**Book Review**

*Statistical Methods for Groundwater Monitoring (Statistics in Practice)*, edited by Robert Gibbons, Dulal Bhaumik, and Subhash Aryal (2009), is published by John Wiley & Sons (ISBN 978-0-470-16496-9).

__Laura L. Sanders, Book Editor__

If you are a groundwater scientist working with data from chemical analyses of groundwater samples, you need a copy of

*Statistical Methods for Groundwater Monitoring, 2nd Edition*, on your bookshelf. This second edition is a significant update of the first. The last 15 years have brought a vastly different statistical computing landscape, even if the regulatory situation has not changed much. This edition includes substantial new material, focused as much on the environmental scientist as the statistician. Whatever you are doing or will be doing to analyze your groundwater data, you will be able to connect those processes directly to one of the chapters in this book.

The book is an instruction manual on how to use data to validate environmental monitoring programs. It is not necessarily the one you will want to read cover-to-cover if you do not have more than an introductory background in statistics. Authors Robert Gibbons, Dulal Bhaumik, and Subhash Aryal comment that the statistical theory in this book is “included for completeness,” but much of it is beyond the reach of a typical nonstatistician.

It may be most helpful for groundwater scientists to begin with Chapter 19, “Regulatory Practices.” This chapter explains how the book originated in response to regulatory requirements and statistical issues with groundwater data. After that, read the Preface for additional valuable information on how to use the book. Next, peruse the overview of each chapter and perhaps subsequent sections that do not delve deeply into statistical equations. Finally, read the brief summary chapter.

This second edition contains five new chapters. Chapter 4, “Gamma Prediction Intervals,” offers the gamma distribution as an alternative skewed distribution to the usual lognormal choice, a nice use of the type of statistical computing power readily available in the 21st century, where basic normality assumptions are less important. Chapter 8, “Inter-laboratory Calibration,” reaches out into another aspect of many monitoring programs, the use of multiple laboratories across the same or related sites for the requisite chemical analyzes.

This adds another source of variability to the location, sampling, and testing variability that already need to be managed. Chapter 13, on normal prediction limits for left-censored data (nondetects), supports the preceding chapter on censored data and will be most accessible to statisticians. Chapters 17 and 18, “Surface Water Analysis” and “Assessment and Corrective Action Monitoring,” are specifically written for environmental scientists and engineers.

As I am a statistician with more than 40 years of experience in the oil and gas industry, you may think it self-serving, but unless you have an uncommonly thorough statistical background I would advise that you not to attempt actually to use the statistical technology described in the book without support from a statistician.

Much of the angst in working with groundwater data comes from two aspects: the need to worry about establishing detection limits and the necessity of working with analytical results that are below those limits. If these are the problems you are encountering, work with your statistician, utilize the methods in this book with appropriate computing support, and provide a complete explanation for the suitability of the methods in reports to your state department of natural resources or USEPA.

There is one drawback in this book: it includes no statistical computing. It discusses DUMPStat, a software package the authors wrote in the 1990s to support the first edition, and notes that there is an alternative, GRITS/STAT. Even with the availability of a large programmable system such as SAS or R, a statistician will need good programming skills to use the concepts in this book, as these groundwater monitoring tools are all nonstandard statistical applications. The authors would no doubt note that space did not permit providing this type of guidance. I would have dropped some of the theory to make room for it and used references for the interested advanced audience.

This edition incorporates the work of two new authors, which may promote the ongoing development of a third edition, hopefully one with more attention to statistical computing.

Source: Ground Water Volume 49, Issue 2, page 128, March/April 2011.

**Book Review:**Statistical Methods for Groundwater Monitoring (Statistics in Practice).