Prof Paul Garthwaite
- Long-standing interest in subjective probability assessment, both the theoretical development of elicitation methods, experimental work with substantive experts to improve methods, and their application in practice.
- General interest in Bayesian methods. Recent work uses MCMC methods for multivariate meta analysis, the analysis of misclassified binary data, and the analysis of experiments to determine the selectivity of fishing nets.
- Monte Carlo and bootstrap methods. Joint work with Stephen Buckland developed a method for constructing Monte Carlo confidence intervals that is attracting attention. For example, the method is examined and recommended by Carpenter (JRSSB, 61 (1999), 159–172) and Matsui and Ohashi (Statist. Med., 18 (1999), 2409–2420).
- Collaborative work with Professor John Crawford to develop statistical techniques of special relevance to psychologists.
- Analysis of mark-recapture experiments.
- The evaluation of new health technologies, particularly the development of statistical methods that allow for the effect of learning during trials. Work was supported by a grant of £98000 in 1998-2000 (other co-holders were A M Grant, I T Russell and A Monk) and it is continuing through a PhD studentship funded by the MRC.
- Partial least squares and other biased regression methods, together with the analysis of near infrared spectra.
- Inter-disciplinary research, notably in animal and human nutrition, medicine and psychology.