Research Seminar

Correlated error terms in structural equation models for categorical data


Johan Braeken


KU Leuven

Abstract: For structural equation models with categorical data, there is one practice regulary applied for Gaussian data that is not easily implemented:the technique of correlated errors. The problem lies mainly in the absence of a categorical analogue to the multivariate normal distribution and the presence of not-closed form formulas in structural equation models for categorical data. We will present a novel method to handle correlated errors that has a parsimonious structure and leads to straightforward interepretations without a too excessive computational burden. This method is based upon the concept of copula functions and will be introduced with a dataset of ordinal responses originating from a contextualized personality study.

Date: Tue Nov 25, 12:15 pm - 1:15 pm
Place: room 00.60 (Department of Psychology, Tiensestraat 102, 3000 Leuven)