HOC - 2-stage approach - weighting scheme/CTA

Frequently asked questions about PLS path modeling.
Post Reply
Stadlmann
PLS Junior User
Posts: 2
Joined: Tue Jan 24, 2017 2:15 pm
Real name and title: Christian Stadlmann

HOC - 2-stage approach - weighting scheme/CTA

Post by Stadlmann » Fri Jun 16, 2017 9:01 am

Hello,
in my model I have got 4 HOCs as independent variables consisting each of 2 LOCs. The HCM is a reflective-formative construct (mode A). In the newst book (Advanced issues in PLS-SEM) a factor weighting scheme is recommended. I am using the 2-stage approach, combining the repeated indicators approach (RIA) with an analysis of latent variable scores (Hair, Joe, Jr.. Advanced Issues in Partial Least Squares Structural Equation Modeling). However, the recommendation of using the factor weighting scheme is mentioned in the context of RIA.
Do I need to use the factor weighting scheme also for the settings in second stage (i.e. in the PLS-analysis with the latent variable scores)? I have not found any comment on that so far. I guess to use the same setting (weighting scheme) to be consistent, correct?

Next, I thought about checking the reflective-formative construct with a CTA. Can I do that although I have got only 2 LOCs for each HOC? In the new book at least 4 indicators are mentioned as minimum requirement. On the other hand I have found a similar model constellation (2 LOC for one HOC) in the study of Klarner, Sarstedt, Hoeck and Ringle (2013) in which a CTA was also conducted. However, I guess this CTA was rather checking the Mode A for the HOC than the formative relationship between the 2 LOCs and the HOC.
Can I conduct the CTA for checking the formative nature of the LOC-HOC relationship in SmartPLS3?
Above that also a different weighting scheme (path weight) is proposed for conducting the CTA in the new book. I guess I need to calculate the CTA separately (another time) with the path weighting scheme and not the factor weighting scheme, correct?

Thank you so much for any advise.
Christian
Christian

jmbecker
SmartPLS Developer
Posts: 842
Joined: Tue Mar 28, 2006 11:09 am
Real name and title: Dr. Jan-Michael Becker

Re: HOC - 2-stage approach - weighting scheme/CTA

Post by jmbecker » Wed Jun 21, 2017 10:22 am

1) If you do a two-stage approach you should use the same weighting scheme in both stages. Factor weighting is probably the best, although there are only minor differences. Exception: If you model your higher-order construct with repeated indicator approach the weighting mode has larger implications (results differ sometimes substantially). In Becker et al. (2012) in Long Range Planning, we actually showed that path weighting scheme is usually the better approach for higher-order constructs.

2) For CTA you don't need to specify the weighting scheme.

3) You can do CTA for higher-order constructs if you use the two-stage approach and do the CTA on the second-stage. However, the requirement of 4 indicators remains. There are some strategies to borrow indicators from other constructs that are described in the original article by Gudergan et al., but they are sometimes not easy to implement.
Dr. Jan-Michael Becker, University of Cologne, SmartPLS Developer
Researchgate: https://www.researchgate.net/profile/Ja ... v=hdr_xprf
GoogleScholar: http://scholar.google.de/citations?user ... AAAJ&hl=de

Stadlmann
PLS Junior User
Posts: 2
Joined: Tue Jan 24, 2017 2:15 pm
Real name and title: Christian Stadlmann

Re: HOC - 2-stage approach - weighting scheme/CTA

Post by Stadlmann » Thu Jun 22, 2017 8:22 am

Dear Prof. Becker,
thank you for the clarification.
Best regards,
Christian Stadlmann
PS: I hope to meet you at the next seminar in Hamburg.
Christian

Post Reply