Demystifying Moderators and Mediators in IDD Research

Background: Relative to other areas of psychosocial research, moderator and mediator analyses are sparse in intellectual and developmental disability (IDD) research. Moderator and mediator variables are necessary for understanding how and why relationships exist between variables, and are incredibly useful in treatment research. They allow researchers to hone in on causal processes and be more efficient in selecting study groups and independent variables. It is not clear why IDD research has been relatively slow to adopt the search for moderators and mediators, although several potential roadblocks are readily identified. First, the jargon of social psychology, where moderators and mediators were first widely-used, can be confusing and daunting. Second, even following the seminal Baron and Kenny (1986) paper, the distinction between the variables themselves is often confusing. Third, technical papers that describe the mechanism for testing the statistical significance of moderators and mediators appear in journals not frequently accessed by IDD researchers and are difficult for non-statisticians to understand.

Objectives: In this presentation, I will explain moderators and mediators in terms familiar to IDD psychologists, using examples from IDD literature. I will also propose that the MacArthur guidelines (Kraemer et al., 2001) be incorporated into IDD research to limit the confusion between moderator and mediator variables, and I will describe in detail the best practice for statistical analysis of moderators and mediators. Finally, I will review the IDD literature to identify strengths and weaknesses in our current utilization of moderators and mediators.

Methods: First, rules and tips for defining, classifying, and quantifying moderators and mediators will be set forth. Second, the field will be assessed with respect to these “best-practice” guidelines. Five major IDD journals were searched for studies that investigated moderators or mediators, and the resulting articles were evaluated in the following areas: (a) publication year, (b) journal, (c) topic, (d) statistical method, (e) quantification of the mediated effect (for mediation studies), (f) criteria used for moderation/mediation (including temporal and causal relationships), and (g) “pluralism,” a term used here to mean inappropriately testing the same variable as both a moderator and a mediator.

Results: Although the Baron and Kenny (1986) definitions are the most widely-used, the MacArthur guidelines (2001) represent a more useful approach to moderation and mediation. Additionally, recent statistical publications in other fields have made the quantification of moderator and mediator effects less daunting to the typical IDD researcher. Although the number of moderator/mediator papers published in the past few years exceeds previous decades, our field is still far behind others in the use of such “third variables.” Only 10% of moderator analyses and 23% of mediator analyses adequately fulfilled all of the criteria set forth in this review.

Conclusions: Although moderators and mediators are very useful for understanding causal processes in greater detail than afforded by most commonly-used statistical procedures, IDD researchers have been more reluctant than others to analyze these variables. The field as a whole will advance if we expand our respective repertoires to include the recent theoretical and technical advances outlined in this paper.