Hi

AMOS is used for large sample size(more than 200 samples) and PLS is used for small sample size.

PLS differs from SEM (as implemented in AMOS, Stata, Lisrel, Mplus, EQS, etc.) in many important ways.

See:

https://www.researchgate.net/publication/263090609_Reflections_on_Partial_Least_Squares_Path_Modeling

Article Reflections on Partial Least Squares Path Modeling

Also see:

http://www.sciencedirect.com/science/article/pii/S0272696316300237

Please use AMOS or liseral

Fathuma wrote:  "Please use AMOS or liseral". 

Why, Fathuma?  Do you have any experience with Stata's SEM command?  What about it makes you judge it inferior to AMOS and Lisrel? 

Theresa,

Stata's SEM command works just fine for me, but then, so do many other packages!  Stata's implementation doesn't have the graphic, 'build-a-model' feature that AMOS, LISREL, and smartPLS do (EQS might also; I haven't used that in decades, so can't say).  The parameter estimations appear sound.  If you're already accustomed to Stata, I'd say go with it. 

The implicit premises of smartPLS are, however, very different from those of SEM packages--as indicated by Heiko and Cameron in earlier comments--they're not interchangeable.

I will also recommend that readers look seriously at the lavaan package for R; it's rapidly becoming my new favorite (and, of course, it's free).

Best of luck with your work!

Hi David.  What version of Stata are you using?  The current version has something called SEM Builder, which allows one to specify the model graphically -- see YouTube below.  Three more video tutorials for Stata's SEM can be found at the other link below.  HTH.

https://www.youtube.com/watch?v=Xj0gBlqwYHI

http://www.stata.com/links/video-tutorials/