Multilevel Regression and Multilevel Structural Equation Modeling
Joop J. Hox · Oxford University Press eBooks · 2013
Multilevel modeling in general concerns models for relationships between variables defined at different levels of a hierarchical data set, which is often viewed as a multistage sample from a hierarchically structured population. Common applications are individuals within groups, repeated measures within individuals, longitudinal modeling, and cluster randomized trials. This chapter treats the multilevel regression model, which is a direct extension of single-level multiple regression, and multilevel structural equation models, which includes multilevel path and factor analysis. Multilevel analysis was originally intended for continuous normally distributed data. This chapter refers to recent extensions to non-normal data but does not treat these in detail. The end of the chapter presents some statistical issues such as assumptions, sample sizes, and applications to data that are not completely nested.