In testing a model with reflective constructs only, i noticed that the values of the indicators of some constructs will change dramatically (e.g., .8->.4 or even from positive to negative such as .7->-.1), when i change the way how other constructs relate to these strange constructs. Say, if i add a new path to allow construct B to affect construct A (one questionable construct), item loading values of the indicators of construct A will see dramatic changes --- which makes the results totally unreliable.
I am using smartpls 3. Any help is appreciated!!!!!!!
Frequently asked questions about PLS path modeling.
1 post • Page 1 of 1