Some Versatile Tests Based on the Simultaneous Use of Weighted Log-Rank Statistics

Abstract

There are many situations where investigators are not able to specify in advance the specific type of the survival differences that may exist between two groups. In this article, some versatile test procedures sensitive to a wide range of types of survival differences have been proposed. Either the maximum or a linear combination of selected members of the family of weighted log-rank statistics is considered. Simulation studies provide some insights into the properties of the proposed statistics across several types of survival differences and degrees of censorship.