RSS02.D10 Advanced Structural Equation Modelling

We, along with others, offer introductory structural equation modeling courses. This course starts where those courses ended.  If you would like to learn more about advanced tricks SEM can offer, apply SEM to longitudinal or multilevel data structures or simply would like to have in depth discussions about advanced SEM topics or deep considerations of model building, this is the course for you.  (On the other hand, if you do not have such foundations in SEM, please take a look at our introductory SEM class.)  We cover a large number of topics typical introductory SEM classes do not go into expanding both the modeling flexibility and proficiency of class participants.

We start with ESEM (e stands for exploratory), a technique designed to merge exploratory factor analysis with structural equation models. Given that we will be discussing factors and factor structures, we will also find the time discuss creative modeling approaches to response styles, reverse coded items and other complications common in CFA, SEM and ESEM.

On the second day we move to longitudinal data structures. We look at how granger causality is modeled in the SEM framework and how these techniques can be expanded to account for measurement error in longitudinal data without the acquisition of multiple indicators. We point to how SEM can actually model most longitudinal econometric models more flexibly than the status quo techniques. Secondly, we build and discuss growth and trend models, most often seen in the multilevel modeling framework, using latent variables and show how SEM offers added flexibility to some of these models.

On the third and fourth day we move to a quick introduction of multilevel structural equation modeling.

On the final day, we will briefly discuss a large number of advanced techniques, and depending on interest go deeper with some of them. These techniques could include the modeling of latent heterogeneity, missing data, different estimators and link functions, the relationship between CFA and IRT, and/or latent interactions.