forum.smartpls.com

Hi,

Can we do Confirmatory Factor Analysis in SmartPLS? How?

Thank you.

Eins

Hi,
Yes.

1) Connect all LV one with each other (take care to have not recursive arrows). Some LVwill send arrows and other will receive, but this doesn’t matter.
2) Use the “factor weighting scheme” in the PLS algorithm.
3) Assess the measurement model (outer loadings, crossloadings, AVE, reliability…), and the correlations between LVs (results of the CFA).
4) Don’t use the results of path coefficients (meaningfulness).

Best regards,

Bido

Bido,

Thank you for your reply.

Do you have references on "scale development and validation" that use SmartPLS.

Eins

Dear all,

I would like to state one follow-up question: I am currently (and for the first time) using smartPLS to engage in a CFA. I have checked for the reliability and validity measures of all my (reflective) latent variables. Now I would like to conduct group comparisons using these latent variables. The group comparisons will be done via factorial ANOVA in SPSS. However, I wonder where I can find the data to be exported in smartPLS for the comparison?
Is this the data shown in the default output in "latent variable scores (unstandardized)" and "latent variable scores" respectively for the standardized scores or does smartPLS not create such information?
When using these standardized and unstandardized scores, I get slightly different results in the follow-up analyses. These differences are in terms of p-values for example 0.099 (unstd score) and 0.102 (std score). Are these deviations "normal" in the sense of common? And would you recommend using the std scores or the unstd scores?


Thank you very much for any helpful comment!

All the best,
Hannes

Hi guys, is there any fit index available (e.g., GoF) that applies to CFA with pls? Thanks in advance.
Germano

Thank you, How can we test the second order CFA in smartPLS? TA mike