RSS01.D9 Structural Equation Modelling - Confirmed

Confirmed

Structural equation modeling (SEM) is a comprehensive method to test (causal) theories. It is popular among psychologists, sociologists, econometrists, and marketing researchers. In SEM a (causal) theory is represented by a model which describes how the concepts that form the basis of the theory are related to each other. The complexity of these relations goes beyond what is possible in multiple regression (or ANOVA). Therefore, with SEM we can study a greater variety of research questions. In addition, with SEM we can overcome some of the problems in multiple regression (or ANOVA) e.g., measurement error in the independent variables.

During the first two days we will start with an intuitive approach to SEM. This will help to understand what SEM is about; namely explaining relationships between (observed and latent) variables. We will cover estimation procedures and assumptions, model evaluation and testing. Model evaluation is of paramount importance, there are more than 50 ways to evaluate models in SEM. This makes it difficult to decide whether a model is to be rejected or accepted. These topics will be explained via in-class illustrations. These illustrations will introduce the topics for the last 3 days of the course. During the third day, causality is covered. The topic of causality is often avoided in courses on SEM. However, at its core SEM is a method to test causal theories (in observational research). Participants will learn how develop and test causal models in a (non-)experimental context. Once they see how causal modeling works, they will understand the importance of a study design that is used to answer a research question. During the fourth day the participants will learn how to test for mediation, moderation (among others using multi-group analysis). On the fifth day we will briefly discuss a number of advanced techniques e.g., measurement invariance, random-intercept cross-lagged panel models, power analysis, and depending on interest go deeper with some of them.

Participants will be able gain experience on the topics discussed in this course via self-study assignments. Participants should install R-Studio and the package lavaan. Lavaan is an acronym for latent variable analysis. It was developed by Yves Rosseel to provide a collection of tools that can be used to explore, estimate, and understand a wide family of latent variable models, including factor analysis, longitudinal, multilevel, latent class, item response, and missing data models.  After this course, participants will be able to use SEM in a variety of common research situations. In addition, they will have a strong critical basis for using SEM in more complex situations. Please note that this course requires basic knowledge of regression analysis