As one of the problems connected with the questionnaire-method in a broad sense, the present writter takes up the item-analysis on an introversion-extroversion test. The introversion-extroversion test arrange many question-items, and count the reponses to each of itms in many cases yes-or-no responses so as to judge the subject to be introversive or extroversive, and if possible, in what degree he is introversive or extroversive. No attempt has get been made to analyse each item and see show how effective each item is to measure introversion or extroversion. The selection of the items is made through the experiences, knowledge and insights of the experts, and these items may be sufficiently significant. But their function comes into play, only when they are printed on paper and shown to the subject, and we cannot be sure whether the purpose intended by the experts can actually be realized. The present writer tries to analyse each item to see what extent it has, in itself, the discriminative power for the judgement of introversion or extroversion. As the procedure of this analysis, it is proposed to pick up among those to whom some introversion-extroversion test, has been administered two groups: Group I considered as “very introversive” and Grop E as “very extroversive” and, then proceed to see how each then of the two groups has reacted upon of the items. According to the intention of test-maker each item is expected to have its special function, but it hsa not yet been proved. If the introversive response shown by Groups I to a certain item is named Ii, the extroversive response by the same group Ie, the introversive response by Group E to the same item Ei, and the extroversive response by the same group Ee, it must be ascertained whether “Ii-Ei” and “Ee-Ie” have significant differences, and agree with the test-maker's intention. The writer has undertaken this analysis of the introversion-extroversion test by the Educational-psychological Laboratory of Tohoku University and the introversion-extroversion test of Dr. Awazi. Such items as have significant differences between Group I and Group E, and agree with the test maker's intension are, it has been found, only 13 out of 40 items in the former case and 4 out of 50 in the latter. This study shows that there is some limitation to be considered when questionnaires for the introversion-extroversion test are made, and also that test-items, in general, often lack such a function as is intended by the test-maker. To make the items effective for the purpose they are intended for special devices, like special devices in verbal expressions, must be made in making the test items.
S. Morinaga 3) studied the assimilatory illusion of concentric circles in which the size of the inner circle was kept constant while that of the outer was varied. (The present writer prefers to call this illusion the displacement effect or simply D-effect.) He found that the size of the inner circle was a little overestimated when the outer circle was relatively small and that the amount of over-estimation increased with the increase in the size of the outer one up to a certain point and then gradually diminished. The maximal effect was obtaind when the ratio between the sizes of the two circles was 2:3. It is not clear, however, in what manner the amount of this D-effect changes when the size of the figures is varied. The present work deals with this aspect of the problem. Experiment I.: Conditions & Precedures: Standard figures consisted of three series of various concentric circles. The sizes of figures are shown in Table 1. (See Table 1, p. 17) Comparison figures consisted of three series of circles. (Table 2, p. 17) The inner circles of Standard figures were compared by turns with Comparison circles by the method of limits and the equivalent values of the inner circles were obtained. Results: Averages of 8 subjects are shown in Table 3 (a), (p. 17). The result of the analysis of cevarian is shown in Table 3 (b). c. d. in this table means critical difference: If the difference of two average values is equal to or larger than c. d., the differnce is statistically significant. The Displacement of the circle (=each equivalent value-the equivalent for “A” figure/2 is shown in Fig. 1. (p. 18) The relative amount of displacement of the inner circle is shown in the lower part of Fig. 2. (p. 18) 1) The inner circle shows D-effect evidently and its absolute amount varies with the size of the Standard figures. 2) The curves of the relative D-effect agree with ore another closely despite the different size of the figures. 3) The maximum of the relative D-effect appeares at the point where the ratio between the two circles is 2:3. 4) Therefore, it may be concluded that Morinaga's finding is verified and validated by this experiment even when the sizes of the figures are different. Experiment II.: How dose the D-effect appear at the outer circle? Conditions: The sizes of three series of Standard figures are shown in Table 6 (p. 21). In this case another newly added series of Comparison figures was used for the series VI of the Standard figures. The other conditions and procedures were similar in Experiment I, except that the outer circles of the Standard figures were compared with the Comparison circles. Results: The averages of 10 subjects and the results of the analysis of variance are shown in Table 7 (a, b, p. 21). The amount of displacement of the outer circle is shown in Fig. 4 (p. 21) and the curves of the relative amount of displacement are shown in the upper part of Fig. 2 (p. 18) 5) The D-effect of outer circle is negative i.e. the outer circle is displaced toward the inner direction and the absolute amount of this D-effect varies with the size of the Standard figures. 6) The relative D-effect curves agree with one another fairly well. 7) The maximal D-effect appears also at the point where the ratio between the two circles is 2:3. 8) Morinaga's finding is thus extended to the case of the D-effect of the outer circle. 9) The Fig. 2 shows, when viewed as a whole, that the inner and outer circles displace themselves interdependently toward each other; they behave as if they were pulling each other. But the D-effect is not a simple function of the distance between objects (circles). Presumably, other variables (e.g. the chracter of configuration) ought to be taken into consideration.
1. This investigation was undertaken 1) to ascertain whether or not the formula f=φ.d (s=perceived saize, φ=visual angle, d=perceived distance) can generally hold trues; 2) to determine the relation between the perceived size and the viewing distance under various conditions; and to find out, if possible, the main factors which determine size constancy. 2. For testing the first problem, values of s and d were measured experimentally, while φ was kept constant. The experimental objects are cardbord discs or lightened discs, as to the size of which observers might have no particular assumption from their past experience. The result showed that there existed a certain condition in which the formula s=φ.d did not hold true (Table 5; Table 6; Table 7), and further that physically larger objects were perceived nearer than the physically smaller objects (Table 9), under the condition in which the perception of distance was ambiguous. 3. To investigate the second problem, observations were first made in commonplacen surroundings, such as an ordinary room, corridor or balcony, which were physically rectangular in shape, homogeneously lightend and frameworks of which were within the observer's visual field. Such spaces have phenomenally different shapes according to the difference of their physical depth : in cases they are long, their frameworks are perceived to be converging towards the end of the spaces. On the contrary, if they are short, the frameworks are perceived to be diverging, when they are observed binocularly keeping the fixation points at the center of the spaces. Results indicated that the types of curves showing the relation between the perceived size of objects and veiwing distance depended upon the physical depth of the spaces to which they belonged : that is, the curve ascended where the space was shorter than 5 meters in length (Table 10; Table 13) and descended where the space was longer than 7 meters (Table 11; Table 12), while it was nearly horizontal and liner where the space was 5.5 meters (Table 14). These facts seem to show that the perceived size of an object is determined by the phonomenal shape of the space to which it belongs (Fig. 2). 4. But, when the space 5.5 meters in length was slightly darkened the curve became convex instead of linear (Table 15, Fig. 3). And almost linear curve was obtained when the observation was made in the bright balcony which had the length of 120 meters (Table 16, Fig. 4). These facts seem to show that the flatness of the curve depends on the phenomenal straightness of the framework of the space. 5. Observations made in a completely darkened room revealed that values os Sc/Ss (Ss=physical size of the standard object, Sc=physical size of the comparison object equalled to the standard object) obtained by the ordinary observers were smaller (Group I in Table 18 and 19) than those obtained by the particular observers who had been accustomed to work in this dark room as the experimenter and the assistant (Group II in Table 18; Table 21) : the average Sc/Ss for Group II was nearly 1 up to 4.5 meters even when observed monocularly and above 1 when observed binocularly (Fig. 6). These facts indicate the presence of the effect of past experience on the perception of size. 6. It is to be presumed that the phenomenon of the size constancy in commonplace surroundings is probably a function of the similar kind of past experience with what Group II had, and, hence, that the curves obtained in commonplaces surroundings may coincide with the curves obtained by Group II in the dark room, provided that proper adjustments are made with respect to the differences between these two conditions. In an attempt to compare these two kinds of curves mentioned above, averagd values of Sc/Ss within a certain range of distances were calculated from the curve obtained by Group II (For explanation of this method, see Fig. 7).