Page 1 of 1

Unstandardized regression coefficients

Posted: Fri Jul 22, 2011 2:37 pm
by eloiacon
I know SmartPLS reports the standardized regression coefficients, but does anyonw know where or how I can get the unstandardized ones?

Re: Unstandardized regression coefficients

Posted: Sun Jul 24, 2011 5:40 pm
by antonioaguiar
eloiacon wrote:I know SmartPLS reports the standardized regression coefficients, but does anyonw know where or how I can get the unstandardized ones?
Dear Prof. Eleanor,
When runing the PLS algorithm choose Original Data instead of Mean 0 Var 1;

My question is: Since you choose Mean 0 Var 1, is it possible to have weights greater than one?

Antonio

Posted: Tue Aug 09, 2011 9:56 am
by jmbecker
If you chose original data the path coefficients will still be standardized. Standardization of the latent variables is always done during the algorithm. Only the outer approximations are different. See also (Lohmöller, 1989) or (Tenenhaus et al. 2005).

If you want the unstandardized path coefficients you should use the unstandardized LV scores from the section “Index Values”. However, currently there is a bug in the calculations of those values so that I would recommend calculating them by hand.
See also viewtopic.php?t=812&highlight=unstandardized

Re: Unstandardized regression coefficients

Posted: Thu Sep 16, 2021 10:21 pm
by jgreig
Hello
I'm wondering if the bug in the calculations for unstandardized coefficients has been fixed?
Thanks, Jenni

Re: Unstandardized regression coefficients

Posted: Tue Jan 11, 2022 8:48 pm
by jmbecker
Smartpls 2 has never been updated and thus still contains the bug. SmartPLS 3 is doing the calculations correctly.