[HELP!!!!!] PLS and IPA matrix
Posted: Fri Oct 20, 2017 11:26 am
Dear colleagues,
I have one very importance question for my work and also to this community.
My paper submitted to a journal was rejected and I got a strong impression that the main reason of rejection is; I failed to convince the reviewer that PLS and IPA matrix can be mixed and useful.
My model has three independent variables (A,B,C) and one mediator (D) and one target construct (E). My research had two data sets. One is from the whole industry and one from one specific firm. I used the same PLS SEM model to generate IPA matrix against these two sets. The importance levels A,B,C showed same patterns in the two data sets (importance level A>B>C). But the performance level of A,B,C from the company was A<B<C (in industry it is A>B>C). I made some arguments the company should make some efforts to make the performance level of A,B,C to A>B>C as the same as the industry data.
To convince readers, I have cited Rigdon and Ringle (2011), Hair et al. (2013) and Ringle and Sarstedt (2016), who combined the analysis of PLS-SEM to IPA matrix. I also addressed how this works (total effects represent importance and their average latent variables scores indicate performance)
However, the reviewer's concern is that
- I did not explain how to combine the two (PLS and IPA) but focused on the interpretation of IPA matrix result even though I cited above articles (In fact, he asked why PLS not ML so I think he might not know IPA is not applicable to ML. But he also mentioned formative construct is available to my model so he might know PLS well).
- The reviewer, however, is very kind person. He suggested a way out (combining the two) as following (but I need your support to digest it).
- calculate index scores in each case for A, B, C that somehow really incorporate the IPA results.
- redo the PLS-SEM analysis using A,B,C indexes (rather than as latent unobservables)
Can you please interpret this for me? I know that in Ringle and Sarstedt (2016), total effects represent the predecessor constructs importance and their average latent variables scores indicate performance. Is the reviewers comments really differ from method covered by Ringle and Sarstedt (2016)? I think the reviewer is suggesting a method within CB SEM context not PLS SEM. But I am not sure as I do not have a good knowledge in CB SEM. Especially, I feel "index scores" and "A,B,C indexes" represent CB SEM method. How I could redo with only 'index scores' and 'indexes of A,B,C'?
I will be waiting for your kind advances and comments.
Thanks in advance. Have a great weekend.
Best wishes,
widefov
Reference
Hair Jr, J. F., Hult, G. T. M., Ringle, C., & Sarstedt, M., 2013. A Primer on Partial Least Squares Structural Equation Modeling (PLS-SEM). Sage, Thousand Oaks.
Rigdon, E., Ringle, C., 2011. Assessing heterogeneity in customer satisfaction studies: across industry similarities and within industry differences. Adv. Int. Mark. 22, 169–194.
Ringle, C.M., Sarstedt, M., 2016. Gain more insight from your PLS-SEM results. Ind. Manag. Data Syst. 116, 1865–1886.
I have one very importance question for my work and also to this community.
My paper submitted to a journal was rejected and I got a strong impression that the main reason of rejection is; I failed to convince the reviewer that PLS and IPA matrix can be mixed and useful.
My model has three independent variables (A,B,C) and one mediator (D) and one target construct (E). My research had two data sets. One is from the whole industry and one from one specific firm. I used the same PLS SEM model to generate IPA matrix against these two sets. The importance levels A,B,C showed same patterns in the two data sets (importance level A>B>C). But the performance level of A,B,C from the company was A<B<C (in industry it is A>B>C). I made some arguments the company should make some efforts to make the performance level of A,B,C to A>B>C as the same as the industry data.
To convince readers, I have cited Rigdon and Ringle (2011), Hair et al. (2013) and Ringle and Sarstedt (2016), who combined the analysis of PLS-SEM to IPA matrix. I also addressed how this works (total effects represent importance and their average latent variables scores indicate performance)
However, the reviewer's concern is that
- I did not explain how to combine the two (PLS and IPA) but focused on the interpretation of IPA matrix result even though I cited above articles (In fact, he asked why PLS not ML so I think he might not know IPA is not applicable to ML. But he also mentioned formative construct is available to my model so he might know PLS well).
- The reviewer, however, is very kind person. He suggested a way out (combining the two) as following (but I need your support to digest it).
- calculate index scores in each case for A, B, C that somehow really incorporate the IPA results.
- redo the PLS-SEM analysis using A,B,C indexes (rather than as latent unobservables)
Can you please interpret this for me? I know that in Ringle and Sarstedt (2016), total effects represent the predecessor constructs importance and their average latent variables scores indicate performance. Is the reviewers comments really differ from method covered by Ringle and Sarstedt (2016)? I think the reviewer is suggesting a method within CB SEM context not PLS SEM. But I am not sure as I do not have a good knowledge in CB SEM. Especially, I feel "index scores" and "A,B,C indexes" represent CB SEM method. How I could redo with only 'index scores' and 'indexes of A,B,C'?
I will be waiting for your kind advances and comments.
Thanks in advance. Have a great weekend.
Best wishes,
widefov
Reference
Hair Jr, J. F., Hult, G. T. M., Ringle, C., & Sarstedt, M., 2013. A Primer on Partial Least Squares Structural Equation Modeling (PLS-SEM). Sage, Thousand Oaks.
Rigdon, E., Ringle, C., 2011. Assessing heterogeneity in customer satisfaction studies: across industry similarities and within industry differences. Adv. Int. Mark. 22, 169–194.
Ringle, C.M., Sarstedt, M., 2016. Gain more insight from your PLS-SEM results. Ind. Manag. Data Syst. 116, 1865–1886.