Control of initialization bias in multivariate simulation response

Publisher Website
Google Scholar
Abstract
A procedure is proposed for controlling initialization bias in multiple response simulation output. The technique is an adaptation of multivariate quality control methods using a two-sample T 2 statistic for testing equality of means. An example taken from an actual simulation study illustrates how initialization bias can go undetected when an aggregate scalar output is used to control initialization effects. A study of a simple computer time-sharing model illustrates that the proposed bias control method can be effective and efficient.