Course 4
Design of complex experiments
Presenters: Andrew Mead and Steven Gilmour Room: Sidney
(Sunday, 10 July 2016 from 9:00 am – 5:00 pm)
Summary: Most biological research obtains data in well-designed experiments to obtain answers to scientific questions. Since Fisher’s initial developments in the 1920s, the statistical principles for designing experiments have had enormous impact, remaining the basis for designing modern biological experiments. However, modern experiments are becoming increasingly complex, generating larger quantities of data than previously, so that applying the general principles to specific technologies or scenarios requires modern methodologies that are not so widely appreciated. This course aims to present some of those methodologies in modern experimental biology contexts. The emphasis will be on direct extensions from the general principles, illustrated using practical applications of the methodologies. Recent research will be put in the context of the standard methods for designing experiments with which course participants will be familiar. The course topics should be of interest to biometricians in most areas of biological experimentation, illustrated with a range of applications.
The aims of the course are to:
- Review the key principles underlying the construction of effective and efficient designs for experiments
- Review the analysis methods for mixed models
- Present and illustrate applications of design principles in four areas of current interest: multifactorial response surface designs for quantitative factors; designs for ‘omics technologies; use and abuse of split-unit designs; sequential aspects of experiments and experimental programmes
At the end of the course, participants should have an improved appreciation of the fundamental statistical concepts required for designing efficient and effective experiments to answer real questions in a range of modern biological science application areas. They should also be able to apply the presented statistical approaches to construct appropriate designs to address experimental objectives in other biological application areas and for experiments using new technologies.
Topics covered
Review of design principles; Multifactorial (response surface) designs for quantitative factors; Designing experiments for ‘omics technologies; Review of analysis methods for mixed models; the use and abuse of split-unit designs; Sequential aspects of experiments and experimental programmes.
Learning strategy
Topics will be introduced by a presentation, with time allowed for group discussions.
Pre-requisites
A good appreciation of the key statistical principles for the design of experiments (though these will be briefly reviewed) and some experience with linear models. A good appreciation of the issues associated with the design of experiments and some experience of designing real experiments in a range of biological application areas.
About the instructors
Andrew Mead is Head of Applied Statistics at Rothamsted Research, the world’s oldest agricultural research institute, having previously worked in the School of Life Sciences at the University of Warwick, and Horticulture Research International.
Steven Gilmour is Professor and Head of Statistics at King’s College London, having previously worked at the University of Reading, Queen Mary University of London and the University of Southampton.
They are co-authors, with Roger Mead, of Statistical Principles for the Design of Experiments: Applications to Real Experiments (2012) on which this course will be partly based.