest for the Lib Dems), but I also doubt it was as stark as @bjornhatten recokns |
|
|
|
Post by bjornhattan on Jan 15, 2024 18:12:36 GMT
On the other hand I have looked at a few cases where I think Ben's model may underestimate differences within seats. One, which due to limited boundary changes, isn't very important, is Sheffield South East. Here Electoral Calculus thinks that the Tories carried Beighton and Mosborough, whereas Ben's model has Labour carrying all five wards. Of course we don't know the truth, but my suspicion would be that Labour carried Darnall by so much that given the size of the majority in the constituency as a whole it's very unlikely that they carried all the other wards, and Beighton and Mosborough would be the two most likely to have gone Tory. Beighton and Mosborough are of course two of many wards around the country which have a strong Lib Dem vote locally which doesn't carry over to General Elections. On Cheltenham: I'm sure I remember seeing a comment somewhere that the Lib Dems had done less well than usual in the more working class parts of the town, which wouldn't be surprising given the general context of the election. I can't remember whether Springbank was specifically mentioned, but it's at least plausible that the strong Lib Dem vote there carried across less well to the General Election than it usually does (though perhaps not to the extent that your model thinks). More generally, the existence of places like Beighton and Mosborough and Cleckheaton and Darton East and Knottingley suggests that even in constituencies where the Lib Dems do quite well nationally the patterns in local elections may not tell us so much. (E.g., on the other side of Sheffield, how well do they do in Stannington in General Elections compared with the other four Hallam wards?) I think that was iain in response to some of the notionals I posted (not sure if it was 2019 or an earlier election). Certainly there was a partial inversion of the traditional voting patterns in Cheltenham (where the working class, council estate wards have usually been the strongest for the Lib Dems), but I also doubt it was as stark as @bjornhatten recokns This post was made at the time: Inspirational summary there Iain , but as has been noted elsewhere Brexit has turned these political allegiances on their head . Hester's Way and Springbank were apparently blowouts for CON back in December , incredible to think really. Obviously the term blowout is subjective; if this rumour is true I expect the results in these western estates wouldn't be landslide (as my model projects) but might instead be on the order of: Con 45-50% LD 35-40% Lab 10-15% |
|
jamie
Top Poster
Posts: 7,065
|
Post by jamie on Jan 15, 2024 18:34:35 GMT
Sometimes you can over-account for local factors. Shard End was 52.7% Labour with the Conservatives at 37.1% in 2017 according to the released ward breakdowns; according to Ben Walker's ward results, its successors voted over 70% Labour in 2019. His also give a Conservative majority of 6% in Coseley East when the released ward breakdowns have it at 35%. Also the Holystone newbuilds are solidly Labour. I seemed to have missed these. Have you got a link? |
|
wysall
Forum Regular
Posts: 326
|
Post by wysall on Jan 15, 2024 18:46:06 GMT
Sometimes you can over-account for local factors. Shard End was 52.7% Labour with the Conservatives at 37.1% in 2017 according to the released ward breakdowns; according to Ben Walker's ward results, its successors voted over 70% Labour in 2019. His also give a Conservative majority of 6% in Coseley East when the released ward breakdowns have it at 35%. Also the Holystone newbuilds are solidly Labour. I seemed to have missed these. Have you got a link? |
|
YL
Non-Aligned
Either Labour leaning or Lib Dem leaning but not sure which
Posts: 4,915
|
Post by YL on Jan 15, 2024 18:55:01 GMT
Sometimes you can over-account for local factors. Shard End was 52.7% Labour with the Conservatives at 37.1% in 2017 according to the released ward breakdowns; according to Ben Walker's ward results, its successors voted over 70% Labour in 2019. His also give a Conservative majority of 6% in Coseley East when the released ward breakdowns have it at 35%. To be fair, I don't think any local election based method is going to be anywhere near the actual result in Coseley East. bjornhattan, what do you have there? |
|
|
|
Post by bjornhattan on Jan 15, 2024 19:05:43 GMT
In fairness, those Birmingham Hodge Hill results are pretty astonishing. As I say, my model tends to exaggerate differences within seats and even they don't pick up on that difference between 94% Labour wards in the inner city and 52% Labour wards further out. There's not a particularly steep divide in terms of the variables my model generally looks at (such as home ownership or unemployment) and it looks as if the polarisation here is basically all down to ethnicity. (Incidentally, I saw the same thing when looking at Walsall and Bloxwich - the relatively affluent but majority minority Paddock ward very nearly voted for Labour according to Walsall's official 2019 figures - something both my model and Ben Walker's missed. Is there a bigger ethnic divide in voting patters in the West Midlands than elesewhere in the country?) |
|
|
|
Post by Pete Whitehead on Jan 15, 2024 19:06:14 GMT
Mine is 19% in Coseley East fwiw
|
|
