ABSTRACT
Cross-cultural research oen deals with hierarchical data structures, either due to the sampling procedure, or to characteristics of sampled units that are related to a grouping variable. Multilevel models oer the possibility to take the resulting dependency between group and individual characteristics into account. In addition to the more traditional technique to deal with hierarchical or nested data, which is multilevel regression analysis, both multigroup structural equation models (SEM) and multilevel SEM are available to analyze such data. Both of these approaches share the well-known advantages of SEM. Among others this means that latent variables can be included in the analysis and that variables can be endogenous (outcomes) as well as exogenous (predictors). is provides very exible model building possibilities, for instance to model mediation or moderation processes. Multilevel SEM combines the advantages of multilevel analysis and SEM. Compared to multigroup SEM, the sample size of the grouping variable is not limited, and it is possible to include
group level variables in the model. An example of a multigroup SEM can be found in Chapter 7 of this book.
