September 2007, Vol. 19, No.9

Water Volumes

Using Statistical Methods for Water Quality Management: Issues, Problems and Solutions

Thomas P. O’Connor

Graham B. McBride (2005). John Wiley & Sons, 111 River Street, Hoboken, NJ 07030, 344 pp., $94.95, hardcover, ISBN 0-471-47016-3.

This book is readable, comprehensible, and I anticipate, usable. The author has an enthusiasm which comes out in the text. Statistics is presented as a living, breathing subject, still being debated, defined, and refined.

Examples address macroinvertebrate studies, pathogenic indicator organisms, and other receiving water quality issues. The author has a firm grasp on statistics and water quality, which provides the reader with greater confidence in the application to water quality analysis.

Advice on how to frame a study, what to expect from statistical analysis, and pointing out the need to clearly identify a hypothesis at the onset is also discussed. The author warns against the practice of “peeking” — essentially continuing to take data points until results are favorable. An understandable description of such common statistical terms as confidence limits, standard error, and probability distribution functions is provided. In one passage, the author points out that “failure to gain p< 0.05 in a point-null test, signifying ‘statistical’ insignificance, does not necessarily mean that population differences or trends are environmentally insignificant.”

The author’s most pertinent argument is that water quality professionals should use a Bayesian approach, rather than the classical (frequentist) approach, the predominant method taught. This claim is supported by examples and tables that show the power of the Bayesian methods over classical methods. What comes through to the reader is the restrictive nature of the frequentist interpretation. The statistical interpretation practitioners want is often Bayesian; however, achieving Bayesian interpretations requires setting the study up for a Bayesian approach in the first place.

The book has many references and an abundance of footnotes. The preface adequately describes the chapter content, directing the reader to read or possibly skip chapters. This is a textbook that provides chapter problems and solutions; it may not be appropriate for undergraduate study unless the curriculum is specifically in water quality. Water quality professionals may benefit from this book, as would regulators and those affected by regulations, given the argument for Bayesian statistics.

The author states, “Many software packages fail to advise the user just what formulae they use.” In this age of canned statistical software with limited descriptions, it is refreshing to see a book that presents statistics as a useful tool for professionals, instead of a burden of proof.
 

Thomas P. O’Connor is an environmental engineer in the U.S. Environmental Protection Agency (EPA) Urban Watershed Management Branch, Water Supply and Water Resources Division, National Risk Management Research Laboratory, Office of Research and Development (Edison, N.J.). Any opinions expressed in this review are those of the author and do not necessarily reflect the official positions and policies of EPA. Any mention of products or trade names does not constitute recommendation for use by EPA.

Thomas P. O’Connor is an environmental engineer in the U.S. Environmental Protection Agency (EPA) Urban Watershed Management Branch, Water Supply and Water Resources Division, National Risk Management Research Laboratory, Office of Research and Development (Edison, N.J.). Any opinions expressed in this review are those of the author and do not necessarily reflect the official positions and policies of EPA. Any mention of products or trade names does not constitute recommendation for use by EPA.