In the past, my teacher told me that if a variable is a mediator, we can not consider it as a moderator. However, currently I find so many published articles to explore that a variable was regarded as both a moderator and a mediator in the model. So are these articles correct? Are there any substantial theoies to support for this？ or if we need to do like this, what should we pay attention to?
I am not sure if a variable can be both, because mediator and moderator variables address different questions. Moderator: does age modify the effect of education on income? Yes, because education has no effect on people below a certain age. Mediator: does age mediate the effect of education on income? I don't think so. Mediation has to do with decomposing an effect into direct and indirect paths. So education may "transmit" its effect directly on income by the knowledge and skills conferred. But it may also influence income indirectly by gaining social connections.
Lloyd Balbuena Thank you for your explanation. However, I heard a theory that when some variables (such as, age, gender) can not be affected by a independent variable, these variable cannot be mediators but can be moderators. However, in some conditons, a variable can be regarded as both a moderator and a mediator. For example, teacher liking had both a mediating effect and a moderating effect on the association between adolescent social behaviors and peer relations. Specifically, if researchers think that adolescent social behaviors will much more be affected by other variables (e.g., academic record) than teacher liking, teacher liking can be regarded as a moderator. However, if researchers think that adolescent social behaviors had an impact on teacher liking and teacher liking will affect the peer relations, teacher liking can be regarded as a mediator. what's your idea for this？ Thank you so much！！！
Wenwang Rao You're welcome and thanks for providing your example. In this case, I agree that teacher liking can be used as either moderator or mediator. This being the case, it would be preferable I think, to have a hypothesis one way or the other, based on the literature. If this is not possible, one can always compare Model 1 (mediator) and Model 2 (moderator).
- Badges
- Science topic
- Similar topics
- Philosophy
- Ethics