Problem: It is hard for preschool children to perceive direction in two-dimensional space. It has repeatedly been pointed out that there is a trend in errors made in their perception of spatial direction. How does their ability to perceive direction correctly develop chronologically? What kind of developmental curve will be obtained? what types of errors do they make along with increasing age? The purpose of this experiment is to answer these questions. Method: As shown in Fig. 1 in the text, a standard figure randomly selected in turn out of the comparison figures, was tachi-stoscopically presented first and then a group of comparison figures was shown below. The subjects were asked to point out the same figure in the latter as the standard one. Three hundreds children ranging in age from 3 to 8 years were used. Results: 1) The developmental curve of the ability of directional perception is contrasted in Fig. 2 in the text with that obtained in the previous study. 2) It was found that the cubic polynomial expression, Y=a+bX+cX2+dX3, was the most appropriate formula to express the functional relation between the age (X) and the ability (Y). 3) The frequency of errors was found highest in a diagonal, less in a horizontal and the least in a vertical direction. 4) A qualitative analysis of errors proved that the errors of mirror drawing type, i.e., the interchange of right and left, were the most frequent, next came those of reversal type or 180° rotation and the least were those of proximate type or 45° rotation. The reason for the highest frequency of mirror drawing errors may be that the concept of right and left is not well established in early ages. Further investigations on children's attitude in perception of figures and the physiological basis for visual perception would clarify the problems.
1) To testify Vinacke's hypothesis which assumed the parallel relationship between intelligence and conceptualization, six groups of feeble-minded children (IQ<80) graded in MA (4:0-9:0) were chosen as the subjects for the classification task of 20 picture-cards each representing some familiar object (e.g. an automobile) and the result was compared with that of the previous experiment in the same condition but in normal children (90<IQ<110) as subjects. 2) The responses of the chidren were, by the qualitative examination of the classification and the verbal expression of the classifying principle by the suject, divided into six categrories of which those belonged to the preconceptual stages were analyzed further. 3) The increase of the responses in the conceptual level with MA, was confirmed and the result was statistically significant, and, as to this tendency, there was no statical difference between the normal and the feeble-minded. Moreover, the successive order in which the various preconceptual responses became dominant with ascending MA, showed remarkable correspondence in both the normal and the feeble-minded group, suggesting a general developmental stages of the conceptualization. 4) Meanwhile the comparison with the normal in the weight of the various preconceptual response categories disclosed some qualitative characteristics of the feebleminded which were not identified by MA as the quantitative, statistical concept; e.g. their stand still mental state, their primitive mentality in contrast to higher CA, and the dominancy of the mechanical verbal habit unsupported by the proper conceptualization.
The purpose of this study is to investigate L. L. Thurstone's theory of absolute scale in intelligence measurement and its modification based on the raw test scores (Modified Absolute Scale Theory), from the standpoint of F. M. Lord's theory of test scores. For empirical verification the protocls collected for the purpose of constructing LIS Measurement Scale for Non-verbal Reasoning Factor were analysed here. In the first section, “item characteristic curve” in Lord's theory, together with the variables and parameters used in the theory, was explained (see Fig. 1, 2 and 3), and the necessary formulae were developed. In the second section, Thurstone's theory of absolute scale was introduced and the main differences between the two theories were pointed out (cf. Fig. 1 and 4). The necessary formulae were developed to show the principal concepts of Thurstone's theory in the system of Lord's theory construction. The two test characteristic curves of LIS Scale on the basis of Lord's theory and of Thurstone's theory were contrasted (see Fig. 5). The value of point estimation of the ability score, c, from each obtained test score, s, and its standard error of measurement were obtained on the basis of Thurstone's theory, and were compared with those obtained on the basis of Lord's theory (see Table 1 and Fig. 6). It was discovered that on the basis of Thurstone's theory of absolute scale the ability score is estimated lower when the subject's ability is high, and it is estimated higher when the subject's ability is low. Theoretical and observed frequencies of subjects acquiring each test score, s, on LIS Scale were compared (see Fig. 7), and it was found out that the fitness of the theoretical frequencies based on Thurstone's theory to the observed frequencies was far worse than that based on Lord's theory. In the third section the data were analysed on the basis of Modified Absolute Scale Theory and the scale value of each test score, s, together with the mean and standard deviation of each examinee group was obtained (see Table 2, 3 and 4, and Fig. 8). It was discovered that the scale value, ts, of the test score gave a very good approximation to the point estimation value of c from the raw score, s, by Lord's method, especially in the case of score where the number of subjects at that level is large (see Table 2 and 3). In the fourth section the data were analysed again on the basis of Thurstone's theory of absolute scale and the mean and standard deviation of each examinee group were estimated and compared with those obtained by the other two methods. It was discovered that the standard deviations of each group estimated by the three methods showed a very good agreement, the means estimated on the basis of Lord's theory and of Modified Absolute Scale Theory coincided with each other, but the interval between the mean scale values of the two examinee groups was a little smaller when they were obtained on the basis of the original absolute scale theory. It is concluded that Modified Absolute Scale Theory will be a hopeful substitute for Lord's theory both in estimating the mean and standard deviation of an examinee group in mental ability and in obtaining the adequate substitute for the ability score, as far as the test consisting of items of high correlation with the ability measured is concerned (see Table 5).