Tuesday 23 July 13.00 - 14.00
Hall 8 (The ICC)
For many trainee scientists, it must seem that the papers they read are dressed up in statistics.
In fact many of the statistical practices that are used are misapplied, and act to conceal rather than reveal the original data. Advice on experimental design, data presentation, and statistics is widely available, but doesn’t seem to be well translated into practice.
This short session cannot provide a cure, but seeks to stimulate a fresh resolve in people too afraid to criticise the Emperor. I shall touch on design and data presentation (get that right and you’re nearly there), the logic or otherwise of statistical “proof”, interpreting tests, absence of evidence, nothing goes wrong, and replicates, what are they? The approach is to avoid equations and symbols as much as possible, and use analogy and pictures to demonstrate the underlying ideas.
This talk is not designed for statistical sophisticates.
Organised by Gordon Drummond, University of Edinburgh, UK and Statistics Editor, The Journal of Physiology