Nine out of ten owners say... don't believe a word of it

My first acquaintance with opinion pollsters came when I was 25, working as a research assistant to the great academic lawyer W A Robson at the London School of Economics. Some old government reports that I needed were not held in the LSE library but in the British Museum. They were housed in what was rather grandly called the State Papers Room, which did not contain documents that were secret or specially significant, but official reports of one sort and another, freely - or, rather, expensively - available from the Stationery Office.

My first acquaintance with opinion pollsters came when I was 25, working as a research assistant to the great academic lawyer W A Robson at the London School of Economics. Some old government reports that I needed were not held in the LSE library but in the British Museum. They were housed in what was rather grandly called the State Papers Room, which did not contain documents that were secret or specially significant, but official reports of one sort and another, freely - or, rather, expensively - available from the Stationery Office.

The room was, I noticed, crowded with middle-aged women, scribbling busily away. Many of them had shopping baskets by their side. For some reason, "housewives, many of them carrying shopping baskets" were a feature of any properly written news story of those days. They would look on admiringly as "firemen wearing breathing apparatus" went bravely about their duties. These housewives, however, were consulting electoral registers and noting down names and addresses from a table of random numbers. They were part-time employees of market research organisations.

I decided to complain to the Trustees of the British Museum, not only because I was being inconvenienced (for these people were occupying all the seats), but also because I thought - as I still think - it wrong for the Reading Room, now the British Library, to provide virtually free facilities for wholly commercial purposes. After some weeks a most civil reply arrived. The Trustees, it said, had discussed my complaint in some detail. After long consideration, they had decided that it would be invidious to distinguish between one researcher and another. Today the reply would, I suspect, be less sympathetic to me, living as we do in an age when the flattery and indulgence of commercial interests is seen as one of the primary purposes of government and its subsidiary institutions.

I next bumped into the pollsters about a year later. By this time I had left the LSE and had just started to work for the Sunday Express. My wife was working for Gallup Poll, at its offices in Regent Street. This poll was published in the News Chronicle, soon to become defunct. The Daily Mail proudly published the National Opinion Poll. The Daily Express published no poll at all. Its proprietor, Lord Beaverbrook, considered polls only scarcely less the handiwork of the Devil than were Lord Mountbatten, the Common Market and commercial television; though whether his dislike derived from a mistaken decision not to join in - which was the reason for his detestation of commercial television - or from other considerations entirely was something I never discovered.

One evening my wife mentioned casually that the Gallup Poll and the National Opinion Poll employed the same interviewers for many of their surveys. Ever the alert reporter, I recognised at once that this was a story deserving of wider circulation. It would certainly be of interest to Express Newspapers. And, while my wife may have owed a duty of confidentiality (as it was not then called) to Gallup, I was under no such obligation. Accordingly I told the story.

I did not tell it to the Sunday Express, which would, I thought, have messed it up, but to the livelier Daily Express. I told it to the paper's managing editor, Mr Robert Edwards, who did me proud, with a front-page lead by the paper's senior reporter. It was all true. But after a few shamefaced noises from the organisations concerned, everyone forgot about it and the polls returned to their old co-operative ways.

The object of the foregoing is not just to go for a trip down memory lane but to demonstrate that the polls are commercial organisations, as much as the newspapers which publish them: more so in fact, for while the papers claim to serve a higher purpose, as some of them do from time to time, the polls have no such exalted object in view. Why should they? Most of their work is about washing-powder, toothpaste and what-have-you. Their relationship to politics is the same as that of Ladbrokes to political betting: the respective activities are a good source of free publicity but a small part of the profits.

Even so, the papers cough up for the results, my colleagues write whole articles around them and the politicians pay them the closest attention, whatever they may say to the contrary. This is odd because the polls have so often turned out to be wrong. Or perhaps it is not so odd after all: for there is nothing that people love more than an apparently firm prediction. In my working life the polls have predicted the wrong result in 1970, February 1974 and 1992.

Sometimes the error has been as great or greater but the result correct: they predicted the winning party but by the wrong margin, as they have recently overestimated Labour's share of the vote. If they end up on the right or the winning side, no one pays any more attention. But if, by miscalculation or mischance, they have come down on the wrong or the losing side, the slickest of public-relations exercises is immediately mounted. They close ranks; show what would in certain circumstances be an admirable loyalty to one another, a sense of solidarity more often to be found in, say, the medical profession than in the newspaper industry.

I made a promise to myself to write this column without quoting any figures, but now find I must break it. On Monday The Independent had a poll purporting to show that more people would vote Labour with Mr Gordon Brown as leader than would with Mr Tony Blair in that position. This was interesting, not so much because of what it appeared to tell us about the relative standing of the two old lags, as because it provided figures for "don't knows" and "won't says": respectively 26 per cent and 11 per cent with Mr Blair as leader, and 20 per cent and 10 per cent with Mr Brown.

Such raw figures are rarely given to us in the final version. Don't knows and won't says are, in the jargon of the trade, "squeezed" by methods of varying degrees of disreputability; while, if all else fails, they can be allocated according to a calculation of simple proportion based on the firmer figures.

We have been so conditioned historically as to think that all tactical voting is a device to be used solely against the Conservatives. This is because it was first thought of (by the extreme Left, in fact) as a device for getting rid of Mrs Margaret Thatcher in the 1980s. It failed then, but it was one of the reasons for the annihilation of Mr John Major in 1997. There is no reason why it should not be used against Mr Blair this year, principally on account of the Iraq war. Mr Blair would not be annihilated, but he might receive a nasty shock all the same.