beginner questions

Xtale · Post by **Xtale** » Thu Apr 24, 2008 9:40 pm

Hi,

Im just learning pls methodology for my thesis and I have a couple of questions:

1) My theoretical model is quite complex with four endogenous variables and I wonder if pls can calculate several DVs simultaneous without decreasing the predictivity of the exogenous variables? Or would it be better to set up a model for each DV to achieve the best model quality?

2) Same for moderators - I would like to test at least ten interaction effects. Can pls handle this or would this create to much "noise"? Is it possible to create multiple moderating effects on one DV?

Regards,
Steve

Diogenes · Post by **Diogenes** » Sun Apr 27, 2008 2:03 pm

Hi Steve,
1) Both ways are Ok.
2) PLS works fine and It is possible.
When I am working in an exploratory context, I prefer to begin with two LVs, and to include others LVs after I could understand how are the realtions between LVs that already are in the model.
Best Regards.
Bido

Xtale · Post by **Xtale** » Tue Apr 29, 2008 12:01 am

Hello Prof. Diogenes,

thank you very much!

Another question: I created a moderation effect on a DV which is significant. Afterwards, I added an additional moderation effect on the same variable (non-significant) which in turn decreased the t-value of the first effect, resulting in two non-significant effects. How can this be explained? I have already read the paper from Chin et al. 2003... anyway, still clueless.

Best,
Stefan Kamin