WarpPLS CFA

[Ahmed Elbaz]

Dear all

Thank you very much for your continuous help. I have a question regarding my
work. I already finished my analysis
using WarpPLS . At the first run of my model I removed 7 indicators because
of notsiginficant or high VIFs. After the second rum after removing the
first 7 indicators, the model is not working good. To make my model fit, I
used trail and error ( I used t-test with my indicators, and I removed
indicators with not significant). This method helped me to fit my model. The
question now is how can I scientifically write this in my thesis? Is this
what is called Post Hoc Analysis?


Best Regards


Ahmed Elbaz
Associate Lecturer
PhD Researcher (Tourism Organisations)
School of Tourism & Hospitality
Plymouth Business School
Room 405, Cookworthy Building,
University of Plymouth.
England, United Kingdom
PL6 4AA
Tel: 07762894737
-----
No virus found in this message.
Checked by AVG - www.avg.com
Version: 2012.0.2221 / Virus Database: 2441/5293 - Release Date: 09/26/12

[Ahmed Elbaz]

Dear Prof. Dr Ned Nock

Testing a model with one mediator between a developed and a developing country:
the model: X--->M----->Y where M is the mediator
M in the model refers to adoption levels, in which we have four levels, each level has a number of specific items. That is why we use these levels as formative constructs.
Testing levels in developed and developing contexts, we found that levels in developing countries do not have the same features in developed countries.
To explain more:
Level 1 has five items (x1, x2, x3, x4, and x5)...testing this in developed countries, level 1 has the five items (x1-x5) but in developing countries, level 1 has three items out the five (x1, x2 and x4).
in this case when we compare the models do we have to compare APC, ARS, and AVIF or how can we compare those two models.
i.e., x and y are reflective constructs in the model.

Ahmed Elbaz
Associate Lecturer
PhD Researcher (Tourism Organisations)
School of Tourism & Hospitality
Plymouth Business School
Room 405, Cookworthy Building,
University of Plymouth.
England, United Kingdom
PL6 4AA
Tel: 07762894737

________________________________________
From: Ned Kock [ned...@scriptwarp.com]
Sent: 27 September 2012 15:46
To: Ahmed Elbaz
Subject: RE: WarpPLS CFA

Hi Ahmed.

My guess is that it would depend on the publication outlet (e.g., journal,
journal department) you are targeting.

If they were interested in the final results, they may not want you to go
through all of the details of your analysis; even to the point of telling
your to remove them, if you include those details.

It would be different if they were specifically interested in a
methodological discussion, particularly a model-building one.

Btw, I think this would be a good question to pose to the PLS-SEM group, to
see what others think.

Ned


-----Original Message-----
From: Kock, Nereu F [mailto:ned...@tamiu.edu]
Sent: Wednesday, September 26, 2012 4:00 PM
To: ned...@gmail.com
Subject: FW: WarpPLS CFA




Nereu F Kock
Professor
ned...@tamiu.edu
Tel. (956) 326-2521
Fax (956) 326-2494
School of Business, A.R. Sanchez, Jr.
Texas A&M International University
PH 304B
5201 University Boulevard
,

As pursuant to Texas A&M International University rule 33.04.99.L2
concerning the Use and Disposition of Electronic Communications, this email
is a mechanism for official communication of the University. Electronic mail
(e-mail) should be used only for legitimate academic or state business.
Official email communications are intended only to meet the academic and
administrative needs of the campus community.


________________________________________
From: Ahmed Elbaz [ahmed...@plymouth.ac.uk]
Sent: Wednesday, September 26, 2012 1:34 PM
To: Kock, Nereu F
Subject: WarpPLS CFA

Dear Professor Dr Ned Nock

Thank you very much for your continuous help. I have a question regarding my
work. I already finished my analysis
using WarpPLS . At the first run of my model I removed 7 indicators because
of notsiginficant or high VIFs. After the second rum after removing the
first 7 indicators, the model is not working good. To make my model fit, I
used trail and error ( I used t-test with my indicators, and I removed
indicators with not significant). This method helped me to fit my model. The
question now is how can I scientifically write this in my thesis? Is this

what is called Post Hoc?


Please Dr Ned help me in this issue.

[Ned Kock]

This is an issue that requires an entire article, but in simple terms - in a
multi-group analysis, you need to ensure that the measurement models are
equivalent, so that you can properly compare the structural models. The
equivalence and comparison are based on the measurement (or outer) and
structural (or inner) model coefficients.

Here is a blog post on multi-group analysis:

http://warppls.blogspot.com/2011/06/multi-group-analysis-with-warppls.html

Under publications in www.warppls.com, you'll find the doctoral dissertation
by Dr. Murad Moqbel, who is a former student. We made sure that he conducted
a measurement model equivalence test to set the stage for a multi-group
analysis targeting the structural models. The full reference is:

Moqbel, M. (2012). The effect of the use of social networking sites in the
workplace on job performance. Laredo, TX: Texas A&M International
University.

Btw, I think Murad was the first to do this type of test in the context of a
multi-group analysis. It entailed a lot of extra work, but I think it was
worth it.

Ned