Industrial Hygiene Exam 1 Practice

Non-detects are measurements that fall below the instrument’s limit of detection, so the true value is below a known threshold. Handling them as zeros or by excluding them biases the results: substituting zero underestimates exposure, substituting the maximum detectable value overestimates it, and excluding non-detects wastes information and can distort the distribution. The best approach is to acknowledge the left-censoring and use methods that reflect this uncertainty. Substituting a value like LOD/2 is a simple, common fix, but more robust options exist, such as methods based on maximum likelihood or regression on order statistics that explicitly model the censored data. These approaches keep estimates of the mean, variance, and other statistics more accurate by treating non-detects as values below the detection limit rather than as fixed zeros or fixed maxima.