forum.smartpls.com

Dear all, I just wonder whether SmartPLS4 can create an interaction effect with the product indicator approach. The model that I am working on requires such features, since I am focusing on the complimentarity between two independent variables. But according to the SmartPLS official website (https://www.smartpls.com/documentation/ ... oderation/), it seems as if SmartPLS4 includes only the two-stage approach. The outline of the model is demonstrated in the attached image. Suggestions or advice are much appreciated.

Why do you want to use the product-indicator approach? According to the literature the two-stage approach is almost always preferable over the product-indicator approach in PLS.
Anyways you can also create the product indicators yourself outside of SmartPLS and import a datafile if you need them. However, I would always prefer the two-stage over the product-indicator approach.

Thank you for the advice. Now I have better understanding of the software's moderation features. I chose the product indicator approach due to its alignment with the model specifications. I have also come across articles that have employed the product indicator approach in their studies, such as Cohen & Olsen (2015). Furthermore, Hair et al. (2022) mentioned this approach in their book. Hence, I was curious as to why this feature is absent in the software despite being introduced in relevant literature.

References:
Cohen, J.F. & Olsen, K. 2015. Knowledge management capabilities and firm performance: A test of universalistic, contignency and complementarity perspectives. Expert systems with applications, 42, 1178-1188.

Hair, J. F., Hult, G. T. M., Ringle, C. M., & Sarstedt, M. (2022). A primer on partial least squares structural equation modeling (PLS-SEM). SAGE.

It was one of the first approaches that was proposed and it is also used in CB-SEM sometimes. However, latest PLS literature recommends the two-stage because it has been show in simulation studies that it is the overall best approach in a variety of models.

To quote Becker et al. (2023) p.334:
"How should the interaction term be generated?
The PLS-SEM literature discusses three approaches to generate the interaction term that maps the independent and moderator constructs’ joint impact on the criterion construct. In the past, the SEM literature often relied on the product indicator approach that cross-multiplies all of the moderator construct’s indicators with those of the independent variable to define the interaction term’s measurement model. However, in PLS-SEM, research has shown that this approach lags behind in terms of statistical power and parameter accuracy (Becker et al., 2018; Henseler and Chin, 2010). Researchers should instead rely on the two-stage approach, which uses the construct scores from a model estimation without the interaction term in Stage 1 as input to compute the interaction term in Stage 2. However, this approach cannot be easily implemented manually because in the second stage, the interaction term (the latent variables scores’ product) should not be standardized, which would happen if researchers were to simply enter its scores in a normal PLS path model. Researchers should, therefore, rely on PLS-SEM software that computes this result for them (e.g. SmartPLS, cSEM, SEMinR andWarpPLS)."

Becker, J.-M., Cheah, J.-H., Gholamzade, R., Ringle, C.M. and Sarstedt, M. (2023), "PLS-SEM’s most wanted guidance", International Journal of Contemporary Hospitality Management, Vol. 35 No. 1, pp. 321-346. https://doi.org/10.1108/IJCHM-04-2022-0474
Henseler, J. and W. W. Chin (2010). A Comparison of Approaches for the Analysis of Interaction Effects Between Latent Variables Using Partial Least Squares Path Modeling, Structural Equation Modeling, 17 (1), 82-109.
Becker, J-M., Ringle, C.M., Sarstedt, M. (2018). Estimating Moderating Effects in PLS-SEM and PLSc-SEM: Interaction Term Generation*Data Treatment, Journal of Applied Structural Equation Modeling, 2(2), 1-21. https://jasemjournal.com/wp-content/upl ... -JASEM.pdf