Dr. Crippen makes a really good point in this post. Unfortunately, it occurs half way down the page, so I reproduce it here:
The journalists have just discovered (see the BBC report here) that no one knows the optimum treatment for patients with early prostate cancer. Doctors have been saying this for years but have been ignored by the journalists who have campaigned to increase the demand for routine PSA screening. Trouble with journalists is that once they discover something they did not previously know, they assume that no one else knew either. Now they will be telling us we do too much screening.
I love that line near the end, so I’m going to print it again, bigger:
Trouble with journalists is that once they discover something they did not previously know, they assume that no one else knew either.
And now I turn to Sunday’s edition of the Observer:
Down on the bottom, there is a mathematical formula in the headline, heralding a great new way to predict football results. The formula itself is that for the Poisson distribution. I first learnt about this in my A level maths course at school, but I think that it might be even older than that. In fact, it was originally published in 1838. Hmm. Not as new as we’d first thought.
The journalists, of course, are totally correct. Given an average number of goals in a match, the Poisson can be used to provide probability estimates for the scores. None of this, however, is particularly novel. The article moves on to mention the “advanced techniques” now used to predict goal scoring. The Poisson distribution isn’t mentioned. Of course, that could be because actually the Poisson distribution isn’t particularly suitable. After all, goals can’t even be assumed to be independent events. And the value of lambda (the average number of goals) has to be tailored to the individual match.
Never let the facts get in the way of a good story.