It is best to measure each variable separately to determine the effect, because if not, it is more likely to have a moderating effect by reducing the indirect effect.

Hi Caitlin,

when you set up the model with all mediators included, the total covariance between X and Y is decomposed in the separate pathways. Hence, you get the unique indirect effects. The harder task is to come up with an idea, where the correlations among the mediators (that will surely exist) come from. One option is to simply specify error covariances among them, the other to incorporate effects of some mediator on another. This choice is crucial as going for error covariances may imply missing an effect which biases the indirect effect of X on Y going over the respective mediator. If you don't understand that I can clarify.

Best,

Holger

Good Day Ma'am,

Refer the following:

https://www.tandfonline.com/doi/abs/10.1080/14783363.2018.1426452?tab=permissions&scroll=top