Michael Goldstein

Department of Mathematical Sciences, University of Durham

Email: michael.goldstein@durham.ac.uk

Finger/Talk:dma0mg@bayes.dur.ac.uk

Department Address:

Michael Goldstein

Dept. of Mathematical Sciences

Science Laboratories, South Road

Durham, DH1 3LE

ENGLAND

Phone: (091 or +44 91) 374 2365

My general interests are in the foundations, methodology and applications of Bayesian/subjectivist approaches to statistics, particularly for problems which are sufficiently large and complex to make the usual Bayesian modelling and analysis extremely difficult. Much of my work in this general area concerns the synthesis of expert judgements and experimental data under partial prior belief specification, and comes together under the general structure of

BAYES LINEAR METHODOLOGY

You can find an article (9 pages, postscript) that I wrote to give an overview of this area for the

ENCYCLOPAEDIA OF STATISTICAL SCIENCES (update vol 3, 1998, Wiley, reproduced by permission)

You can find out more about Bayes linear analysis

(including our programming language for carrying out the graphical modelling, belief specification, analysis, and diagnostics, together with tutorial guides to the approach and to the program, and a Bayes linear bibliography)

by going to the Bayes linear homepage.

You might be interested in seeing how Bayes linear methodology works for large scale practical applications. We are carrying out, with EPSRC support, the following project:

Prediction and decision making for large scale physical systems

with particular application to problems in the oil industry.

Find out about this project at the homepage.

Currently, I am involved in a project on the use of Bayesian methods for software testing, with EPSRC and
commercial support.

Find out about this project at the homepage.

My publications are listed here.