This essay addresses the need for media effects researchers to decompose their structural equation models. We highlight the importance of studying specific indirect effects within a conditional effects framework and discuss how the lack of analysis of this type of effect in structural equation modeling does not fit well with the discipline's theoretical foundations. We summarize several classes of mediation formulas and make recommendations for the estimation and testing of mediating relationships. Finally, an argument is made that the study of mediation is a necessary but not sufficient condition for better understanding media influence.