Factorial experimental designs as a means of improving experiments with animals

Michael FW Festing
MRC Toxicology Unit, Hodgkin Building, University of Leicester, PO Box 138, Lancaster Road, Leicester LE1 9HN, UK.

It is important that animal experiments are well designed, giving the maximum amount of information and using the minimum number of animals consistent with achieving the scientific objectives. Factorial experimental designs, in which more than one experimental factor is varied at a time, make it possible to study many different treatment combinations simultaneously, without using excessive numbers of animals. As an example, we recently completed an experiment to study chemoprevention of lung tumours in mice using six chemopreventive agents, two carcinogens and four mouse strains, or a total of 48 treatment combinations. Useful and repeatable results were obtained even though the experiment only involved 72 mice, or an average of 1.5 mice per treatment combination. Factorial designs are quite widely used, though they are often analysed incorrectly. Use of an analysis of variance makes it relatively easy to interpret the results of such experiments even if there are large numbers of treatment combinations.
Back to the Scand-LAS 1998 Symposium programme.
This page is provided by the server of the Laboratory Animal Unit, at the Norwegian College of Veterinary Medicine, Oslo.