Mediation and moderation analysis from the perspective of behavioral science / Sharifah Khairol Musairah

Mediation and moderation are theories for understanding causal relationships. The purpose of this article is to make researchers aware the difference between mediator and moderator. Research that combines mediation and moderation are common in either basic or applied behavioral research. Usually, this kind of research is structured in terms of mediated moderation or moderated mediation. Unfortunately, many researchers use the terms mediator and moderator interchangeably; not realizing the difference between the two. Therefore, in this article, the differences between a mediator and a moderator are outlined. This article described conceptual basis, research model, data analysis and construct validity issues that are necessary for making inferences in mediation and/or moderation analysis. Then, this article defines the mediating variable and how it differs from moderating variable. Next, longitudinal mediation model, model with moderators as well as mediators, and causal inference for mediation models are described. An empirical illustration is provided using structural equation modeling (SEM) techniques. Statistical methods to analyze mediation and moderation as well as newer techniques are described. Future directions using mediators and moderators with more elaborate models such as moderated mediation and mediation moderation are discussed conceptually.