PSYCtIOMETRIKA~VOL. 16, NO. 1 ~ARCH, 1951 THE DIMENSIONS OF TEMPERAMENT* L. L. THURSTONE THE UNIVERSITY OF CHICAGO The correlations among the thirteen personality scores yielded by the Guilford schedule for factors STDCR, and the Guilford-Mar- tin schedules for factors GAMIN, and O, Ag, and Co, as reported by Lovell, were factored by the centroid method. The purpose was to see how many factors were represented by the thirteen scores; therefore the test reliabilities were used in the diagonal cells. It was found that the scores represent not more than nine linearly independent factors. The orthogonal factor matrix was rotated to oblique simple structure. Seven of the oblique factors were given tentative interpretation. Two factors were regarded as residual factors because of the small variance which they represent. The seven factors have been named Active, Vigorous, Impulsive, Domi- nant, Stable, Sociable, and Reflective. The purpose of this study was to determine the number of factors or dimensions that are implied in current personality schedules, and also to ascertain the nature of each factor or type. The several sched- ules of Guilford were chosen for this purpose because they represent careful analytical work. Each of his schedules has previously been analyzed factorially, and correlations have been determined between the separate scores for his schedules. The various personality schedules cover a wide range of personal characteristics, including those which are relatively permanent for each person as well as those which change more or less from one year to the next because of social experience. Most of the scores derived from the Guilford schedules represent relatively permanent character- istics of a person which may be called temperamental trails. Some personality scores, such as appraisals of attitudes on controversial so- cial questions, represent only partly the temperamental characteristics of a person. Such scores also reflect his recent social experience, his social identifications, and the propaganda to which he may have been exposed. They are less stable as indicators of temperamental types. Our interest here is in those non-inteUective traits of personality which are relatively stable, the temperamental types, and which are *This study was supported in part by a research grant from Sears Roebuck and .Company. The writer wishes to acknowledge in particular the interest and assistance of Mr. J. C. Worthy of the National Personnel Department at Sears Roebuck and Company. The writer also wishes to acknowledge the assistance of Mr. James Degan who was responsible for the computing in this study. 1112 PSYCHOMETRIKA not often markedly changed in social experience. Hence we refer to this problem as the dimensions of temperament rather than the much larger domain that is called personality. Guilford has produced three personality schedules that were used in the present study. These were Guilford's schedule for the scores STDCR, the Guilford-Martin schedule for the scores GAMIN, and the Guilford-Martin schedule for the scores O, Ag, and Co.* Each of the first two schedules gives five scores, and the third schedule gives three scores. Hence the schedules give thirteen separate scores, all of which were used in the present study. The correlations among the thirteen scores were reported recently by Lovell who gave all three schedules to 213 subjects.t She made a factor analysis of the thirteen scores in which the communalities were determined by their intercorrelations. This is the usual procedure, but in the present case it should be recalled that the thirteen scores were themselves determined as factor scores from the original question- naires that contained many hundreds of items. Hence the procedure of Lovell was essentially to investigate the second-order domain in the thirteen factor scores. This is an interesting and important problem. The second-order domain in the traits of temperament may be psycho- logically revealing. But before undertaking such a study, it would be preferable to make sure that the factor scores which enter into a sec- ond-order analysis are linearly independent. Lovell questions the linear independence of the thirteen scores in her opening statement. She says: "The original studies showed that the thirteen factors were not completely inde~ende~t of each other though they were sufficiently separate to make individual scores helpful." Test scores may be very useful even though they are not linearly independent, but such a situ- ation introduces reservations about a second-order analysis. In the present study we direct ourselves first to the main problem, namely, to determine the number of dimensions or factors in these personality schedules which are represented by thirteen separate scores. This is the same problem that Lovell mentions in introducing her study. Instead of dealing with the thirteen scores as variables whose common factors are to be ascertained, we want to know how many factors are represented in the thirteen scores. Fer this purpose we make the factorial analysis with the test reliabilities in the diagon- *Guilford, J. P., and Guilford, R. B. Personality factors, S, E, and M, and their measurement. J. Psyehol., 1936, 2, 107-127: Personality factors, D, R, T. and A. J. abnor~n, soc. PsychoL, 1939, 34, 21-36; Personality factors N and GD. J. abnorm, sac. Psychol., 1939, 34, 239-248. tLovell,.Constance. A study of factor structure of thirteen personality vari- ables. Educ. psyvhol. Meas., 1945, 5, 335-3.~0.L. L. THURSTONE 13 al cells. If a second-order analysis is to be made of these thirteen scores, then the common factor variances, the communalities, are recorded in the diagonal ceils as was done in Lovell's paper. The thirteen scores from the Guilford Schedules are listed in Table 1. Each trait is shown by Guilford's name for the trait and by his code symbol. Then follow some items indicative of the presence of the trait (positive items) and some items that indicate the absence of the trait (negative items). Then follow further sample items from the schedules. Most of