Change-Over Trials

Summary A number of designs for experiments in which sequences of treatments are applied to experimental units are considered. These experiments are described as change-over trials. The arrangements are such that estimates can be obtained of direct and first residual effects of treatments. The total number of units is kept as low as possible consistent with a neat analysis. It is found that confounding in factorial arrangements and incomplete block designs can be introduced by an adaptation of the methods of single period experiments. Only 3 and 4 period designs are included, but the extension to larger numbers of periods where this is practicable presents little difficulty. The statistical analysis of data from change-over trials is discussed. The ordinary method based on a straightforward application of the least squares principle is subjected to some criticism and a modification suggested.