The estimation process in the Bayesian quantile structural equation modeling approach

Document Type : Research Paper


1 Technical College of Management, Baghdad, Middle Technical University, Iraq

2 College of Administration and Economics, Department of Statistics, University of Baghdad, Iraq


latent variable models define as a wide class of regression models with latent variables that cannot be directly measured, the most important latent variable models are structural equation models. Structural equation modeling (SEM) is a popular multivariate technique for analyzing the interrelationships between latent variables. Structural equation models have been extensively applied to behavioral, medical, and social sciences. In general, structural equation models includes a measurement equation to characterize latent variables through multiple observable variables and a mean regression type structural equation to investigate how the explanatory latent variables affect the outcomes of interest. Despite the importance of the structural equations model, it does not provide an accurate analysis of the relationships between the latent variables. Therefore, the quantile regression method will be presented within the structural equations model to obtain a comprehensive analysis of the latent variables. we apply the quantile regression method into structural equation models to assess the conditional quantile of the outcome latent variable given the explanatory latent variables and covariates. The posterior inference is performed using asymmetric Laplace distribution. The estimation is done using the Markov Chain Monte  Carlo technique in Bayesian inference. The simulation was implemented assuming different distributions of the error term for the structural equations model and values for the parameters for a small sample size. The method used showed satisfactorily performs results.


[1] R. Alhamzawi and H.T. Mohammad Ali Brq: an R package for Bayesian quantile regression, METRON 78 (2020) 313-–328.
[2] D.F. Andrews and C.L. Mallows, Scale mixtures of normal distributions, J. Royal Stat. Soci. Ser. B 36(1) (1974) 99–102.
[3] D.F. Benoit and D. Van den Poel, Binary quantile regression: a Bayesian approach based on the asymmetric Laplace distribution, J. Appl. Economet. 27 (2012) 1174–1188.
[4] B.D. Dunson, J. Palomo and K. Bollen, Bayesian Structural Equation Modeling, Stat. Appl. Math. Sci. Institute, Technical Report, 2005.
[5] R. Everett, An Introduction to Latent Variable Models, Springer, 2003.
[6] A. Gelman, J.B. Carlin, H.S. Stern, B.D. Dunson, A. Vehtari and D.B. Rubin, Bayesian Data Analysis, Third Edition, A Chapman & Hall Book, 2014.
[7] R. Koenker and G. Bassett, Regression quantiles, Economet. 46 (1978) 33–50.
[8] R. Koenker, Quantile Regression, Cambridge University Press, London, 2005.
[9] H. Kozumi and G. Kobayashi, Gibbs Sampling Methods for Bayesian quantile regression, J. Stat. Comput. Simulat. 81 (2011) 1565–1578
[10] S.Y. Lee, Structural Equation Modeling: A Bayesian Approach, John Wiley & Sons, 2007
[11] X.Y. Song and S.Y. Lee,Basic and Advanced Bayesian Structural Equation Modeling: With Applications in the Medical and Behavioral Sciences, John Wiley & Sons, 2012.
[12] Y. Wang, X.N. Feng and X.Y. Song, Bayesian quantile Structural equation models, Struct. Equ. Model. A Multidisciplinary J. 23(2) (2016) 246–258 .
[13] Z. Yanqing and T. Niansheng, Bayesian empirical likelihood estimation of quantile structural equation models, J. Syst. Sci. Complex 30 (2017) 122–138.
[14] F. Yanuar, The estimation process in Bayesian structural equation modeling approach, J. Phys. Conf. Ser. 495 (2014) 012047.
[15] K. Yu and R.A. Moyeed, Bayesian quantile regression, Stat. Probab. Lett. 54 (2001) 437–447.
Volume 13, Issue 1
March 2022
Pages 2137-2149
  • Receive Date: 01 December 2021
  • Revise Date: 06 December 2021
  • Accept Date: 28 December 2021