ABSTRACT
Ideally, you will have data that represent your endpoint entities and attributes and that are generated by studies in conditions that are representative of the environment being assessed. More often, you must estimate the quality or quantity of interest by extrapolating from available but substantially different data. Extrapolations are made between similar entities such as from a measurement of the sensitivity of a tested fish species to an untested fish species. Extrapolations extend the applicability of information. Estimating a property of an entity from a very dissimilar entity is not extrapolation because it is a change of type. For example, using a chemical’s structure to estimate the sensitivity of fish (i.e., a Quantitative Structure–Activity Relationship) is simply empirical modeling (Chapter 24). Extrapolation can be performed by analogy, application of a factor, or statistical analysis.
