Workshop on Structural Equation Modeling through Partial Least Squares
Partial Least Squares (PLS) path modeling is a structural equation modeling technique which is especially suitable when the assumptions for applying Maximum Likelihood based techniques (such as LISREL) do not apply, like in the case of exploratory research or small samples. In such cases, PLS is far less likely to lead to inappropriate solutions or non-convergence. The aims of this course are (1) to understand the basic intuition behind PLS, (2) to be able to apply the technique in testing a structural equation model, and (3) to be able to interpret the results.
The course is open for all ERIM research master students, Ph.D. students, and faculty members. It is assumed that participants have basic knowledge of multivariate statistical methods, such as multiple regression and factor analysis. Participants are encouraged to bring with them a data set that they are using in their own research, which they can then use to test a PLS structural equation model. Participants who want to use their own data should preferably have at least some ideas about (1) the structural model they want to test (i.e., which constructs are related to which other constructs) and (2) which indicators measure each of the constructs (the so-called measurement model). Another data set, plus the corresponding structural and measurement models, will also be made available.
On the first day of this workshop, we will discuss the principles of SEM through PLS in the morning, and apply these principles in a practice session in the afternoon. On the second day, we focus on some special topics, which we will again apply in the afternoon.
The schedule is as follows:
- Overview of Structural Equation Modeling
- Partial Least Squares versus LISREL
- Basic PLS algorithm
- When to use PLS?
- Model validation in PLS
Practice session: Testing and validating a basic PLS model using SmartPLS
- Formative indicators
- Higher-order factors
- Interaction effects
- Multi-group analysis
Practice session: Formative indicators, higher-order factors, interaction effects, and multi-group analysis using SmartPLS
For Research Master and Ph.D. students, there is the possibility to earn 2 ECTS as course credits. To do this they should write a short (5-10 page) report describing and discussing the application of PLS modeling to an existing data set.
Chin, Wynne W. (1998). The partial least squares approach to structural equation modeling. In Modern methods for business research, George A. Marcoulides, Ed. Mahwah, NJ: Lawrence Erlbaum.
Dijkstra, Theo K. & Henseler, Jörg (2015). Consistent partial least squares path modeling. MIS Quarterly, 39(2), 297-316
Hair, Joe F., Marko Sarstedt, Christian M. Ringle, and Jeannette A. Mena (2012). An assessment of the use of partial least squares structural equation modeling in marketing research. Journal of the Academy of Marketing Science, 40 (3), 414-433.
Henseler, J., Hubona, G., & Ray, P. A. (2016). Using PLS path modeling in new technology research: updated guidelines. Industrial Management & Data Systems, 116 (1), 2-20.
After enrolling in the course, participants ought to take the following steps.
First, you all need to download the SmartPLS software. To do so, you need to go to the website www.smartpls.com. There are four options:
- a free 30-day trial
- a free student version (limited to 100 cases)
- a commercial version (€400 a year)
- the older version (SmartPLS 2) (see www.smartpls.com/smartpls2)
Unless you already know that you will use SmartPLS in the future, I would not recommend to buy the commercial version at this point, but to choose the 30-day trial or the student version. A version of the example data set with 100 cases will be available, so the student version should work. It will be possible to switch between versions later on. If you decide later on that you will use SmartPLS in your research, but don’t want to pay €400 a year, you can also switch to the older version. It has fewer options than the newer version (particularly, no Consistent PLS and no Confirmatory Tetrad Analysis), but works fine. However, it is no longer supported and you will need to obtain approval from the SmartPLS administrators before you can use it (this approval is typically given within one or two days).
Second, you ought to install the SmartPLS program on your own laptop computer if you have one, and bring this computer with you to the afternoon practice sessions. The reason is that the university’s IT security policy does not allow files of the type that SmartPLS is, to be installed on computers in the PC labs. (The same holds for all other PLS software, as far as I know.) If you do not have a laptop you will need to share one with another participant during the practice session.
The timetable for this course can be found in the EUR course guide.
ERIM PhD candidates and Research Master students can register for this course via Osiris Student.
External (non-ERIM) participants are welcome to this course. To register, please fill in the registration form and e-mail it to the ERIM Doctoral Office by four weeks prior to the start of the course. For external participants, the course fee is 260 euro per ECTS credit.