Mediation Analysis: Expanding From One Mediator to Multiple Mediators
Mediation analysis is the process of determining whether or not variables acting as an in-between step, called mediators, are present when looking at the relationship between an independent variable X and a dependent variable Y. An educational study was proposed to analyze the effect of different classroom environments on students performance. This study also measured if the student is on-task or off-task, and if off-task what type of distraction is the student distracted by. It is believed that the particular types of distractions are the mediators and act as an in-between step between the type of classroom environment and how well the students perform. This is what motivated the idea of multiple mediators and the question of how to deal with them. However, previous research that has been done has only looked at single mediators. My research is based on the Product of Coefficients approach (Sobel, 1982). Multiple mediators were analyzed separately or used principal component analysis. Simulated data was used to determine how well these tactics accurately detect mediation. It appears that detecting mediation when there are multiple mediators present is very feasible, and this paper will show the results of this analysis. Finally, this paper also compiled the functions into a statistical package in R, making it easier to analyze data with multiple mediators.
History
Date
2012-05-01Advisor(s)
Howard SelmanDepartment
- Statistics