Parametric Bootstrap Approach for Comparing the Means of Two Independent Lognormal Distributions

• Activity Number: 613
• Type: Contributed
• Date/Time: Wednesday, August 12, 2015 : 2:00 PM to 3:50 PM
• Sponsor: Section on Statistical Computing
• Author(s): Ahmet Sezer* and Berna Yazici and Evren Özkip
• Companies: Anadolu University and Anadolu University and Ankara Police Collage
• Keywords: Bootstrap ; Lognormal ; Generalized p value

Abstract

Faced with positive right-skewed data, the lognormal distribution is frequently used. It is difficult to construct exact tests and confidence intervals for the ratio of two lognormal means. The parametric bootstrap approach is a type of Monte Carlo method applied on observed data. This approach can be applied in situations where samples or sample statistics are not easy to derive. In this study, we use a parametric bootstrap approach for comparing the means of two independent lognormal distributions. We also include Z test based on large sample method and generalized p-value approach. A Monte Carlo simulation study is conducted to evaluate type I error probabilities and powers of these tests under different scenarios.