Questions about Repeated indicator approach

[Yvone]

hi all. I constructed a model that is two-order (first-order reflective and second-order formative). And I use repeated indicator approach(model B). However, I am not sure whether I use the approach in a correct way. It is about overall perceived value and behavioral intention. There are 7 dimensions under overall perceived value. And each dimensions has 3-4 reflective indicators. Thus i simply connect the reflective indicators to corresponding dimension and connect each dimension to the overall perceived value formatively. Finally I connect the indicators of each dimension to the overall perceived value reflectively and run the PLS algorithm. Could anyone let me know if I use this approach correctly? the conceptual model can be seen as the table below.
Another question. After running the PLS Algorithm, I need to analyze the indicator reliability based upon outer loadings. However, there are two outer loadings for each indicator, probably because I repeated the indicator. So, which outer loading should I use?


Thanks in advance!!!!

[jmbecker]

1) Your conceptual model looks ok.

2) You should use the loadings that is attached to your first-order constructs and not the loading on the second-order construct.

[agalvez]

1) Your conceptual model looks ok.

2) You should use the loadings that is attached to your first-order constructs and not the loading on the second-order construct.

Dear jmbecker

If you let me ask, If one item has a outer loading less than 0.708, then, it is removed from the LOC and also from the HOC and run PLS again? Or it should be deleted only from the LOC? I think I have to delete it from both LOC and HOC, but only check if it loading is >0.708 from the LOC.

Thank you.

[jmbecker]

Yes, you are right:
You check the loadings only for LOC, but you would delete them from both LOC and HOC if you delete.

Rather than deleting, I would always recommend to look at the other reliability measures first. If your indicator has a small loading, but overall AVE, composite reliability, etc. are good, I would keep the item.

[agalvez]

Thank you again !