Chapter 3Laboratory Detection Limits, Nondetects, and Data Analysis
3.1 Introduction and Overview
Contaminant sampling data from ambient air, soil, ground or surface water, or other environmental media often include data values that are not directly usable for statistical analysis. This usually arises when the levels or concentrations of the contaminants in some sampling locations or during some sampling periods are too low to be reliably measured and reported by the analytical laboratories. In these cases, the data values are described as nondetects (NDs) and usually reported as “less than” the laboratory detection limit (DL), quantitation limit (QL), reporting limit (RL), or other specified threshold, indicating to the user that the concentrations are not necessarily zero but just not sufficiently high to be quantified. NDs are also referred to as left-censored data in the sense that we do not know the actual value, but we know that it lies to the left of the detection or reporting limit and is somewhere between zero and this limit. In some cases, the contaminant may be detected but due to lack of confidence in the actual amount present in the medium, it is reported as an “estimated” value.
There usually is a natural inclination to completely exclude these types of data or regard them as essentially equal to zero, so that the statistical analysis can proceed. It turns out that for some highly toxic contaminants, the health-based target concentrations (i.e., the concentrations ...
Get Statistical Applications for Environmental Analysis and Risk Assessment now with the O’Reilly learning platform.
O’Reilly members experience books, live events, courses curated by job role, and more from O’Reilly and nearly 200 top publishers.