11
Analysis of Longitudinal Data
In health-related studies, researchers often collect data from the same unit
(or subject) repeatedly over time. Measurements may be taken at different
times for different subjects. These are called longitudinal studies. Diggle,
Liang, and Zeger (1994) offer an excellent exposition of the issues related
to the design of such studies and the analysis of longitudinal data. They
also provide many interesting examples of data. We refer to their book
for a thorough treatment of the topic. The purpose of this chapter is to
introduce the methods based on recursive partitioning and to compare the
analyses of longitudinal data using different approaches.
11.1 Infant Growth Curves
The data for this example were collected from a retrospective study by Dr.
John Leventhal and his colleagues at Yale University School of Medicine,
New Haven, Connecticut. Their primary aim was to study the risk factors
during pregnancy that may lead to the maltreatment of infants after birth
such as physical and sexual abuse. The investigators recruited 298 children
born at Yale–New Haven Hospital after reviewing the medical records for
all women who had deliveries from September 1, 1989, through September
30, 1990. Detailed eligibility criteria have been reported previously else-
where such as Wasserman and Leventhal (1993) and Stier et al. (1993).
The major concern underlying the sample selection was the ascertainment
of cocaine exposure. Two groups of infants were included: those whose
©
H. Zhang and B.H. Singer, Recursive Partitioning and Applications,
Springer Series in Statistics, DOI 10.1007/978-1-4419-6824-1_11,
163
Springer Science+Business Media, LLC 2010
剩余35页未读，继续阅读
评论0