Third Order construct

Small talk on SmartPLS that does not correspond to the other forums!
Post Reply
elmanzani
PLS Junior User
Posts: 2
Joined: Tue Feb 21, 2017 2:33 am
Real name and title: Younès El Manzani

Third Order construct

Post by elmanzani » Sun Apr 08, 2018 10:48 pm

Hey,

Please, I need your help.

What are the adequate steps to establish the third order construct (M) bellow in the picture?

If I use the two-stage approach, are the steps bellow correct :

1- use the repeated indicators in second (A, B) and third (M) order constructs simultaneously.
2- save the latent variable scores of the reflective second-order constructs.
3- use these latent variable scores as indicators of the reflective third order construct in a new model.

OR

1- use the repeated indicators only in the second order constructs (A, B) without the third order construct (M).
2- save the latent variables scores of the reflective first-order constructs (1....8).
3- Then use these latent variable scores as repeated indicators of the reflective second-order and third-order construct in a new model.
4- save the latent variables scores of the reflective second-order constructs.
5- use latent variables scores of the second order constructs as indicators for the reflective third-order construct in a final model.

The reliability and validity of the reflective second and third order constructs should be assessed in every step, or after assessing the reflective first-order constructs we move directly to the assessment of the reflective third order construct?

Thank you.
Attachments
Exempl.png
Exempl.png (71.7 KiB) Viewed 158 times

jmbecker
SmartPLS Developer
Posts: 820
Joined: Tue Mar 28, 2006 11:09 am
Real name and title: Dr. Jan-Michael Becker

Re: Third Order construct

Post by jmbecker » Fri Apr 13, 2018 9:19 am

I think there is no specific guideline in the literature on how to use the two-stage approach correctly in the case of a third-order construct.

Usually using the repeated-indicator in the first stage of the repeated indicator approach might also not be the best solution. Usually it is better to model the effects of the first-oder constructs on the other variables in the model and then use these LV scores. The repeated-indicator approach without context of the rest of the model is sometimes not very stable and might produce unexpected and implausible results.
Dr. Jan-Michael Becker, University of Cologne, SmartPLS Developer
Researchgate: https://www.researchgate.net/profile/Ja ... v=hdr_xprf
GoogleScholar: http://scholar.google.de/citations?user ... AAAJ&hl=de

Post Reply