CMStatistics 2017: Start Registration
View Submission - CMStatistics
Title: Testing the equality of coefficients of variations with right-censored data Authors:  Nihan Potas - AHBV University (Turkey) [presenting]
Abstract: The coefficient of variation is commonly used as a measure of relative variability in clinical trials and medical sciences. To combine or compare variability from several clinical trials has become necessary. Under these circumstances, parametric or non-parametric tests for the equality of coefficient of variations must be widely considered. In the past most of the papers concentrated upon testing coefficient of variations from $k$ normal distributed populations. However, for most of the clinical trials, medical sciences data are longitudinal, failure time observation and it is unlikely to omit the censoring cases. According to that, the aim is to present the test statistics for right censored data to test the equality of coefficient of variations. The empirical sizes and powers of the tests statistics are computed and compared with a simulation study. We provide examples using right censored survival data from oncology clinic. Based on the simulation results right censoring case, likelihood ratio tests give quite good results compare to score tests.