ABSTRACT
This chapter explores the basic idea of the cross-over model, and shows how the concept of carry-over simplifies treatment interactions over time. Cross-over designs are used for experiments in which subjects are expensive and few in numbers, requiring a design that yields as much as possible from each. Subjects typically receive two or more treatments in succession, with possibly very little time between treatments. The cross-over basically involves a sample of subjects chosen at random from some population and assigned at random to a sequence of treatments to be given over several time periods, one treatment per time period. A cross-over design is similar to repeated measures except that the factor is assigned across the time periods. With more than two periods, different variants can arise depending on how the possible sequences of treatments are assigned to subjects across the groups.
