Bootstrap sample size in Henseler's PLS-MGA

Questions about the implementation and application of the PLS-SEM method, that are not related to the usage of the SmartPLS software.
Post Reply
jialin3105
PLS Junior User
Posts: 8
Joined: Sun Mar 27, 2011 2:28 am
Real name and title:

Bootstrap sample size in Henseler's PLS-MGA

Post by jialin3105 »

Hello everyone,
Maybe I did not interprete the nonparametric multigroup analysis approach of Henseler thoroughly. By using the excel spreadsheet found on the Internet, we can just calculate the P-henseler (the probability that group 2 is significantly lager than group 1) by using bootstrap data with sample size 100. Is there any other spreadsheet that can be used to calculate bootstrap data with other sample size, such as 500 or 5000? Or you adjust the original spreadsheet to calculate by using bootstrap data with other sample size such as 500 or 5000? I appreciate if you can give me some inspiration! Thank you.
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 Lin,

Non-parametric approach (PLS-MGA) required 5.000 sample bootstrap. So, you run model with 5.000 resampling. (case = 100 -> original sample, sample = 5000). If, P group 2 > group 1 = significant.

Best Regards,
Hengky
User avatar
Diogenes
PLS Super-Expert
Posts: 899
Joined: Sat Oct 15, 2005 5:13 pm
Real name and title:
Location: São Paulo - BRAZIL
Contact:

Post by Diogenes »

Hi,

1) run the bootstrap in the usual way
Cases = the sample size (rows in the data set)
Samples = 500 or 1000 or 5000 or more
Sign = construct level changes or individual level changes

2) Copy the results of the bootstrap to a spreadsheet
Report
Bootstrapping / Bootstrapping / Path coefficients /
Select this result / Copy
Past it in a spreadsheet

3) Copy the first 100 results (Sample 0 to Sample 99) to the Henseler’s spreadsheet.
Column (A9:A108) = 100 Bootstrap values of group 1

4) Repeat steps 1 to 3 for the other dataset (second group).
Column (C9:C108) = 100 Bootstrap values of group 2

5) Repeat steps 1 to 4 for each path coefficient (or outer weight or outer loading or total effect) that you want to compare.

Done!

You had confused “sample size” with “100 Bootstrap values of group 1”

The value 100 = the first 100 results of the bootstrap procedure.



Best regards,

Bido
Barbara Stahl
PLS Junior User
Posts: 3
Joined: Sun Jun 13, 2010 10:16 am
Real name and title:

Where to find the Henseler Excel spreadsheet

Post by Barbara Stahl »

Could you kindly let me know where to find the mentioned Excel sheet for doing the multigroup ananlsis?

Thanks!
hxin009
PLS Senior User
Posts: 25
Joined: Mon Aug 13, 2012 4:23 am
Real name and title:

Post by hxin009 »

yes, i also wanna have a copy of that excel sheet. Could you please email me a copy? here is my email nyp8061@aut.ac.nz

thank you so much!

Best regards,
Catherine
christian.nitzl
PLS Expert User
Posts: 248
Joined: Sat Jul 25, 2009 1:34 pm
Real name and title:

Post by christian.nitzl »

You can get the excel sheet directly from Jörg Henseler when you send him an email to:

joerg.henseler@pls-school.com

Cheers,
Christian
jialin3105
PLS Junior User
Posts: 8
Joined: Sun Mar 27, 2011 2:28 am
Real name and title:

Post by jialin3105 »

Thank all of you to response. Just as Dr. Nitzl said Dr. Henseler is a very good professor. He is so kindly that he sent me the excel spreadsheet in which I can copy the 5000 data into the group 1 and group 2. However, I think maybe it is impolite to send out the spreadsheet without the permission of Dr. Henseler. Please email Dr. Henseler for the spreadsheet. Thanks!
Barbara Stahl
PLS Junior User
Posts: 3
Joined: Sun Jun 13, 2010 10:16 am
Real name and title:

Thanks to Prof. Henseler for the spreatsheet

Post by Barbara Stahl »

I fully agree - and it was not a problem at all to get this Excel sheet.

This is really kind1 Many thanks to Prof. Henseler!

Cheers, Barbara
Post Reply