Context:
An industrial hygienist from a consulting firm submitted a question regarding a puzzling result from a recent exposure assessment. The dataset included 15 concentration measurements of a chemical substance with an occupational exposure limit (OEL) of 50 ppm. Most values ranged between 0.7 and 2.0 ppm, but one reading stood out dramatically at 340 ppm—more than six times the OEL.
Challenge:
When analyzed using Expostats, the results indicated a probability of overexposure of 8.74% : 8.74% chance that >5% of exposure levels exceed the OEL. This would make the situation labelled as “tolerable” according to the decision scheme proposed in the AIHA video series about exposure decisions. The hygienist was confused, as the high single value in his dataset seemed inconsistent with that conclusion.
Resolution:
A member of the Expostats development team reviewed the case and noted that the geometric standard deviation (GSD) exceeded 5 suggesting extremely high variability. Upon further inspection, it became evident that the elevated GSD was driven entirely by the 340 ppm outlier. Once this value was removed, the GSD dropped to 1.5, indicating a much more, and even highly homogeneous exposure profile.
The team hypothesized that the 340 ppm measurement might reflect an unusual event or context—possibly not representative of typical worker exposure (and as a consequence not coming from the same distribution). The hygienist confirmed that this value had been recorded during a special monitoring activity on equipment rather than during routine personnel sampling. As a result, the value was excluded from the exposure assessment.
Outcome and Insight:
This case illustrates the critical role of professional judgment when interpreting statistical outputs. While Expostats correctly modeled the dataset based on the input values, identifying the outlier as contextually irrelevant was essential to producing a meaningful risk characterization. This example underscores how Expostats can support decision-making but must be paired with contextual expertise for accurate interpretation.