linear or non-linear pls sem

Questions about the implementation and application of the PLS-SEM method, that are not related to the usage of the SmartPLS software.
Post Reply
chandu
PLS Junior User
Posts: 4
Joined: Sun Oct 26, 2014 1:34 pm
Real name and title: Praveen Doctoral Student

linear or non-linear pls sem

Post by chandu »

Hi
As I understand, smartpls uses linear pls regression and warppls uses non-linear pls regression.

How should one choose between linear and non-linear regressions?
I have a likert scale data. Which one is more applicable in this case?

WHat is the interpretation of non-linear regression coefficient, which warppls gives?

Can someone help with the above question?

Thanking you

Praveen S
Doctoral Student in marketing
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:

Re: linear or non-linear pls sem

Post by Hengkov »

Hi,

According to Herman Wold, term Non-Linear in NIPALS algortihm have ambiguity, because PLS always can used in linear condition. So, this is answer why Wold change name NIPALS to PLS based this problem. WarpPLS software try to adopt this term "Non-Linear" but you just can run it using (Warp3 algortihm). Non-Linear means, your model have quadratic or interaction effect. In WarpPLS 4.0 software, you have options to run model with "linear relationship", and the result simillar with SmartPLS. You can start to understanding WarpPLS software with read my book.

"Partial Least Squares: Concepts, Methods and Appliactions using WarpPLS 4.0 (2nd), Diponegoro University Press".

Best regards,
chandu
PLS Junior User
Posts: 4
Joined: Sun Oct 26, 2014 1:34 pm
Real name and title: Praveen Doctoral Student

Re: linear or non-linear pls sem

Post by chandu »

Thanks Hengky

I will definitely go through the book. I will need it because I am using it for my dissertation..


One more doubt.
How can one judge the acceptability of different PLS softwares?

Coefficients are similar in smartplS and Warp-pls. Only 1of 6 coefficients is significant in smartpls. But, I get all the coefficients significant in Warppls.

Can I trust Warppls?
jmbecker
SmartPLS Developer
Posts: 1282
Joined: Tue Mar 28, 2006 11:09 am
Real name and title: Dr. Jan-Michael Becker

Re: linear or non-linear pls sem

Post by jmbecker »

SmartPLS implements the bootstrapping procedure to test the significance of path coefficients. It is a very established method that is well published and widely used, not only in the PLS context.
As far as I know, WarpPLS uses several methods to test the significance of coefficients. Some of them are not published are extensively researched. You will find some resources/publications about them on the WarpPLS websites.
I’m not saying that they are be generally bad. Yet, I won’t trust them as much as the bootstrapping procedure as long as I don’t see an extensive assessment of them in a good journal.
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
chandu
PLS Junior User
Posts: 4
Joined: Sun Oct 26, 2014 1:34 pm
Real name and title: Praveen Doctoral Student

Re: linear or non-linear pls sem

Post by chandu »

Thanks JMbecker

WarpPLS also has bootstrapping option. But, the results for significance are very different from SmartPLS.

WarpPLS also has Jackknife option. I think this is also widely accepted method.
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:

Re: linear or non-linear pls sem

Post by Hengkov »

Hi,

Yes, WarpPLS 4.0 provide four options to test significance of path coefficients (Bootstrapping, Jackknifing, Stable and Parametric). I think this is a good innovation by WarpPLS software and not bad. Many researcher and reviewer few confuse about the result bootstrapping is not stable (because every run the result is change). This is make possible to inflation type I or II error. For example, if the running result one is T-statistic = 1.968 and the running result two is T-statistics = 1.784, the researcher will confuse to make the judgment (two-tailed test) is reject or eccept hyphotheses. Based on this problem, I think the developer WarpPLS develop stable method to test significance of path coefficients. Stable method will result the stable analysis (T-statistics or P-value) if you run more than one. With another word, this method simillar with many program statistics (such IBM SPSS, STATA, LISREL, AMOS, Mplus etc). Next, about the parametric, this method test significance of path coefficients under normality assumption. This is a good idea, because some techniques in PLS context required normality assumption (such FIMIX-PLS, Multigroup Analysis Parametric etc). For detail information about WarpPLS software, you can read my book WarpPLS (2nd ed). I also prepare SmartPLS 3 book (2nd ed) for publish next year. I very like both software, and I not happy if the both software bettle or claim one is batter than other. Both software is the best for me because very innovative and have good ideas.

Best regards,
chandu
PLS Junior User
Posts: 4
Joined: Sun Oct 26, 2014 1:34 pm
Real name and title: Praveen Doctoral Student

Re: linear or non-linear pls sem

Post by chandu »

Thanks Hnegky
bradley7680
PLS Expert User
Posts: 51
Joined: Tue Nov 04, 2014 5:27 am
Real name and title: Dr. Bradley Wilson. Senior Lecturer in Advertising. School of Media and Communication. RMIT University.

Re: linear or non-linear pls sem

Post by bradley7680 »

have a footnote in your thesis if you using different software you can comment on the differences

i did this for my thesis... if it is out at third decimal place then for most social science problems that is good enough...

see researchgate for how I reported.

Brad
Bradley Wilson. Ph.D. Senior Lecturer in Advertising. School of Media and Communication. RMIT University. Melbourne, Australia.
SEE: http://www.researchgate.net/profile/Bradley_Wilson2 http://scholar.google.com.au/citations? ... AAAJ&hl=en
Post Reply