Exploring the “Fourth Drawback”: Overlap Between Mediator and Criterion in Mediation Analysis

Abstract

In mediation analyses, several assumptions are routinely made. One such assumption is that the mediator and criterion variables are distinct, and not essentially the same variable with overlapping content. When this assumption is strongly violated, traditional mediation analysis is likely to falsely indicate that mediation has occurred. Spencer et al. refer to this as the “fourth drawback” of mediation analysis. It is argued here that this fourth drawback is not confined to cases of nearly complete overlap between mediator and criterion; mediation may often be falsely inferred when such overlap is partial. The results of a series of simulation studies are reported that corroborate this conclusion. Some potential responses to the risks of the fourth drawback are discussed.