This short handbook is a practical and accessible guide to the statistical design and analysis of 2-level, multi-factor experiments of the kind widely used in industry and business. Written for technologists and researchers, it forgoes the usual heavy statistical overlay of typical texts on this subject by focusing on a limited catalog of standard designs that are useful for commonly encountered problems. These design choices are based on relatively recent developments in design projectivity, and their analysis requires nothing more than simple plots of the data: neither special expertise nor complex software is needed. Numerous examples show how to carry out this program in practice.
Even though the statistical content of the handbook has been deliberately limited, it nevertheless discusses several practical matters that are rarely included in more comprehensive treatments, but which are vital for experimental success. Among these are the realities of randomization versus split-plotting, the importance of identifying the experimental unit, and a discussion of replication that argues that it is generally not worth the effort. Readers with some prior statistical exposure — and statisticians — may also be surprised to find that p-values do not appear anywhere in the book, and that in fact the authors explicitly argue against their use.
Those new to the ideas of Statistical Design of Experiments (DOE)– or even those who have some familiarity but would like greater insight and simplicity — should find this handbook an effective way to learn about and apply this powerful technology in their own work.