I have two second order variable (A and B) which are predictor variables for one construct (C). Moreover, I created an interaction effect (additional Variable) with both variables by using the product of the latent scores of variables A and B. Is it a problem or redundant to additionally use B as a mediation variable? I have tested the Mediation effect of variable B by using a Sobel Test. However, I am not sure if this somehow bears redundant information if I use both, the interaction variable and the mediation. I haven't found any clear information on that. I appreciate any information
- Interaction
- Interaction.png (176.17 KiB) Viewed 3108 times
- Interaction + Mediation
- Interaction + Mediation.png (178.78 KiB) Viewed 3108 times