A comparison of some confidence intervals for estimating the population coefficient of variation: a simulation study

Abstract

This paper considers several confidence intervals for estim ating the population coefficient of variation based on parametric, nonparametric and modified m ethods. A simulation study has been conducted to compare the performance of the existing and new ly proposed interval estimators. Many intervals were modified in our study by estimating the va riance with the median instead of the mean and these modifications were also successful. Dat a were generated from normal, chi-square, and gamma distributions for CV = 0.1, 0.3, and 0. 5. We reported coverage probability and interval length for each estimator. The results were app lied to two public health data: child birth weight and cigarette smoking prevalence. Overall, go od intervals included an interval for chi-square distributions by McKay (1932), an interval esti mator for normal distributions by Miller (1991), and our proposed interval