A COMPARISON BETWEEN COVARIANCE - BASED (CB) AND PARTIAL LEAST SQUARES (PLS) STRUCTURAL EQUATION MODELING : APPLICATIONS, ASSUMPTIONS, AND APPROPRIATENESS

บทคัดย่อ

This study compares Covariance-Based Structural Equation Modeling (CB-SEM) and Partial Least Squares Structural Equation Modeling (PLS-SEM). We emphasize the distinct scopes, techniques, and applications of each. CB-SEM is specifically designed to evaluate hypotheses and validate them. It is effective with substantial sample sizes and data sets commonly disseminated and extensively used chiefly in social sciences, psychology, and education. The maximum likelihood estimation approach estimates parameters, emphasizing fit indices such as Chi-square, RMSEA, CFI, and TLI. In contrast, Partial Least Squares Structural Equation Modeling (PLS-SEM) is particularly suitable for doing exploratory research or making predictions. It can handle lower sample sizes and data that does not follow a normal distribution. Furthermore, it is prevalent in business, marketing science, information systems, and management research. An iterative technique optimizes the explained variance in Partial Least Squares Structural Equation Modeling (PLS-SEM). This algorithm focuses on key features such as R-squared, Q-squared, Average Variance Extracted (AVE), and Composite Reliability (CR), which are indicators of prediction quality. This research examines how measurement error is addressed in CB-SEM by explicitly including terms, whereas PLS-SEM takes an implicit method. Furthermore, it examines model fit indices and data needs to assist researchers in selecting between these two methodologies, providing them with a practical reference. This document offers comprehensive guidance that considers the study goals and the specific features of the accessible data sets. The preceding discourse enhances the use of rigorous techniques and offers valuable direction to researchers.