The following is the concluding paragraph from my 2900-word review published in: Limnology and Oceanography Bulletin 22(2): 62-65, 2013.
"This edition of SR, like earlier editions, does generally provide clear, readable, step-by-step instructions for carrying out the tests and procedures it recommends. Doubtless most of those instructions are correct. But also like earlier editions, this one provides few conceptual frameworks for understanding when and why, if at all, a particular procedure is needed or appropriate, it gives negligible attention to the basic concepts of sampling and experimental design, it does not recognize that design and statistical analysis are two distinct aspects of a study, it muddies the distinction between experiments and observational studies, it dedicates large amounts of space to matters of trivial importance, and it contains many more simple errors than are listed here. There truly is nothing to recommend SR as a course text. Experienced researchers, statisticians, and editors with a critical eye might benefit from having a copy at hand as a compendium of mostly accurate `recipes'. It will also help them understand one origin of the discombobulated statistical approaches found in many manuscripts that are crossing and will continue to cross their desks. After three further decades of heavy reliance on SR by biologists since Mead's review of the 2d edition, this book alone must account for a large fraction of the uncritical statistical thought and poor statistical advice found both in the primary literature of biology and as well as in other biologist-authored statistics books."