A1157
Title: On the multi-sample problem in the presence of random right censoring
Authors: Jaco Visagie - North-West University (South Africa) [presenting]
James Allison - North-West University (South Africa)
Anke Steyn - North-West University (South Africa)
Abstract: A classical problem in survival analysis is testing whether two or more independent samples are realized from the same distribution. In practice, this problem is often complicated by the presence of random right censoring. A new test statistic is introduced based on empirical characteristic functions. The proposed test statistic is a weighted L2-type distance between the empirical characteristic functions of the observed samples. The counterpart of the empirical characteristic function in the presence of censoring is based on the Kaplan-Meier estimator. The finite sample properties of the newly proposed test are investigated via simulated as well as observed data.
