I fisked the Sunday Express’ story on Roger Coghill’s notion that the Bridgend suicides are linked to mobile phone masts here. Since then, Ben Goldacre has covered the issue twice, including asking Mr Coghill how he calculated the average distance a home is from a phone mast.
Not only did Roger Coghill refuse this basic information, he got extremely stroppy about it. On Comment is Free, which ran for over 380 posts, he constantly refused to give this information, preferring to verbally abuse his critics, such as calling one a “spongiform creature” in the pay of”commercial interest”. Since he did not know who the critic was, this is clearly without foundation.
Interestingly, in the comments following Ben Goldacre’s blogposts linked to above, he returns to this theme, accusing Ben of reflecting “the ideology of powerful industrial, technological and political vested interests” and of having a “hidden agenda”. He gives the impression that he can concieve of no reason other than personal gain for holding an opinion. I cannot imagine why this should be.
He refers to his critics as “all Ben’s creatures” and accuses us of making ad hominem attacks. He does not appear to see the irony in him making such an accusation.
Finally, he does show his working here. He begins by assuming the area of the UK to be 30 million hectares (it is actually 24 million). He divides by the number of masts in the country to get an area per mast, treats this as a circle, calculates the radius and halves this to obtain his average.
The first problem with this is that it could only work if people and phone masts were evenly distributed across the country. In fact they tend to cluster in towns and cities.
The second problem is that while this gives a mean, it does not give a distribution curve of distances, thus no standard deviation and thus it is impossible to say whether or not a particular home is significantly nearer a mast.
A third problem is that, as Phayes points out, the average distance by this method would be 2r/3, not r/2.
El Pollo Diablo hammers the final nail into the coffin of Mr Coghill’s idea when he uses his methods to calculate that homes in Manchester are about the same distance from masts as are those in Bridgend. Given the population of Manchester, he points out that if Mr Coghill were correct in linking masts to suicides there would have ben 1760 suicides in Manchester over the last 18 months. The UK as a whole over that period had approximately 10,000 suicides, so Manchester – with less than 1% of the population – would have had one sixth of the suicides. Such an enormous cluster would not have escaped anyone’s notice if it existed.
It does not, of course, thus demonstrating that Roger Coghill’s idea is arrant nonsense.