Tony O'Hagan - Academic pages

Tony O'Hagan's Research

My research interests are in Bayesian Statistics. For those unfamiliar with the Bayesian approach to statistics, I have prepared a reading list. Potential PhD students might like to view some possible research topics.

There are links to various resources in statistics, and particularly Bayesian statistics, on my other links page. I am involved with applications in the environment, industry, finance and health.

My publications page contains abstracts of all my papers over the last few years, and you can download copies of those that are not yet in print.

My CV (last updated 6 March 2008, file size 101KB, in pdf format, see my publications page if you do not have the Adobe Acrobat pdf viewer) also includes details of published books and papers.

The three main research topics I have been engaged in recently are as follows. (Letter codes CC, HE and EL identify relevant papers on my publications page.)

• Bayesian analysis of computer code outputs. (CC) This is concerned with a variety of problems arising in the use of complex computer models.

  Links: general information; MUCM, the Managing Uncertainty in Complex Models project.

  See my publications page for abstracts and downloads of recent papers (coded CC). In particular, my key papers in this field are as follows: [1] O'Hagan, Kennedy and Oakley, J. E. (1999) reviews the field up to 1999; [2] O'Hagan (2006) is a relatively non-technical introduction aimed at model users; [3] Kennedy and O'Hagan (2001) is the seminal paper on calibration which introduces the idea of model discrepancy; [4] Kennedy, O'Hagan, Anderson, Lomas, Woodward, Heinemeyer and Gosling (2008) is an application of uncertainty and sensitivity analysis techniques in an important area, with various innovative features.

• Bayesian methods in health economics. (HE) This topic is concerned with evaluating the cost-effectiveness of medical treatments.

  Links: general information; CHEBS, the Centre for Bayesian Statistics in Health Economics.

  See my publications page for abstracts and downloads of recent papers (coded HE). In particular, my key papers in this field are as follows: [1] O'Hagan and Stevens (2002) is a review of work on the assessment of cost-effectiveness using data from clinical trials; [2] Stevens, O'Hagan and Miller (2003) is a tutorial paper based on an application of those methods, which was awarded a prize for the best paper in the journal Pharmaceutical Statistics in 2003; [3] Kharroubi, O'Hagan and Brazier (2005) introduces a Bayesian method for estimating health-related quality of life; [4] O'Hagan, Stevenson and Madan (2007) provides efficient ways to estimate cost-effectiveness using patient-level simulation models.

• Elicitation of expert knowledge. (EL) A key task in applications of Bayesian methods is the expression of expert knowledge and opinions in the form of probability distributions.

  Links: general information; BEEP, the Bayesian Elicitation of Experts' Probabilities project; the book that came out of the BEEP project; SHELF, the Sheffield Elicitation Framework.

  See my publications page for abstracts and downloads of recent papers (coded EL). In particular, my key papers in this field are as follows: [1] O'Hagan (1998) sets out my early work in the field; [2] Oakley and O'Hagan (2007) introduces a novel approach to quantifying uncertainty in elicitation.

I have also done research in various other areas in the general field of Bayesian statistics. Some examples are given below, and there are papers on these and other topics to be found on my publications page.

• Bayesian inference for unknown functions. There are many situations where we wish to estimate or make other inference about an unknown function. The topic of Bayesian analysis of computer code outputs is actually one example of this, where the function of interest is the computer code itself. Other examples that I have worked on include inference for the radiocarbon calibration curve (see the Bayesian archaeology research pages in the Department of Probability and Statistics web site) and interpolation of pollution monitoring stations.

• Robust Bayesian analysis. I have worked for some time with modelling using heavy-tailed distributions, which produce a kind of automatic robustness to outlying observations or parameters. The theory of these models concentrates on asymptotic properties as the conflict between data sources (e.g. conflict between outliers and other observations) becomes extreme. Whereas earlier research dealt with conflict in location parameter models, some new work with Ailton Andrade (see my publications page) is addressing scale parameter conflicts. Another aspect of robust Bayesian analysis is model criticism.

• Bayesian methods in auditing. In a research contract from the National Audit Office (NAO), David Laws and I developed Bayesian methods for auditing complex organisations. This work continued with a CASE PhD studentship supported by the NAO. I have recently been advising the Audit Bureau of Circulation on Bayesian auditing methods, and a paper is in preparation concerning repeated audits with application to monitoring of health care provision in New York state.

• Fractional Bayes factors. The Fractional Bayes Factor is a device I developed for Bayesian comparison of models when the prior information about parameters within the models is weak. It has become established as one of the leading techniques in this area.