Construct validity theory holds that (a) validity is a property of test score interpretations in terms of constructs that (b) reflects the strength of the evidence for these interpretations. In this paper, we argue that this view has absurd consequences. For instance, following construct validity theory, test score interpretations that deny that anything is measured by a test may themselves have a high degree of construct validity. In addition, construct validity theory implies that now defunct test score interpretations, like those attached to phlogiston measures in the 17th century, ‘were valid’ at the time but ‘became invalid’ when the theory of phlogiston was refuted. We propose an alternative view that holds that (a) validity is a property of measurement instruments that (b) codes whether these instruments are sensitive to variation in a targeted attribute. This theory avoids the absurdities of construct validity theory, and is broadly consistent with the view, commonly held by working researchers and textbook writers but not construct validity theorists, that a test is valid if it measures what it should measure. Finally, we discuss some pressing problems in psychological measurement that are salient within our conceptualization, and argue that the time has come to face them.
Stored only when both papers are in the corpus — 130 citations exist upstream.