Bayesian Structural Equation Modeling

In August, I organised a three-week summerschool with my colleagues in Utrecht on new features in MPLUS 7, the software we use to build Structural Equation Models. Video’s of all lectures can be found here.

The latest issue of Psychological Methods contains background reading of Bayesian SEM, which is the main new feature of SEM. I find this fascinating stuff, and can think of hundreds of articles that could be written about replications of Maximum Likelihood based approaches.

I think this is gonna be extremely interesting to survey methodologists as well, because any parametric model (for example models where variances are assumed to have a normal distribution, or factor loadings are assumed to be equal across the sample), can now be estimated as a non-parametric model. This will be huge, I promise.