Assessment of Variability and Uncertainty Distributions for Practical Risk Analyses

TLDR

The EPA is moving beyond single‑point risk assessment to consider distribution‑based analyses, distinguishing inter‑individual variability from uncertainty, which can be reduced with better data. The study evaluates when distribution‑based analyses are warranted for baseline Superfund site assessments and illustrates the distinction between variability and uncertainty. The authors review distribution‑based risk assessment methods and present case studies on variability/heterogeneity and uncertainty analyses.

Abstract

In recent years the U.S. Environmental Protection Agency has been challenged both externally and internally to move beyond its traditional conservative single‐point treatment of various input parameters in risk assessments. In the first section, we assess when more involved distribution‐based analyses might be indicated for such common types of risk assessment applications as baseline assessments of Superfund sites. Then in two subsequent sections, we give an overview with some case studies of technical analyses of (A) variability/heterogeneity and (B) uncertainty. By “inter‐individual variability” is meant the real variation among individuals in exposure‐producing behavior, in exposures, or some other parameter (such as differences among individual municipal solid waste incinerators in emissions). In contrast, “uncertainty” is a description of the imperfection in knowledge of the true value of a particular parameter or its real variability in an individual or a group. In general uncertainty is reducible by additional information‐gathering or analysis activities (better data, better models), whereas real variability will not change (although it may be more accurately known) as a result of better or more extensive measurements. The purpose of the rather long‐winded exposition of these two final sections is to show the differences between analyses of these two different things, both of which are described using the language of probability distributions.