Dear experts:
I am testing the moderating effect of a moderator variable (response cost) on a relationship between sanction severity and compliance. All my latent variables are measured with reflective indicators. When I ran the moderating test in SmartPLS3, I first used the two-stage approach, the result is no moderating effect. Then, I ran the test again using the product indicator approach, the result shows a significant moderating effect. The data has 232 records. I set the resampling at 5000. Please see the following two images for the two results. I am not sure how to interpret the results. Would you help me to comprehend the two different results?
Thank you very much!
Xavier
The two stage result:
The product indicator result:
Moderating effect test: different results
-
- PLS Junior User
- Posts: 4
- Joined: Sat Oct 14, 2017 1:36 am
- Real name and title: Xavier Chen, assistant professor
Re: Moderating effect test: different results
I am puzzled by the different results from the two approaches. I thought that the two stage approach has higher statistic power than the product indicator approach. But the results seem showing the other way around: the product indicator approach shows the significant moderating effect, but the two stage approach shows insignificant effect. It seems to me that in this case product indicator approach has higher statistical power.
-
- SmartPLS Developer
- Posts: 1282
- Joined: Tue Mar 28, 2006 11:09 am
- Real name and title: Dr. Jan-Michael Becker
Re: Moderating effect test: different results
Which options do you use with the product indicator approach. It is usally recommended to use standardzed variables before multiplication in the product terms because all other (latent) variables in the model are also standardized.
Dr. Jan-Michael Becker, BI Norwegian Business School, SmartPLS Developer
Researchgate: https://www.researchgate.net/profile/Jan_Michael_Becker
GoogleScholar: http://scholar.google.de/citations?user ... AAAJ&hl=de
Researchgate: https://www.researchgate.net/profile/Jan_Michael_Becker
GoogleScholar: http://scholar.google.de/citations?user ... AAAJ&hl=de
-
- PLS Junior User
- Posts: 4
- Joined: Sat Oct 14, 2017 1:36 am
- Real name and title: Xavier Chen, assistant professor
Re: Moderating effect test: different results
Dr. Becker,
Thanks for your reply. I did use the Standardized option for Product Term Generation when setting up the moderator.
Should any other settings I need to change?
Thanks,
Xavier
Thanks for your reply. I did use the Standardized option for Product Term Generation when setting up the moderator.
Should any other settings I need to change?
Thanks,
Xavier
-
- SmartPLS Developer
- Posts: 1282
- Joined: Tue Mar 28, 2006 11:09 am
- Real name and title: Dr. Jan-Michael Becker
Re: Moderating effect test: different results
I think a potential problem is that all the effects are very small. Therefore, it can easily happen that you just find some chance effects that might not be really there. I don't know how the rest of your model looks like, but given the snippet of results that you showed us, I would not trust any of the two results.
Dr. Jan-Michael Becker, BI Norwegian Business School, SmartPLS Developer
Researchgate: https://www.researchgate.net/profile/Jan_Michael_Becker
GoogleScholar: http://scholar.google.de/citations?user ... AAAJ&hl=de
Researchgate: https://www.researchgate.net/profile/Jan_Michael_Becker
GoogleScholar: http://scholar.google.de/citations?user ... AAAJ&hl=de
-
- PLS Junior User
- Posts: 4
- Joined: Sat Oct 14, 2017 1:36 am
- Real name and title: Xavier Chen, assistant professor
Re: Moderating effect test: different results
Dear Dr. Becker,
Thank you so much for your reply. I will take your suggestion into consideration when I explain the result.
Thanks again,
Xavier
Thank you so much for your reply. I will take your suggestion into consideration when I explain the result.
Thanks again,
Xavier
Double Moderation effect
, In my model I have two IVs single DV and two moderating variables. how can I check for the combine effect of these two moderating variables