2nd order LV - computing AVE and Composite Reliability

Feel free to post your requests for new SmartPLS features and other suggestions.
Post Reply
User avatar
Diogenes
PLS Super-Expert
Posts: 905
Joined: Sat Oct 15, 2005 5:13 pm
Real name and title:
Location: São Paulo - BRAZIL
Contact:

2nd order LV - computing AVE and Composite Reliability

Post by Diogenes » Sat Feb 09, 2008 2:23 pm

I have computed by hand the quality criteria (Overview output) of the 2nd order LV and confirmed that the SmartPLS uses the outer loadings of the MVs to have it done.

However, the indicators of the 2nd order LV are the 1st order LVs (not the MVs), then the correct way to compute AVE, Composite reliability... will be using the path coefficients between them.

My experience until now is that the results using paths coefficients are bigger than using outer loadings (as SmartPLS does).

Best regards.
Bido

sudzina
PLS Junior User
Posts: 6
Joined: Tue Apr 20, 2010 8:16 pm
Real name and title:

Post by sudzina » Sat May 08, 2010 1:25 pm

Maybe it's a bit methodological question - but is it OK that AVE = communality? (I see the same column in a different report under a different heading.)

User avatar
Diogenes
PLS Super-Expert
Posts: 905
Joined: Sat Oct 15, 2005 5:13 pm
Real name and title:
Location: São Paulo - BRAZIL
Contact:

Post by Diogenes » Sat May 08, 2010 3:27 pm


Mrs Tourky
PLS User
Posts: 11
Joined: Sat Dec 03, 2011 3:31 pm
Real name and title:

Post by Mrs Tourky » Tue Jan 03, 2012 6:19 am

Hi all,


I am confused from all the posts regarding Communalities, loadins, AVE - for MV, Latent variable!!!

I am working on a model: my Independent variable is a 3rd order reflective construct - which includes four 2nd-order constructs. Each 2nd order construct includes two 1st-order latent variables. latent variables are measured by indicators/items.

The dependent variable is a 2nd-order construct which contains two 1-st order latent variables.

I have run the algorithm for the whole model to check the loadings and cross loadings of each MV/items. I have also checked the composite reliability, AVE and Cronbach alpha of each 1st-order LV...Is this correct?

Wouls you please let me know what is the next step and how to perform it in order to ensure the validity of the whole model?? Thanks

I really appreciate you help.

Best,
Marwa

coser72
PLS Junior User
Posts: 6
Joined: Tue Dec 13, 2011 5:02 pm
Real name and title:

2nd order LV - computing AVE and Composite Reliability

Post by coser72 » Wed Jan 25, 2012 8:43 pm

Hello,

In my 2nd order reflective model one of my 2nd order LVs is giving AVE = 0,46 (I've calculated AVE by hand considering the paths to the 1st order LVs as factor loadings).

As I won't be able to increase this AVE, must I remove the 2nd order LV from the model or is 0.46 a value still accepctable?

Thanks in advance for any help.

Regards,

Adriano

Post Reply