Better confidence intervals for the population coefficient of variation

Abstract

The paper considers different confidence intervals for the population coefficient of variation based on the resampling methods, the ranked set sampling, the resampling methods for the ranked set samples and the partial ranked set sampling. We modify the existing confidence intervals for the population coefficient of variation using the trimmed mean in order to get better coverage accuracy. The results for the Gamma, Weibull, Log-normal and skew-normal distributions and for the real data set are reported. The best coverage probabilities are obtained using the confidence intervals based on the trimmed mean.

Keywords:

• Confidence interval for the population coefficient of variation
• Coverage accuracy
• Resampling methods
• Ranked set sampling
• Trimmed mean

JEL CLASSIFICATION:

• C13
• C63

Notes

1 Coverage probability represents the proportion of confidence intervals which contain the parameter value.

2 α is the ponder that is used to compute a coefficient on the basis of which we can determine which part of the partial ranked set sample is simple random sample.

3 We considered the trimmed mean 5% and 10% and decided to present the results for the trimmed mean 10%, because there wasn’t the significant difference between the results.