currently, I am trying to test effects of corporate reputation and customer satisfaction on customer citizenship behaviour. customer citizenship behaviour is measured using 4 separate constructs that is "helping other customers", tolerance towards the firm", "advocating for the firm" and "providing feedback". I have run the CFA and SEM and in SEM the model appears to be not good unless I add "helping other customers" as a mediator between independent (customer satisfaction and corporate reputation)and other dependent variables (Tolerance, Feedback, advocacy). my thought is that there is another variable that is similar to " helping other customers" that actually mediate between all these. such as Empathic or altruism traits. My question is how to deal with such results, given all four dependent variables are supposed to make a second-order construct that is customer citizenship behaviour.
If, as you mentioned, there is a reference of conceptual similar variables and theoretical frameworks to your varibale mediator that fulfill that role, and you have some study that supports that, you can propose the mentioned mediation analysis, if your objective is based on fulfilling only the predictive model you can consider to remove some item or factor that does not adjust to the model or also correlation of covariance error in the model according to the modification indexes.
- Badges
- Science method
- Similar topics
- Mathematics
- Statistics
- Statistical Techniques
- Structural Equation Modeling