A Case Study in Sample Size
Abstract: Statisticians are often asked questions related to sample size. However, they are often not well prepared to answer them. There are simple approximate sample-size formulas for such as two-sample t tests, and those can be pressed into service. But most studies are more complicated than that. In fact, often several effects come into play, and thus there are several aspects of the power/sample-size question. This article presents an example of a sample-size problem in a semiconductor experiment having a somewhat complex design. It is hoped that it helps to give some practical guidance in this confusing area.
Keywords: Sample Size - Semiconductor experiments - Split-lot designs