Paired Observations from a Survival Distribution

Abstract

Hypothesis tests are developed to investigate the equality of mean survival times when the observations from the population under study come in pairs, for example length of survival of a tumor remission when a patient receives a standard treatment versus length of survival of a tumor remission when the same patient, at a later time, receives an experimental treatment. Two cases are considered: (i) the case where the paired observations are assumed to be statistically independent; (ii) the case where the independence assumption is relaxed. The underlying survival distribution was first developed by Block and Basu (1974, Journal of the America> Statistical Association 69, 1031-1037).