Page 1 of 1

Testing multi-collinearity for formative measurement models

Posted: Sun Mar 31, 2013 2:35 pm
by moftec
Hello together,

who can help me out with the calculation of the Konditionsindex?
I have a formative measurement model that I want to check for multi-collinearity.
As far as I know, I can not calculate the KI in PLS but only in SPSS. But how?
I have the formular (KI = SquareRoot(Eigenwertmax/Eigenwerti)
But I don't what the Eigenwert is?
Can someone tell me step by step what I have to do?

My approach:
1. run a regression in SPSS with the indicators (raw data or estimation results???) of the construct (= independet variables)?
2. what is my dependent variable?
3. select "collinearity diagnosis", however then I receive a KI for each indicator, but as far as I understood, the KI indicates the multi-collinearity of the entire construct. So which one is the "right" one or how do I calculate it? Or is it the max?

Thanks so much in advance!

Best regards.

Posted: Mon Apr 01, 2013 11:26 pm
by hanifmauludin
using spss..you can test multicol...if VIF < 5 or 10...there is multicol

Posted: Mon Jul 08, 2013 10:25 am
by DrNeurocooper
Hi all,
I would like to renew this topic since I have the same questions as Jana!

Thanks in advance

Posted: Mon Jul 08, 2013 1:32 pm
by DrNeurocooper
Hi all,
I found a referred topic to this issue:
viewtopic.php?t=2003&highlight=collinearity+spss

And the path to find the values for the Latent Variables are here: SmartPLS -> standard report ( PLS > Calculation results > Latent Variable Scores).

Thanks all