Who won the local elections - a mathematical odyssey

Well, they all did according to the main parties, and none of them did according to the main parties. This hardly makes our task clearer. Tony Blair, whose heart was evidently not in it, claimed it was a springboard to victory at the next election. The cyanide capsules are out in the Fuhrerbunker as I write, mate.

Commentators on the night (including me, actually, and Nosemonkey) thought Labour were doing better than expected and the Lib Dems worse, with the Tories having a good night of it.

However, watching the results come in is inevitably going to lead to bias - you see a sea of blue and assume the Tories are winning, you don't see many 'Lab lose to Con' flashes and assume that Labour are doing well. Let's be a bit more stringent here. Over the night I amused myself seeing how the parties were doing as a percentage loss in council seats - basically how well they were hanging on to their seats. The results were interesting - early on Labour and the Lib Dems were losing seats at about the same pace, but later on (in the early hours and the following day) the Labour loss rate accelerated and the Lib Dems pulled out of the nosedive slightly. This of course meant that if you watched the first results come in and then listened to the following days spin (Tony Robinson saying 'it's not as bad as some people predicted - some said we'd lose 300-600 seats) you got a better view of Labour's results than if you stayed up all night, after which advanced analysis was probably beyond you.

Time for some numbers - here's the Top Three from the BBC results page:

Party	Gain	Final
CON	911	5315
LAB	-505	1877
LD	-246	2171

2007 Council Elections

Immediately we need to add a column for the number of councillors previously held, plus the percentage change:

Party	Gain	Final	Former	%
CON	911	5315	4404	+21
LAB	-505	1877	2382	-21
LD	-246	2171	2417	-10

2007 Council Elections

This is awfully interesting now. Labour lost councillors at twice the rate of the Lib Dems.
For my next trick we need to go back to the 2003 elections in the same areas (I'm a bit dubious about the merits of comparing disparate elections).

Party	Gain	Final	Former	%
CON	566	4423	3857	+15
LAB	-833	3001	3834	-22
LD	193	2624	2431	+8

2003 Council Elections

So overall:

* The Conservatives have improved twice from a low base.
* Labour are in a spiral dive with the wings coming off
* The Lib Dems, having marched up the hill in the years to 2003, have marched halfway back down. The Grand Old Duke of Ming has some work to do.

As someone will no doubt point out, we have a Labour Government since 2005, despite the 2003 disaster, which lends some credence to the idea that council elections are worth buggery-ding-dong as predictors. Consider the same table as applied to the 2005 General Election:

Party	Gain	Final	Former	%
CON	33	198	165	+20
LAB	-47	356	403	-12
LD	11	62	51	+22

2005 General Election

Tories up, Labour down. At best, Labour did as badly last week from a lower base. Applied to the General Election that's a springboard out of office, not to re-election. Still, having shat on his party, at least Tony won't be around to see it, eh?

UPDATE: A graph showing % loss of councillors at each of the last 9 elections