Abstract
Understanding the factors that mediate the effect of educational or behavioral intervention is critical to advancing both research and practice. When properly implemented, mediators add depth to the results of intervention research, indicating why a program works, highlighting ways to enhance its effectiveness, and revealing the elements that are essential to successful implementation. However, many researchers find mediation a difficult topic and struggle to implement it properly in statistical models of effects from between-groups randomized studies. In an effort to bring clarity to the topic of mediation and encourage its use where appropriate, this article lays out the requirements for evidence of a causal-mediated effect. An example of a randomized trial of an intervention targeting self-regulation and student behavior is used to illustrate the process of conceptualizing and testing for mediation of treatment effects. Statistical considerations also are addressed.