Tuesday, September 29, 2009

A journey of a thousand miles...

‘Can Labour win?’ is, for party supporters, a deeply depressing question. Look at this chart of opinion polls:

(Monthly averages of YouGov, ICM and Populus polls, adjusted for accuracy at the last election. See footnote for explanation.)

The latest figures (a 17-point Tory lead) would give the Tories a majority of 122, according to Electoral Calculus. Can Labour win? Don’t be a fool. The gap is just too big to close. Go back to your constituencies and prepare for opposition.

So it seems a simple question with a simple and utterly negative answer, but it’s actually composed of a number of smaller questions – whose answers are a bit more hopeful.

For instance, could Labour take, say, two percentage points of support off the Tories? Surely yes. And that would cut the Tory majority by more than half, to 54.

A couple more percentage points? Probably manageable. That would take us – just – into a hung parliament.

Then another two-point swing beyond that? Well, maybe. And if so, Labour would be only three seats behind the Tories.

But let’s not go wild. Each step of political recovery will be harder than the last. And it’s a lot easier to point out that victory and defeat are matters of degree than it is to execute a strategy that could seriously narrow the gap.

And while it’s true that ‘something might turn up’ to help Labour and damage the Tories, it’s as likely that unexpected events would push things the other way. The Tories have proved many times that they’re simply better at responding in a politically astute way (expenses, Gurkhas… the glaring exception being last autumn’s financial turmoil).

I still expect a Tory majority, but if Labour simply tries to do as well as possible – and gives the unforced errors a rest – then a lot of people may be surprised at how far that possibility can go.

Footnote: where the numbers come from
ICM, YouGov and Populus are the only pollsters whose methods are unchanged from before the 2005 election. I took their average figures from during the 2005 campaign and compared those with the result.

ICM understated the Tory and Lib Dem votes by 0.7% each, and overstated Labour by 3%. YouGov overstated the Tory and Labour votes by 0.4% each and the Lib Dems by 0.1%. Populus understated the Tories by 1.9% and the Lib Dems by 1.6%, and overstated Labour by 3.7%.

Then I added and subtracted the relevant errors from each pollster’s monthly average, and then took the average of those. The overall effect is to add about 3% to the Tory lead from the published figures (via UK Polling Report). On the admittedly uncertain assumption that these pollsters are as accurate now as they were in 2005, this should give us a better picture.

2 comments:

Paul said...

a simplistic reading, perhaps, as I tend not to study polls, but surely the big issue here is whether the other parties' rise from 8% to around 18%, judging by the graph, will be sustained in a general election, when the question is suddenly much starker - do you want a Labour or a Tory govt? and of course much will depend on marginal lib dem/tory battles, which will be much less linked to national polling data

anticant said...

Why would anyone WANT Labour to win, with their abysmal record?