Jan Moir on Stephen Gately: serious statistical flaws
Gately I am deeply dismayed by Jan Moir's article on Stephen Gately, suggesting that the poor man died of homosexuality.
For Moir's piece contained within it extraordinary statistical flaws.
How did it transpire that everybody on the Daily Mail failed to pick her up on it?
Here is her argument:
Another real sadness about Gately's death is that it strikes another blow to the happy-ever-after myth of civil partnerships.
Gay activists are always calling for tolerance and understanding about same-sex relationships, arguing that they are just the same as heterosexual marriages. Not everyone, they say, is like George Michael.
Of course, in many cases this may be true. Yet the recent death of Kevin McGee, the former husband of Little Britain star Matt Lucas, and now the dubious events of Gately's last night raise troubling questions about what happened.
There seem to me to be three statistical mistakes here.
The first is that you can learn something useful from a sample size of two.
The second is that you select your sample by reading back copies of the Daily Mail and finding famous people who fall into the category you wish to study who have been in front page news stories in the last month.
There is, you see, a chance that this method will bias the sample.
And third, Ms Moir appears to have forgotten how useful it is to have a comparison group.
It did not strike her that by employing the same sample selection method (Daily Mail stories) she could have found two marriages to compare with the civil partnerships.
All in all I think the statistical errors are so great as to leave open the possibility that Ms Moir's piece of analysis was not designed to get at the truth.
Posted by Daniel Finkelstein on October 16, 2009 at 04:44 PM