Improved simplified T2 test statistics for a mean vector with monotone missing data.

In this study, the distribution of the simplified Hotelling's statistic for testing the mean vector when a data matrix has a monotone missing pattern is considered. The test statistic is decomposed, and an asymptotic expansion of the null distribution of each term is derived. Using the result, approximations to the upper percentiles of this statistic are derived. Furthermore, transformed test statistics based on the Bartlett adjustment are proposed. Finally, using a Monte Carlo simulation it is shown that the transformed statistics usually perform better than the original statistic with respect to the speed of convergence to the chi-squared limiting distribution. [ABSTRACT FROM AUTHOR]