Experimental design

Experimental design is the definition of the experimental conditions for which the responses of interest will be, or have been, measured in an experimental study. Each condition defines the levels of each factor that is varied in the experiment. Note that it is important for the experimenter to be able to set the factors at the levels specified by the experimental design.

Types of designs

There are many standard experimental designs, e.g. factorial and fractional factorial designs, composite designs, etc. However, it is often better to use a tailor-made design for the study where it will be used. Such a design can be constructed using computer algorithms as this will ensure their efficiency.

Important principles are followed in the construction of the designs in order to ensure their efficiency. Use of randomization is used in order to minimize the risk of bias and blocking is used to control variability when possible or when this guarantees the required validity of the results.

Statistics

Statistics plays a very important role in the analysis of the data and their interpretation. Therefore it is important to ensure statistical support for every experimental study from the moment it has been decided that the study is necessary.

Experts/users at Manchester

• Dr Alexander Donev

References

Atkinson, A. C., Donev, A. N. and Tobias, R. D. (2007). Optimum Experimental Designs, in SAS. Oxford University Press.