2005 Volume 17 Issue 2 Pages 75-85
When we recognize a significance difference between two population mean vectors by the statistical test, it is often necessary to investigate whether the variates in all variates have the significance difference. In this study, we partition all variates into several partial variates and consider the group sequential procedure to test sequentially the difference between two partial population mean vectors corresponding to each group of partial variates. Then we propose two group sequential procedures based on the statistic by using the design matrix and based on the Rao's statistic. In the simulation, we calculate the repeated confidence boundaries based on two statistics when we change the number of all variates and the number of the group of partial variates. Furthermore we compare the power of the test on two group sequential procedures. In a case study, we apply two group sequential procedures to the significance test after partitioning several groups of partial variates based on weather data of two cities.
