Treatment of Non-Detects Can Lead to Inaccurate Forensic Conclusions

Collecting samples from a contaminated site, characterizing background, and considering potential sources are integral parts of both traditional site assessments and environmental forensic investigations, yet the objectives of these investigations differ. Whereas traditional site assessments generally aim to characterize the nature and extent of contamination and assess the resultant risks, the goal of forensic investigations is to identify contributing sources of contamination, including influences from local or regional background. Typical methods for handling non-detect (ND) results for site characterization work are less appropriate for forensic investigations because their influence on the forensic analyses and final conclusions is unpredictable. Detection limits (consistency, values, and treatment of) are of particular importance in forensic studies where presence or absence of one contaminant may be the characteristic that distinguishes different sources. Although substitution methods for handling ND results may be appropriate in the context of risk assessment, this approach may obscure the true forensic characteristics of the underlying sources and measured sample compositions if applied in a forensic investigation, ultimately resulting in erroneous conclusions about sources. Through case studies, we demonstrate how various methods for handling detection limits can change the conclusions drawn from common forensic analyses. The effect of substituting numeric values for NDs on the results of common chemical forensics multivariate analyses is illustrated using simulated data and example soil samples from Colorado and Washington. The analyses discussed demonstrate a variety of impacts on multivariate analyses that stem only from changes in how ND results were included and illustrate the potential for inaccurate conclusions about the true forensic features within the data. Recommendations are provided for how to investigate and identify the impact of NDs on chemical forensic multivariate analyses.