The Prowess and Pitfalls of Bayesian Structural Equation Modeling

Abstract

Muthén and Asparouhov introduced an approach for conducting Bayesian inference in the context of structural equation models that they termed Bayesian structural equation modeling (BSEM). In this article, we provide an overview of the BSEM technique, illustrate how this technique relates to confirmatory and exploratory factor analysis, and highlight several key problems with using the BSEM approach as it is currently advocated. Utilizing data from a large-scale study of entrepreneurial self-efficacy, we develop a modified approach for applying the BSEM technique in a manner that is more consistent with accepted principles of reflective measurement, factor analysis, and model selection. We devise a series of recommendations to guide future use of the BSEM technique to help ensure that mainstream use of this approach heralds the coming of a new day in measurement development rather than a false dawn.

A Primer on Moderated Mediation Analysis: Exploring Logistics Involvement in New Product Development

“…Given there is no “right” approach for model selection in moderated mediation, we echo Cudeck and Henly's () and Stromeyer et al. 's () recommendations for scholars to articulate the criteria utilized when making model selection decisions. Beyond selecting a mathematical model from a set of alternatives, scholars face the challenge of justifying the chosen structure vis‐à‐vis a universe of potential alternatives.…”

Investigating the theoretical structure of the DAS‐II core battery at school age using Bayesian structural equation modeling

“…Because of this, Bayesian estimation may permit an otherwise nonidentified model to be identified. The practice of estimating all correlated residuals is currently a topic of debate with some suggesting it should be avoided (Stromeyer, Miller, Sriramachandramurthy, & DeMartino, ) and others contending that it better clarifies an instrument's structure (Muthén & Asparouhov, ). Stromeyer et al.…”

“…() criticized the use of simultaneous estimation of small but informative correlated error terms; however, Asparouhov, Muthén, and Morin () suggested that Stromeyer et al. () may have misapprehended the approach and provided additional guidelines for the use of small variance correlated residuals. Asparouhov et al.…”

“…() concluded that instead of adding statistical noise to the model as Stromeyer et al. () suggested, the use of correlated residuals can be used to improve upon an understanding of a structural or measurement model. Despite Asparouhov's et al.’s (2015) clarification regarding how to properly estimate correlated residuals, generalized apprehension about its use remains, notably that its inclusion may result in models with limited theoretical meaning (Rindskopf, ) and cumbersome additional computations that does little to improve structural clarity (Stromeyer et al., ).…”

Bayesian latent variable models for the analysis of experimental psychology data

“…Some authors of the recent papers refer to a need for fully automated software that fits latent variable models via MCMC, with Mplus (Muthén & Muthén, 1998) providing SEM functionality and BUGS (Lunn et al, 2012), JAGS (Plummer, 2003), and Stan (Stan Development Team, 2014) providing general functionality. This automation has the reward of easily estimating complex models coupled with the risk of estimating inappropriate models or drawing inappropriate conclusions (also see MacCallum, Edwards, & Cai, 2012;Steiger, 2001;Stromeyer, Miller, Sriramachandramurthy, & DeMartino, 2014).…”