ABSTRACT
Ever since Robert M. Yerkes tested a million World War I recruits with his Army Alpha Intelligence Test, multiple-choice items have dominated educational selection, placement, and assessment applications. Occasional criticism has marked their reign, from observers including no less than Banesh Hoffman, Ralph Nader, and (Educational Testing Service's own!) Norman Frederiksen. But the character of today's debates strikes at the very heart of the enterprise: The view of human ability that spawned multiple-choice tests no longer holds universal currency among psychologists and educators. The ascendant view originates from a perspective more attuned to instruction than to selection or prediction. Learners increase their competence not simply by accumulating new facts and skills at rates determined by relatively immutable “aptitudes,” but by reconfiguring knowledge structures, by automating procedures and chunking information to reduce memory loads, and by developing strategies and models that tell them when and how facts and skills are relevant.
