Truth and Consequences in Validation: An Ecological Theory of Validity

• In Event: Validity in Systems of Data Use

Abstract

The presenter argues that when Borsboom et al. (2004) argue for a causation based model of validity, they unintentionally call for a greater importance of social consequences.Taking an ecological view of testing and society, the presenter argues that well researched phenomena such as self-fulfilling prophecy, stereotype threat, teacher expectation, etc., reveal that test scores also cause the very attributes that cause test-score variation. Thus, the presenter argues that, following Borsboom et al.'s argument for a causation based model of validity, the evidence for a testing procedure's validity must actually be expanded, not narrowed; to the extent that fairness and bias, e.g., play a causal role in the existence of a test-taker's attribute of interest, they become vital validation evidence.

Author

• Josh Lederman, Wellesley College