I recently discussed1 some common difficulties with employing continuous outcomes and how established conversion methods might help. Space limits precluded some worthwhile considerations, and I highlight a few here.

Although continuous outcomes carry certain inferential considerations,1 this is not an insurmountable ‘problem’ per se. An estimated proportion of people who will achieve some degree of effect may seem easier to understand. However, one must respect: (1) what such estimation entails when the underlying data are continuous, and (2) that continuous data are not just irrelevant or inapplicable; they may just require additional effort …