Practical Experiment Designs: for Engineers and Scientists

This book is a great introduction to ballanced orthogonal experimental designs. The so-called Scientific Method (tweak one parameter while holding the other variables fixed) is not used that much where there are complicated interactions between variables. This Designed Experimental (DOX) approach is used more frequently in the Semiconductor industry. This approach envolves a series of experiments where all the variables are potentially changed form one trial to another, but in such a way that one can resolved the effects of one variable from another AND understand interactions and non-linear effects. It is simply the best way to optimize a complicated system.
OK, if you have SAS or RS/1 or …packages you may ask yourself “why bother learning the theory.” I would say for 2 reasons – if you need to manually intervene or supplement an existing experiment AND seeing some more designs that are very effecient (John’s 3/4 designs and Plackett-Burman designs) for initial screening that is not ususally part of the packages.
Compared to Box, Box, Hunter this book is easier to read, has a huge summary fold out sheet and is geared to a practicing engineer or scientist not a professional mathematician. Best introduction and most likely the only book you will ever need on the subject.