MODEL SELECTION TECHNIQUES FOR THE COVARIANCE-MATRIX FOR INCOMPLETE LONGITUDINAL DATA

In longitudinal studies with incomplete data, where the number of time points can become numerous, it is often advantageous to model the cov ariance matrix. We describe several covariance models (for example, mi xed models, compound symmetry, AR(1)-type models, and combination mode ls) that offer parsimonious alternatives to unstructured <(Sigma)over cap>. We evaluate each covariance model with longitudinal data concern ing cholesterol as the repeated outcome measure. We discuss strategies for deciding the 'best' model and show a graphical technique for judg ing goodness-of-fit of covariance models.