STATISTICS FOR CENSORED ENVIRONMENTAL DATA USING MINITAB~® AND R, 2ND EDITION

摘要

How many of us have pondered what to do with nondetect (i.e., left-censored) data in our environmental data sets? For years, a default practice adopted by many scientists was to substitute zero, one-half the reporting limit, or the full reporting limit for nondetect values. This was a quick fix that allowed us to blithely move on to statistical evaluations, risk assessments, and/or modeling of our data sets. However, this quick fix is rife with problems that introduce bias in analyses and could lead environmental managers down the wrong path for resolving contamination problems. In Dennis Helsel's 2nd edition of Statistics for Censored Environmental Data Using Minitab~® and R, he provides a straightforward description of the statistical tools one can use (without substitution) to properly analyze environmental data sets containing nondetect data with 1 or more detection limits.