|
|
Post by afleitch on Apr 30, 2021 16:40:16 GMT
|
|
J.G.Harston
Lib Dem
Leave-voting Brexit-supporting Liberal Democrat
Posts: 14,771 
Member is Online
|
Post by J.G.Harston on Apr 30, 2021 16:53:45 GMT
Interesting. I went to the website all prepared to be disparaging "ha! we have actual real polling district data". But I was surprised how close their statistical modelling was to my actual box counts. |
|
J.G.Harston
Lib Dem
Leave-voting Brexit-supporting Liberal Democrat
Posts: 14,771
Member is Online
|
Post by J.G.Harston on Apr 30, 2021 16:56:41 GMT
But looking at Sheffield shows how sharply some areas vote differently for Parliament and for Council.
|
|
|
|
Post by afleitch on Apr 30, 2021 17:09:12 GMT
But looking at Sheffield shows how sharply some areas vote differently for Parliament and for Council. There was work done by Danny Dorling et al looking at 1992 notionals and testing the established method (local election patterns) against regression analysis. They looked at changes between party shares in seats with no or little change v significant change and noted error which they wanted to correct by using a different method. When I looked ages ag at the actual ward data for 2007 vs a dummy run of using local election results from the same day you could see the limits to the established method. Regression works well and is much easier to do, even if the results might seem a little counter intuitive to local election results. |
|
|
|
Post by greatkingrat on Apr 30, 2021 17:59:33 GMT
Am I missing something? I can see the maps by constituency for each election, but there doesn't appear to be any ward data?
|
|
|
|
Post by bjornhattan on Apr 30, 2021 18:46:09 GMT
That's quite something - you have to give them credit for trying and I appreciate the fact their estimates go back to 2010 (which is particularly interesting in Scotland and seats which formerly had a strong Lib Dem challenge).
It does seem to get most things right, and it feels a little churlish to point out the mistakes it's made. But there are a few which stand out a mile: they have Labour winning deeply rural wards like Wolds in Ryedale and Worth Valley in Bradford. I don't think it would work because of their modelling method, but I would suggest incorporating population density when modelling the main parties. I'm also concerned about how they use seperate models for each region. That will have benefits in picking up trends which might show up differently across the country but could lead to overfitting if they're not careful. Still, I think they've done a remarkably good job considering just how ideosyncratic some seats can be!
|
|
|
|
Post by hullenedge on Apr 30, 2021 18:51:32 GMT
Interesting but incredible. Some Calderdale wards are way out. Queensbury too. Turnouts exceeding 100% for the Referendum etc. edgbaston is unlikely to agree with their Kirklees estimates. |
|
|
|
Post by No Offence Alan on Apr 30, 2021 19:04:10 GMT
Am I missing something? I can see the maps by constituency for each election, but there doesn't appear to be any ward data? On the front page leantossup.ca/uk-ward-election-atlas/wait for the spinny thing around the little "x" to stop spinning it may take a while |
|
|
|
Post by bjornhattan on Apr 30, 2021 19:05:10 GMT
Interesting but incredible. Some Calderdale wards are way out. Queensbury too. Turnouts exceeding 100% for the Referendum etc. edgbaston is unlikely to agree with their Kirklees estimates. I did notice they had Hebden Bridge down as an unlikely Conservative bastion, which seems completely implausible. |
|
|
|
Post by hullenedge on Apr 30, 2021 19:07:00 GMT
Interesting but incredible. Some Calderdale wards are way out. Queensbury too. Turnouts exceeding 100% for the Referendum etc. edgbaston is unlikely to agree with their Kirklees estimates. I did notice they had Hebden Bridge down as an unlikely Conservative bastion, which seems completely implausible. Looking at the census stats/home ownership they'd be right. Knowing the residents...definitely not one of our bastions. |
|
edgbaston
Labour
Posts: 4,376
Member is Online
|
Post by edgbaston on Apr 30, 2021 19:53:33 GMT
Interesting but incredible. Some Calderdale wards are way out. Queensbury too. Turnouts exceeding 100% for the Referendum etc. edgbaston is unlikely to agree with their Kirklees estimates. Well I’d love to have a look but the website keeps crashing! Is the entirety of vote UK trying to load the page? |
|
|
|
Post by hullenedge on Apr 30, 2021 20:00:03 GMT
Interesting but incredible. Some Calderdale wards are way out. Queensbury too. Turnouts exceeding 100% for the Referendum etc. edgbaston is unlikely to agree with their Kirklees estimates. Well I’d love to have a look but the website keeps crashing! Is the entirety of vote UK trying to load the page? Probably. Denby Dale 2019 Con 5251 Lab 2575! |
|
edgbaston
Labour
Posts: 4,376
Member is Online
|
Post by edgbaston on Apr 30, 2021 20:44:22 GMT
Well I’d love to have a look but the website keeps crashing! Is the entirety of vote UK trying to load the page? Probably. Denby Dale 2019 Con 5251 Lab 2575! I’m guessing they are massively underestimating the Tory vote in Dewsbury east then. |
|
|
|
Post by hullenedge on Apr 30, 2021 21:02:12 GMT
Probably. Denby Dale 2019 Con 5251 Lab 2575! I’m guessing they are massively underestimating the Tory vote in Dewsbury east then. Lab 4994 Con 3566 |
|
|
|
Post by Pete Whitehead on Apr 30, 2021 21:19:01 GMT
Looking at the Hertfordshire results, its total bollocks to be honest
|
|
|
|
Post by lancastrian on Apr 30, 2021 21:26:01 GMT
I can't see it either, but on the section of map on the how it works page Hebden Bridge really stands out. Not sure why Worth Valley goes the other way. Some of the wards Labour get in Lancashire seem rather odd. Maybe the I'm underestimating the Brexit effect but it feels like the Tories are overstated in some (but by no means all) urban working class wards and underestimated in traditionally stronger areas.
|
|
Deleted
Deleted Member
Posts: 0
|
Post by Deleted on Apr 30, 2021 21:45:43 GMT
I admire what they've done but some of their Stroud estimates are.....off |
|
iain
Lib Dem
Posts: 11,435
|
Post by iain on Apr 30, 2021 21:59:06 GMT
I admire what they've done but some of their Stroud estimates are.....off Yeah, I took a quick look at Gloucestershire and concluded it was complete crap |
|
jamie
Top Poster
Posts: 7,054
|
Post by jamie on Apr 30, 2021 22:02:09 GMT
They have the Ponteland wards as all voting leave, which strikes me as... dubious. CC bjornhattan |
|
Deleted
Deleted Member
Posts: 0
|
Post by Deleted on Apr 30, 2021 22:05:03 GMT
I admire what they've done but some of their Stroud estimates are.....off Yeah, I took a quick look at Gloucestershire and concluded it was complete crap Obviously I can only really speak for my own patch but the Tory wins in some of the Stroud wards (except F&P), Dursley voting remain and Painswick and Hardwicke voting Labour in 2010 are fantasy land. What did you make of the Cheltenham results? The idea of the Tories winning 50%+ in Oakley and St Paul's seem fairly ridiculous but I remember a few people saying that the GE produced some surprising results in areas like that |
|
