Level: Beginner
Job Function: Analyst / Scientist / Engineer
Industry: Life Sciences
Ruth Hummel, JMP Academic Ambassador, SAS
Julia Sharp, Associate Professor and Franklin A. Graybill Statistical Laboratory Director
Many academics across a wide spectrum of fields, especially non-engineering fields, are not exposed to efficient design of experiments. These researchers tend to rely on “first course” standard classical designs, and they often find themselves lacking the tools and knowledge to design efficient experiments. We will describe a short stepping-stone course of study for researchers and graduate student statistical consultants, using examples that take a statistical practitioner from “just the basics” to the needed design knowledge and implementation skills that will help them save time and money and maximize insight gained through efficient DOE. Motivated by humanity’s great need for a softer Lego brick that is less painful underfoot, we will review how to create some standard classical designs and then move into the Custom Design platform, pointing out connections and differences between the Classical and Custom Design platforms. We will talk briefly about optimality criteria, and then generate several potential designs out of the Classical and the Custom Designer. We will spend time in Compare Designs to think about how to choose appropriate effect sizes for power calculations, and we will provide guidelines for choosing a “best” design.
