Page 1 of 1
How to use the correlation matrix to run???
Posted: Sun Jan 09, 2011 6:26 am
by cucupaul
Dear all,
please help me
I have the correlation matrix data(I don't have the raw data),
how to use this correlation matrix data to run?
thank you very much!!!
Posted: Sun Jan 09, 2011 12:20 pm
by Diogenes
Hi Lin,
PLS-PM is based in the explanation of the variance of the endogenous LV and need the raw data.
If you have just the correlation matrix, use LISREL, EQS, AMOS, etc.
Best regards,
Bido
Posted: Wed Mar 02, 2011 2:57 pm
by Tobjoh
Hi,
Perhaps a bit late.
I have never done it myself, but I know that it is possible to simulate a data set if you know the correlations, means, SD. I am not absolutely sure of its applicability to your specific case though.
A quick search on the www gave me this:
http://www.uvm.edu/~dhowell/StatPages/M ... rrGen.html
Posted: Tue Mar 08, 2011 6:10 am
by mronkko
A couple of the Herman Wold's original papers argue that PLS model can be estimated with covariance matrix as the data. Also, the LVPLSC program in the LVPLS software package should be able to estimate such models.
Unfortunately, LVPLS does not seem to be available anymore and the description of how to run PLS with coveriance data has been very difficult to find.
If anyone knows a way to get LVPLS or knows the algorithm used to run models with covariance matrices, I would be really interested to hear about this.