Britain's pollsters are an odd breed and certainly not a representative sample of the population. But Daniel Butler's findings reveal a surprising degree of similarity in their attitudes and views as a general election looms.
As he sits in his wood-panelled office overlooking St James Park, Bob Worcester, chairman and founder of MORI, tries not to look smug. A general election is looming and he can feel his media desirability rising already. During the campaign itself, the nation will be hypnotised by his and his fellow pollsters' findings. Politicians will hang on every statistic he can produce and in the City fortunes will be made and lost on rumours about his work.
All the while he will make money and MORI's name will be splashed across every front page in the county. To hard-pressed ad-men and PR executives across the country, it must seem like the proverbial golden goose.
Worcester snorts at the suggestion: 'We work seven days a week, 18 hours a day. We move things in 24 hours that would normally take a fortnight or a month,' he barks in a strong American accent that three decades of living in Britain has done nothing to diminish. 'Anyone can make more money if they work 80 - rather than 40-hour weeks,' he growls, estimating that his workload goes up by 80% during a campaign and claiming that he slept in the office for most of the 1987 election.
'Bob just does that because he gets a kick out of it,' laughs his rival, Nick Moon, head of political polling at NOP. 'Of course you put in a few extra hours, but Bob has a romantic notion about sleeping in the office because he likes telling people about it afterwards.' On the scruffy side of casual, the bearded Moon's Covent Garden office, with its chipped paintwork and dingy lighting, could hardly be more of a contrast to Worcester's panelled formality. But then, for statisticians obsessed with representative samples, Britain's pollsters are an odd breed and far from a typical population sample - all five are men, three are bearded, three called Bob and the other two, Nick.
Their companies - MORI, Gallup, ICM, Harris and NOP - are also a tiny minority of the 385 members of the Market Research Society (MRS).
Even MORI, the largest independent pollster, had a 1990 turnover of less than £9 million.
The amounts of money pollsters earn from political work are surprisingly small. All five estimate that polls generate about £200,000 in a non-election year, probably doubling in the heat of a campaign. Harris's Bob Waller even claims to lose money on polls - or at least to run on tiny margins of 5%. He believes they undercharge and give the increasingly cost-conscious newspapers good value for money.
A headline grabbing poll can provide a sensational lead story for a typical cost of £5,000 and, of course, get the newspapers the promotional benefit of a mention on television news.
Polls naturally become more expensive in an election campaign and up to £10,000 can be eaten up in providing a handful of answers as fast as possible, which may require 50 fieldworkers.
'If you really want to make money out of polling, use the stock market,' Waller continues. He is making a bitter joke, for the prospect of polls being used to manipulate the market strikes terror into all pollsters. If any of them has a heart attack during the coming campaign it will not be from overwork, but from fury at a story planted in the markets.
They have cause to be annoyed. During the 1987 election, on Black Thursday, one rumour-monger claimed Marplan was about to cut the Tory's lead to just two points. ICM's Nick Sparrow was running Marplan's polls at the time: 'If anyone is going to make money out of my polls, I want it to be me,' he complains, adding bitterly that while he may charge a few thousand for a poll, the rumour's author probably made millions.
The incident gave polling a bad name: 'That night Sir Nicholas Goodison, then chairman of the Stock Exchange, had the gall to call for a close look at the ethics of the polling industry,' says Sparrow vehemently. 'He should have stayed in the City if he wanted heads to roll.'
In fact, the pollsters point out, no one knows the result of a poll before the client. Nevertheless, elections are spent fending off non- stop phone enquiries ('That's where my work load really does go up,' laughs Moon).
Polls represent only a small part of the workload of all five companies, the bulk coming from private commercial research. For most it represents about 5% of turnover, rising slightly in an election year. Even for ICM, the most dependent on polls, they only represent 15% or 20% in an election year. There's a downside too, says Gallup's Bob Wybrow. If polls were flavour of the month during the four-week campaign, for the next six months they were as popular as turkey in January.
'The big money in opinion polling is working for the political parties, not the newspapers,' says Worcester. The Labour Party commissioned £250,000 worth of private polls from him in 1987. Worcester can be open now, having lost the Labour Party account after 'disagreements' with Peter Mandelson, former head of communications at Walworth Road. His rivals find the subject more delicate, although Moon (who won the Labour account) concedes that 'serious amounts of money' are spent.
Waller, who works for the Conservatives falls unusually quiet, pleading a duty of confidentiality. Inquiries about polling are greeted with awkward silences at party headquarters. It is difficult for a Conservative Parry chairman to dismiss unpleasant poll results when he is spending thousands on private surveys from the same source. And Labour activists would be angered to discover how much of its new rhetoric and image came from MORI research. 'I never discussed our polls when at Walworth Road and I'm not going to start now,' snaps Mandelson in an uncharacteristically unhelpful mood.
But however lucrative the Conservative and Labour contracts might appear, only two out of the five pollsters can win them. Similarly, there are finite numbers of papers and programmes with the budgets to fund national polls. Even five companies begin to look as if they are crowding the pond of available money. Anyone looking for the pool to empty has a long wait however. The real attraction of polling is not financial: 'It isn't big business,' says Wybrow.'Publicity, that's the real name of the game.
It is an undeniable attraction. Until the late '50s Gallup had the field almost exclusively to itself. For years its name was synonymous with political research. Even today, with five players, the marketing value is powerful. Sparrow smiles when he remembers a recent write-up in The Mail on Sunday: 'It described ICM as one of the most prestigious polling organisations - that pleased me,' he says with obvious understatement.
NOP's Moon can think of only one contract won because of the publicity, while Worcester guesses that 'maybe 10%' of his clients are attracted in this way. And all agree that as many clients may be put off, not realising that pollsters do a wider range of commercial and financial work. Publicity is a double-edged sword they point out. It can be a useful marketing tool, but the perils of making a mistake are great. 'Nobody shoots craps with their careers the way we do,' growls Worcester.
He has a point. Most market research is conducted over a prolonged period with countless interviews. Mistakes are more easily avoided and results are difficult to compare with reality. But polls must be done quickly and there's always a yardstick to check them against, whether a rival poll or election result. Errors, or 'rogues' as they are called, do occur. But public ignorance is a greater worry.
Worcester likes to explain polling with a photographic analogy: he collects 'snapshots' of opinion. Like a photograph of a crowded Blackpool Beach on a sunny June afternoon with the tide out, the picture can change quickly, and a small last-minute shift can turn an outcome on its head. As a result the only 'snapshot' that is a firm prediction is the exit poll (taken as voters leave the booth).
Unfortunately, all too often the public remembers an earlier poll and blames the pundits for 'getting it wrong'- in the same way that disgruntled holiday makers, lured by the sun-drenched brochure, demand refunds when it rains non-stop.
Worcester is the most evangelical pollster ('publicity seeking more accurately,' say his many critics). He says he's 'a man with a mission - to make people understand polling'. The theory is simple, he claims. If a large group of 'typical' voters is asked unbiased questions, their answers will reflect the views of the entire population at the moment the question is asked. This normally means at least 1,000 interviews, spread across the country and 'weighted' by factors such as age, sex and class.
But even when conducted by experts, the results are not infallible. Statistics say that such a poll should be correct to within 2%, 95% of the time. And on paper the track record is impressive, with almost all results within the margins of error. Naturally, it is the mistakes that are remembered. Two of the most notable were the 1970 and February 1974 elections: in 1970 Labour were forecast to win (and in fact did get the higher percentage of the votes) while in 1974 (the 'Who Governs Britain?' election) it was the Conservatives who were wrongly forecast as the winners. There was public derision but in fact the pollsters in each case were within their margins of error. Their problem was that the result was too close to call.
The risks that mistakes present explains the rather cosy relationship between the five men: 'Of course I have a good giggle when someone produces something out of line, but you try to repress it as much as possible,' confesses Moon. 'It damages us all if one of us has their credibility seriously dented.'
If the pollsters' treatment of each other seems curiously gentle to the outsider, he should see the vein on Worcester's temple as he discusses non-scientific surveys. These come every election in the form of thinly-disguised party canvass returns and phone-this-number results.
'It's not even a Lada versus a Rolls Royce,' snorts Waller. 'What you're looking at is a pile of junk.' Serious amateurs are a threat too. A 1987 exit poll for Newsnight's Vincent Hannah, conducted by polytechnic students was one particularly damaging example - it almost destroyed some work Harris was doing, says Wallet: 'One of their constituencies coincided with us. The SDP rang up Channel 4 to complain that we were using Labour activists and to try to get the poll pulled. It very nearly was.'
The dangers of bad publicity from inaccurate findings have led to a strengthening of the Market Research Society which now insists on published polls being accompanied by the interviewing dates, sample size and number of locations where these took place. Moon claims that theirs is one of the few industries where, instead of patenting new techniques, there's a rush to give the details to arch-rivals.
From the way they describe the high risks and low rewards, you might think that pollsters would dread the coming campaign. Not a bit of it: 'I'd love a hung parliament which staggered on for a couple of years - it's wonderful to have something fun to do,' declares Moon mischievously. The idea makes Wallet bounce with excitement: 'How about annual parliaments,' he chortles. Whatever they say, it is hard to escape the conclusion that fun is the real motive. As one observer says: 'They're just academics playing in a commercial sandpit.'
A snapshot of the famous five.
Gallup - founded 1937; 1990 turnover £5.2 million; 80 employees. Main media client is The Daily Telegraph but it does a limited amount of work for the Tories.
Harris - born 1969; 1990 turnover £6.6 million; 80 employees. Clients include The Observer, Channel 4, Thames and TV AM. It is also chief pollster for the Tories.
ICM - founded 1989 ;1990 turnover £1.5 million; 10 employees. Clients include The Guardian, The Daily Express, The Sunday Express, The Sunday Express, The Daily Mail, The Scotsman.
MORI - founded 1969; 1990 turnover just under £9 million; 110 employees. Formerly Labour's chief pollster, now its political work is done mainly for The Times, The Sunday Times and The Economist.
NOP - founded 1957. Clients include The Mail on Sunday, The Independent, The Independent on Sunday, The Sun, The Evening Standard and the BBC. It also acts as pollster for the Labour Party.