Why I Became a Scientist
Evidence-based health care relies on accurate information which in turn needs to be extracted from large amount of research data, using rigorous statistical methods. I became a scientist in biometrical research because I want to help medical researchers by providing principled statistical tools. I have developed several useful methods useful for medical research. Of most notable is the development of a modified Poisson model for assessing effect of risk factors on human disease in 2004. This model has now been widely used in a wide range of disciplines.
As a scientist of biometrics, I focus on developing and evaluating statistical methods that are useful for medical research, with an emphasis on estimation in the context of the following substantive areas:
- Cluster randomization trials;
- Prospective epidemiological studies;
- Inter-observer agreement and instrument reliability;
- Genetic association studies;
- Health related quality of life;
- Health economic evaluation.
How to design and analyze cluster trials which randomize intact social units, rather than individual patients?
Randomization by cluster is useful in avoiding logistic, administrative or ethical issues which may otherwise arise. It inevitably creates new methodological issues. My research will provide statistical tools in correctly analyze such trials, as demonstrated in a recent trial randomizing physicians in evaluating a new care algorithm for hypertension patients.
How to use research data to draw valid conclusion, since what you see is not what you get in research?
Collecting research data is expensive and time-consuming. Yet useful information does not jump out dat a automatically, it requires rigorous statistical tools to draw inference about subjects that are not considered in the research.
How to unify the statistical significance and scientific significance?
Many researchers have been relying on statistical significance i.e., whether P-value is less than 5%, to draw scientific conclusion. My research will provide alternative ways of summarizing data without using P-values. This approach will ultimately help researcher in all disease areas to interpret research results.
- BSc (1985, Agronomy, Southwest University, Chongqing, China)
- MSc (1995, Soil Science, McGill University, Montreal, Canada)
- MSc (1998, Statistics, University of Connecticut, Storrs, USA)
- PhD (1998, Plant Science, University of Connecticut, Storrs, USA)
- PhD (2002, Biostatistics, The University of Western Ontario, London, Canada)
- Early Researcher Award, Ontario Ministry of Research & Innovation (2007)
Zou GY, Donner A (2006). The merits of testing Hardy-Weinberg equilibrium in the analysis of unmatched case-control data: A cautionary note. Annals of Human Genetics 70: 923-933.
Zou GY, Donner A (2008). Construction of Confidence limits about effect measures: A general approach. Statistics in Medicine 27: 1693-1702.
Zou GY (2008). On the estimation of additive interaction using the four-by-two table and beyond. American Journal of Epidemiology 168: 212-224.
Zou GY, Taleban J, Huo CY (2009). Confidence interval estimation for lognormal data with application to health economics. Computational Statistics & Data Analysis 53: 3755-3764.
Donner A, Zou GY (2011). Estimating simultaneous confidence intervals for multiple contrasts of proportions by the method of variance estimates recovery. Statistics in Biopharmaceutical Research 3: 320-335.
Dept of Epidemiology & Biostatistics,
Robarts Clinical Trials of Robarts Research Institute
Schulich School of Medicine & Dentistry
The University of Western Ontario
London, ON, Canada N6A 5C1
Phone: 519-661-2111 ext. 86298