ABSTRACT
The advantages of split-plot design are that the effects of subplot treatments and their interaction with the whole-plot treatment, sex, can be estimated more precisely than can the main effect of sex, which is assumed to be of no major direct interest in the experiment. The design necessitates two different estimates of error. The key first step in the interpretation of factorial systems such as this is the calculation and inspection of mean responses corresponding to the various combinations. Further analysis is concerned partly with assigning standard errors to the resulting contrasts and partly with finding simple summaries of the many comparisons possible. Comparisons of treatments within males, or within females, are independent of systematic effects between days. A further general question concerns the testing of interactions for significance. Interpretation is, of course, much simpler in the absence of interactions; only marginal means need to be thought about.
