Good Practice in Experimental Moderation Designs

Siegfried Dewitte, Catholic University Leuven, Belgium
The field’s favorite way of gaining insight in the processes underlying a causal effect is by means of statistical mediation tests, which is not without its problems. One powerful alternative is the moderation-of-process design, in which the process, rather than measured, is experimentally manipulated. However, three mistakes can be made. Rather than changing the process, the moderation manipulation may be associated with the independent variable, the dependent variable, or even change the process all together. A set of simple checks helps to make sure that that the moderator is exclusively related to the process of interest.
[ to cite ]:
Siegfried Dewitte (2009) ,"Good Practice in Experimental Moderation Designs", in NA - Advances in Consumer Research Volume 36, eds. Ann L. McGill and Sharon Shavitt, Duluth, MN : Association for Consumer Research, Pages: 82-85.