Abstract
In a 2-step monotone missing dataset drawn from a multivariate normal population, T2-type test statistic (similar to Hotelling’s T2 test statistic) and likelihood ratio (LR) are often used for the test for a mean vector. In complete data, Hotelling’s T2 test and LR test are equivalent, however T2-type test and LR test are not equivalent in the 2-step monotone missing dataset. Then we interest which statistic is reasonable with relation to power. In this paper, we derive asymptotic power function of both statistics under a local alternative and obtain an explicit form for difference in asymptotic power function. Furthermore, under several parameter settings, we compare LR and T2-type test numerically by using difference in empirical power and in asymptotic power function. Summarizing obtained results, we recommend applying LR test for testing a mean vector.
Appendix A Proofs and auxiliary results A.1. Proof of Theorem 2.1
We define the following random variables: where is a matrix, and is a matrix. Furthermore, are mutually independent. Using these random variables, we obtain the following stochastic representations: where
Here,
First, we obtain the stochastic expansion of Define and respectively as where
Then the stochastic expansion of is (A.1) (A.1) where
Next, we derive the stochastic expansion of The result is as follows: (A.2) (A.2) where
Here,
Finally, we derive the stochastic expansion of T2. From (A.1) and (A.2), the stochastic expansion of T2 is obtained as where
We define where and Then, we express Q0, Q1 and Q2 as follows: where
The characteristic function of T2 can be expressed as
To evaluate the expectation (A.1), we consider the following transformation:
The Jacobians are and Thus we obtain where denotes the probability density function of and
Through this variable transformation, the characteristic function (A.1) is rewritten as follows:
This notation expresses the expectation of and
We now evaluate the expectation of higher-order terms in the characteristic function (A.1). Using the moments stated in subsection A.2, the expectations of each term are obtained as follows:
Organizing by the power of we obtain (A.3) (A.3)
Performing an inverse Fourier transform on each term in (A.3), we finally obtain (i) in Theorem 2.1.
Next, we will show (ii). The stochastic expansion of Λ is obtained as where
Here,
The expectations of is obtained as follows:
Let be the characteristic function of Λ. Organizing by the power of we obtain (A.4) (A.4)
Performing an inverse Fourier transform on each term in (A.4), we finally obtain (ii) in Theorem 2.1.
A.2 Proof of Theorem 2.2
Let be the characteristic function of Organizing by the power of we obtain where
Hence if Performing an inverse Fourier transform, we finally obtain Theorem 2.2.
A.3 Some moments
Assume that and and are p × p symmetric matrices. We divide into p1 and p2 as follows:
Then it holds that
These results can be directly derived from some results of the moments of the Wishart matrix. For some results of the moment about the Wishart matrix, see Gupta and Nagar (Citation2000).