UK - Alternate Coloring

[DistingFlyer]

Here's 1987:

[DistingFlyer]

And 1992:

(As a side note, I've also made a slight change to the Buckingham-Milton Keynes border for 1983 & 1987, which you can now see in the maps for those years)

[DistingFlyer]

1997:

[DistingFlyer]

And here's February 1974:

[DistingFlyer]

Plus the swings of 1979:

[DistingFlyer]

2001:

[DistingFlyer]

Now, to cap things off, 2005 & 2010:

[DistingFlyer]

1992 swings:

[DistingFlyer]

Here are the constituency swings for 1987:

[DistingFlyer]

And for 1997:

[Antonio the Sixth]

Wow, it's impressive to see LibDem gains overshadowed Labour ones in such a sizable share of the country.

[YL]

Wow, it's impressive to see LibDem gains overshadowed Labour ones in such a sizable share of the country.

I assume this partly reflects the usual UK approach to calculating swing, which is to calculate it between the top two parties in any constituency, so that seats where the Tories and Lib Dems are the top two will be either yellow or blue even if there was a big increase in the Labour vote.

E.g. in Skipton & Ripon, there was a 5% swing from Con to Lib Dem, but actually the Lib Dem vote went down: the changes were Con -11.9%, LD -1.8%, Lab +7.8% (the Wikipedia figure is obviously wrong), Referendum +5.9%.  From that you can get a Con to Lab swing of nearly 10%, but because Labour didn't overtake the Lib Dems it'll be the Con to Lib Dem swing which is normally given.

However, there were some constituencies (such as Harrogate & Knaresborough and Sheffield Hallam) where the Lib Dems marshalled the anti-Tory vote and got a big swing themselves, and there was little or no Labour advance.

[Antonio the Sixth]

Wow, it's impressive to see LibDem gains overshadowed Labour ones in such a sizable share of the country.

I assume this partly reflects the usual UK approach to calculating swing, which is to calculate it between the top two parties in any constituency, so that seats where the Tories and Lib Dems are the top two will be either yellow or blue even if there was a big increase in the Labour vote.

So what do you do when the top two parties change between two elections? For example, a seat which LibDems won both times, but with a Torie candidate coming second a 1992 but being supplanted by the Labour guy in 1997.

[YL]

Wow, it's impressive to see LibDem gains overshadowed Labour ones in such a sizable share of the country.

I assume this partly reflects the usual UK approach to calculating swing, which is to calculate it between the top two parties in any constituency, so that seats where the Tories and Lib Dems are the top two will be either yellow or blue even if there was a big increase in the Labour vote.

So what do you do when the top two parties change between two elections? For example, a seat which LibDems won both times, but with a Torie candidate coming second a 1992 but being supplanted by the Labour guy in 1997.

You use the top two at the election at which you're talking about the swing in.  In your example, it would be the swing between Lib Dem and Labour.

(This is why Sheffield Heeley, for example, looks so pale on the map.  The Labour vote went up by 5%, and the Tory vote down by over 10%.  But the Lib Dem vote went up too, by 2.9%, and they overtook the Tories for second place.  So the swing reported is just over 1% Lib Dem to Labour, although it's clear that isn't a very good description of what really happened.)

[minionofmidas]

You get some bizarre results with that sometimes: http://news.bbc.co.uk/2/shared/vote2005/html/252.stm http://news.bbc.co.uk/2/shared/vote2005/html/44.stm

[Filuwaúrdjan]

Though the really traditional way of calculating swing is to ignore entirely parties that aren't Labour or the Tories (normally this would now be called a 'Butler swing' - after David Butler - rather than just 'swing').

[EPG]

That's exactly right. Swing was designed to summarise behaviour in a two-party system, back in the days when England gave 97% of votes to LabCon. It's not great when a third party can intervene, and it's going to give very strange results at the next general election.

[DistingFlyer]

Wow, it's impressive to see LibDem gains overshadowed Labour ones in such a sizable share of the country.

I assume this partly reflects the usual UK approach to calculating swing, which is to calculate it between the top two parties in any constituency, so that seats where the Tories and Lib Dems are the top two will be either yellow or blue even if there was a big increase in the Labour vote.

So what do you do when the top two parties change between two elections? For example, a seat which LibDems won both times, but with a Torie candidate coming second a 1992 but being supplanted by the Labour guy in 1997.

You use the top two at the election at which you're talking about the swing in.  In your example, it would be the swing between Lib Dem and Labour.

(This is why Sheffield Heeley, for example, looks so pale on the map.  The Labour vote went up by 5%, and the Tory vote down by over 10%.  But the Lib Dem vote went up too, by 2.9%, and they overtook the Tories for second place.  So the swing reported is just over 1% Lib Dem to Labour, although it's clear that isn't a very good description of what really happened.)

The only exception to this would be a seat like Oldham East & Saddleworth (or Stevenage in 1983) where the incumbent party drops to third place - in those cases, I'd use Labour-Tory swing even though the Liberals rose to second in both seats. I would not calculate the swing the way they did it in the BBC link provided by only back for the worldcup.

[EPG]

What would a better version of swing look like?

Possibly an index of winnability/safety.

Say party W is the winner at this election, and party N was the next-highest party after W at the previous election (i.e. N either won or came second after W).

The change in winnability/safety would be the average of W's current majority and the gap between N and W at the previous election. It's not quite an average of majorities, because W could have previously finished third, fourth, or nowhere.

[DistingFlyer]

More swings:

2001


2005


2010

[DistingFlyer]

And here are the swings for 1983:

[DistingFlyer]

Here's October 1974:

[Free Bird]

Wouldn't it be "colouring?"

[DistingFlyer]

Sorry for not adding anything new in a while but I've been at sea for some time - here are maps for the 1974 & 1979 elections, showing the Labour & Tory vote in each constituency (whether they won the seats or not):

Feb 1974


Oct 1974


May 1979

[DistingFlyer]

And here are similar maps for the next three Tory victories:

Jun 1983


Jun 1987


Apr 1992