Abstract
In a prospective study in which females giving birth in a specified small interval of time are followed to obtain the distribution of the duration of postpartum amenorrhea (PPA), the observed distribution involves a selection bias because of an overrepresentation of smaller PPA. In this article, the nature of this selection bias is investigated and an analytical procedure is outlined to adjust the observed distribution of PPA in order to obtain the true distribution. The proposed technique is illustrated with one set of observed data.