Electoral Calculus

[Deleted]

Nov 12, 2019 21:02:35 GMT andrew111 said:

Nov 10, 2019 17:57:50 GMT Sandy said:

The usual caveat about how Electoral Calculus is full of sh*t, but I have also noticed they tend to have the SNP on the high side - so having us on five Scottish seats and several others very close may not be a bad sign.

You think SNP up 0.1% since 2017 is "on the high side"? I would have said it was on the low side..

I was referring to seat numbers.

[manchesterman]

last on of those wholly before the actual election.

November polls average:

CON 41%

LAB 29.9%

LD  15.1%

Electoral Calculus (with the usual caveats) gives us:

CON 339

LAB 221

LD 20

SNP 47

PC 4

Grn 1

NI 18

Con majority of 28

Delicately balanced now in that a tiny swing from this position either way results in either a comfortable 5-year majority or a hung parliament!

[bjornhattan]

Electoral Calculus have updated their maps and statistics for the new election.

I'm mentioning this because they're actually far better at this side of things than predicting elections. Their clustering of voters and analysis of demographics are usually pretty sound.

[jamie]

They clearly struggle with estimating how wards voted when theres been a big swing and seem to assume relative uniformity in movement which leads to obviously safe Labour wards being predicted to be marginal/vote Tory. Also obvious issue of under-contested/indy wards in places like County Durham.

[jamie]

Just had a look at Putney, LOL.

[pragmaticidealist]

I've noticed that the ward-by-ward figures don't actually seem to be an estimation of what happened at the election but based upon a 'prediction' of what will happen at the next election (which is odd, as I don't think there has yet been any published voting intention polls since last week).

[Pete Whitehead]

Dec 19, 2019 13:27:19 GMT pragmaticidealist said:

I've noticed that the ward-by-ward figures don't actually seem to be an estimation of what happened at the election but based upon a 'prediction' of what will happen at the next election (which is odd, as I don't think there has yet been any published voting intention polls since last week).

You have to click on show workings evidently to reveal the estimates of the actual result

[bjornhattan]

Dec 19, 2019 13:27:19 GMT pragmaticidealist said:

I've noticed that the ward-by-ward figures don't actually seem to be an estimation of what happened at the election but based upon a 'prediction' of what will happen at the next election (which is odd, as I don't think there has yet been any published voting intention polls since last week).

If you click on "Show working" you can see their estimates for 2019 (the map on each constituency is based on the 2019 estimates not their "predictions").

There's a second method too, look at new boundaries instead of old ones.

[clyde1998]

My constituency's estimates are based on the 2015 local election results, despite the most recent local election being in May, and the ward boundaries being redrawn since 2015. It leaves the figures looking wrong on the basis of the 2019 local results, a quick look shows the best Green areas in 2019 (even in wards outside the local Green-Lib Dem pact) not being the best Green areas in their estimates; the same being true for the best Lib Dem areas.

I'm guessing other areas are using outdated local election results as their basis which, even discounting the massive realignment since the 2016 referendum, makes them not particularly useful or accurate, IMO, right now.

[jamie]

clyde1998 Its even worse in other councils. Many county/unitary councils are based off 2013!

[bjornhattan]

Dec 19, 2019 16:06:42 GMT jamie said:

clyde1998 Its even worse in other councils. Many county/unitary councils are based off 2013!

It does seem to correct these clear errors by using demographics, though that only partially works.

Though I was impressed by some of their estimates - for example in Blaydon (a seat we both know well), they have Winlaton and High Spen as being about as marginal as the two north Whickham wards. That would make sense based on the kind of realignment we saw, and wouldn't be clear from local results alone.

[Clark]

This site is currently projecting a reduced CON majority of 12 based on recent polling

[nodealbrexiteer]

Sept 2, 2020 21:23:28 GMT Clark said:

This site is currently projecting a reduced CON majority of 12 based on recent polling

Party like it's 2015 

[clyde1998]

[The Bishop]

And still 50-something seats predicted for the SNP, its actually becoming amusing at this point.

[iain]

Oct 18, 2022 11:43:01 GMT The Bishop said:

And still 50-something seats predicted for the SNP, its actually becoming amusing at this point.

Oct 14, 2022 9:53:10 GMT iain said:

Oct 14, 2022 9:33:32 GMT The Bishop said:

There is absolutely no way the SNP get anything like that many seats if Labour basically sweeps England/Wales. In the bin.

Unless you change the Scotland specific prediction, Electoral Calculus has a 'locked in' Scottish prediction regardless what is entered for the national numbers.

Currently these are:
SNP 45%, Lab 30.7%, Con 14%, LD 7%, Grn 1%


Which results in:
SNP gaining all 6 Tory seats
SNP gaining Caithness and NE Fife from Lib Dem
Labour gaining East Lothian, Glasgow NE, Kirkcaldy, Rutherglen from SNP

Whatever national numbers you enter, the Scottish prediction will not change.

[willpower3]

Oct 18, 2022 11:43:01 GMT The Bishop said:

And still 50-something seats predicted for the SNP, its actually becoming amusing at this point.

The Lib Dem figure is even more absurd.

[The Bishop]

Oct 18, 2022 12:24:09 GMT iain said:

Oct 18, 2022 11:43:01 GMT The Bishop said:

And still 50-something seats predicted for the SNP, its actually becoming amusing at this point.

Oct 14, 2022 9:53:10 GMT iain said:

Unless you change the Scotland specific prediction, Electoral Calculus has a 'locked in' Scottish prediction regardless what is entered for the national numbers.

Currently these are:
SNP 45%, Lab 30.7%, Con 14%, LD 7%, Grn 1%


Which results in:
SNP gaining all 6 Tory seats
SNP gaining Caithness and NE Fife from Lib Dem
Labour gaining East Lothian, Glasgow NE, Kirkcaldy, Rutherglen from SNP

Whatever national numbers you enter, the Scottish prediction will not change.

Well that is obviously pretty nonsensical anyway, but even a split of SNP 45 Lab 30 surely wouldn't give as lopsided a picture as they are claiming.

[michaelarden]

Oct 19, 2022 10:30:24 GMT The Bishop said:

Oct 18, 2022 12:24:09 GMT iain said:

Whatever national numbers you enter, the Scottish prediction will not change.

Well that is obviously pretty nonsensical anyway, but even a split of SNP 45 Lab 30 surely wouldn't give as lopsided a picture as they are claiming.

It's a tricky one though. With the SNP on 45% it means they're going to be over 40% in pretty much every constituency - usually enough in a 4ish party system to win. Labour got 7 seats on 27% in 2017 - when the SNP were on 37%. With the SNP 15 points ahead rather than 10 under these figures it's difficult to see Labour getting into double figures in terms of seats. And I guess much of the swing to Labour is from Conservative unionist voters - which helps the SNP win in the 6 Tory seats and is only half as effective in picking up SNP central belt seats. 

Think 52 seats is a bit on the high side but not out of the question but something like Con 0, Libs 2-3, Lab 6-11 and SNP 45-51 seems pretty reasonable on these sort of figures. If the SNP fall back below the low 40s then it's game on for Labour in Scotland - but with such a block vote for the SNP, Labour are up against it.   

[afleitch]

Oct 19, 2022 12:08:59 GMT michaelarden said:

Oct 19, 2022 10:30:24 GMT The Bishop said:

Well that is obviously pretty nonsensical anyway, but even a split of SNP 45 Lab 30 surely wouldn't give as lopsided a picture as they are claiming.

It's a tricky one though. With the SNP on 45% it means they're going to be over 40% in pretty much every constituency - usually enough in a 4ish party system to win. Labour got 7 seats on 27% in 2017 - when the SNP were on 37%. With the SNP 15 points ahead rather than 10 under these figures it's difficult to see Labour getting into double figures in terms of seats. And I guess much of the swing to Labour is from Conservative unionist voters - which helps the SNP win in the 6 Tory seats and is only half as effective in picking up SNP central belt seats. 

Think 52 seats is a bit on the high side but not out of the question but something like Con 0, Libs 2-3, Lab 6-11 and SNP 45-51 seems pretty reasonable on these sort of figures. If the SNP fall back below the low 40s then it's game on for Labour in Scotland - but with such a block vote for the SNP, Labour are up against it.   

To add to this,

In terms of flow of the vote from the five recent published polls, rather than direct swing, the SNP and Labour have the same retention rate of 86% and flow to each other of 9%. Numerically yes, there are more SNP to Labour switchers as there are more SNP voters to start with. A direct flow from Tory to Labour (about 25% based on the polls) disproportionally would advantage Labour more in seats where they are in third or even fourth where there is a sizable Tory share. Movement in Lib Dems depends on each seat.

So based on purely on that model, the only seats Labour win are Kirkcaldy and East Lothian (currently Alba held) and the SNP take every seat from the Tories.