Psychometric Tests as Cognitive Tasks: A New "Structure of Intellect"

Abstract

From its beginnings, psychometrics has had a split personality. On the one hand,\nit has been concerned with practical means of measurement and prediction,\nincluding not only the construction of instruments but also the mathematical\nand statistical bases for obtaining reliable and valid measurements-or what is\ncommonly called "test theory." On the other hand, the very notion of validityparticularly the notion of "construct validity" (Gulliksen, 1950)-implies that\none be at least somewhat bothered by the problem of what a test measures.\nTests of "intelligence" have always been the most prominent type of psychometric instrument. However great their interest in practical matters, all the\nleading figures in psychometrics-Binet, Spearman, Thurstone, and Guilford (to\nname but a few)-have had an abiding concern for the nature of intelligence; all\nof them have realized that to construct a theory of intelligence is to construct a\ntheory of cognition. It is not without significance that one of Spearman&s;s (1924)\nmajor works bore the title The Nature of Intelligence and the Principles of\nCognition. The same theme was carried by the titles of books by Thurstone\n(1924) and Guilford (1967).