Poll positioning

by Stephen Tall on July 27, 2006

The headline results in today’s ICM poll for thegrauniad don’t make the happiest reading for the Lib Dems – we’re at 17%, less than half Labour’s rating, and trailing the Tories by over 20%. But to live your life by the polls is to condemn yourself to politico-manic depression.

It’s one of the features of the hyper-kinetic, hyper-connected digital age we live in that each and every poll is poured over by geek-obsessives slavering over the latest data set, trying to interpret, extrapolate and predict. How useful is their analysis? Let’s take a look at a recent historical example.

In April 1997, two consecutive ICM polls (taken just a fortnight away from the historic Labour landslide so many of us cheered in) showed the following:

13th-14th April, 1997: Con 31%, Lab 45%, Lib Dem 19%
20th-21st April, 1997: Con 37% (+6%), Lab 42% (-3%), Lib Dem 14% (-5%)

What would the contributors to, for example, PoliticalBetting.com make of such a poll in today’s climate? “Tories on the up, as Lib Dems slump.” Would Iain Dale be able to avoid creaming himself at the prospect of the Lib Dems losing some of the 23 seats the party then held? What angst would be provoked among Lib Dem bloggers, desperate to see the party’s election campaign gain traction? Would there be criticism of the leadership, and calls for the campaign team to be replaced?

Too many hypotheticals; let’s deal in reality. This is how The Observer reported the Lib Dems’ chances, on 16th April, 1997:

[Paddy Ashdown’s] party stands at around 15 per cent in the polls [and] unless his recent improvement accelerates, his gains are likely to be few and a large Labour majority may leave the Liberal Democrats as marginal as ever at Westminster.

Well, half right, I suppose – the Lib Dem actually doubled their number of MPs, but the size of Labour’s majority thwarted any prospect (however remote it was in reality) of the Ashdown-Blair flirtation getting to first base, let alone being consummated.

It’s often forgotten today how sceptical many people were that Labour would really win in 1997. The spectre of John Major’s surprise 1992 triumph, when he confounded the pollsters, still hung heavy. I won a sweep-stake by predicting a Labour majority of 146, which was 33 seats shy of Mr Blair’s eventual margin of victory. No-one else guessed three-figures, and several pessimists felt sure the Tories would pull a rabbit out of the hat once again.

Pollsters will rightly note that they do not pretend they can predict results – all they can do is take a snapshot of public opinion, with a 95% confidence level that their findings are an accurate reflection of the national mood to within a margin of error of +/-3%.

But both old and new media are often less than scrupulous in recognising polling’s statistical frailties, while the 24×7 news agenda’s voracious appetite can only be satisfied by whipping up a morsel of bad news into a three-course crisis to be feasted upon.

All this displacement activity is, of course, merely distracting from the fact that no-one has a bloody clue when the next general election might be, or what will happen when it’s called – least of all Gordon Brown, David Cameron or Ming Campbell.

It also absolves the media of their responsibility to have to focus on the boring crap – like, y’know, public policy – when they could be talking shit about who’s hot and who’s not.