ABSTRACT
In Chapter 3 we described the development of QLS that was originally presented
in three manuscripts (Chaganty, 1997; Chaganty and Shults, 1999; Shults and Cha-
ganty, 1998). Here we describe the implementation of QLS for the basic correlation
structures first described in the seminal paper on GEE (Liang and Zeger, 1986), as
well as for the Markov structure, which was not presented in Liang and Zeger (1986).
The Markov correlation structure is appropriate for analysis of longitudinal data that
are unequally spaced in time because it models the correlation as a function of the
time gaps between measurements. In Section 5.1 we discuss the different ways that
unequal spacing of data can arise.