Tuesday 22 July 2008

Dodgy Mathematics Exposed: #2, The Danger Drawing of Conclusions From Absolute Values

Every year around about now the TV Licence people bring out their latest figures on how many people are avoiding paying for a TV licence while still watching TV, and they rank the list according to city. And every year I listen in disbelief when they announce that London is the city with the most TV Licence evaders. If I happen to hear this gem on local radio, it is usually followed by something along the lines of 'and the city in Northern Ireland with the highest number of people avoiding paying the licence was Belfast'.

And every year I dig around in vain to try to find out what point on earth they are trying to convey to us with this information. It is, of course, always said in a tone that implies that the entire population of Belfast always looked a bit dodgy and if it weren't for the moderating influence of the honest villagers then the whole province would be racing headlong to hell in a handcart, so I assume this is somehow related to whatever point they're trying to make.

Now obviously you wouldn't catch 30,000 people in Portavogie failing to pay their TV Licence. But that's mainly because you'd be lucky to find 30,000 people in Portavogie with mains electricity, or indeed 30,000 people in Portavogie at all, outside of The Twelfth.

I dare say that if we did a survey of 'number of people with 2 arms' or 'number of people who had breakfast yesterday', or 'number of people who pretty much anything', Belfast would come first and Drumbo wouldn't. It might not be a winner in absolutely everything - 'number of people who own sheep' being an obvious example - but these are what are known as 'statistical blips' and do not prove anything.

So of course the wider point here is that if there are more people here than there then here will always come higher up than there in most leagues that count the number of people in any given category, and that a far more informative picture would be gained by providing us with the number of people who do these things 'per head of population'. But they don't do that.

No comments: