ABSTRACT
We come now to what is perhaps the most satisfying, and sometimes the most satisfactory, approach to repeated measurements, at least from the regression modelling point of view. It is essentially a generalization of the basic Anova model (3.1.1) of Chapter 3. The fixed effects means μ ij = Ε(y ij ) are extended to a general regression model μ i = X i β as in Chapter 5, and the random effects α ij similarly to Z i b i . The resulting covariance structure is more general than that of compound symmetry in Chapter 3, but more naturally restricted than the completely undisciplined form of sections 5.1 and 5.2. The ideas will be introduced via an idealized example.
