Practical Implications of Design-Based Sampling Inference for Thematic Map Accuracy Assessment

The conceptual basis of design-based inference is described, followed by discussion of practical implications of the framework, including the population to which the inferences apply and the role of probability sampling.

Abstract

Sampling inference is the process of generalizing from sample data to make statements or draw conclusions about a population. Design-based inference is the inferential framework commonly invoked when sampling techniques are used in thematic map accuracy assessment. The conceptual basis of design-based inference is described, followed by discussion of practical implications of design-based inference, including (1) the population to which the inferences apply, (2) estimation formulas and their justification, (3) interpretation of accuracy measures, (4) representation of variability, (5) effect of spatial correlation, and (6) role of probability sampling. Design-based inference is contrasted with model-based inference, another inferential framework frequently invoked in statistics.