WITHIN the MARGIN of ERROR

I am hugely enjoying the contortions through which the members of the commentariat are putting themselves to make any kind of comment on the recent opinion poll findings. Said pundits decided more than two years ago that Labour “couldn’t” win the next election, that David Cameron was a shoo-in, that the only hope Labour had was to ditch Gordon Brown in favour of David Miliband or Ed Miliband or Alan Johnson or indeed Harriet Harman but even then the party was destined to be out of office for a generation.

They were perfectly categorical about it. As recently as November 28th last year, the main headline on the front page of The Daily Telegraph was: “Tories head for landslide”. Don’t expect to see any paper risk that one again this side of the election. The new conventional wisdom, rapidly agreed in whatever are the modern equivalents of El Vino and The Cheshire Cheese, is that it’s “going to be” a hung parliament (again, the revised result is not in doubt), that the Tories may be the largest party but will fall short of an overall majority.

As JK Galbraith, who coined the term in the first place, once observed: “circumstances have been dealing the conventional wisdom a new series of heavy blows” [The Affluent Society 1958]. That is always the danger in embracing the conventional wisdom. Being conventional, the reading isn’t very imaginative or bold. It heavily depends on what your own waters tell you chiming with what their waters are telling others with whom you are anyway liable to agree. Before long, any other projection of a result that is still weeks away seems merely maverick. Well, for what little it is worth, I have always felt sure that Labour can (I put it no higher) win the election outright. But I’m not going to the wall for it. What would be the point? We will know the actual result soon enough.

And on what ground are these two successive and – their expounders conveniently forget – strikingly different consensuses constructed? Why, on the evidence of a few hundred people asked how they think they will vote, data that is then massaged (“weighted” in the jargon) allegedly to render the respondents more representative.

I have always argued that market research is a pseudo-science, a profoundly misleading exercise, much more likely to elicit the answers that the pollsters want to hear than anything truly accurate on which to build a theory about public opinion. The polling organisations themselves, who always trumpet their methods as if they are inviolate, do all sorts of things that – it seems common sense to me – must skew the findings. Some exclude those who “don’t know” and “wouldn’t say”, as if these two responses are the same and equally irrelevant. Some give more weight to those who say they’ll “definitely” vote, as if such people are more credible and more reliable in their habits than others who perhaps are more realistic.

Some polls ask if people are optimistic or pessimistic about their economic prospects and then translate that directly and most crudely into a judgment on the present government. But if you were a convinced Tory who felt sure that the Tories would win the election, wouldn’t you by definition feel optimistic about the economic prospects? If the main breadwinner in your household had just died, you might well feel pessimistic about the prospects, while not yet knowing how far state benefits might ease the situation. Stark beauty contests are held in these polling exercises: who would make the best prime minister, Brown, Cameron or Clegg? Well, you might think Clegg but vote for Cameron anyway because the Libs Dems don’t stand a chance in your closely-fought constituency. You might think that Michael Gove or Tessa Jowell would actually make the best prime minister and then have to calculate which answer is the one most likely to bring such an event about.

But what folly to parlay even a 16-point lead for the Tories in September into an immutable Tory electoral victory in May. Indeed, if Cameron has, say, a comfortable-looking five-point lead on the morning of the election, there can still be no certainty until the count is done. Furthermore, exit polls after the voting booths closed have proved misleading and even directly wrong in the past. You don’t have to be as old as me to remember the confounded projections of the British elections in 1992 and 2005 or of the US elections in 2000 and 2004. You perhaps do have to be as old as me to remember that neither the 1970 nor the first of the two elections in 1974 turned out quite as billed.

Why anyway is it deemed so necessary to anticipate news rather than reporting it? Why is it so important to all news-gathering organisations – from The Sun to The Independent, from the Today programme to ITV News – to get their unreliable prediction in first? That Polly Toynbee or Nick Robinson discounts the chances of a fourth Labour term in succession has no bearing upon whether such a result will come about, unless significant numbers of voters determine that they will vote to outface these pundits in greater numbers than those who feel they need to take their voting guidance from someone in the public eye.

What you can be sure of is that, following a sensational Brown victory, the media will not be eating humble pie and asking themselves how they could have got it so wrong. Rather, they will be finding new material for speculation and what they are pleased to call analysis: the newly discovered fact that the electorate is – who would have thought it? – unpredictable.