Analysis of split‐plot designs: an overview and comparison of methods

Abstract Split‐plot designs are frequently needed in practice because of practical limitations and issues related to cost. This imposes extra challenges on the experimenter, both when designing the experiment and when analysing the data, in particular for non‐replicated cases. This paper is an overview and discussion of some of the most important methods for analysing split‐plot data. The focus is on estimation, testing and model validation. Two examples from an industrial context are given to illustrate the most important techniques. Copyright © 2006 John Wiley & Sons, Ltd.