I have tested models for mediation. Because my individuals are children that are nested within families, I conducted my models with a multilevel structure. In the full model, in which I included the direct + the mediated pathways, my estimates for the x-->y and the m-->y pathways become larger. This is strange, because I think they should become smaller. Can you help me out?
Dear Dr. Muthen, I am considering a multilevel mediation analysis. These are my variables: LC_P is the predictor x, FWE_P the mediator m, and JP_P JP_C the outcomes y. I have a diary study for four weeks. Would you please kindly let me know the syntax to run the multilevel mediation analysis? Thanks you a lot.