Best Practices for Using Median Splits, Artificial Categorization, and their Continuous Alternatives

Abstract

Methodologists have long discussed the costs and benefits of using medians or other cut points to artificially turn continuous variables into categorical variables. The current paper attempts to provide a perspective on this literature that will be of practical use to experimental psychopathologists. After discussing the reasons that clinical researchers might use artificial categorization, we summarize the arguments both for and against this procedure. We then provide a number of specific suggestions related to the use of artificial categorization, including our thoughts on when researchers should use artificial categories, how their use can be justified, what continuous alternatives are available, and how the continuous alternatives should be used.