Dynamic models with latent variables from a system theoretic perspective: theory and applications

Abstract

In the paper Jöreskog's general static model with latent variables (LISREL) and extensively used in social sciences will be generalized to a dynamic version which, in structure, is equivalent to a stationary version of a discrete state-space model. Parameter identifiability and estimation aspects will be briefly discussed and several examples of applications are given such as dynamic factor modelling, errors-in-variables, models with missing observations, control models with latent variables, .... The so-called state-space approximation approach of time series modelling will be emphasized.