New Tests for Normality
提出一族正态性检验统计量，包括针对偏度、峰度和双峰性备择假设的新检验，使用简便且功效与Oja(1981)的U统计量相当。
A family of statistics for testing normality is presented which includes new tests for skewness, kurtosis and bimodality alternatives. The tests are simple to use and their power properties seem to be as good as those of the very similar U-statistics given by Oja (1981).