Dear Sabri

Conventional SEM Method is reffered to covariance-based SEM.

Regarding the :

1. Assumption: the assumption of normality and big sample size are not essential while using PLS.

2. PLS-SEM are exploratory while CB-SEM is comfirmatory analysis.

3. PLS-SEM could deal with small sample and is friendly user. CB-SEM is very accurate in term of confirming the hypothesis.

4. To use CB-SEM lots of assumption must taken under consideration.

5. PLS-SEM is easy to use and interpret.

6. the use of each one could be helpful its depends on characteristics of the study.

7. Both methods based on GLM modeling and could deal with linear models.

I recommend the Hair (2016) book. A Primer on PLS- SEM .

Dear Sabri,

I suggest that you follow:

• For a big picture: H. Olya, " Partial Least Squares Based Structural Equation Modeling (PLS-SEM) ", 2017 - http://gloserv.org/wp-content/uploads/dr.olya_.pdf
• For all the rest: my prior post on https://www.researchgate.net/post/How_PLS_based_approach_is_different_from_co-variance_based_approach_to_structural_equation_modeling_And_when_to_adopt_which_approach with a particular focus on Hair et al., " Mirror, mirror on the wall: a comparative evaluation of composite-based structural equation modeling methods ", 2017

The set of tools {CB-SEM, PLS-SEM, GSCA, etc.} is the subject of controversy. As a result, the preferences of each other do not matter. The choice of a tool must be mainly conditioned by the joint needs of your analysis and requirements.

Hoping it will be helpful to you.

Dear Sabri

The fundamental difference between CB-SEM with the PLS-SEM approach and the number of samples.

1. CB-SEM approach uses the basis of covariance analysis so that the construct of the latent variable allows it to be removed if the loading factor does not meet the minimum requirements (> 0.4) and does not change the meaning of the latent variable. Instead PLS-SEM uses the basis of variant analysis so that the constuct of the latent variable when omitted / cut off will eliminate the meaning of the latent variable.

2. CB-SEM requires a lot of statistical requirements, data normality, minimum sample size of 100. In contrast, PLS-SEM is not as strict as CB-SEM, a minimum sample of 35, does not require data normality.