problem with second order construct

Questions about the implementation and application of the PLS-SEM method, that are not related to the usage of the SmartPLS software.
Post Reply
Tina28
PLS Junior User
Posts: 1
Joined: Mon Dec 04, 2006 2:21 pm
Real name and title:

problem with second order construct

Post by Tina28 »

Hi,

I have a question about second order constructs: I dont`t want to use the method described by Wold (1982). I want to handle second order constructs like Yi/Davis (2003).

For this method I have to calculate the factor scores for each dimension (reflective first order constructs) of the second order construct at first. For calculating these values for a dimension I "duplicate" the dimension in SmartPLS and connected both constructs, which are (of course) identical (I do this because SmartPLS doesn`t calculate single latent constructs). I don`t know why, but the results (loadings, weights etc.) are correct (that means I get the same values like I get in PLSGraph, when I calculate the single latent construct) ;-)

My problem is the next step:
These factor scores become formative indicators of the second order construct. Then they are multiplied by their loadings and summed up to derive a composite second order score for the second order construct. With this score the second order construct will be integrated in the structural model.
In order to achieve these loadings I tried to calculate this formative construct as described above: I duplicated the construct and then connected both constructs. But with formative constructs I didn`t get the same results (loadings and weights) in SmartPLS compared to PLSGraph. And my question is why?!?
How do I have to use a formative construct (in SmartPLS) in this case?

Thanks a lot!

Tina28
viswadatta
PLS Expert User
Posts: 139
Joined: Wed Jul 05, 2006 1:43 pm
Real name and title:
Location: Coimbatore, India

single construct scores

Post by viswadatta »

You cannot solve single construct models or models with unconnected constructs in smartPLS. Try visual PLS instead, but take care with second order constructs and missing values.
Post Reply