Simultaneous Confidence Intervals for All Differences of Means of Two-Parameter Exponential Distributions

Abstract

For constructing simultaneous confidence intervals for the differences of means of several two-parameter exponential distributions, generalized confidence interval (GCI) approach and method of variance estimates recovery (MOVER) approach are proposed. The performance of the proposed approaches is compared with parametric bootstrap (PB) approach. Simulation studies showed that the MOVER approach performs better than the other approaches: its coverage probability is close to the nominal confidence level and average length is shorter than the other approaches. Three approaches are illustrated using an example.