Hypotheses between 1st order and 2nd order construcs

Frequently Asked Questions about the SmartPLS software. Please refer to this section first, if you have problems with the software.
BochAbdel
PLS Junior User
Posts: 4
Joined: Sat Sep 15, 2018 9:21 pm
Real name and title: Boch Abdel

Hypotheses between 1st order and 2nd order construcs

Dear all,

I have a reflective-reflective model with three 2nd order variables A, B and C, and D which is a reflective construct.

My model is complex because I have hypotheses between A, B, C and D as well as hypotheses between the 1st order constrcuts of A (A1-A4) and the 2nd order construct B (Please see the attached image).

If I want to test the hypotheses between A, B, C and D, I should test the whole model. But how to test the hypotheses between A1-A4 and B?
Should I test a submodel composed only of A1-A4 directly linked to B or should I test the direct links between them within the whole model?

Thank you very much for your help.

Boch
Attachments
Forum 1.png (30.27 KiB) Viewed 10985 times

BochAbdel
PLS Junior User
Posts: 4
Joined: Sat Sep 15, 2018 9:21 pm
Real name and title: Boch Abdel

Re: Hypotheses between 1st order and 2nd order construcs

Any help please?

jmbecker
SmartPLS Developer
Posts: 1094
Joined: Tue Mar 28, 2006 11:09 am
Real name and title: Dr. Jan-Michael Becker

Re: Hypotheses between 1st order and 2nd order construcs

Direct connections between the lower-order constructs (e.g., A1-A4) and other higher-order (second-order) constructs (e.g., B) are unusual.

A formative higher order (second-order) constructs assumes full mediation of the effects of its underlying lower-order constructs on the other constructs in the model related to the higher-order construct.
For your reflective-reflective type constructs it makes even less sense, because you assume a causal direct from the higher constructs (A) to the lower-order constructs. So there is not even this full mediation assumption because the lower-order constructs are effects and do not effect the constucts. So why should they then also influence another higher-order construct directly? Isn't that what the higher-order construct is for?
Dr. Jan-Michael Becker, University of Cologne, SmartPLS Developer
Researchgate: https://www.researchgate.net/profile/Jan_Michael_Becker
GoogleScholar: http://scholar.google.de/citations?user ... AAAJ&hl=de