I am currently analyzing a model in which my exogenous variable is type II reflective-formative construct. The HOC has three formative dimensions (3 LOCs) which are individually measured by reflective items. The reflective items of each LOC are not equal so I am inclined to use two-stage model. I am not able to attach the diagram so here is the explanation of the model
My HOC is endogenous that is connected to Y1 and Y2, whereas Y1 is the mediator between HOC and Y2. The relationship between HOC and Y1 is being moderated by M.
I have three questions
1. I understand that for two stage model, I model the relationships with LOCs only . i.e. all the three LOCs have relationships to the Y1 and the Y2 and HOC is not in the structural model. Then calculate the latent variable scores for the three LOCs and put them in as formative items for the HOC in the second stage (following Becker et al 2012, Figure B3). My question is when and how to include the moderator. Should that be only included in second stage or am I to include three moderator arrows to three LOC relationships to Y1 in first stage too (the latent variable calculation stage)?
2. Also I understand that I need to use Mode A for the first stage calculation but for the latent variable relationship to HOC at second stage I should be using Mode B as this is formative. Is this correct? Furthermore is path model setting fine for all calculations?
3. As my exogenous variable is a reflective formative construct, is it necessary to use the two-stage approach to moderation? Hair et al (2017) recommends to use two-stage approach to moderation if the moderator or exogenous variable is formative? But I have seen various videos etc that state that product indicator is better and should be used. Any insights on both these issues will be very helpful.
Looking forward to your help.
Thanks in advance
Frequently asked questions about PLS path modeling.
1 post • Page 1 of 1