I am relatively new to using PLS-SEM, so my problem might be some basic error of thought.
I am currently trying to model a third order reflective-reflective-formativ model with Smart PLS (repeated indicator aproach). I have six first order constructs (Mode A), two second order constructs (Mode A) and one higher order construct (Mode B). My problem is, that although my values for Cronbachs alpha, composite reliability or discriminant validity seem to be alright, there are way to low values for three out of six first order constructs and both second order constructs. There are many indicator loadings in the range of .4 to .7, but eliminating these indicators from the analysis leads to a reduction of Cronbachs alpha and only slight improvements for AVE.
So my questions is: Are there any other ways to improve AVE? And if not: Are violations of AVE really serious for the model or can they be considered as a minor problem?
Thanks for your help in advance.
Frequently asked questions about PLS path modeling.
1 post • Page 1 of 1