Local Council by-elections 9th December 2021

[jm]

It's worth bearing in mind that the majority of votes cast in these by-elections will have been cast before the events of the last week.

[evergreenadam]

Dec 10, 2021 16:37:09 GMT jm said:

It's worth bearing in mind that the majority of votes cast in these by-elections will have been cast before the events of the last week.

So next week will be even more fun!

[jamesdoyle]

Good Week/Bad Week Index (henceforth GWBWI)

Grn +120
Lab +68
LDm +15
Con -306

Remember that last week's -233 for the Cons (on more elections) had been the worst up till now. I've also spent a bit of downtime this afternoon inputting some data prior for July and August, and although not finalised, it's really noticeable how different the picture is for the Tories, then and now.

[tonyhill]

I still think that your model is exaggerating how bad it is for the Conservatives. OK, they lost five seats and gained one, which looks pretty bad, but their vote share was up in four of the seven contests which was better than Labour. Admittedly, there was a lot of interference in the overall picture due to changes in non-party candidature, and I'm sure your point about the change in Conservative fortunes from earlier in the year is correct, but Conservative support is not in freefall, and were they to get shot of Johnson it would almost certainly consolidate again.

[jamesdoyle]

Dec 10, 2021 19:05:51 GMT tonyhill said:

I still think that your model is exaggerating how bad it is for the Conservatives. OK, they lost five seats and gained one, which looks pretty bad, but their vote share was up in four of the seven contests which was better than Labour. Admittedly, there was a lot of interference in the overall picture due to changes in non-party candidature, and I'm sure your point about the change in Conservative fortunes from earlier in the year is correct, but Conservative support is not in freefall, and were they to get shot of Johnson it would almost certainly consolidate again.

Hmmm. Yes their vote share was up in four seats, but in two of those four they lost ground on the party in first place; and in fact lost both those seats. How can it possibly be a good result when you held the seat and now don't?

And in all four seats where their vote share went up, their increased vote share was based  not on taking it from one of the major parties, but:
a: Aughton & Swallowsnest: sweeping up votes from further right parties (RDP, and even then Labour seemed to do better from the redistribution)
b: Kings Hill: sweeping up votes from further right parties (UKIP) and independents
c: West Malling & Leybourne: sweeping up votes from further right parties (UKIP) and independents, and this time it was the LibDems doing better out of it than the Cons
d: Northam: sweeping up votes from further right parties (UKIP) and independents, and here it was the Greens doing better out of it than the Cons.

Maybe rather than my model exaggerating how poor a performance it was, perhaps yours underestimates how poor it was?

[tonyhill]

Well, as I said, there was a lot of interference this week which makes the trends less clear, and I need to do more work on my model so that I have better tracking information, and I hope to be able to input that data over the Christmas break. If I worked in the local government section of Conservative Central Office (assuming they have one) I would not be too dismayed by last night's results though.

[andrew111]

Dec 10, 2021 19:23:39 GMT jamesdoyle said:

Dec 10, 2021 19:05:51 GMT tonyhill said:

I still think that your model is exaggerating how bad it is for the Conservatives. OK, they lost five seats and gained one, which looks pretty bad, but their vote share was up in four of the seven contests which was better than Labour. Admittedly, there was a lot of interference in the overall picture due to changes in non-party candidature, and I'm sure your point about the change in Conservative fortunes from earlier in the year is correct, but Conservative support is not in freefall, and were they to get shot of Johnson it would almost certainly consolidate again.

Hmmm. Yes their vote share was up in four seats, but in two of those four they lost ground on the party in first place; and in fact lost both those seats. How can it possibly be a good result when you held the seat and now don't?

And in all four seats where their vote share went up, their increased vote share was based  not on taking it from one of the major parties, but:
a: Aughton & Swallowsnest: sweeping up votes from further right parties (RDP, and even then Labour seemed to do better from the redistribution)
b: Kings Hill: sweeping up votes from further right parties (UKIP) and independents
c: West Malling & Leybourne: sweeping up votes from further right parties (UKIP) and independents, and this time it was the LibDems doing better out of it than the Cons
d: Northam: sweeping up votes from further right parties (UKIP) and independents, and here it was the Greens doing better out of it than the Cons.

Maybe rather than my model exaggerating how poor a performance it was, perhaps yours underestimates how poor it was?

I have a friend who just works out the weighted vote share in all by-elections and plots an average for each month on a graph. He says it tracks the opinion polls pretty well, but with LD and probably now Greens doing a bit better than the polls all the time. (just like May local election results

[jamesdoyle]

Dec 10, 2021 19:56:01 GMT andrew111 said:

Dec 10, 2021 19:23:39 GMT jamesdoyle said:

Hmmm. Yes their vote share was up in four seats, but in two of those four they lost ground on the party in first place; and in fact lost both those seats. How can it possibly be a good result when you held the seat and now don't?

And in all four seats where their vote share went up, their increased vote share was based  not on taking it from one of the major parties, but:
a: Aughton & Swallowsnest: sweeping up votes from further right parties (RDP, and even then Labour seemed to do better from the redistribution)
b: Kings Hill: sweeping up votes from further right parties (UKIP) and independents
c: West Malling & Leybourne: sweeping up votes from further right parties (UKIP) and independents, and this time it was the LibDems doing better out of it than the Cons
d: Northam: sweeping up votes from further right parties (UKIP) and independents, and here it was the Greens doing better out of it than the Cons.

Maybe rather than my model exaggerating how poor a performance it was, perhaps yours underestimates how poor it was?

I have a friend who just works out the weighted vote share in all by-elections and plots an average for each month on a graph. He says it tracks the opinion polls pretty well, but with LD and probably now Greens doing a bit better than the polls all the time. (just like May local election results

Interesting. How do they weight the vote?

[tonyhill]

Next week promises to be worse for the Conservatives and they ought to lose six out of the seven seats they are defending.

[andrew111]

Dec 10, 2021 19:58:15 GMT jamesdoyle said:

Dec 10, 2021 19:56:01 GMT andrew111 said:

I have a friend who just works out the weighted vote share in all by-elections and plots an average for each month on a graph. He says it tracks the opinion polls pretty well, but with LD and probably now Greens doing a bit better than the polls all the time. (just like May local election results

Interesting. How do they weight the vote?

Just by the number of votes cast in each election.
If you like it is just adding up all the votes cast for each party each month and converting to %. Since by-election triggers are pretty random events it should over a period average out the noise. But it is not going to pick up short term spikes

[hullenedge]

Does anyone recall the Rallings & Thrasher weekly local by-election tracker (published in the New Statesman/Guardian)? They were upfront that 'odd' results could occur like 9% Tory leads in 1998/9.

[grahammurray]

Dec 10, 2021 20:56:28 GMT hullenedge said:

Does anyone recall the Rallings & Thrasher weekly local by-election tracker (published in the New Statesman/Guardian)? They were upfront that 'odd' results could occur like 9% Tory leads in 1998/9.

Yes.  For a while they even made it onto the Friday night edition of Newsnight for a few weeks.

[Deleted]

Dec 10, 2021 20:08:13 GMT andrew111 said:

Dec 10, 2021 19:58:15 GMT jamesdoyle said:

Interesting. How do they weight the vote?

Just by the number of votes cast in each election.
If you like it is just adding up all the votes cast for each party each month and converting to %. Since by-election triggers are pretty random events it should over a period average out the noise. But it is not going to pick up short term spikes

I used to something similar but also work out % change on the last time each seat was contested 

[Davıd Boothroyd]

They still do a regular byelection monitor report for 'First', the LGA magazine for councillors. It's not particularly detailed but if you want to read it go to www.lgafirst.co.uk.

[Andrew_S]

Dec 10, 2021 21:03:59 GMT grahammurray said:

Dec 10, 2021 20:56:28 GMT hullenedge said:

Does anyone recall the Rallings & Thrasher weekly local by-election tracker (published in the New Statesman/Guardian)? They were upfront that 'odd' results could occur like 9% Tory leads in 1998/9.

Yes.  For a while they even made it onto the Friday night edition of Newsnight for a few weeks.

I recall that now you mention it. One of those things I'd never remember unless someone reminded me.

[andrew111]

Dec 10, 2021 23:58:12 GMT Davıd Boothroyd said:

They still do a regular byelection monitor report for 'First', the LGA magazine for councillors. It's not particularly detailed but if you want to read it go to www.lgafirst.co.uk.

Thanks, there is a big Rallings and Thrasher spreadsheet of by elections going back years which has a lot of calculations in it, and could have others. It does include change in % since last election, for example

[grahammurray]

Dec 11, 2021 9:30:33 GMT andrew111 said:

Dec 10, 2021 23:58:12 GMT Davıd Boothroyd said:

They still do a regular byelection monitor report for 'First', the LGA magazine for councillors. It's not particularly detailed but if you want to read it go to www.lgafirst.co.uk.

Thanks, there is a big Rallings and Thrasher spreadsheet of by elections going back years which has a lot of calculations in it, and could have others. It does include change in % since last election, for example

Is there a link to that spreadsheet, Andrew?

[andrew111]

Dec 11, 2021 10:37:44 GMT grahammurray said:

Dec 11, 2021 9:30:33 GMT andrew111 said:

Thanks, there is a big Rallings and Thrasher spreadsheet of by elections going back years which has a lot of calculations in it, and could have others. It does include change in % since last election, for example

Is there a link to that spreadsheet, Andrew?

It is in the LGA website that David Boothroyd posted, in the set of monthly Rallings and Thrasher summaries

[The Bishop]

Dec 10, 2021 19:05:51 GMT tonyhill said:

I still think that your model is exaggerating how bad it is for the Conservatives. OK, they lost five seats and gained one, which looks pretty bad, but their vote share was up in four of the seven contests which was better than Labour. Admittedly, there was a lot of interference in the overall picture due to changes in non-party candidature, and I'm sure your point about the change in Conservative fortunes from earlier in the year is correct, but Conservative support is not in freefall, and were they to get shot of Johnson it would almost certainly consolidate again.

Yes, but simply citing vote share rises/falls without any context is close to meaningless.

In the two Rotherham contests, for example, a formerly significant vote for the populist right fell to almost nothing. There were still swings against the Tories in both.

[ilerda]

There is an argument that two party swing where both top parties increase their vote share at the expense of others is also close to meaningless.