ABSTRACT
In this chapter we develop analyses for continuous outcomes from cross-over trials. These will largely be parametric, using conventional linear regression and analysis of variance, as introduced in Section 1.7. For particular settings these models will be extended to incorporate other random terms, and other covariance structures, i.e., we will be using linear mixed models. We have already met simple versions of such models with fixed and random subject effects in Chapters 2 and 3. Here we provide a systematic development. In addition we will briefly describe some methods of analysis that are robust to the covariance structure of the measurements from one individual.
