A Model for Longitudinal Data

Abstract

In practice, longitudinal data are often highly unbalanced in the sense that not an equal number of measurements is available for all subjects and/or that measurements are not taken at fixed time points. In the rat data set and the toenail data set, presented in Section 2.1 and in Section 2.2, respectively, a fixed number of measurements was scheduled to be taken on all subjects, at fixed time points. However, during the study, rats died, and patients left the toenail study prematurely, implying unbalance. This is different from the prostate data and the hearing data (Sections 2.3.1 and 2.3.2, respectively), where the unbalance is an immediate result from the fact that the volunteers participating in the BLSA were asked to return approximately every 2 years for medical examination.