Journal of Statistics: Advances in Theory and ApplicationsAuthor's: Nobumichi Shutoh
Pages: [55] - [72]
Received Date: August 30, 2017
Submitted by:
DOI: http://dx.doi.org/10.18642/jsata_7100121871
We propose a method of approximation for the percentiles of the sample coefficient of variation based on the samples drawn from a normal population via Cornish-Fisher expansion. We derive the asymptotic expansion of the distribution of the sample coefficient of variation by using the normal approximation of the Chi-square variable and some useful expectation formulas. Finally, we observe the superiority of our result on the percentiles of the test statistic for one-sample problem by conducting Monte Carlo simulation under some selected parameters.
Cornish-Fisher expansion, coefficient of variation, normal approximation.
