Multi-variate statistics & Mediation

Research topics can be discussed in this area.
Post Reply
kamal123
PLS User
Posts: 14
Joined: Mon Feb 25, 2013 8:27 am
Real name and title:

Multi-variate statistics & Mediation

Post by kamal123 »

Hi,

I would like to mention that, in my model, I have multiple independent variables and I conduct a bootstraping and mediation analysis for each independent variable (IV) separately; my question is there a way to have a mediating test in presence of all IVs? and how?

It will be highly appreciated if you could provide information regarding this concern.
Kamal
User avatar
Hengkov
PLS Super-Expert
Posts: 1599
Joined: Sun Apr 24, 2011 10:13 am
Real name and title: Hengky Latan
Location: AMQ, Indonesia
Contact:

Post by Hengkov »

Hello,

Just connect all construct in model and PLS Algorithm + Boostrapping. Check significant direct and indirect effect.

Regards,
Hengky
kamal123
PLS User
Posts: 14
Joined: Mon Feb 25, 2013 8:27 am
Real name and title:

Post by kamal123 »

Hello,

I would like to thank you very much,

How we can find a references that can support the SmartPLS obtained values, for example: if the p-value of Sobel test < 0.05 then we can conclude that there is a mediation role; as another example if the AVE > 0.5 then we have a reasonable convergent validity. Please would you provide me with some reference that can be cited within the discussion of the results.

Great thanks for you
kamal123
PLS User
Posts: 14
Joined: Mon Feb 25, 2013 8:27 am
Real name and title:

Post by kamal123 »

Hi, Second Question,

Is there a test in SmartPLS equivalent to Multiple regression in order to check the R square, significance and the Beta value of each independent variables.

Best regards
kamal123
PLS User
Posts: 14
Joined: Mon Feb 25, 2013 8:27 am
Real name and title:

Post by kamal123 »

Hi,
Third Question, if we have the following values after applying online Sobel test
Sobel test statistic: 1.84357663
One-tailed probability: 0.03262243
Two-tailed probability: 0.06524486

Can we conclude that there is a mediation, because one-tailed < 0.05, regardless the value of Two-tailed?

Fourth Question
How we can ensure the significance of correlation in multiple regression using SmartPLS (P-Value)?

Thank you in advance
User avatar
Hengkov
PLS Super-Expert
Posts: 1599
Joined: Sun Apr 24, 2011 10:13 am
Real name and title: Hengky Latan
Location: AMQ, Indonesia
Contact:

Post by Hengkov »

Hi,

1. Many articles and books in PLS Literature provide information about evaluation outer & inner model, and also mediation analysis please check it.

2. Sobel-Test required normality assumption, not same with character PLS (distribution-free). One-tailed or Two-tailed, you must justify it (used one level, no mixed).

3. Yes, you can see R^2 for variance explain, but now I recommend used Adjusted R^2. Next compute predictive validity Q^2 and also f^2 + GoF.

4. For correlation rule of thumb P < 0.05.

Best Regards,
Hengky
kamal123
PLS User
Posts: 14
Joined: Mon Feb 25, 2013 8:27 am
Real name and title:

Post by kamal123 »

Thank you very much

best regards
Post Reply