ABSTRACT
A number of estimation procedures have been suggested for the analysis
of repeated measurements. These include likelihood, pseudo-likelihood, and
generalized least squares methods. In this chapter, we attempt to link
together these different methods using the principle of least squares (LS) and
generalized estimating equations ( G E E ) as our guide. W e start by assuming
that the first two moments of the response variable are known. As we add
further assumptions (e.g., normali ty) , we find these methods reduce to more
familiar forms of estimation such as max imum or pseudo max imum likelihood.