It is hard to argue against using ‘scientific’ evidence to decide policy; which health care interventions should be employed or which policy actions should be used in the social sphere. Randomised controlled trials give clear, accurate results … if everything works out perfectly. However everything does not always work out perfectly. There are big problems from studies even in the tightly regulated medical arena. (Recently, 3rd Jan 2014, the Public Accounts Committee published a damning report on the multi-million pound acquisition of the drug ‘Tamiflu’).
It seems that many reports, written in the language of science, can have totally misleading or at least questionable conclusions. (It should be said at the outset that non-randomised studies have additional problems and so it is even more tricky to ensure reliable results.)
This talk will discuss these issues and give some examples of the sorts of things which can go wrong. It will mention the intriguing problem of determining the effect of spending money on road lighting, an issue the speaker discussed on Radio4’s ‘More or Less’; the matter also made it into the investigative magazine Private Eye. Ideas will be given on what needs to be done to improve the quality of scientific evidence.
Paul Marchant is a Chartered Statistician of the Royal Statistical Society. He originally gained a PhD from research in astrophysics, subsequently getting an MSc in applied statistics. He holds visiting fellowships at both Leeds and Leeds Metropolitan universities.