SILPAKORN UNIVERSITY SCIENCE AND TECHNOLOGY JOURNAL, Vol 7, No 2 (2013)

Double Bootstrap-t One-Sided Confidence Interval for Population Variance of Skewed Distributions

Wararit Panichkitkosolkul




Abstract


This paper proposes a double bootstrap-t one-sided confidence interval for population variance of skewed distributions. The upper endpoint and lower endpoint confidence intervals are studied. The onesided confidence intervals based on the chi-square statistic, bootstrap-t method and double bootstrap-t method are compared via Monte Carlo simulations. The simulation results indicated that the coverage probabilities of bootstrap-t confidence interval can be increased by using double bootstrap resampling. The upper endpoint confidence interval using double bootstrap-t method predominates the other methods with respect to the coverage probability criteria. The performance of the proposed one-sided confidence interval is illustrated with an empirical example.

Key Words: Double Bootstrap-t; Confidence Interval; Variance; Skewed Distribution

Full Text: PDF