Home

Abstract: This paper discusses methodological and analytic issues related to the study of causal structures in empirical educational, behavioral, and social research. Following an introductory discussion of the increasingly popular latent variable modeling (LVM) methodology, the concern is with its application for testing causal theories. The possibility of using LVM for evaluation of consistency between empirical data and implications of causal hypotheses is then focused on. The method of propensity score analysis is subsequently discussed, and an approach to its utilization for treatment effect examination within the framework of LVM is outlined and illustrated on data. A concluding section deals with limitations of this approach to studying causality.

Key words: causality – hypothesis testing – latent variable modeling – propensity score analysis

[1]This paper represents the contents of a workshop offered to doctoral students and junior faculty participating in the Firth Meeting of the Section “Empirical Educational Research” in March 2009, organized by the Central Institute for Empirical Educational Research (ZEPF) at the University of Koblenz-Landau. The support by the Organizing Committee, which made this workshop possible, is gratefully acknowledged. I am indebted to P. Nenniger and J. C. Anthony for valuable discussions on causality studies using structural equation modeling.

Analysis of causal structures using latent variable modeling