T Statistics for Indirect Effect

Eins · Post by **Eins** » Mon Jun 18, 2012 7:32 am

Dear SmartPLS Experts,

May I know why the Bootstrap report of SMARTPLS does not include the T-Statistics for INDIRECT/Mediation EFFECT?

Thank you.
Eins

Hengkov · Post by **Hengkov** » Tue Jun 19, 2012 8:19 pm

Hi Eins,
Compute by hand using Sobel test, Ok?;-)
Regards,
Hengky

saloualabib · Post by **saloualabib** » Tue Nov 06, 2012 2:14 am

Please, i have the same problem.
How can i calculate t test for indirect effect (mediator effect), and how can i integrate the mediator variable in my model in pls?

Hengkov · Post by **Hengkov** » Tue Nov 06, 2012 3:15 am

Hi,

Mediator variable = Endogenous variable (for eample: X -> M -> Y)
For test indirect effect, used Sobel-Test (Normality required) or Bootstrapping Approach.

Best Regards,
Hengky

HYWang · Post by **HYWang** » Mon Mar 11, 2013 8:44 am

Hengkov wrote:Hi,

Mediator variable = Endogenous variable (for eample: X -> M -> Y)
For test indirect effect, used Sobel-Test (Normality required) or Bootstrapping Approach.

Best Regards,
Hengky

Hi Hengkov,

As my understanding, the ceofficient should be the unstrandardized one when use Sobel test to test the indirect effect. But, in the report of SmartPLS, there is only standardized coefficient. How we can get the unstrandardized coefficient from smartpls report?

Thanks in advance for your instruction.

Hengkov · Post by **Hengkov** » Wed Mar 13, 2013 12:11 am

Hi,
Just select "original" in PLS Algorithm. Mean 0 Var 1 for standardized, original for unstandardized OK. But Sobel required normality assumption.

Regards,
Hengky