PhD in Statistics

Bayesian Statistics

The Bayesian approach to statistics has long been considered theoretically sound and has more recently made great inroads into practice. Current interests include systems risk and software project risk assessment, operational risk in finance, decision analysis with Bayesian networks, outliers and diagnostics for model choice, degradation models and inference for stochastic processes, with applications in medicine and engineering.

Design of Experiments Planning investigations so that they will produce useful data is at least as important as analysing the data which are collected. Research interests in this area include: experiments with multiple phases (eg a field phase followed by a laboratory phase), doseescalation designs in clinical trials, experiments in genomics and proteomics, industrial experiments with hard-to-set factors, design of measurement schedules for communication networks, experiments in enzyme kinetics and pharmacokinetics, discrete choice experiments in market research, design for generalised linear mixed models and computer experiments.

Econometrics and Time Series The science of economics is based largely on data arch in time series methodology continues todeal with the larger and more complex data sets which have become common in practice. Interests at Queen Mary include the econometric analysis of present value models, theory and application of non parametric methods to high frequency financial data, parametric and semiparametric estimation for weakly and strongly dependent time series models, ARCH type models, econometrics of auctions, adaptive nonparametric specification testing, nonlinear stationary processes and estimation of dynamic panel data models.

Medical Statistics The medical field continues to be a source ofchallenging statistical problems, as well as a major area of application of statistical methods. Particular interests at Queen Mary include cancer prevention and screening, design and analysis of cluster randomised trials, the assessment and communication of risk-benefit of medicines, systematic reviews of evidence, spatial epidemiology, regional demography, epidemiology of dense, highrisk breast patterns, evaluating the effectiveness of prenatal screening strategies and the prevention of cervical cancer in both the developed and the developing world.

Statistical Inference Frequentist and likelihood-based inference continues to require development in response to the ever morecomplex studies being carried out in applications. Current areas of interest include sequential analysis, asymptotic approximations, inference, medical applications and multivariate analysis.