Structural equation modeling depicts an extension of the classical factor analysis, which allows explaining relationships amonglatent variables (constructs). It allows to empirically validate theoretically established causal models in the various social science disciplines such as marketing (e.g., to perform research on brand equity, consumer behavior, and customer satisfaction) or management (e.g., to evaluate factors that influence of alliance networks on firm performance). Covariance-based structural equation modeling (CB-SEM) and partial least squares SEM (PLS-SEM) constitute the two matching statistical techniques for estimating structural equation models. Both apply to the same class of models—structural equations with unobservable variablesand measurement error—but they have different structures and objectives.

CB-SEM is usually used in social sciences to empirically estimate relationships in causal models. Apparently, there has been little concern about the frequent inability of empirical data to meet methodological requirements or about the common occurrence of improper solutions. In comparison with CB-SEM, HermannWold’s basic PLS-SEM design or basic method of soft modeling rather represents a different statistical method. Soft modeling refers to the ability of PLS-SEM to be more flexible in handling various statistical modeling problems insituations where it is difficult or impossible to meet the hard assumptions of more traditional multivariate statistics. Within this context, "soft"is only attributed to distributional assumptions and not to the concepts, the models or the estimation techniques. The goal of PLS-SEM is the explanation of variances (prediction-oriented character of the methodology) rather than explaining covariances (theory testing via CB-SEM). The application of the PLS-SEM method is of special interest if the premises of CB-SEM are violatedand the assumed relations of cause-and-effect are not sufficiently explored. An additional advantage of the PLS-SEM method is the unrestricted incorporation of latent variables in the path model that either draws on reflective or formative measurements models.

Presentations and SmartPLS exercises. Most of the workshop will involve “hands-on” analysis of the dataset using the SmartPLSsoftware. SmartPLS software output diagnostics and proper interpretation of the results will be covered. Potential obstacles and “rules-of-thumb” to ensure appropriate application of the technique will be covered. The dataset from the book on PLS-SEM (Sage, 2017) by Hair, Hult, Ringle, and Sarstedt will be used for demonstration purposes.

Grading is based on in-class participation (50%) and a take-home exam (50%), which will be administered after the course.

The lecturer will be available after class and in the time following the workshop.