B0559
Title: Variational inference for structural equation models
Authors: Khue-Dung Dang - University of Melbourne (Australia) [presenting]
Luca Maestrini - The Australian National University (Australia)
Abstract: Structural equation models (SEMs) are commonly used to study the structural relationship between observed variables and latent constructs. Recently, Bayesian fitting procedures for SEMs have received more attention, as they overcome the issues of frequentist approaches when the number of observations is small and facilitate the adoption of more flexible model structures. Markov chain Monte Carlo procedures for Bayesian inference of SEMs have been developed, however, they are usually computationally expensive for complex structures. Variational approximations have been shown to be a fast alternative, but their application has been limited to very simple SEMs. Variational Bayes algorithms are developed, that tackle more challenging settings involving non-normal data and missing values. Their performance is then investigated in a simulated data study and a real data application.
