Confidence Intervals for the Mean of Delta-Lognormal Distribution

Abstract

Focusing on delta-lognormal distribution, confidence intervals for the mean are proposed in this research. This will be achieved using a method of variance estimates recovery (MOVER) and generalized confidence interval (GCI) based on weighted beta distribution by Hannig, and MOVER based on variance stabilized transformation (VST). These are then compared with GCI based on VST. The coverage probabilities and average lengths are performances from the presented methods, computed via Monte Carlo simulation. Our results showed that MOVER based on VST is the recommended method under situations of slight probability of being zero and large coefficient of variation in small to moderate sample sizes. Ultimately, rainfall data in Chiang Mai was used to illustrate all of the presented methods.