A polemical take on politics, economics and foreign affairs

RSS

Beware groupthink

The perils of “following the crowd” in this unpredictable election.

Yesterday, I flagged up Alex Barker's "unpacking" of the consensus view among Britain's top pollsters that the Tories will indeed secure a Commons majority on 6 May -- despite a narrowing in the polls.

Writing on the FT blog, Alex pointed out that "a unanimous consensus is always something to be wary of, particularly when it doesn't quite reflect the evidence available".

Today, the excellent Aditya Chakrabortty has a piece in the Guardian dissecting the cult of the "expert commentator" and dismissing the "high-intensity forecasting" of the nation's political pundits as "largely worthless".

Aditya writes:

Cast your mind back to June last year, when yet another Labour putsch was being launched against the Prime Minister. Trawling through the comment pages published in the week when the coup was at its height -- with ministers resigning, and local and European elections looming -- I found 20 columns and leading articles in the Times, Telegraph, Independent and Guardian discussing whether Brown would survive. Of those, half predicted he would go, while only a quarter thought he might stay (the rest, perhaps wisely, didn't chance their arm). If the broadsheet fortune-tellers could not assess the outcome of a backroom plot featuring a few ministers and MPs, how far should we trust their judgement on what tens of millions of voters will do?

This is probably unfair, and certainly unscientific. Brown did come close to being toppled last summer, and cabinet conspirators tend to duck out of ICM-style opinion polls. But, on the broad question of whether we should set much store by political predictions, the answer is a flat no.

Aditya also flags up a fascinating book by the University of California, Berkeley psychiatrist Phil Tetlock, based on his 20-year study of predictions by political, media and economic experts, which found that they are no better than the rest of us at prognostication.