Why do betas change when using moderator?

Questions about the implementation and application of the PLS-SEM method, that are not related to the usage of the SmartPLS software.
Post Reply
mpalanski
PLS Junior User
Posts: 4
Joined: Tue Feb 28, 2006 7:32 pm
Real name and title:

Why do betas change when using moderator?

Post by mpalanski »

I have a model with 4 LVs: exogenous variables X1, X2, M and endogenous variable Y. When I run a direct effects model, only the beta for X2 is significant (X1 and M are not significant) and the r-square for Y is .163.

When I run a model with a moderator (X1*M), now only the beta of M is significant (X1, X2, and X1*M are not significant) and the r-square for Y is now .231.

Can anyone explain to me why this occurs? Thank you.

Regards,

Mike
Michael Palanski
Rochester Institute of Technology
Top
mpalanski
PLS Junior User
Posts: 4
Joined: Tue Feb 28, 2006 7:32 pm
Real name and title:

Post by mpalanski »

After thinking about my question, I am fairly certain that it is due to a multicollinearity problem with the exogenous variables.
Michael Palanski
Rochester Institute of Technology
Top
viswadatta
PLS Expert User
Posts: 139
Joined: Wed Jul 05, 2006 1:43 pm
Real name and title:
Location: Coimbatore, India

multicolinearity

Post by viswadatta »

Multicolinearity means that exogenous constructs are actually one construct. This is verified using discriminant validity of constructs. If two constructs have a very high correlation (more than RMS value of AVE values) then there is no discriminant validity, and the two constructs become one.

This should answer your question
Top
Post Reply

Return to “Method and application”