Regression score or composite scores

[a.bury]

Hi,

I am used to work with AMOS and I am passing to AMOS these days.
My problem is this...
I have seven constructs which represent seven independent variables. I would like to use them in order to realize a cluster analysis with SPSS or another software.
In AMOS, I would estimate a measurement model and then I would calculate regression scores ( Data imputation ---> Regression imputation ).
How can I calculate regression scores in SmartPLS ? Perhaps, Should I use latent variable ? In this case, which model should I estimate ?
Can I estimate a model with independent variables only in SmartPLS ?

[jmbecker]

Yes and No.

Latent variable scores in PLS are similar to the regression scores in AMOS. It would be the value that you want to extract from a PLS analysis in your case.

However, you cannot estimate a model where the variables are not connected in PLS. PLS requires a path model. Otherwise it would be a PCA (Principal Component Analysis) for each set of indicators. That could of course be done in SPSS and the scores could be extracted from there. But it is not PLS anymore.