Avoidance of representativeness in presence of effect modification

Abstract

We read with interest the contribution by Rothman et al. ^{1, 2} who believe that representativeness of a target population should be avoided. Avoidance of representativeness is an active process and obviously means that the investigator intentionally selects a subgroup of the target population that is not representative of the target population. As a result, the distributions of exposures and disease risks do not represent those of the target population any more.We believe that Rothman et al.’s suggestion is only reasonable for associations that are not modified by selection factors that modify the association between the exposure and disease of interest.