A psychological construct, such as personality, is an abstraction that is not directly seen but inferred through observed regularities in cognitive, affective, and behavioral responses in various settings. Two assumptions give meaning to the idea of construct validity. First, constructs represent real phenomena that exist apart from the potential ways in which they are measured. Second, constructs have a causal relation to their measures (see D. Boorsboom, G.J. Mellenbergh, & J. van Heerden, 2004). According to these twin assumptions, variation in a construct such as personality or intelligence "causes" individual differences in responses to items on measures; it also accounts for performance in real life settings. An alternative perspective equates a construct with the operation used for its measurement (M. Friedman, 1991) without a presumption of causality. This article elaborates on the implications of different assumptions about measurement, "operation-referenced" and "construct-referenced," for advancing the science and practice of psychology. (Contains 1 table and 1 figure.)