Classical Test Theory

Origins and Development of Classical Test Theory

The initial proposals of what we now refer to under the generic term classical test theory (CTT) [Page 193]date from the beginning of the twentieth century. The beginnings of this approach were not particularly easy, as the quantitative orientation of psychology was at that time not the dominant paradigm. Nevertheless, it established itself little by little, and soon the majority of universities were including courses on test theory. As Joncich (1968) recounts in his biography of E.L. Thorndike, when the latter sent a copy of his pioneering work on measurement (Thorndike, 1904) to his maestro William James, he included a note advising him to oblige his students to read the book, but adding that under no circumstances should James himself even open it, as the figures, curves and formulas it contained would drive him mad. This anecdote serves to indicate the, at best, lukewarm reception to be expected from the psychological establishment for these psychometric issues that were taking their first steps. However, the period that followed was one of great activity and progress for psychometrics. New tests were constructed, psychometric technology was developed, and important advances were made in psychological and psychophysical scaling (Thurstone, 1927, 1928; Thurstone & Chave, 1929).

In 1936, Guilford would attempt to synthesize in his classic work Psychometric Methods all the basic developments up to that time in the fields of test theory, psychological scaling, and psychophysical scaling. These three fields share many concepts and models, and at that time it was still possible to treat them jointly, but the development and specialization of each of them has since made it necessary to deal with them separately, the latest edition of Guilford's 1954 book constituting an understandable exception.

At the same time as the test theory corpus of knowledge was becoming consolidated, the first steps were taken toward its institutionalization. The year 1936 saw the formation of the American Psychometric Society, with Thurstone as its president, and whose organ of expression would be the journal Psychometrika. Little by little, more and more journals specializing in psychological and educational measurement would appear, and today they are many. In 1947, Thurstone published his classic work Multiple Factor Analysis, presenting a multivariate technique with its origins in the psychometric field, and which has made an enormous contribution to the construction, analysis and validation of tests. From the publication of Thurstone's book until today, factor analysis has made gigantic strides, thanks to new methods of extraction and rotation of factors, and thanks above all to the power and speed of calculation afforded by modern computers; nevertheless, it is still gratifying today to re-read Thurstone's book and wonder at the wisdom and psychological substance with which it is imbued. Thurstone was without doubt one of the great pioneers and personalities of classical psychometrics. As my own maestro, Mariano Yela, who studied under him in Chicago in the mid-1940s, relates, Thurstone ‘was, above all, a creator…. He always remained the engineer-inventor that as a young man had worked with Edison. He was as clear and incisive as crystal, shy, hard, sarcastic, implacable. With me, he was understanding, tolerant and cordial. He was totally devoted to his specialty, psychology as a rational experimental and quantitative science, and to his photographic interests. Nothing else existed for him’ (Yela, 1996).

...