Confidence Intervals for Single Coefficient of Variation of Weibull Distribution

Abstract

The aim of this paper is to propose the new confidence intervals for single coefficient of variation of Weibull distributions, using the concept of the generalized confidence interval (GCI), the fiducial generalized confidence interval (FGCI), and the bootstrap percentile method. The coverage probabilities and the expected lengths were evaluated via Monte Carlo simulation. The simulation results showed that the coverage probabilities of the GCIs and the FGCIs were greater than or close to the nominal confidence level. Both of them were recommended. The coverage probabilities of confidence intervals based on the bootstrap percentile method were under the nominal confidence level. Regarding the expected lengths, they tended to decrease when the sample size was increased. All proposed confidence intervals were applied to some real world data in this study.