The plural of anecdote is not data — even when it’s ‘crowdsourced’

Tuesday, 11 March 2008, 18:05

I really like what the Times does online, but I must say that Roy Greenslade’s analysis of today’s Times splash is right on the money.

It doesn’t take university-level stats to know that self-selecting samples cannot be extrapolated to populations. An online vote touted as “a new kind of interactive poll” is no different than the sort of phone-in vote that papers have run for years.

It’s a very clever way of creating a very big collection of anecdotes, but to call it a “poll”, interactive or not, is misleading.

My favourite bit (in the printed editon) was this: “The poll, which attracted 2,476 responses, is novel because it reflects not just hard statistical data, but people’s observations and anxieties about the state of the economy”.

Eh? “Hard statistical data”? Where? Methinks the “just” was superfluous.

Further down (in a tiny boxout at the foot of the jump on page 4) communities editor Tom Whitwell provides the disclaimer that should have been right up in paragraph two — the story “does not have the statistical rigour of an opinion poll”.

The Times isn’t the first newspaper to report its online votes as if they were some sort of survey, of course. But to sacrifice the intellectual rigour of a story for the sake of fostering online “community” or experimenting with “crowdsourcing” is a very strange set of journalistic priorities indeed.

I wonder how many letters will flood in tomorrow from the Royal Statistical Society and various OxBridge dons.

The obligatory Google Map is very pretty, though.

Entry Filed under: Journalism, The Times, crowdsourcing, innueracy, polling, science, statistics, timesonline, wtf