TB stands for Troublesome Badgers
There has been controversy in the United Kingdom about the spread of tuberculosis (TB) among cattle. It is a serious economic problem in the country, as TB-infected animals must be slaughtered. Many people claim that cases of the disease in some herds are caused by badgers with TB.
So, over the past few weeks, an experimental cull of badgers has been taking place, authorized by the government, to reduce the population of badgers in two areas of southern England. Unfortunately for the minister concerned, Environment Secretary Owen Paterson, the number of animals killed was substantially less than had been planned. He made a gaffe in an interview when he said that the failure was due to “badgers moving the goalposts”. He explained this comment, “These are wild animals, who live in an environment where their numbers will be impacted by weather and disease.” Describing the outcome, he said “Current indications suggest that the pilot has been safe, humane and effective in delivering a reduction in the badger population of just under 60 per cent. We set ourselves a challenging target of aiming to ensure that 70 per cent of the badger population was removed during the pilot.”
I don’t know whether or not there were any O.R. scientists advising about the cull; in some ways, I hope not. One of the tenets of our discipline is that our models take note of chance and randomness. Therefore, if one is constructing a model of the success rate for an enterprise of this sort, one would start with scenarios describing the population of the wild animal, including disease (as suggested above), and combine that with a model of the weather (as also suggested above, to let you know how many nights of shooting would be likely), and combine these with a model describing the success of each expedition for shooting.
From such a model – probably a simulation model—one could derive a distribution for the number of animals killed by the proposed strategy. With enough runs of the model, one could achieve an expected percentage culled and a confidence interval on that number, based on the best available data. And that probably should have been done – and an outcome of less than 60% when the planned lower limit of the confidence interval was 70%, is so far adrift that somewhere in the preliminary model-building somebody made a gross error, or overlooked some key feature. And that is why I hope there were no O.R. scientists advising about the cull.