Statistical wizardry

The long awaited results on incidence of cattle TB, in the areas of the two pilot culls, or at least the first year of them, are now published. And indeed, Queen May, with guitar in one hand and badger in the other is crowing.

 "No statistical difference" he repeated, between a comparison block and the pilot cull areas in an article, published by Farmers Guardian - [link]. And everyone from Owen Paterson, who set the pilots up through the NFU who backed them, down to the farmers involved are 'lying'.

But can the farmers concerned - [link] be so wrong? And how can the drop of some 60 per cent in herds under restriction - {link] become so skewed after its journey through a computer?

In her paper - [link] reporting the methodology used for the analysis, Professor Christl Donnelly, of Imperial College explains:

"Herds under restriction for four or more months of the reporting period due to an incident that started before the reporting period were excluded from the analyses."

and:

"Baseline date: The date on which the culling is initiated in an intervention area. Cattle bTB incidents prior to this date are not attributable to any effects of culling."

Now this lady and her electronic abacus have history. And unfortunately, many of us have had the misfortune to have lived and watched our cattle die, through all it has churned out. From BSE, to FMD the RBCT and now these pilot badger culls.

Whatever statistical wizardry has been employed, the first question asked of its results should surely be 'does it adequately reflect the situation on the ground?' Because if it does not, it is not only meaningless, it is misleading and wrong.

And it appears that as with the RBCT input data, a cut off date for inclusion of data from farms involved in these pilot culls, specifically excluded farms already under long term TB restriction and the results of culling the wildlife reservoir of infection, had on them.

Wizardry indeed. And if this really is the case, dead wrong.