Diagnostic Test Evaluation or: How I Learned to Stop Worrying and Love Bayesian Statistics
Inaugural Professorial Address by Geoff Jones
Institute of Fundamental Sciences, Massey University, Palmerston North
In some inferential situations, adopting a Bayesian framework makes the analysis much simpler. There are other situations where, because of identifiability issues, Bayesian analysis becomes the only viable possibility.
In this talk I will trace my gradual adoption of Bayesian methods through a number of examples, and focus eventually on the use of Bayesian latent class analysis for data from diagnostic testing when a “gold-standard” test is not available. Issues to be discussed include the use of multiple tests, the assumptions of conditional independence and homogeneity of test performance, sampling from under-identified models, and modelling between-group heterogeneity in prevalence.
Geoff Jones was recently made Professor of Biostatistics at Massey University Palmerston North, where he has been working (or at least has been employed) since 1997. He completed his PhD at the University of California, Davis in 1996, after many years teaching in secondary schools in the UK and Malaysia. He has collaborated widely with scientists in a range of disciplines, but particularly with the Veterinary Epidemiologists at Massey’s Epicentre, most recently in the area of diagnostic testing. He is currently a member of the OiE (World Organization for Animal Health) Collaborating Centre for Diagnostic Test Validation Science. He is particularly proud of his work with Professor Steve Haslett for the UN World Food Programme, on small area estimation of deprivation indices to guide the allocation of food aid in Third World countries.
This plenary address will be delivered in AH1 on Wednesday 28 November at 2:00.