Abstract
When the data dimension is large relative to the sample size, some of the conventional multivariate testing procedures cannot be applied. Recent studies have proposed some test statistics applicable to high-dimensional data. In this thesis, a new test for testing the equality of the mean vectors of two independent normal distributed populations is proposed. Furthermore, the asymptotic power function is obtained. Some simulations are carried out to compare its performance with some existing tests under null and alternative hypotheses, respectively. Finally, the test is applied to Plato's works data and DNA microarray gene expression data of colon cancer tissues to see if it can distinguish the difference between two population mean vectors.