Hi!

I ran a meditational model using SEM. I have an independent variable (X), a mediator (M), and two outcomes (Y1 and Y2).

Using bootstrap I found a mediation effect of M between X and both Y1-Y2 (i.e., full mediation, the direct paths between X and Y1-Y2 become n.s. when in the model is introduced the mediator).

Using SEM, I want to deepen this result highlighting for which outcomes M is the best mediator.

Is it enough to check the magnitude of the two indirect effects?

Alternatively, I thought to constrain the coefficient of the paths M->Y1 and M->Y2 to be equal and then check the models' fit.

If the model fit of the constrained model would result worsen compared to the unconstrained model could help me to sustain that the two paths are likely different in favor of the strongest ones? Thus, if the coefficient of M->Y1 is stronger than M->Y2, then I would sustain that for Y1 (vs. Y2), M is the stronger mediator.

Thanks