Bayesian Methods for Analyzing Structural Equation Models With Covariates, Interaction, and Quadratic Latent Variables

Comparing intention to avoid malware across contexts in a BYOD-enabled Australian university: A Protection Motivation Theory approach

“…Before concluding with the multiple-group SEM results, Bayesian analysis was performed to overcome the limitations of the sample size (n = 252) and the non-normality distributions while adding rigour to such results with their posterior distributions (Lee et al, 2007). This process was conducted using Bayesian SEM function of AMOS software.…”

Maximum Likelihood and Bayesian Estimation for Nonlinear Structural Equation Models

“…Implentation of Mplus for a similar interaction model can also be seen in the Mplus Userguide (Muthen and Muthen 2007) Example 5.13. A similar implementation of Winbugs for the second-order cross product model has been previously demonstrated by Lee et al (2007) and Lee (2007) Chapter 8. In a slightly different set-up, the implentation of PROC NLMIXED for nonlinear SEM has been previously demonstrated by Patefield (2002).…”

“…Over the last 20 years though there have been great advances in the statistical computation methods for maximizing intractable likelihoods and generating from intractable posterior distributions. Building on these computational methods, there is a growing literature focused on using direct maximum likelihood and Bayesian methods for estimation specifically for different forms of nonlinear structural equation models: using full maximum likelihood there is, e.g., Klein, et al (1997), Klein and Moosbrugger (2000), Amemiya and Zhao (2001), Lee and Zhu (2002), Lee and Song (2003a), and Lee et.al (2003); and using Bayesian methods there is, e.g., Wittenberg and Arminger (1997), Arminger and Muthén (1998), Zhu and Lee (1999), Lee and Zhu (2000), Song and Lee (2002), Lee and Song (2003b), Lee et al (2007), and Lee (2007).…”

Bayesian analysis of non-linear structural equation models with non-ignorable missing outcomes from reproductive dispersion models

“…In modern educational, medical, social and psychological studies, various structural equation models have been developed to identify the latent variable from the manifest variables, and to assess the relationships of latent variables among themselves [1][2][3][4][5]. A lot of theories and methods have been proposed to analyze structural equation models in various fields on the basis of the assumptions of manifest variables from normal distribution.…”

“…Recently, Lee & Tang [4] proposed a novel Bayesian method to analyze non-linear structural equation models with manifest variables from an exponential family. In particular, Lee, Song & Tang [5] introduced a Bayesian method to analyze a general structural equation model that accommodates the general non-linear terms of latent variables and covariates. Also, there are more than a dozen statistical software packages that have been developed to satisfy the strong demands in various fields, for example, EQS6 [1], LISREL [3], Mplus [6] and WinBUGS [7].…”