Abstract
Recent simulation studies have pointed to the higher power of the test for the mediated effect vs. the test for the total effect, even in the presence of a direct effect. This has motivated applied researchers to investigate mediation in settings where there is no evidence of a total effect. In this paper we provide analytical insight into the circumstances under which higher power of the test for the mediated effect vs. the test for the total effect can be expected in the absence of a direct effect. We argue that the acclaimed power gain is somewhat deceptive and comes with a big price. On the basis of the results, we recommend that when the primary interest lies in mediation only, a significant test for the total effect should not be used as a prerequisite for the test for the indirect effect. However, because the test for the indirect effect is vulnerable to bias when common causes of mediator and outcome are not measured or not accounted for, it should be evaluated in a sensitivity analysis.
Highlights
The Baron and Kenny (1986) description of how mediation can be statistically assessed has led to a massive number of applied publications in the psychological literature over the last 25 years
MacKinnon (2008, pp. 38–40) distinguishes two critical parts in this mediation model: the first part is referred to as “action theory,” which describes how the independent variable or intervention X changes the mediating variable M; and the second part as “conceptual theory,” which specifies how the mediator affects the dependent variable
We show that the circumstances under which the power of the test for the mediated effect vs. the power of the test for the total effect is largest, are relatively vulnerable to violations of the no unmeasured M-Y confounding assumption
Summary
The Baron and Kenny (1986) description of how mediation can be statistically assessed has led to a massive number of applied publications in the psychological literature over the last 25 years. The estimation of direct and indirect effects may be biased in such randomized experiments This may happen when a variable other than the independent variable affects both M and Y and is not controlled for (e.g., because it is unmeasured). Very few applications control for variables that may affect both M and Y, nor do they discuss how plausible it is to assume the absence of such variables, an assumption often referred to as no unmeasured confounding of the M-Y relationship This is daunting: even when in reality there is no effect of M on Y at all, and no indirect effect, an analysis that ignores common causes of M and Y may reveal a spurious effect of the mediator on the outcome. One can pro-actively think about potential common causes of www.frontiersin.org
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callDisclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
