Types of SEM: Covariance-based SEM (CB-SEM) and Partial Least Squares SEM (PLS-SEM)

Structural Equation Modeling (SEM) encompasses different variations that are commonly used in research. Two widely known types of SEM are covariance-based SEM (CB-SEM) and partial least squares SEM (PLS-SEM). Let’s explore each type and their respective applications:

Covariance-Based SEM (CB-SEM):

CB-SEM, also referred to as variance-based SEM, focuses on modeling the covariance relationships among observed variables. It assumes that the relationships between variables are linear and that the data follow a multivariate normal distribution. CB-SEM is primarily used to confirm theoretical models and test complex relationships.

Applications of CB-SEM:

• Confirmatory factor analysis (CFA): CB-SEM is frequently used to assess the measurement properties of latent variables by examining the factor loadings, measurement errors, and model fit.
• Path analysis: CB-SEM allows researchers to test hypotheses about direct and indirect relationships between variables in a structural model.
• Mediation and moderation analysis: CB-SEM can explore the mediating and moderating effects between variables.
• Structural equation modeling for latent growth curve analysis: CB-SEM can be applied to analyze longitudinal data and model growth trajectories over time.

Partial Least Squares SEM (PLS-SEM):

PLS-SEM is a non-parametric approach that focuses on predictive modeling and provides a flexible framework for analyzing complex relationships. Unlike CB-SEM, PLS-SEM does not rely on assumptions of multivariate normality and linearity. It emphasizes the prediction of latent variables rather than confirmatory testing.

Applications of PLS-SEM:

• Exploratory factor analysis: PLS-SEM can be used to identify latent variables and their indicators in an exploratory manner.
• Formative measurement models: PLS-SEM allows for the modeling of formative measurement models where the latent construct is defined by its indicators.
• Complex models with small sample sizes: PLS-SEM is suitable for models with small sample sizes, as it requires fewer observations compared to CB-SEM.
• Moderation and interaction analysis: PLS-SEM can handle interactions and nonlinear relationships between variables.

It’s worth noting that the choice between CB-SEM and PLS-SEM depends on the research context, data characteristics, and specific research objectives. Both approaches have their strengths and can be applied to various fields such as social sciences, business research, psychology, and engineering.

Researchers should carefully consider the underlying assumptions, data requirements, and the nature of the research questions when selecting the appropriate type of SEM for their study.