STATISTICAL PROCEDURES FOR ADDRESSING NONDETECT RESULTS IN ENVIRONMENTAL-ANALYSES

Censor data sets, those containing observations that are less than the analytical chemistry limit of detection (LOD) value, are becoming mor e common as regulatory agencies and dischargers strive to reduce conce ntrations of trace organic contaminants in treated effluent. The recen tly reissued ''Technical Support Document For Water Quality Based Toxi cs Control,'' a frequently used reference during NPDES permit developm ent, states that the mean and variance of censored data sets may be de termined by assuming the data conform to what EPA has termed the ''del ta log-normal distribution'' (D-LOG). The USEPA D-LOG statistical proc edure along with two commonly used alternatives (Cohen's MLE and regre ssion of normal order scores) are described, illustrated with example calculations and evaluated using Monte Carlo simulation experiments. U sing minimal absolute bias as the evaluation criteria, Cohen's method [Technometrics 1:217 (1959)] had superior predictive capability overal l, particularly for real world data sets. For mean and standard deviat ion estimators, Cohen's method was at least 2.5 and 1.2 times less bia sed than those of the EPA recommended methodology, respectively.