内田クレペリン作業曲線の因子分析的研究

抄録

The present study was so designed as to verify the factorial validity of the “work curve” of the Uchida-Kraepelin Psychodiagnostic Test (U-K Test) chiefly comparing with Tsujioka's previous studies (3, 8). In 1952 Tsujioka factorized the work curves of this test (Tables 1 and 2, Fig. 1) by means of a sort of T-technique for the purpose of verifying Kraepelin's hypothetical work-factors such a Uebung, Gewoehnung, Ermuedung, Antrieb, and Anregung. Tsujioka's biased sampling of subjects, however, would seem not to be adequate to the experimental purpose, and it may have need of clarifying the factorial meaning of his percentization of performance scores by which he aimed at neglecting the effect of the intelligence in the scores. The sampling in the present study was cared to be unbiased, and for the raw scores the performance scores were directly used.
In“Analysis I”the whole battery of 25 rows of the test was factorized by Thurstone's complete centroid method. The result (Table 4, Fig. 2) was compared with Tsujioka's (Table 2, Fig. 1). The main finding was that, in spite of difference in sampling and scoring, both centrold factor structures were similar except the first centroid factor of‘Analysis I’which seemed to be identified with the factor eliminated by Tsujioka's percentization of scores.
Soma (3, 4) investigated the statistic, practical validity in the second half of this U-K Test. Its factorial validity was examined here in‘Analysis II’where the partial matrix of the second half in Table 3 was factorized into the centroid factor matrix (Table 5). The comparison of this result with that of‘Analysis I’suggests that the total variance of the first centroid factor in‘Analysis II’(Table 5, Fig. 3) includes that of the second centroid factor in“Analysis I”(Table 4, Fig. 2).
Tsujioka's interpretation was based upon the centroid factors (Fig. 1). But these centroid factors would be desirable to be properly rotated. In“Analysis III”some reference-tests (Table 6) were added to the partial battery of the second half in order to make them the pilot of the rotation and interpretation of factors, though the addition of these reference-tests would cause the condition of Thurston's Experimental Dependence (6). The results were as follows: 1) Four significant factors were extracted (Table 8). 2) Rotated primary oblique factors and their intercorrelations are given in Tables 10 and 11 respectively. 3) Second-order general factor and orthogonalized primary factors are shown in Table 13, and their profiles in Fig. 4. 4) Each factor was tentatively named as Numerical ability, Mental fatigue, Volitional-tension, Excitation, and Second-order mental energy level (cf. Fig. 4). Tsujioka's percentization would have omitted the factor of numerical ability which contributes to more than 90% of variance in the common-factor-space. Table 14 shows the corresponding relations among factors of each analysis.