Post Your Answer

Help Please! I need to develop a PLS-SEM using EFA

SEM Analysis

I am really confused at this moment after going through various studies and so, I finally thought to get answers from here. I will try to explain my question as easily as I can. Please bear with me.

Okay so, I have to do a factor analysis of reflexive scales and only after that, I can move on to the PLS analysis. I was looking for approaches suggested in other studies.

In some cases, the authors and researchers have recommended to conduct an SPSS based EFA (exploratory factor analysis) first and then to use AMOS software for carrying out CFA (confirmatory factor analysis) based on covariance. The items are then recommended for PLS in the order of results obtained.
In other cases, some researchers said that there could be issues in the method mentioned above and it is better to carry out EFA separately (before the PLS analysis) and that CB-CFA can be conducted along with the PLS analysis.
Moreover, I am very much confused about the process of EFA alone as there are different ways in which it is carried out.For example,

there are different extraction methods (principle axis, principle component, and so on).
there are different rotation measures as well like varimax and oblique.

So, I am confused as to which of these approaches will be the most suitable for PLS analysis.

This is the whole situation and I would really like it if someone can point out whether there is a general rule of using EFA and CFA when it comes to PLS analysis and modelling.
If you can also provide a reference research document or article to support the rule, then also I will be grateful to you. Please provide your valuable suggestions.

All Answers (3 Answers In All)

By Mehak Chaudhary Answered 3 years ago

Hello John, the approach can be decided based on the scale you are considering. If it is a new scale then you have to conduct the EFA first and use those results to carry out CFA and PLS analysis. On the contrary, if you are using an adopted scale (which has been already used in the previous studies or the studies you are considering), then you don't have to carry out the exploratory factor analysis and can move forward with the other two analyses and do the SEM

By Tom Answered 3 years ago

Hi, I am attaching a few links below for you to check out different approaches. These include the PLS-SEM analysis through EFA and CFA methods and also, show the use of different softwares including Amos. Please go through these as I believe they can clear at least some of your basic doubts to move further with your process.

By Arjun Bijlani Answered 3 years ago

Hello John, I don’t know if I will be of much help or not as I am yet new to the concept of SEM analysis but still I would like to point out some of the points which were asked about by my lecturer. CFA is a mandatory step (always theoretically and mostly practically) because in order to analyse, you must have had an idea of what to expect from the outcome or at least you should know what you are looking for in the result. For the extraction method, I have studied that the principal axis method is better and should be chosen and the Kaiser Gutman rule (components or factors that have eigenvalues higher than 1 and are in favor of the scree plot need to be retained) should be avoided. As for the rotation measure, I know that orthogonal rotation assumes that the factors are not correlated; whereas, the oblique rotation approach assumes otherwise. So, I guess you need to decide this based on your research question. (Note: mostly systems have orthogonal rotation as the default approach) I only know this much as I have not worked on Amos. But I heard Mplus is also a good option for SEM. I am attaching a link for a better understanding and easy break points. Please let me know if it proves to be helpful. Goodluck.

A Quick Primer on Exploratory Factor Analysis

View Related Questions