07 April 2010

Playing with the swings... and the polls.... etc.

Only 24 hours into the campaign and I'm already addicted to the Guardian's 3-way swingometer. It's a nice site - the swingometer looks a bit like the rear light assembly on a Mk1 Ford Cortina (which nobody remembers, of course...) and you drag the black circle towards blue for a swing towards the Tories, yellow for Lib Dems. (I'm assuming there won't be a swing towards Labour in the election, surprise surprise).

The 'baseline' result of no swing is complicated by boundary changes this time round which means Labour has a notional majority of 48 over all other parties. For the record, the results last time were roughly speaking:

Labour - 36%
Tory - 33%
Lib Dems - 23%

The polls have, generally speaking, been all over the place these last few days. Yesterday, ICM/Guardian showed a 4-point deficit - 33% Labour, 37% Tory - the lowest on this poll since late 2008. But YouGov in the Sun and Opinium (sounds like opium?) in the Express had a 10-point Tory lead. I'm generally discounting anything published in the Sun or the Express in this election as they are rabidly biased. For example, the YouGov tracker poll a couple of weeks back showing the Tory lead cut to 2 points wasn't even reported in The Sun. Therefore, I'm going to make the assumption that ICM is more accurate than the other two polls and that the 'true' Tory lead is somewhere between 2 and 5 points. I could of course come a total cropper with this assumption but that's what I'm going with for now and we'll see where it takes us.

My prediction at this stage, then, is that the ICM poll is very close to what the final result will be. Let's give Labour an extra percentage point to reflect some pre-election tightening of the race - making for Labour on 34%, Tories on 37%, Lib Dems 20%. From 2005, that gives:
0.5% swing Labour to Tory
3% swing Lib Dem to Tory

Feeding that into the Guardian calculator (or as close as I could - it's very difficult to fine-tune the mouse enough to get the swings exactly right), we get:
Labour - 324 seats (-25)
Tory - 248 seats (+38)
Lib Dem - 48 seats (-14)

That would leave Labour about 1 seat short of an overall majority, but almost certainly able to stay in power, as the likelihood of all other parties being able to co-ordinate on a no confidence vote is slim, to say the least. At least until by-elections start to push Labour further into minority... very much a re-run of the 1974-9 Labour government situation.

This prediction may change if there's an obvious, and strong, movement one way or the other in the polls. But for now, I'm predicting Labour on the cusp of an overall majority. There are of course other factors to consider - e.g. will the swing be uniform or will the Tories perform better in the marginals than elsewhere? I'll come back to that later in the campaign as time permits.


Hymnal said...

Some interesting comment from Nick Cohen on how to interpret the polls here:


If he's right, you're being a bit optimistic on Labour's behalf.

giroscoper said...

IF he's right. But all the pollsters include corrections for non-response and potential bias, which are adjusted after each election. So it's by no means obvious that there's any systematic bias towards Labour.

But it IS clear that there is publication bias in some of the papers that favour the Tories - for example if a YouGov poll shows a small Tory lead the Sun just doesn't bother to report it.

Tiddlemouse said...

I do remember the Ford Cortina - I recently had to make some spare glass for it which was a royal pain.

giroscoper said...

Great stuff. What about the very curvy Mark 3 model? The ones that were available in acid yellow and just fell apart? The seventies were both great and awful at the same time.