Adaptive Robust Profile Analysis of a Longitudinal Data

This paper uses nine winsorized scores in the adaptive test of Hogg, Fisher and Randles and deals with its extension to hypothesis testing in profile analysis of a balanced longitudinal data. Simulation studies are conducted to evaluate the efficiency of the adaptive test procedure relative to the traditional ANOVA-F test for different non-normal data sets. To illustrate the feasibility of the test, we analyzed a real data set from the study of tumor sizes in mice.