Interview: Peter Kellner, YouGov (Evening Standard)
By David Rowan
SOME things are simply not fair. Peter Kellner, an Evening Standard political analyst until two years ago, is about to find himself almost a million pounds richer. YouGov, the implausibly named polling firm he now chairs, floats on the stock market next Monday, valued at around £18 million.
With a six per cent stake, Kellner is naturally feeling rather pleased with himself in his art-filled office overlooking Smithfield market. It's only when probed about his polls' alleged Tory bias that he turns stony-faced and threatens to sue.
"I will do fairly well out of this, I can't deny that," he admits sheepishly, adding that if his journalist friends are jealous, they have not shown it. "But though it's a large gain on paper, as a director I can't yet convert it into a Ferrari. If the company doubles in value next Tuesday, we can't simply cash in our chips and retire to the Cayman Islands." John Humphrys, too, stands to make around £250,000 from the flotation.
Still, as a career gamble it has paid off. Kellner, once Newsnight's political commentator, was brought in by the company's founders, Stephan Shakespeare and Nadhim Zahawi, who had run Jeffrey Archer's unsuccessful mayoral campaign. In May 2000, just as the internet bubble was bursting, they launched YouGov as an online-only polling firm. While critics sneered at its methods, the company accurately predicted voting in the 2001 General Election (forecasting Labour's 9.3 per cent lead as 10 per cent), the Australian election, even Pop Idol. And while corporate clients such as Asda provide its main income, it forecasts voting intentions for papers ranging from The Sun to The Daily Telegraph.
The attraction, Kellner says, is that the internet drastically cuts costs. With a traditional pollster, a typical largescale poll would cost £20,000, of which £8,000 would go on asking the questions and £9,000 on overheads, leaving £3,000 profit. "But we would charge £15,000," he says. "And there would be just £2,000 in direct costs and £7,000 in indirect costs, leaving us with £6,000 - twice the profit."
YouGov has a panel of 90,000 internet users, paid 50p for each survey. For a political poll, it emails a group deemed "representative of the country", selecting them and weighing their verdicts according to gender, age, class and region. "We also do something no other pollster does," Kellner says. "We weight by newspaper readership. I'm not saying every Sun reader is identical, but it helps ensure your sample is attitudinally representative. And with one exception, we've got elections pretty much spot on."
That exception was the last European election, an error Kellner dismisses as a mistake over where UKIP was placed on the questionnaire. Still, there remain concerns about the reliability of internet polls. The columnist Stephen Pollard claims to have fraudulently registered with YouGov under a dozen fake names. Kellner replies that security has since been tightened. "Even if Pollard had taken every poll he could have, there was just a 1 in 500 chance of him moving one figure by one percentage point," Kellner says.
Yet YouGov's results generally favour the Conservatives far more than rival pollsters'. Its latest Sunday Times survey has Labour ahead by just one point, while a Populus poll in The Times puts the gap at nine points, and Mori in the FT has the Tories eight behind.
"That happened, too, at the last three General Elections, when almost all surveys exaggerated Labour's support and understated the Conservatives'," he replies calmly. "The problem proved to be with the other pollsters. We got it about right." Conventional polls, he explains, undercount the "shy Tories", those reluctant to share their allegiance, and fail to account for Labour's lower turnout rates.
In the murky, factional world of party polling, rivals have dared to suggest that YouGov - or "Anything You Want Guv" - intentionally favours the Conservatives. Lord Bell, formerly Iain Duncan Smith's media adviser, was an early investor, they say - and Shakespeare and Zahawi have been party activists. How would Kellner respond?
"If you put that in print, we'll sue," he says, suddenly colder. "It is not true, it is plainly defamatory, and we'd have no compunction about suing."
Shakespeare, who has been working quietly elsewhere in the office, gets up to add: "And those who suggested it in the past, we have taken legal action against. It would mean we were twisting results to suit someone, which is patently crazy."
Clearly a touchy subject. In fact, Kellner's history suggests these conspiracy theories to be farfetched. Now 58, he stood for Labour in local Westminster council elections in 1978, and remains a party member. He also happens to be married to Catherine Ashton, a Labour minister in the Lords (they have a son and a daughter, and live in St Albans).
"Equally irrelevant. There is no connection between any of our political views and the results we put out. Look, a 2003 poll for the Telegraph showed how unpopular Iain Duncan Smith was with both the public and party members, yet he had hired us for internal party polls. So we were actually producing figures against the interests of one of our main clients!" Michael Howard scrapped the contract.
THE real problem is that newspapers "overinterpret" polls. "They provide a lot of knowledge, but are incapable of providing certainty. Newspapers want certainty. The mathematics of polls are identical to the mathematics of electrons. Just as when you try to map out what's happening to electrons, you get a blurred picture. It's Heisenberg's uncertainty principle."
Excuse me? "Look, we're showing Labour leads on average of three or four per cent. Some show seven per cent. But statistically that's a tiny difference, because of the two or three per cent margin of error for each party."
Does he miss newspapers? "To my surprise, I miss practically nothing," he reflects. "I'm still pursuing my interests in political analysis. The only difference is that on election day, for the first time, I'll be the one getting nervous as we wait for the results."
Although this time, envious excolleagues might point out, Kellner the millionaire cannot fail to win.
(Evening Standard, April 20 2005)
SOME things are simply not fair. Peter Kellner, an Evening Standard political analyst until two years ago, is about to find himself almost a million pounds richer. YouGov, the implausibly named polling firm he now chairs, floats on the stock market next Monday, valued at around £18 million.
With a six per cent stake, Kellner is naturally feeling rather pleased with himself in his art-filled office overlooking Smithfield market. It's only when probed about his polls' alleged Tory bias that he turns stony-faced and threatens to sue.
"I will do fairly well out of this, I can't deny that," he admits sheepishly, adding that if his journalist friends are jealous, they have not shown it. "But though it's a large gain on paper, as a director I can't yet convert it into a Ferrari. If the company doubles in value next Tuesday, we can't simply cash in our chips and retire to the Cayman Islands." John Humphrys, too, stands to make around £250,000 from the flotation.
Still, as a career gamble it has paid off. Kellner, once Newsnight's political commentator, was brought in by the company's founders, Stephan Shakespeare and Nadhim Zahawi, who had run Jeffrey Archer's unsuccessful mayoral campaign. In May 2000, just as the internet bubble was bursting, they launched YouGov as an online-only polling firm. While critics sneered at its methods, the company accurately predicted voting in the 2001 General Election (forecasting Labour's 9.3 per cent lead as 10 per cent), the Australian election, even Pop Idol. And while corporate clients such as Asda provide its main income, it forecasts voting intentions for papers ranging from The Sun to The Daily Telegraph.
The attraction, Kellner says, is that the internet drastically cuts costs. With a traditional pollster, a typical largescale poll would cost £20,000, of which £8,000 would go on asking the questions and £9,000 on overheads, leaving £3,000 profit. "But we would charge £15,000," he says. "And there would be just £2,000 in direct costs and £7,000 in indirect costs, leaving us with £6,000 - twice the profit."
YouGov has a panel of 90,000 internet users, paid 50p for each survey. For a political poll, it emails a group deemed "representative of the country", selecting them and weighing their verdicts according to gender, age, class and region. "We also do something no other pollster does," Kellner says. "We weight by newspaper readership. I'm not saying every Sun reader is identical, but it helps ensure your sample is attitudinally representative. And with one exception, we've got elections pretty much spot on."
That exception was the last European election, an error Kellner dismisses as a mistake over where UKIP was placed on the questionnaire. Still, there remain concerns about the reliability of internet polls. The columnist Stephen Pollard claims to have fraudulently registered with YouGov under a dozen fake names. Kellner replies that security has since been tightened. "Even if Pollard had taken every poll he could have, there was just a 1 in 500 chance of him moving one figure by one percentage point," Kellner says.
Yet YouGov's results generally favour the Conservatives far more than rival pollsters'. Its latest Sunday Times survey has Labour ahead by just one point, while a Populus poll in The Times puts the gap at nine points, and Mori in the FT has the Tories eight behind.
"That happened, too, at the last three General Elections, when almost all surveys exaggerated Labour's support and understated the Conservatives'," he replies calmly. "The problem proved to be with the other pollsters. We got it about right." Conventional polls, he explains, undercount the "shy Tories", those reluctant to share their allegiance, and fail to account for Labour's lower turnout rates.
In the murky, factional world of party polling, rivals have dared to suggest that YouGov - or "Anything You Want Guv" - intentionally favours the Conservatives. Lord Bell, formerly Iain Duncan Smith's media adviser, was an early investor, they say - and Shakespeare and Zahawi have been party activists. How would Kellner respond?
"If you put that in print, we'll sue," he says, suddenly colder. "It is not true, it is plainly defamatory, and we'd have no compunction about suing."
Shakespeare, who has been working quietly elsewhere in the office, gets up to add: "And those who suggested it in the past, we have taken legal action against. It would mean we were twisting results to suit someone, which is patently crazy."
Clearly a touchy subject. In fact, Kellner's history suggests these conspiracy theories to be farfetched. Now 58, he stood for Labour in local Westminster council elections in 1978, and remains a party member. He also happens to be married to Catherine Ashton, a Labour minister in the Lords (they have a son and a daughter, and live in St Albans).
"Equally irrelevant. There is no connection between any of our political views and the results we put out. Look, a 2003 poll for the Telegraph showed how unpopular Iain Duncan Smith was with both the public and party members, yet he had hired us for internal party polls. So we were actually producing figures against the interests of one of our main clients!" Michael Howard scrapped the contract.
THE real problem is that newspapers "overinterpret" polls. "They provide a lot of knowledge, but are incapable of providing certainty. Newspapers want certainty. The mathematics of polls are identical to the mathematics of electrons. Just as when you try to map out what's happening to electrons, you get a blurred picture. It's Heisenberg's uncertainty principle."
Excuse me? "Look, we're showing Labour leads on average of three or four per cent. Some show seven per cent. But statistically that's a tiny difference, because of the two or three per cent margin of error for each party."
Does he miss newspapers? "To my surprise, I miss practically nothing," he reflects. "I'm still pursuing my interests in political analysis. The only difference is that on election day, for the first time, I'll be the one getting nervous as we wait for the results."
Although this time, envious excolleagues might point out, Kellner the millionaire cannot fail to win.
(Evening Standard, April 20 2005)
posted by David Rowan at Wednesday, April 20, 2005
<< Home