Abstract
Mediation and moderation analyses are increasingly being used in accounting research. In this manuscript, we explain mediation and moderation analyses in plain-English, and demonstrate how to use an efficient tool to help build your theory, provide inferential tests of indirect (mediated) effects, and interpret your results. Briefly, mediation is an expansion of the relation between an independent variable (X) and a dependent variable (Y); it seeks to explain how or why X influences Y, i.e., the mechanism underlying the basic hypothesized relationship. An independent variable can affect the dependent variable directly (as in most models), indirectly through mediators, or both. Moderation is a conditional analysis (equivalent to an interaction in regression analysis); it seeks to understand when X influences Y or for whom the relationship exists or varies in strength or sign. Moderated mediation is when the mediated relationship between X and Y is conditioned by variable W. Many theories in accounting research can be conceptualized as mediated, moderated or moderated-mediation models to investigate both simple and complex hypothesized relationships as they seek to understand how and when X influences Y. Analyses using these models capture the dependent nature of an entire set of relationships rather than attempting to make piecemeal inferences from a series of univariate regressions that may not be as revealing and may even yield misleading inferences. Tools such as the PROCESS macro (Hayes 2020) facilitate examination of a set of conditional relationships, reduce the number of inferential tests that are relied on, and use bootstrapping for inferential tests of moderated mediation that do not rely on distributional assumptions. We provide two examples from published research to illustrate these concepts and analysis using the PROCESS macro.
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have