Second order moderating model

Frequently asked questions about PLS path modeling.
Post Reply
hidayah
PLS Junior User
Posts: 5
Joined: Mon Jul 15, 2019 4:44 am
Real name and title: Nurhidayah Yahya

Second order moderating model

Post by hidayah » Thu Jul 18, 2019 7:49 am

Hello,
I have a doubt on my moderator analysis in my model.

My model is a second-order and all of them are reflective formative.
For the moderator analysis, there is no need to do the repeated indicator approach to analyse the model since the software has a calculation method of two-stage.
However, when I run bootstrapping, the results are not favourable; but when I run an analysis of the model using the latent score, the result is different and more favourable.
Model 1, t-value.png
Model 1, t-value.png (78.67 KiB) Viewed 538 times
model 1=t-value.
Model 2, t-value.png
Model 2, t-value.png (22.5 KiB) Viewed 538 times
model 2=t-value.
Model 3, t-value.png
Model 3, t-value.png (41.66 KiB) Viewed 538 times
model 3=t-value.

How the results can be different though the data and method are the same?
So, which one I should report? Can I report the bootstrap result base on the model that uses latent score (model 2 or 3) instead of model 1?

Thank you.

wyfwong
PLS Junior User
Posts: 2
Joined: Thu Mar 28, 2019 2:14 am
Real name and title: WYF Wong

Re: Second order moderating model

Post by wyfwong » Thu Jul 18, 2019 10:16 am


wyfwong
PLS Junior User
Posts: 2
Joined: Thu Mar 28, 2019 2:14 am
Real name and title: WYF Wong

Re: Second order moderating model

Post by wyfwong » Thu Jul 18, 2019 10:39 am

Also see https://youtu.be/LRND-H-hQQw
Run PLS Algorithm, select Weighting Scheme: Factor (not Path)
Start Calculation
Click on Latent Variables, Excel Format to copy the Latent Variable scores
Run Excel to create a new Excel, paste the results to Excel and save to CSV
Import the newly saved CSV as the default dataset.

hidayah
PLS Junior User
Posts: 5
Joined: Mon Jul 15, 2019 4:44 am
Real name and title: Nurhidayah Yahya

Re: Second order moderating model

Post by hidayah » Fri Jul 19, 2019 2:02 am

Thank you wyfwong for your guidance and help. God bless you.

However, I had run such procedures to get to latent scores for model 2 and 3 data.

It's just that the results are different if using latent score in model 2 and 3 as compared to using items in model 1 though the data are the same.
Hence, I need advice on which model should I use and report?

Thank you.

hidayah
PLS Junior User
Posts: 5
Joined: Mon Jul 15, 2019 4:44 am
Real name and title: Nurhidayah Yahya

Re: Second order moderating model

Post by hidayah » Mon Jul 22, 2019 2:51 am

Ok, I get it.
I must use model 2 as the R2 in model 1 will be 1.00 because it is formative.

Thank you so much wyfwong.. God bless.

jmbecker
SmartPLS Developer
Posts: 1046
Joined: Tue Mar 28, 2006 11:09 am
Real name and title: Dr. Jan-Michael Becker

Re: Second order moderating model

Post by jmbecker » Mon Jul 29, 2019 3:32 pm

I would actually prefer model 3 using a disjoint two-stage approach for the second-order model combined with a two-stage approach for the moderating effects.

You should also read:
https://www.researchgate.net/publicatio ... in_PLS-SEM
https://www.researchgate.net/publicatio ... ype_Models
Dr. Jan-Michael Becker, University of Cologne, SmartPLS Developer
Researchgate: https://www.researchgate.net/profile/Jan_Michael_Becker
GoogleScholar: http://scholar.google.de/citations?user ... AAAJ&hl=de

hidayah
PLS Junior User
Posts: 5
Joined: Mon Jul 15, 2019 4:44 am
Real name and title: Nurhidayah Yahya

Re: Second order moderating model

Post by hidayah » Wed Jul 31, 2019 3:15 am

Thank you Dr Becker.
Will read the suggested article especially the latest 2019 for my further understanding.

Thank you..God bless.

scottsen11
PLS Junior User
Posts: 1
Joined: Fri Nov 01, 2019 11:46 am
Real name and title: scott sen

Re: Second order moderating model

Post by scottsen11 » Fri Nov 01, 2019 11:55 am

Before you start the analysis for moderator analysis, you have to ensure that you have a theory to justify that this variable is suit for moderator variable. I dont think we can apply the same variable can be handled for second order and moderator simultaneously because the equation for the contruct will be redundant. Plus, before you interest to apply sem in your study using SmartPls or Amos package, you have to know the characteristics of your research. If you interest for SmartPls package, means that, you choose for theory development in your study but if your study about to confirm of your theory, you have to apply Amos that already implemented for covariance based sem. hopefully this information can help you.
Best regards shareit apk
Last edited by scottsen11 on Wed Nov 06, 2019 9:55 am, edited 1 time in total.

hidayah
PLS Junior User
Posts: 5
Joined: Mon Jul 15, 2019 4:44 am
Real name and title: Nurhidayah Yahya

Re: Second order moderating model

Post by hidayah » Tue Nov 05, 2019 6:58 am

Thank you, God bless.

Post Reply