Location: Room 1.62, Franklin-Wilkins Building, Waterloo campus
The analysis of variance and the analysis of causes
Speaker: Professor Neil Pearce, Professor of Epidemiology and Biostatistics, London School of Hygiene and Tropical Medicine
This seminar discusses the difference between the analysis of variance (the assessment of variation) and the analysis of causes (causal inference), and discusses the theoretical and practical implications of this distinction for epidemiological research, including data analysis and modelling strategies.
The seminar will address such questions as: Is it possible to have a strong and important causal factor for a disease outcome which has small predictive ability (i.e. explains a small portion of the variability) for that same outcome? If we are interested in explaining disease aetiology, how do we make sure that we do not miss the causal factor, when applying standard statistical modelling? In building a model to include confounders, is it possible to actually choose the wrong confounder through widely used variable selection processes (like e.g. stepwise regression)?
The presentation will give evidence of how it is possible, under different situations of exposure, outcome and confounder distributions, through the application of widely used methodology, to actually miss the point and will provide alternative pathways for a better understanding and more appropriate applications.