Nondetects and Data Analysis: Statistics for Censored Environmental Data (Statistics in Practice)

Rating:
Summary: Sensible advice for analyzing non-detects with environ. data
Review: Anyone involved in the analysis of environmental data will benefit from this easy-to-follow book. Measurements of environmental data often contain "less than" values below the laboratory limit of detection. This book provides a sensible approach for analyzing this type of information and, importantly, informs the reader of the inadvertent bias that can result from some of the commonly used substitution methods (e.g, replacing non-detects with one-half the detection limit). In addition to his own original research, Dr. Helsel extensively surveyed the scientific literature in this area and provides enough detail to properly support the recommended procedures. This book covers the handling of non-detects in many areas, including univariate data analyses, time series, interval estimates, comparing two or more groups, correlation and regression. Another useful feature of this book are the MiniTab Statistical Software macros available from the practicalstats.com website. These macros allow non-statisticians to readily perform the described analyses without creating unique computer programs or complicated spreadsheets.

As an environmental scientist with a western state regulatory agency, I have followed the statistical and environmental journal literature in this area for many years. I recently incorporated many of the concepts from this book into regulatory requirements for coastal water quality standards.


Rating:
Summary: Fills Void - Proper Handling of Censored Environmental data
Review: The concentration of a chemical compound may be so low that it cannot be quantified by an analytical method. The result from a chemical analysis attempting to quantify such an analyte concentration is generally reported as a nondetect (ND), rather than as zero or not present, and the appropriate limit of detection is usually provided. In cases where measurement data are described as not detected, the concentration of the chemical is unknown, though its possible concentration ranges from zero to the sample-specific detection limit. Analytical testing results that are reported as nondetects (NDs) are encountered frequently in environmental work. The proper handling of such ND results has been discussed within the environmental community, including the United States Environmental Protection Agency (USEPA) for many years.
The most common method of handling NDs is to substitute ½ the detection limit. This book shows that this method can lead to incorrect conclusions. Proper statistical approaches have been used in other scientific disciplines; these approaches required computing resources that were not readily available. Now, however, statistics software that can make the required calculations is available on personal computers.
This book fills a huge gap by providing proper statistical methods from the fields of medical and industrial research to the analysis of censored environmental data. Included are examples that can be computed using Minitab, a commercially-available statistics software package.
I have taken several courses taught by Dr. Helsel and have read and used the book. The book is well organized and provides many useful and informative examples. I have found the information to be extremely helpful in my work as an in-house environmental consultant working for a large manufacturing company.