A Note on Testing for Skewness Persistence
指出Muralidhar的偏度持续性检验存在第一类错误膨胀问题，并提出了一个能控制第一类错误的新检验方法，报告了对数正态分布下的检验功效。
This paper shows that the tests of skewness persistence considered by Muralidhar (Muralidhar, K. 1993. The bootstrap approach for testing skewness persistence. Management Sci. 39 487–491.) far exceed the true Type I error. That is, the probabilities of detecting an increase (decrease) in skewness from one time period to another when in fact there is no change are inflated. Consequently, higher power achieved by these tests comes at the cost of a higher than specified level of Type I error. We propose a new test which maintains the specified Type I error levels. Additionally, the power of this test for lognormal distributions is reported.