P values differences between PLS and SPSS

Questions about the implementation and application of the PLS-SEM method, that are not related to the usage of the SmartPLS software.
Post Reply
JLArregle
PLS Junior User
Posts: 6
Joined: Mon Aug 02, 2010 11:55 am
Real name and title:

P values differences between PLS and SPSS

Post by JLArregle »

Dear all,

I have just imported my PLS standardized LV scores in SPSS (wanted to check VIF) and run a regression.

Results: I have the same standardized coefficients in the 2 models (which is normal) BUT I have different p values for some coefficients (lower p values in SPSS compared to PLS). For instance, one variable has a p value of 0.08 in PLS and 0.01 in SPSS.
A precision: I have 637 observations.

I guess the differences come from the fact that the significances (t-values) in PLS is estimated with bootstrap, for this reason it is "distribution free".

Which results should I use: PLS or SPSS?

Thanks a lot for your help.

Jean-Luc
Top
kerim
PLS Junior User
Posts: 3
Joined: Sun Apr 10, 2011 5:47 pm
Real name and title:

Post by kerim »

Hi,
I think here you have to use lisrel because, PLS is much for a small/moderate sample size (<200)..
Top
User avatar
Diogenes
PLS Super-Expert
Posts: 899
Joined: Sat Oct 15, 2005 5:13 pm
Real name and title:
Location: São Paulo - BRAZIL
Contact:
Contact Diogenes

Post by Diogenes »

Hi,
you are right.

PLS-PM --> nonparametric (bootstrap)
SPSS --> parametric

If you intend to use SPSS results, just show that all assumptions of regression are ok: normality of residuals, etc.

best regards,

Bido
Top
JLArregle
PLS Junior User
Posts: 6
Joined: Mon Aug 02, 2010 11:55 am
Real name and title:

Post by JLArregle »

Dear Bido,

Ok.
Thanks a lot for your reply and help.

Best regards.

Jean-Luc
Top
Post Reply

Return to “Method and application”