Murray Donald Smith
Associate Professor, Faculty of Science
Contact
Biography
BEc(Hons) and PhD (Monash University, Australia)
I am an econometrician with interests in microeconometric modelling and health informatics. Specifically, the development of statistical methodology for the modelling of variables associated with health and health-related behaviours. Data for my analyses are sourced from large, individual-based secondary datasets such as country-wide health surveys, hospitalisation episodes and morbidity records, and gp prescribing records. The experience that lead to this direction of research had its origins at Monash University where, in 1983, I gained a first class honours degree in Economics majoring in econometrics and operations research. By 1987 I had gained my PhD having studied aspects of exact finite-sample statistical theory under the supervision of Grant Hillier. My first academic appointment was in the School of Economics at the University of New South Wales, and my second, spanning almost 20 years of service, was at the University of Sydney in the Department of Econometrics and Business Statistics. I then moved from Australia to the UK where I held the post of Senior Research Fellow at the University of Aberdeen in their Health Economics Research Unit from 2007-9. On January 1, 2010, I took up my current position at the University of Nottingham.
I have held a number of visiting research positions including: the University of East Anglia in 1993 where I was a Leverhulme Trust visiting research fellow in the Department of Economics; the University of Munich (Institute for Statistics) in 1998-9; and the University of Dortmund (Institute for Economics and Social Statistics) in 2001. My visits to Germany were made possible through sponsorship by the Alexander von Humboldt Foundation: http://www.avh.de.
I serve as a committee member on a National Institute for Health and Clinical Excellence (NICE) Technology Appraisal Committee (September 2010 - present).
Teaching Summary
My main involvement in teaching has been at the universities of Sydney and New South Wales. Because of my background in econometrics and statistical theory I have been able to teach across a range of… read more
Research Summary
My current research focuses on the use of copulas to represent statistical dependence structures in multivariate microeconometric modelling settings. Development of this methodology in the context of… read more
Recent Publications
ALYAMANI, N., SMITH, M. D., WILLIAMS, D. and PETTY, R., 2010. Predictive biomarkers for personalised anti-cancer drug use: discovery to clinical implementation European Journal of Cancer. 46(5), 869-879 SMITH, M. D., 2008. Stochastic frontier models with dependent error components The Econometrics Journal. 11, 172-192 DANCER, D., RAMMOHAN, A. and SMITH, M. D., 2008. Infant mortality and child nutrition in Bangladesh Health Economics. 17, 1015-1035
My main involvement in teaching has been at the universities of Sydney and New South Wales. Because of my background in econometrics and statistical theory I have been able to teach across a range of topics including: principles of economic statistics; introductory statistics; introductory, intermediate, and advanced econometrics; computing; and operations research. Advanced courses that I have taught include limited and qualitative dependent variable modelling (econometric modelling and computational methods), statistical modelling and computation, and time series analysis (ARIMA modelling). My preferred area of teaching reflects my research interests in microeconometrics, with emphasis on issues surrounding limited and qualitative dependent variable models.
At the University of Nottingham my teaching covers introductory statistics, final year project supervision and PhD supervision.
Current Research
My current research focuses on the use of copulas to represent statistical dependence structures in multivariate microeconometric modelling settings. Development of this methodology in the context of potential sample selection influences on the variables of interest is my main focus of attention. Such influences are frequently present in microeconometric datasets where individual responses on lifestyle choices and opinion are recorded alongside other socioeconomic and demographic data. Participatory decisions, intensity decisions, and so on can all influence, and be influenced by, other outcome variables. Similarly, individual decision making in regard to medicine use can influence the risk and time to incidence of, for example, a disease outcome, and so is a further area of study to which this methodology can be applied. By taking a flexible, copula approach to statistical modelling - in a nutshell individually modelled components are bound together by families of copulas - a wider range of model constructions, even within the confines of fully parametrised models, are often estimable with standard optimisation algorithms.
This methodology is being used in a 30-month research project that commenced in November 2008 in which I am a Co-Investigator. Others in the project team are located at the universities of Aberdeen and StAndrew's, NHS Scotland and the Medical Research Council (MRC). This is a National Preventative Research Initiative Phase 2 grant administered by the MRC, value £243,926.
Our contention is that the likelihood that conditions such as heart disease, cancers, diabetes and strokes will occur can be affected by the lifestyles that we adopt - such as smoking, over eating, not taking enough exercise, etc. We aim to provide information for policy makers, the NHS and individuals about which interventions to change unhealthy behaviours are effective, as well as cost-effective. Such information will assist in making decisions about how best to use the limited resources available to improve the health of the nation. The idea is that the consequences (in terms of ill health and costs) of different types of lifestyle are affected by an individuals personal circumstances and the environment in which they live. It is these factors that either promote or provide barriers to an individual changing their behaviour.
The main research questions concern: establishing the contribution of different lifestyle factors to the ill-health and mortality from CHD, cancer, stroke and diabetes that might be prevented; how this preventable ill-health varies between different areas and different social groups who differ in the amount of deprivation they suffer; and the relative costs and benefits of alternative ways of changing behaviour and their impacts on health inequalities.
Past Research
One long-standing piece of research that has greatly influenced my approach to computational work in econometrics has been the use of symbolic computer algebra, oriented especially to tackle problems in mathematical statistics. In a decade-long collaboration with Colin Rose (http://www.tri.org.au) outcomes produced include the mathStatica software package that adds on to the Mathematica programming language, with uses that are illustrated extensively throughout our book Mathematical Statistics with Mathematica (2002); for details visit http://www.mathstatica.com.