Power comparison between two multivariate permutation tests for matched-pairs designs

A permutation maximal paired t-test (MPT) and permutation maximal parried signed rank test (MPSR) for testing problem in matched pairs designs are compared. A simulation study is conducted to estimate the power of these two tests for various parent populations. The effect of the dimension of the response variable on the power of these tests is studied. It has been observed that the power of these tests increases as the dimension of the response variable increases. In addition, the power of these two tests increases as the correlation between the components of the response variable decreases. It has also been noticed that when the underlying distribution is symmetric, the MPT test pro forms better than the MPSR test, while their performances are close when the distribution is skewed. (author). 18 refs., 12 figs., 5 tabs