STRUCTURAL EQUATION MODELLING (SEM): OVERVIEW, APPLICATIONS, AND MISAPPLICATION IN SCIENCE AND TECHNOLOGY EDUCATION

Abstract

Structural Equation Modelling (SEM) has become an increasingly important 
multivariate statistical technique in educational and social science research due to 
its ability to model complex relationships among observed and latent variables 
simultaneously. Traditional techniques such as multiple regression analysis and 
analysis of variance (ANOVA) are limited to directly measured variables and are 
less effective in handling measurement error, indirect effects, and complex model 
structures. This paper provides a comprehensive overview of SEM with special 
emphases on its conceptual foundations, common terminologies, types, and 
implementation procedures. Special attention was also given to the distinction 
between covariance-based SEM (CB-SEM) and variance-based SEM, also known 
as partial least squares SEM (PLS-SEM), including their underlying assumptions, 
objectives, and appropriate conditions for use. The paper further outlines the major 
steps involved in SEM implementation which include model specification, 
measurement model evaluation, and structural model assessment within the PLS
SEM framework. Most importantly, frequent misapplications of SEM in 
educational researches were highlighted. This paper aims to guide researchers 
toward more rigorous, theoretically grounded, and meaningful applications of SEM 
in educational research.