I agree with  Florian's explanation, PLS does not offer a model test. Therefore a chi-square test is not available and also all fit indices based on chi-square,  such as RMSEA,CFI, TLI, etc.

Regards, Karin 

Hey,

PLS does not offer a chi square test but the bootstrap-based SRMR does a similar job (Dijkstra & Henseler, 2015) and can be expanded to a variety of distance measures. The implementation in SmartPLS does not offer the bootstrap version and furthermore it only calculates the SRMR either for a pure common factor moder or a pure composite model but not for a mixture of both. You could use ADANCO where the SRMR is well implemented otherwise you could switch to R, I don't know if this is an option.

To my knowledge there is no CFI in PLS as the idea in PLS is a different to CBSEM same for RMSEA.

Kind regards,

Florian

I agree with  Florian's explanation, PLS does not offer a model test. Therefore a chi-square test is not available and also all fit indices based on chi-square,  such as RMSEA,CFI, TLI, etc.

Regards, Karin 

Dear Karin,

yes PLS by itself doesn't offer a overall model fit as it is based on several regressions. But this doesn't mean that the overall model fit can be assessed. As PLSc provides consistent estimates for common factor models, a bootstrap-based procedure provided by Yuan & Hayashi (2003) can be used in combination with various distance measures to measure if the underlying common factor model fits the data. Dijkstra & Henseler (2015) did a small simulation and it seems to work quite well. I'm currently doing further investigations and I'm looking forward to discuss the results with you in Zurich.

But of course, you are right that PLS does not offer a chi-square test and therefore no fit indices based on it.

Best regards,

Florian

Yuan, Ke‐Hai, and Kentaro Hayashi. "Bootstrap approach to inference and power analysis based on three test statistics for covariance structure models." British Journal of Mathematical and Statistical Psychology 56.1 (2003): 93-110.

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

Thanks a lot  Florian and Karin ! Today I started to explore ADANCO, and I found also some references that I am reading (Dijhstra and Henseler, 2015, and others).

Guillermo.

Dear Guillermo,

at the PLS Symposium in Sevilla 2015, Jörg Henseler also gives a talk about composite factor model at minute 20:30 he also talks about assessing the overall fit in composite models.

Here is the link to the video of his talk: http://tv.us.es/videoembed/?numberpost=30277

Kind regards,

Florian

Thanks Florian! I am watching the video now.

By the way, do you have any Guide or Article about ADANCO ?

Best regards,

Guillermo.

Hey Guillermo,

unfortunately, there is no explicit article about ADANCO. Furthermore, ADANCO does not provide an instruction. But the usage is very intuitive and very very similar to SmartPLS and as far as I know ADANCO can do everything that SmartPLS can do and even more. So you can take the instructions from SmartPLS. I guess the bigger problem is the misuse of PLS in general independent from the software you use. Henseler et al (2016) provide, to my opinion, a very good paper which provides several insights in PLS and its usage: http://dx.doi.org/10.1108/IMDS-09-2015-0382 it's open access.

I hope this helps.

Best regards,

Florian