Thurstone, L. L.
Thurstone, L. L.
Louis Leon Thurstone (1887–1955), American psychometrician and psychologist, was born in Chicago. Both of his parents had been born in Sweden, and for a period Thurstone himself attended school in Stockholm. At Cornell University, which he entered in 1908, he first studied civil engineering but changed to electrical engineering. Even before graduating from college he patented a model motion–picture projector; it attracted the attention of Thomas A. Edison, who offered him an assistantship in his laboratory in East Orange, New Jersey. However, Thurstone did not remain long in Edison’s laboratory, leaving in the fall of 1912 to teach descriptive geometry and drafting in the College of Engineering at the University of Minnesota.
Thurstone’s interest shifted to the experimental study of learning, and in the summer of 1914 he enrolled for graduate study in psychology at the University of Chicago. While still a graduate student he accepted an assistantship in Walter V. Bingham’s newly established division of applied psychology at the Carnegie Institute of Technology. Thurstone continued at Carnegie after receiving his doctorate from Chicago, eventually becoming professor and chairman of the department of psychology.
Early in 1923 Thurstone left Carnegie for Washington, to work for the foundation–supported Institute for Government Research and specifically to prepare instructional manuals and specimen materials that would assist civil–service agencies in using objective examinations. With the assistance of Thelma Gwinn, who became his wife in 1924, he prepared a psychological test for selecting or classifying college students; he and his wife were responsible for 24 successive annual editions of the American Council on Education Psychological Examinations.
Thurstone returned to the University of Chicago in 1924, as associate professor of psychology, and taught courses in descriptive statistics and mentaltest theory. Three years later he became a full professor, and in 1938 he was named Charles F. Grey distinguished service professor. He established the Psychometric Laboratory at Chicago, and after his retirement in 1952 he re–established his laboratory at the University of North Carolina, where he went as a research professor of psychology and as director of this laboratory.
Although Thurstone credited others with the founding of the Psychometric Society and of its journal, Psychometrika, he was close to the nucleus of 10 or 12 persons who brought these into being in 1936 in order to foster the development of psychology as a quantitative rational science.
Only a few of the many honors that Thurstone received can be mentioned here. He served as president of the American Psychological Association in 1932 and was the first president of the Psychometric Society (in 1936). He was elected honorary fellow of the British Psychological Society, the Spanish Psychological Society, and the Swedish Psychological Society. The University of Goteborg awarded him an honorary doctorate in 1954.
Contributions to psychology
Throughout his career in psychology, Thurstone was primarily concerned with the problem of measurement. He sought to establish valid principles of measurement in such areas as mental abilities, intelligence, attitudes, social judgment, psychophysics, and personality.
Intelligence. Thurstone advanced the thesis that “Intelligence, considered as a mental trait, is the capacity to make impulses focal at their early, unfinished stage of formation" (1924, p. 159); that is, intelligence is “the capacity to live a trial-and–error existence with alternatives that are as yet only incomplete conduct” (ibid., p. xv). In his early approaches, he regarded intelligence as an ability to abstract that involves the inhibition of immediate impulsive behavior and that opens up a number of alternate paths toward impulse satisfaction (need reduction or goal attainment). When abstraction does not occur, when awareness of impulse–driven behavior is limited, the organism is more likely to engage in behavior that is inappropriate to real impulse satisfaction (ibid., especially pp. 123–169).
The area of mental abilities continued to absorb Thurstone. He took issue with the mental–age concept that had become so important to psychologists and educators. In 1926 he pointed out incisively that two different definitions of mental age were being used: one that infers mental age from the chronological age for which a specified test performance is average; and a second that infers mental age from the average chronological age of persons performing at a specific level. Thurstone demonstrated the inadequacy of both definitions for dealing with adult intelligence and suggested that test performance of children as well as of adults should be represented in terms of either percentile ranks or standard scores.
Thurstone’s later work on factor analysis (discussed below) bears directly on the study of mental abilities. In the late 1920s he began an analysis of the correlations between various types of aptitude tests. This area had already been investigated by both C. E. Spearman and E. L. Thorndike, who had come to quite different conclusions. Spearman claimed, as a result of his method of factor analysis, that there is a general factor “g” that characterizes all mental functioning, even though there are many additional specific, unique abilities (Spearman 1904; 1927). Thorndike took issue with Spearman’s position and argued that intelligence is composed of a large number of separate factors or elements and that there is no general intelligence.
Thurstone, using improved statistical techniques, arrived at a position somewhere between these two extremes. He claimed that performance on tests of cognitive abilities results from several factors rather than from one common general factor. After making several factor analyses, he identified a number of factors that he referred to as “primary mental abilities": spatial visualization, perceptual ability, verbal comprehension, numerical ability, memory, word fluency, and reasoning. He saw that some of these could be separated into two or more factors, depending upon the number of tests of the original factors introduced in new analyses and the nature of the specific variations incorporated in the tests. Thus when a number of tests of, for example, the spatial factor, first identified as a primary factor, are included in an analysis, additional or partitioned factors in the realm of space may be identified, as Thurstone was well aware. Noting that the primary factors are positively correlated, Thurstone suggested that a factor analysis of these correlations might reveal a “second–order factor,” similar to Spearman’s “g” (1948).
Measurement theory and psychophysics. Various psychological scales then in use seemed to Thurstone to imply that distributions of scores for various age groups differ only with respect to the mean, and his first paper on measurement theory presented a scaling method that permits both the mean and the dispersion to vary (1925a).
Thurstone approached the area of measurement vigorously. In one year, 1927, he published several articles dealing with various problems of subjective measurement, introducing the concept of discriminal dispersion and the law of comparative judgment, and relating the law of comparative judgment to the classical psychophysics of Weber and Fechner (see 1927; Measurement of Values, pp. 19–81). He also issued a lithoprinted work, “The Reliability and Validity of Tests” (1931a).
Further papers, first published between 1928 and 1932, dealt with the inconsistency of the phigamma hypothesis with Weber’s law (Measurement of Values, pp. 82–91), the limitations of the method of equal–appearing intervals (ibid., pp. 92–99), the method of rank order as a substitute for paired comparisons (ibid., pp. 100–111), and the numerical evaluation of the dispersions of stimuli presented by the constant method (ibid., pp. 112–122).
In 1931 Thurstone wrote an article that was based upon the postulate that motivation toward accumulation of a commodity is inversely proportional to the amount already possessed (ibid., pp. 123–144). At the University of North Carolina, where he went in 1952, he worked with Lyle V. Jones, with whom he had collaborated earlier, on the experimental determination of the zero point on a scale of utility; they demonstrated that subjective values are additive (ibid., pp. 195–210). In a 1945 paper, Thurstone had shown that the dispersion of affective values, as well as the average affective value of a proposal, is significant in the measurement of the social attitudes of a group (ibid., pp. 145–160).
Even in the last years of his life Thurstone continued to investigate approaches to the measurement of subjective attributes. He developed a new scaling method that avoided the assumption of the normality of the subjective distribution for each stimulus. He also explored the less restrictive assumption that repeated judgments by the same individual will be normally distributed on the subjective continuum. A 1954 paper, “The Measurement of Values,” reviewed fundamental concepts of subjective measurement (ibid., pp. 182–194).
The measurement of attitudes. Perhaps the most popular application of Thurstone’s work in subjective measurement is his contribution to the scaling of social attitudes. Although he recognized the inadequacy of methods of attitude measurement current in the 1920s, he held that, in principle, attitudes were subject to measurement. Thurstone had earlier, in 1928, advanced the notion of equal–appearing intervals for subjective measurement, and later he extended it to development of an attitude scale. “The scale is so constructed that two opinions separated by a unit distance on the base line seem to differ as much in the attitude variable involved as any other two opinions …also separated by a unit distance” (Thurstone & Chave 1929, pp. xi–xii).
Thurstone and Chave presented a series of 130 opinions about the church to 300 “judges” who were asked to arrange them in 11 piles ranging from extremely unfavorable to extremely favorable. An analysis of these judges’ arrangements produced a final set of 45 items that had been rated relatively unambiguously and consistently and, most important, that represented a more or less uniformly graduated series of scale values. Subsequently, individual responses to this final opinion scale could be converted into meaningful scores and could be treated mathematically. The scale had internal reliability. Furthermore, on the basis of its ability to discriminate particular groups whose attitudes toward the church were assumed to be different, its validity was demonstrated.
The uniqueness and importance of Thurstone’s contribution to attitude scaling are still appreciated, as others continue to extend psychophysical methods to social phenomena that lack the simple scalable dimensions of physical stimuli.
Multiple–factor analysis. Despite widespread interest in the application of attitude–measurement techniques to all sorts of issues and groups, Thurstone abandoned this field in the early 1930s to work on the development of multiple–factor analysis. Although later psychologists may place more emphasis upon Thurstone’s contributions to psychophysics, his contemporaries probably paid more attention to his work in factor analysis.
Thurstone was impatient with the debate on the merit of Spearman’s single–factor method, the universality of a general factor, and the role of group factors. Instead of asking whether correlation coefficients support a general factor, he wondered how many factors must be postulated in order to account for observed correlations. The power of this approach lay in its ability to establish in the case of any particular study whether or not one factor should be regarded as general.
Thurstone’s excursions into multiple–factor analysis also embraced the concepts of communalities, rotation of the reference frame, oblique reference axes, and factorial invariance. The concept of simple structure—his solution to the problem presented by the infinite number of positions of reference axes—may well be his most noteworthy contribution to factor analysis. He was also concerned with second–order factors and with the effects of selection upon factorial structure.
Multiple–factor analysis was the subject of Thurstone’s 1933 presidential address to the American Psychological Association, “The Vectors of Mind” (1934); he also wrote two major books on the subject (1935; 1947). He continued his work on factor analysis after he moved to North Carolina, and at about the same time that others began to work on this problem, he devised an analytical method of rotating the reference axes (1954). He also developed a new method of factor analysis designed to avoid the communality problem (1955).
At the same time that he was refining the methodology of factor analysis, he was making important applications of the approach. His first large study entailed 57 tests of cognitive functions, administered to 240 subjects (1938). This research was followed by other studies, several of which were completed in his laboratory as student dissertations. An experimental battery of tests of primary mental abilities for use in schools was made available in 1938. In these and many other efforts, Thurstone was fortunate in having the collaboration of his wife.
The study of personality. A recurrent interest of Thurstone’s was the elusive realm of personality. He and his wife at one time developed a personality schedule patterned after Woodworth’s questionnaire (1930). Returning to this earlier interest after World War n, he identified the psychological hypotheses implicit in such tests as the Rorschach and assembled more than sixty tests representing hypotheses concerning the manifestation of personality traits in performance on objective tests. A fairly short temperament schedule for use with normal persons resulted from factorial studies of personality questionnaires (1951). Although he himself did not pursue work in this field to its limits, Thurstone regarded objective laboratory tests of temperament as one of the most challenging areas of research.
Thurstone not only contributed abundantly to the development of psychology but also exhibited a rare ability to capture the imagination of university colleagues and administrators, students, military leaders, industrialists, and the representatives of foundations. He had infinite skill in imparting ideas to others and inspiring them, too, to creative accomplishment.
Dorothy C. Adkins
[For the historical context of Thurstone’s work, see the biographies of Spearman; Thorndike;for discussion of the subsequent development of Thurstone’s ideas, see Aptitude Testing; Factor Analysis; Intelligence and intelligence testing; Psychometric; Psychophysics; Scaling; and the biography of Kelley.]
WORKS BY THURSTONE
1924 The Nature of Intelligence. London: Routledge; New York: Harcourt. → A paperback edition was published in 1960 by Littlefield.
1925a A Method of Scaling Psychological and Educational Tests. Journal of Educational Psychology 16: 433–451.
(1925b) 1930 The Fundamentals of Statistics. New York: Macmillan.
1926 The Mental Age Concept. Psychological Review 33: 268–278.
1927 The Unit of Measurement in Educational Scales. Journal of Educational Psychology 18:505–524.
1928 The Absolute Zero in Intelligence Measurement. Psychological Review 35:175–197.
(1929) 1937 Thurstone, L. L.; and Chave, E. J. The Measurement of Attitude: A Psychophysical Method and Some Experiments With a Scale for Measuring Attitude Toward the Church. Univ. of Chicago Press.
1930 Thurstone, L. L.; and Thurstone, T. G. A Neurotic Inventory. Journal of Social Psychology 1:3–30.
1931a The Reliability and Validity of Tests. Unpublished manuscript, Ann Arbor, Mich.
1931b Multiple Factor Analysis. Psychological Review 38: 406–427.
1931c A Multiple Factor Study of Vocational Interests. Personnel Journal 10:198–205.
1934 The Vectors of Mind. Psychological Review 41:1–32.
1935 The Vectors of Mind: Multiple–factor Analysis for the Isolation of Primary Traits. Univ. of Chicago Press.
1938 Primary Mental Abilities. Psychomatic Monographs, No. 1. Univ. of Chicago Press.
1947 Multiple–factor Analysis: A Development and Expansion of The Vectors of Mind. Univ. of Chicago Press.
1948 Psychological Implications of Factor Analysis. American Psychologist 3:402–408.
1951 The Dimensions of Temperament. Psychometrika 16:11–20.
1954 An Analytical Method for Simple Structure. Psychometrika 19:173–182.
1955 A Method of Factoring Without Communahties. Pages 59–62 in Invitational Conference on Testing Problems, 1954, Proceedings. Princeton, N.J.: Educational Testing Service.
The Measurement of Values. Univ. of Chicago Press, 1959. →> All the papers but one were first published between 1927 and 1955; the remaining one was completed by Lyle V. Jones and first published in this volume.
SUPPLEMENTARY BIBLIOGRAPHY
Adkins, Dorothy C. 1964 Louis Leon Thurstone: Creative Thinker, Dedicated Teacher, Eminent Psychologist. Pages 1–39 in Norman Frederiksen and Harold Gulliksen (editors), Contributions to Mathematical Psychology. New York: Holt.
Spearman, C. E. 1904 “General Intelligence” Objectively Determined and Measured. American Journal of Psychology 15:201–293.
Spearman, C. E. 1927 The Abilities of Man: Their Nature and Measurement. London: Macmillan.