ABSTRACT
Personal Reections on the Origins of the Model When item response theory (IRT) originated and was developed in psychology and sociology in the 1940s, 1950s, and the first half of the 1960s, the theory only dealt with dichotomous responses, where there are only two item score categories, for example, correct and incorrect in ability measurement, true and false in personality measurement. As a graduate student I was very much impressed by Fred Lord’s (1952) Psychometric monograph “A Theory of Mental Test Scores” and could foresee great potential for latent trait models. It seemed that the first thing to be done was to expand IRT to enable it to deal with ordered, multicategory responses and enhance its applicability, not only in psychology, sociology, and education, but also in many other social and natural science areas. That opportunity came when I was invited to spend one year as visiting research psychologist in the Psychometric Research Group of the Educational Testing Service (ETS), Princeton, New Jersey, in 1966. The essential outcomes of the research conducted during my first year in the United States were published in Samejima (1969).
