Thongkam Maiklad

pp. 547 - 552

Abstract

The purpose of this paper is to propose a new variable for the generalized confidence interval method to estimate the confidence interval of the log- normal mean. In order to evaluate  the efficiency of this new method, here called t-generalized  method, a simulation study was conducted to examine and compare the coverage probability, interval width,and  relative bias of this new method and three other methods, the generalized confidence method of Krishnamoorthy and Mathew, the Modified  Cox  methods and Angus's conservative methods.The results show that at small sample sizes with large variances, only the t- generalized  methods and generalized confidence method of Krishnamoorthy and Mathew provide coverage probabilities greater than the nominal level.The t- generalized method is more accurate with a shorter confidence interval than the old generalized confidence method in the case of small samples that have a lot of variances.