On the Causal Interpretation of Randomized Interventional Indirect Effects
share
Identification of standard mediated effects such as the natural indirect effect relies on heavy causal assumptions. By circumventing such assumptions, so-called randomized interventional indirect effects have gained popularity in the causal mediation literature. In this article, I introduce an essential criterion (the sharp null criterion) that any indirect effect measure ought to satisfy in order to have a true mediational interpretation. Namely, an indirect effect measure must be null whenever there is no individual-level indirect effect. I show that without stronger assumptions, randomized interventional indirect effects do not satisfy this criterion. I additionally discuss alternative causal interpretations of such effects.
READ FULL TEXT