prof Ali orozi Sougui this table can help you to choose. 

PLS-SEM is an exploratory methodology that relies on primary or secondary data (Not suitable for CB-SEM-Hair 2015).

The PLS approach is suitable with respect to the researcher’s prediction-oriented objective, does not require normal data distribution and accommodates small sample sizes (Chin & Newsted, 1999). Where else, CB-SEM need bigger sample size (minimum >400).

PLS-SEM also provides R2 values and indicates the significance of relationships among constructs in order to demonstrate how well the model is performing. CB-SEM on the other hand only provides path modeling (coefficient and CR). Bear in your mind that PLS is best for prediction-based research, while CB-SEM is more on modelfit.

One of the main advantages of PLS-SEM over CB-SEM is that PLS-SEM can handle numerous independent variables at the same time, even when these display multicollinearity (Hair, Ringle, & Sarstedt, 2011).

If your study is related on theory development or theory building, you have to choose PLS-PM over CBSEM. But when your study are more on theory testing, you have to choose CBSEM over PLS-PM. In your case, i believe your study is may appropriate for CBSEM technique.   

You should use CB-SEM as you are going to test the theory rather than development. 

prof Ali orozi Sougui this table can help you to choose. 

I concur with Asyraf and Imtiaz as your objective is to testing the effect of mediation construct. 

CBSEM/VSEM-AMOS Analysis.

•  Using CBSEM-AMOS you need up-to  200 sample size above, proper data screening. Amos suitable when when you like 5 constructs 
• CBSEM-AMOS estimators based on CFI, RMSEA, GFI, AGFI, Cmin etc
• Using VBSEM-AMOS you need at least 100 sample size depending on your constructs as well
• VBSEM-AMOS estimators based on square multiple regression (R2), critical ratio, p-values

SMARTPLS Analysis.

• Like using any statistical tools, you need to understand how to used it, using PLS seems not complex on like AMOS, the data screening also important.
• Most Estimator include bootstrapping and Algorithm (path coefficients, t-statistic, R2 )

Sir, you need to seek assistance from experts,  read texts and articles on AMOS and SmartPLS..

Contact ([email protected]) for assistance

How to chose SEM-AMOS or SEM-SmartPLS for data analysis?

Agreed with the comments shared by the above scholars.  At high level - the usage of SEM-SmartPLS (Variance-based SEM) is to explore theory whereas SEM-AMOS (Covariance-based SEM) is to test / confirm theory i.e. need strong prior theory & established questionnaire.

You can also refer to the following RG links for related questions being asked.

https://www.researchgate.net/post/What_is_Structural_Equation_Modeling2

https://www.researchgate.net/post/Are_SPSS_Amos_SmartPLS_Wrap_PLS_PLS_Graph_30_used_for_same_purpose2

BUT... the consistent PLS estimator now mirrors or uses correction factors to give the same or similar results to CB SEM.

So to get the benefits of PLS and flexibility and the theory testing potential this may be worth looking into

best regards

Brad.

I agree with Bradley and furthermore bootstrapped based methods can be used to evaluate the overall model fit in PLS which seems to work quite nice, see Dijkstra & Henseler (2015). The procedure is still implemented in ADANCO as well as SmartPLS.

Best regards,

Dijkstra, Theo K., and Jörg Henseler. "Consistent and asymptotically normal PLS estimators for linear structural equations." Computational statistics & data analysis 81 (2015): 10-23.

it is always hard to answer such broad brush questions when people do not put their research questions in there.

AND of course everything is model dependent. some techniques are better for recursive and non recursive models. other techniques are useful for interactions or moderation analysis versus mediation testing etc.

AND then there are issues to consider regarding discipline space and journals. Some reviewers still favour certain methods.

so it is all about your goals at the end of the day!

brad

what should be minimum sample size for PLS when unit of analysis is organization and gathering primary data?

We are studing 6 variables with Structured equations

Please anyone can suggest me which tool  to chose, SEM-AMOS or SEM-SmartPLS for data analysis?

How to chose SEM-AMOS or SEM-SmartPLS for data analysis?. 

We want to analyze 6 variables to analyze structured equations, I was recommended to use this software but I do not know what the best possibility of analsis.

 The below answer given by Dr.Han Ping Fung :

SEM generally divided into Variance-based SEM / PLS-SEM (e.g. SmartPLS) and Covariance-based SEM (e.g. AMOS which is an extension module from SPSS).  Variance-based SEM e.g. SmartPLS can be used for exploratory research whereas Covariance-based SEM is meant for confirmatory research / analysis.

Following shows the differences between the 2 SEMs in general:

1) Objective

Variance-based SEM - Focus on construct’s prediction i.e. variance explained

Covariance-based SEM - Focus on covariance i.e. explanation of items’ relationships

2) Theory Treatment

Variance-based SEM - Theory Exploration

Covariance-based SEM - Theory Confirmation i.e. need strong prior theory & established questionnaire

3) Sample Size

Variance-based SEM - Support small sample e.g. < 100

Covariance-based SEM - Need sizeable sample e.g. > = 300

4) Data Type

Variance-based SEM - Support both metric & non-metric data types

Covariance-based SEM - Only support metric data types

5) Construct Type

Variance-based SEM - Support both reflective & formative constructs

Covariance-based SEM - Support only reflective constructs

6) Construct-Item Relationship

Variance-based SEM - Support Construct with single-item

Covariance-based SEM - Support Construct with three-items & above

7) Missing Values

Variance-based SEM - Support data sets with missing values

Covariance-based SEM - Need to address missing values before analysis

8) Existence of Multi-collinearity

Variance-based SEM - Support data sets with multi-collinearity

Covariance-based SEM - Need to address multi-collinearity before analysis

What is Structural Equation Modeling?. Available from: https://www.researchgate.net/post/What_is_Structural_Equation_Modeling2 [accessed Jul 8, 2017].

These links may help you

http://proceedings.utwente.nl/357/1/UsePLS_2015_submission_61.pdf

https://www.researchgate.net/post/How_to_decide_which_package_is_best_for_my_research_Amos_or_Smart_PLS

https://www.researchgate.net/post/SEM_with_AMOS_or_LISREL

https://www.researchgate.net/post/Which_software_gives_more_accurate_result_AMOS_OR_Smart_PLS

By having a look into your objectives of the research, i would suggest you to use CB SEM

Check this out:

https://www.smartpls.com/resources/ebook_on_pls-sem.pdf

If you ask people to rate a bride in her wedding day for how beautiful she was ? Guess who will say she is very gorgeous ? -Her sister, mother and maternal aunt.

The Wisdom : Those who used the PLS egregiously in the past and it showed them artificial fit when their models were misscpecified tend to say PLS is good, Good and for ever Good. Many assumptions get violated very often by those who use PLS for theory testing. In plan PLS is NOT suitable for theory testing when data , factors are all in reality latent factors. How could you surmise PLS is suitable for theory and hypothesis testing, people ? Read More . Theory and hypothesis testing is best under structural equation modelling .

Dear Mohammad,

Interesting view point. As you referred to people, I feel addressed in a sense as I gave an answer previously to this question.

Why do you think PLS is not suitable for theory testing? PLSc is capable to consistently estimate LV models and Bollen-Stine bootstrap can be used to assess the overall model fit. Inference about model parameters can be made by bootstrap confidence intervals. Moreover, PLSe can be used to obtain asymptotically efficient estimates.

I agree that in applied research PLS is often not properly applied, but that is a different topic.

Best regards,

Florian

Thank you Dear Florian for your kind message .

To clarify, I did Not address any particular person or comments in my response to be precise I did Not read every comments.

My point was that people adapt methodologies, scales and questionnaires made by their manufacturers as ( latent factors ) with presumed distribution in different geographies and societies . The Birth certificate of each scales should aid in deciding what type of analysis applies. I have seen hundreds of clients ( in my part of the world ) requesting an PLS to predict an outcome measured with scales that do Not suite the PLS at all, and this is becoming a very epidemic . Those scales often get misspecified and tried in the ladder of testing up to SEM or PLS methods before testing their structural validity or even the factorial validity , just automatic knee-jerk use . That is my point .

Moreover, bootstrapped PLS estimates ( significance) has different meaning and interpretation, evidence suggests they could Not be treated and interpreted as if they're parametric bootstraps that generalize to the population , however those bootstraps can apply to the data itself Not the population, particularly in PLS , to my best of knowledge.

Thank you again

Warmest regards,

Mohamed

Also, believe it or Not? some of my clients come to me with fake data ( fake = fiddled and fudged data ) in various disciplines , those data fail all the tests of person-fit, quality, outliers, odd responses and they just end as trash data in COV-SEM , they just torture us with those faked data and go . They try it with PLS And it damn works for them with excellent fit parameters ,

can someone do an experiment and examine why such fake data fits and nicely in PLS-SEM, when the clients confess they have made the entire data by themselves and wives while eating nuts and watching movies ?

I am Not condemning PLS-SEM, but despite its ease and sense sometimes I will Not use it for anything until it proves otherwise -as an applied statistician I do Not think it makes sense for theory testing .