Abstract

Within statistical methods, design of experiments (DOE) is well suited to make good inference from a minimal amount of data. Two types of designs within DOE are screening designs and optimization designs. Traditionally, these approaches have been necessarily separated by a gap between the objectives of each design and the methods available. Despite being so separated, in practice these designs are frequently connected by sequential experimentation. In fact, from the genesis of a project, the experimentor often knows that both designs will be necessary to accomplish his objectives. Due to advances in the understanding of experimental designs with complex aliasing and their analysis, a current topic within statistics is how to desegregate these methods into a more unified and economical approach. This project is one treatment of that issue.

Degree

College and Department

Physical and Mathematical Sciences; Statistics

Rights

http://lib.byu.edu/about/copyright/