A novel approach for investigating moderated-mediation

What is it about?

Interventions are designed to increase health promoting behaviors. It is often of interest to investigate if these intervention effects are different for different groups of people. For example, does an intervention designed to increase healthy eating behaviors work better for individuals who start out with poor eating behaviors compared to those that start out with adequate eating behaviors? The study of how intervention effects differ across groups of individuals is termed moderation analysis. Currently, moderation analysis is well-equipped to investigate if intervention effects differ across subgroups of individuals when individuals are categorized into discrete groups such as the assigned sex categories male and female. However, there is little guidance on investigating if intervention effects differ across subgroups when individuals are not easily categorized into discrete groups. For example, individuals may vary across a continuum on healthy eating behaviors prior to the start of an intervention. This study aims to demonstrate the application of a statistical method that can estimate treatment effects for subgroups of individuals when they are not easily categorized into discrete groups.

Featured Image

Photo by Michael Carruth on Unsplash

Photo by Michael Carruth on Unsplash

Why is it important?

The method described in this paper can also be extended to help determine if mechanisms targeted by an intervention are affected differently for different groups of individuals. In other words, did the intervention increase healthy eating behaviors through its effect on increasing the intent to eat healthy and does this effect depend on how healthy the individuals were prior to the start of the intervention?