ABSTRACT
This chapter presents the sphericity assumption and allows the covariance matrix to be arbitrary. Moreover, it permits individuals to be measured on different numbers of occasions. In particular, this has the merits that individuals with an incomplete set of measurements can be straightforwardly included in the analysis, and that individuals can have different covariates. Interest centres on whether the three groups grow in different ways. In particular, it might be natural to be interested in growth rate here. Therefore, the fitted regression models with a separate regression intercept and slope for each group. The BMDP5V commands for this are shown in the chapter. The equivalent to assuming a particular structure for the covariance matrix of observations is discussed. BMDP5V gives the observed covariance and correlation matrices for the data as well as estimates obtained using the particular form specified. This permits a comparison to be made between the observed and hypothesized forms of the matrix.
