Crossover designs provide another form of repeated measurements. In a crossover design, subjects serve as their own controls and receive two or more treatments or conditions in two or more consecutive periods. You can use the GEE method to analyze such data, managing the subjects as clusters and managing the treatment as a time-varying covariate.
The following data are from a two-period crossover study investigating three treatments. These data were analyzed with conditional logistic regression in Chapter 10.
As described in Chapter 10, this is ...