Page 1 of 1
Problem with Multicollinearity in covariance based analysis
Posted: Mon Feb 13, 2012 1:14 am
by achmadghazali
I have data with certain indicator that has multicollinearity. The analysis using covariance based show some problems however PLS shows no problem at all. Can I use this as a reason to use PLS for my analysis? Thanks
Posted: Tue Feb 14, 2012 6:06 pm
by mronkko
Probably not. There is generally very little research showing that PLS would be able to reject false models and there is at least one study that shows that this is indeed the case
http://aisel.aisnet.org/icis2010_submissions/21/
How exactly to treat multicollinearity depends on your data and research question.