Analysing data with repeated observations on each experimental unit

Summary Split-plot (or split-block) analyses are commonly applied to experimental results where several successive observations of the same variable have been recorded on each experimental unit. The assumptions required for such analyses receive scant attention and it often seems unlikely that these assumptions would be satisfied in experimental situations. Five sets of results are presented to support this proposition. An alternative analytical approach is suggested in which contrasts over time are analysed; such a method is always valid, computationally simple, and readily interpretable, and may also be used to gauge the validity of the split-plot analysis.