I removed a bit of a conjouring trick from our story on Hotel-fairy.com's Facebook poll, which suggests that only 14% of the site's users frequent high-street travel agents.
The poll itself is basically sound. Survey of Facebook users, carried out on Facebook, covering the standard 1,000-person sample size. Fine.
But there's sleight-of-hand in the press release: 1,000 Facebook users become 'the Facebook Generation'.
'Generation' is a bit of a fuzzy word, inviting you to assume that what is true of users of a particular website is true of an entire age demographic.
It's also simply the wrong word, since stats indicate that Facebook is increasingly multi-generational, and there's no indication that the poll was targeted to a particular age group.
If anything, the assumption should be that only 14% of a fairly broad swath of the UK use high-street agents.
But that doesn't work either, since Facebook has (acc. to April 09 figures from O'Reilly Research) 18m users in a population of over 60m - and the very fact that poll respondents are Facebook users means they're the sort of people who are likely to book online.
So why publish the story at all? Well, because the point about Facebook users is useful on its own terms. If you're an agent and you're considering Facebook as a marketing tool, figures about its users' buying habits are going to help you make a decision.
Everybody loves a trend story. But sometimes extrapolation just muddies things up.
(NB: But it's fine if you're honest about it.)