Local Council by-elections 23rd February 2023

[Tony Otim]

Actually I don't think the transfers are particularly that surprising or interesting here. Once Labour got that close to the SNP on first preferences with a sizeable Lib Dem and Conservative vote to redistribute they were always strong favourites. The things to really not here are the significant fall in the SNP vote - whilst they clearly lost some to Labour and Alba, on raw numbers they also lost a lot to not voting at all. What those voters do in a general election may be relevant. Also, the Conservative vote held up well under the circumstances - that will be encouraging for them. Be interesting to see if it's replicated in the other elections to come.

[clyde1998]

Feb 24, 2023 17:47:32 GMT johnloony said:

So the SNP got 91 out of 191 votes from Alba, and 48 out of 121 from Green. Those proportions are about what I would have expected. It is too naive to think that the SNP would (or “should”) get all (or most) votes from Alba, just because of where the Alba party split from. The very fact that Alba split away from the SNP at all is a clue that there is a reason for not simply giving their votes back when it comes to transfers. (Similarly I am reminded of a transfer rate of only about 1-in-4 votes being transferred mutually to and from the SSP and Solidarity about 15 or whenever years ago).

From the 'limited' information we've got so far, Alba's transfers to the SNP were the highest of any single party to another single party, with that figure being south of 50% - which really tells a lot about how transfers aren't monolithic. I'd want to see the csv file of preferences to see what the Alba transfers would've been without any party being excluded. That transfer rate does seem to be broadly in-line with the Alba's national second preferences from last year's local elections (about 37% going to the SNP in wards where all five Holyrood parties were standing; 24% went to unionist parties).

Of course at a local election different issues are at play than a national election, so transfer patterns could be very different between the two. The above article shows a quarter of SNP second preferences were for unionist parties (where both the Greens and Alba were standing).

[jamesdoyle]

Feb 24, 2023 16:51:10 GMT nobodyimportant said:

Feb 24, 2023 12:31:06 GMT jamesdoyle said:

GWBWI

PC +41
Lab +38
Con +3
Grn -3
LDm -4
SNP -16

ASV

PC +0.7
Lab +0.6
Con +0.0
Grn +0.0
LDm -0.1
SNP -0.3

So the Conservative score is positive when they went from a weak result to no candidate in Wales and lost vote share in Scotland, and the Lib Dem score is negative when we went from no candidate to a weak result in Wales and gained vote share in Scotland.

Neither party scores anything, positive or negative, in the Wrexham contest because of going from no candidate to having a candidate, or vice versa. In the first iterations of GWBWI, it became apparent that these situations gave very distorted results, so they are zeroed out. 
In Scotland - I use average vote, and that has the LibDems losing vote share very slightly, while the Tory average goes up - but is counterbalanced for GWBWI scores by losing the seat. And I've got no idea how BBS calculated those vote swings, because they don't correspond to anything I can work out - top vote, average vote, anything.

[Tony Otim]

Feb 24, 2023 18:26:33 GMT jamesdoyle said:

Feb 24, 2023 16:51:10 GMT nobodyimportant said:

So the Conservative score is positive when they went from a weak result to no candidate in Wales and lost vote share in Scotland, and the Lib Dem score is negative when we went from no candidate to a weak result in Wales and gained vote share in Scotland.

Neither party scores anything, positive or negative, in the Wrexham contest because of going from no candidate to having a candidate, or vice versa. In the first iterations of GWBWI, it became apparent that these situations gave very distorted results, so they are zeroed out. 
In Scotland - I use average vote, and that has the LibDems losing vote share very slightly, while the Tory average goes up - but is counterbalanced for GWBWI scores by losing the seat. And I've got no idea how BBS calculated those vote swings, because they don't correspond to anything I can work out - top vote, average vote, anything.

How can you use average vote for an STV election? I'm confused - then you end up halving the SNP vote - because c.2600 voters first preferenced them last May split between 2 candidates, not 1300 voters 1st preferencing both candidates?

There aren't really competing methods for working out voteshare in an STV election unlike multi-member FPTP, you just add the votes of all the candidates for that party, which for 2022

gives:
SNP 41.3;
Con 26.2;
Lab 17.8;
LD 8.8
Grn 4.5
SFP 1.4

Matching the BBS vote changes

[jamesdoyle]

Feb 24, 2023 18:29:09 GMT Tony Otim said:

Feb 24, 2023 18:26:33 GMT jamesdoyle said:

Neither party scores anything, positive or negative, in the Wrexham contest because of going from no candidate to having a candidate, or vice versa. In the first iterations of GWBWI, it became apparent that these situations gave very distorted results, so they are zeroed out. 
In Scotland - I use average vote, and that has the LibDems losing vote share very slightly, while the Tory average goes up - but is counterbalanced for GWBWI scores by losing the seat. And I've got no idea how BBS calculated those vote swings, because they don't correspond to anything I can work out - top vote, average vote, anything.

How can you use average vote for an STV election? I'm confused - then you end up halving the SNP vote - because c.2600 voters first preferenced them last May split between 2 candidates, not 1300 voters 1st preferencing both candidates?

It's not ideal, but the complications arising from having a different system in the spreadsheet for a very small number of STV by-elections where most parties game the system anyway would be enormous. And most of the time - *most* of the time - it makes no major difference.

[Tony Otim]

Feb 24, 2023 18:34:03 GMT jamesdoyle said:

Feb 24, 2023 18:29:09 GMT Tony Otim said:

How can you use average vote for an STV election? I'm confused - then you end up halving the SNP vote - because c.2600 voters first preferenced them last May split between 2 candidates, not 1300 voters 1st preferencing both candidates?

It's not ideal, but the complications arising from having a different system in the spreadsheet for a very small number of STV by-elections where most parties game the system anyway would be enormous. And most of the time - *most* of the time - it makes no major difference.

Surely you just need to add the votes of candidates from the same party together before you enter them in the spreadsheet and your problem is solved?

[jamesdoyle]

Feb 24, 2023 18:39:06 GMT Tony Otim said:

Feb 24, 2023 18:34:03 GMT jamesdoyle said:

It's not ideal, but the complications arising from having a different system in the spreadsheet for a very small number of STV by-elections where most parties game the system anyway would be enormous. And most of the time - *most* of the time - it makes no major difference.

Surely you just need to add the votes of candidates from the same party together before you enter them in the spreadsheet and your problem is solved?

To coin a phrase, it's a bit more complicated than that

[No Offence Alan]

Feb 24, 2023 18:54:52 GMT jamesdoyle said:

Feb 24, 2023 18:39:06 GMT Tony Otim said:

Surely you just need to add the votes of candidates from the same party together before you enter them in the spreadsheet and your problem is solved?

To coin a phrase, it's a bit more complicated than that

No, it is really not.

[Deleted]

Feb 24, 2023 17:50:47 GMT michaelarden said:

Feb 24, 2023 17:26:13 GMT YL said:

	Count 1	McLean transfers	Count 2	Family transfers	Count 3	Green transfers	Count 4	Alba transfers	Count 5	LD transfers	Count 6	Con transfers	Count 7
Lab	1227	6	1233	22	1255	27	1282	24	1306	153	1459	512	1971
SNP	1455	6	1461	4	1465	48	1513	91	1604	70	1674	55	1729
Con	1190	2	1192	9	1201	4	1205	11	1216	94	1310
Lib Dem	452	8	460	4	464	16	480	11	491
Alba	178	3	181	2	183	8	191
Green	111	5	116	5	121
Family	60	8	68
McLean	52
Non-transferable		14	14	22	36	18	54	54	108	174	282	743	1025

Interesting transfers - particularly Alba and the Tories.

Also that a plurality of transferred SFP votes went to Labour, and that figure was over twice the next highest

[batman]

Feb 24, 2023 17:47:32 GMT johnloony said:

So the SNP got 91 out of 191 votes from Alba, and 48 out of 121 from Green. Those proportions are about what I would have expected. It is too naive to think that the SNP would (or “should”) get all (or most) votes from Alba, just because of where the Alba party split from. The very fact that Alba split away from the SNP at all is a clue that there is a reason for not simply giving their votes back when it comes to transfers. (Similarly I am reminded of a transfer rate of only about 1-in-4 votes being transferred mutually to and from the SSP and Solidarity about 15 or whenever years ago).

this comment totally lacks lunacy, but instead is sensible, well-informed and well-considered. 

[andrewp]

Feb 24, 2023 19:17:55 GMT batman said:

Feb 24, 2023 17:47:32 GMT johnloony said:

So the SNP got 91 out of 191 votes from Alba, and 48 out of 121 from Green. Those proportions are about what I would have expected. It is too naive to think that the SNP would (or “should”) get all (or most) votes from Alba, just because of where the Alba party split from. The very fact that Alba split away from the SNP at all is a clue that there is a reason for not simply giving their votes back when it comes to transfers. (Similarly I am reminded of a transfer rate of only about 1-in-4 votes being transferred mutually to and from the SSP and Solidarity about 15 or whenever years ago).

this comment totally lacks lunacy, but instead is sensible, well-informed and well-considered. 

Some of the PR backers on Twitter don’t understand at all that even when a party is close to another party there is never a block of 2nd preference votes which just move across.  The Alba example is similar to the assumption that 100% of UKIP voters would have a 2nd preference of the Tories and the assumption that every Green and Lib Dem voter would cast a 2nd preference vote for Labour always amuses me.

[andrewteale]

Feb 24, 2023 17:26:13 GMT YL said:

	Count 1	McLean transfers	Count 2	Family transfers	Count 3	Green transfers	Count 4	Alba transfers	Count 5	LD transfers	Count 6	Con transfers	Count 7
Lab	1227	6	1233	22	1255	27	1282	24	1306	153	1459	512	1971
SNP	1455	6	1461	4	1465	48	1513	91	1604	70	1674	55	1729
Con	1190	2	1192	9	1201	4	1205	11	1216	94	1310
Lib Dem	452	8	460	4	464	16	480	11	491
Alba	178	3	181	2	183	8	191
Green	111	5	116	5	121
Family	60	8	68
McLean	52
Non-transferable		14	14	22	36	18	54	54	108	174	282	743	1025

Trying to model these votes for an ordinary four-seat election is giving me a headache. The top three parties all have a quota each: the SNP have 1.54 quotas, Lab 1.30, C 1.26 and LD 0.48. At the moment I have the final seat as too close to call between the SNP and Labour. If the Lib Dems can stay ahead of the second Conservative candidate then they might be in the mix for the final seat as well, but I'd have to see the preference profile to call that.

[J.G.Harston]

Feb 24, 2023 19:28:07 GMT andrewp said:

Feb 24, 2023 19:17:55 GMT batman said:

this comment totally lacks lunacy, but instead is sensible, well-informed and well-considered. 

Some of the PR backers on Twitter don’t understand at all that even when a party is close to another party there is never a block of 2nd preference votes which just move across.  The Alba example is similar to the assumption that 100% of UKIP voters would have a 2nd preference of the Tories and the assumption that every Green and Lib Dem voter would cast a 2nd preference vote for Labour always amuses me.

The SNP is even more a party of internal coalitions than "normal" parties, as their sole unifiying factor is independence. You can be a free-market indy, a state control indy, a gay rights indy, a traditional values indy, a laissez-faire indy. As discussed before, in an independent Scotland the SNP would fragment into its component "sub-parties". How many Sinn Féin governments has Ireland had?

[johnloony]

Feb 24, 2023 19:28:07 GMT andrewp said:

Feb 24, 2023 19:17:55 GMT batman said:

this comment totally lacks lunacy, but instead is sensible, well-informed and well-considered. 

Some of the PR backers on Twitter don’t understand at all that even when a party is close to another party there is never a block of 2nd preference votes which just move across.  The Alba example is similar to the assumption that 100% of UKIP voters would have a 2nd preference of the Tories and the assumption that every Green and Lib Dem voter would cast a 2nd preference vote for Labour always amuses me.

There is a similar naive mentality among some centre-left people who seem to assume that:

(1) the Lib Dems will always, and in all circumstances, prefer Labour rather than Conservative
(2) the Lib Dems have an automatic duty to support / prop up a minority Labour government in the event of a hung parliament

and then go further to think that:

(3) every election (if held by PR) would result in people voting for the same parties, in the same proportions, as they did in FPTP elections

and that therefore

(4) PR would have resulted in a permanent progressive centre-left coalition between Labour and Liberals / SDP / LD ever since WW2.

Some of those people never seem to have considered the reality that people would vote in different numbers for a wider range of parties, and that coalitions would vary according to circumstances. Those sorts of people were the ones who were bewildered and uncomprehending when the Con/LD coalition was formed in 2010, as if they thought it was automatically some sort of betrayal.

[batman]

Indeed, you're totally right about 1) just for starters. This is one of the major reasons why a "Progressive Alliance" just can't work - while Labour supporters would overwhelmingly vote Lib Dem against Conservative if it came to it, the reverse just isn't the case to anything like the same degree. But there are other major reasons too. What would be the point in Lib Dem candidates being told to pull out if more of their supporters voted Conservative than Labour? In some constituencies, I think that's not as improbable as some think it is.

[johnloony]

Feb 24, 2023 18:26:33 GMT jamesdoyle said:

Feb 24, 2023 16:51:10 GMT nobodyimportant said:

So the Conservative score is positive when they went from a weak result to no candidate in Wales and lost vote share in Scotland, and the Lib Dem score is negative when we went from no candidate to a weak result in Wales and gained vote share in Scotland.

Neither party scores anything, positive or negative, in the Wrexham contest because of going from no candidate to having a candidate, or vice versa. In the first iterations of GWBWI, it became apparent that these situations gave very distorted results, so they are zeroed out. 
In Scotland - I use average vote, and that has the LibDems losing vote share very slightly, while the Tory average goes up - but is counterbalanced for GWBWI scores by losing the seat. And I've got no idea how BBS calculated those vote swings, because they don't correspond to anything I can work out - top vote, average vote, anything.

Why on Earth would you want to do that?

[grahammurray]

Feb 24, 2023 22:14:35 GMT batman said:

Indeed, you're totally right about 1) just for starters. This is one of the major reasons why a "Progressive Alliance" just can't work - while Labour supporters would overwhelmingly vote Lib Dem against Conservative if it came to it, the reverse just isn't the case to anything like the same degree. But there are other major reasons too. What would be the point in Lib Dem candidates being told to pull out if more of their supporters voted Conservative than Labour? In some constituencies, I think that's not as improbable as some think it is.

A simple and obvious solution if that's the case.  Labour stand down in Con/Lib Dem seats.

[timmullen1]

Feb 24, 2023 19:28:07 GMT andrewp said:

Feb 24, 2023 19:17:55 GMT batman said:

this comment totally lacks lunacy, but instead is sensible, well-informed and well-considered. 

Some of the PR backers on Twitter don’t understand at all that even when a party is close to another party there is never a block of 2nd preference votes which just move across.  The Alba example is similar to the assumption that 100% of UKIP voters would have a 2nd preference of the Tories and the assumption that every Green and Lib Dem voter would cast a 2nd preference vote for Labour always amuses me.

I know the system is somewhat different but the principle holds, Antony Green has calculated (prior to last year’s election for which I’ve not seen an analysis) that Labor pick up somewhere in the region of 45% of Pauline Hanson’s One Nation Party second preferences despite the obvious chasm in policy terms between the two parties.

[carlton43]

Feb 24, 2023 16:05:17 GMT minionofmidas said:

Feb 24, 2023 15:37:08 GMT yellowperil said:

The Lab majority at stage 7 was rather bigger than I had expected- indeed I had on balance thought the SNP would scrape home, and indeed that had been the majority view on the prediction competition. Neverthless several of us had qualified that prediction by saying it would be likely very close. In the event,a majority of 242 wasn't that close, and it was predominantly down to the 5012 Con>Lab transfers at stage 7.

Or rather the massive 457 netted at that stage. Really not a lot of Tory-Tartantory transfers.

I suppose it does indicate how for many Scottish Tories the constitutional question trumps everything else?

One further thought- had the Tories stayed ahead of Labour in the early stages- and there really wasn't that much in it- I do wonder what the result would have been if it had been Labour going out at stage 7. My guess is that more Labour votes would have transferred to SNP and probably the latter would have got home reasonably comfortably? 

Probably. But note Labour pulling gradually away from the Tories.

I am acquainted with that area but do not have the gut feel that our Kincardinshire/Fife correspondent might care to add.  But it occurs to me that both SNP and Conservatives will have suffered a significant loss of votes because of the names of their candidates. 

[riccimarsh]

Feb 24, 2023 14:12:45 GMT aidypiez said:

Feb 24, 2023 14:09:49 GMT John Chanin said:

As usual with Scottish elections what we need to see is the transfers. It is a constant annoyance that they are so hard to come by.

Ask and ye shall receive...

Another couple of rounds and "Didn't Transfer" could have sneaked a victory here.