Introduction to Structural Equation Modeling
This course introduces students to structural causal models and structural equation modeling (SEM). SEM can handle multi-equation models, and allows estimation among latent (unobserved) and observed variables of multiple effects transmitted over combinations of paths. It combines measurement models (as in psychometrics) with path-models (as in econometrics). Using graphical interfaces to describe the models makes relationships more transparent and usable for policy analysis, for example the model for a Food Price Intervention where a healthy food discount impacts diabetes via a path through BMI.
We will present an introduction to structural causal models, an overview of latent variables, measurement error, SEM equations, subset applications in latent growth curve modeling, and latent class analysis. By using prepared datasets and exercises students will gain experience with limitations/pitfalls of SEM as well as the challenges that SEM can overcome that traditional regression cannot.