Applications of Structural Equation Models: Case Studies in

In scientic research, data acquired in time-series and cross-sectional form or from experiments with repeated measures often exhibit strong association in the response variables. The phenomenon sometime may be explained by theoretical reasoning or can be attributed to the existence of some structural relations that linked the variables together. When latter is alleged, one usually attempts to discover the underlying structure and hence provide an explanation for the relationships. It is then natural to specify a model that will allow adequate description of interdependency of these variables. Structural equation models, in many instances, can be employed to achieve this purpose. In this talk, we illustrate the model building and estimation process by studying three types of data. The rst type is the usual experimental data collected from a human research study designed to clarify central and peripheral mechanisms that govern the modulation of human autonomic cardiovascular rhythms [2]. The second type is the retrospective observational data extracted from air trac management database to study the causal relationship between air trac performance and contributing factors [3]. And the third type is the experimental data from an industrial human factor experiment designed to understand the eect of cabin altitude and arterial oxygen saturation level (SaO2) on passenger comfort during a prolonged ight [4]. In the study of cardiovascular data, theory in physiology and results from exploratory data analysis are used to hypothesize the dynamic structure of the system. Structural equation models are employed to validate th