Scroll to Section:

Social science researchers need to use modeling to understand complex real-life phenomena. But how does a researcher decide which of the available models is most appropriate? In this video, MARCO SARSTEDT analyzes the metrics employed by researchers in assessing PLS (Partial Least Squares) models, outlining how such assessments can be optimized. Running a Monte Carlo simulation study, Sarstedt explains the inadequacies (for PLS researchers) of commonly used metrics like R² and the Goodness of Fit index by comparison with information criteria like BIC and GM. Offering suggestions as to how these metrics should ideally be implemented, Sarstedt notes that further work is required to assess whether their advantages extend to studies which employ more complex modeling.


Marko Sarstedt is Chaired Professor of Marketing at Ludwig-Maximilians-University München and won the 2018 Research Award. He is also an Adjunct Professor at Babeș-Bolyai University, Cluj. Sarstedt has previously worked at the University of Newcastle (Australia) and Ludwig Maximilian University of Munich. His research focuses on consumer behavior and on the improvement of marketing decision making. The winner of five Emerald Citations of Excellence and two AMS William R. Darden awards, in 2020, Sarstedt was judged the second most influential business researcher in Germany, Austria and Switzerland (F.A.Z.-Ökonomenranking).


Original publication

PLS-Based Model Selection: The Role of Alternative Explanations in Information Systems Research

Sharma P. N., Sarstedt M., Shmueli G., Kim K. H. and Thiele K. O.
Journal of the Association for Information Systems
Published in 2019