Lectures and seminars SfoEpi Seminar Series in Biostatistics: Dr Tim Morris

Abstract:

There are plenty of statistical methods out there. Some are very old and some are very new. Some recent methods are used all the time in medical research and others are ignored for all time. Inevitably, some poor methods are overused and other excellent methods are underused. In this talk, I will consider what makes a new statistical method fit for use. I will talk through some examples of statistical methods that are in widespread use and attempt to sketch some of the efforts it took to get them there. There is no simple answer but bearing this question in mind benefits methodologists - who should want to put their method through serious testing - and applied statisticians - who should want to understand how well probed a new method is before they use it in earnest.