Special Topics Regarding the Analysis

Abstract

Throughout this book we have used Lenth t statistics to conduct individual significance tests for unreplicated designs. Here we discuss modifications to Lenth’s method when there is minimal replication.We also discuss alternatives to Lenth’s method, especially for controlling the experimentwise error. In the previous chapter, we discussed structured sampling within runs in order to study different variance components. Here we illustrate the analysis of such data. This final chapter also discusses analysis for weighted least squares, mixed-model methods, and applications with too many response variables to analyze individually. We conclude with four tips to help one avoid common data analysis errors.