其他摘要：The human growth and development has various aspects which are more or less associated with each other. In order to investigate this process of growth and development more precisely, these various aspects must be treated as the associated attributes in the same space. Now, let X be the measure vector (X_1X_2X_3.........X_n), X can be expressed as a point in the n dimensional variable space. Testing the difference between the mean vectors of two groups in deferent growth level is much more efficient than testing the one between the mean values of measures, separately, However, the n dimensional space can not be observed intuitively, so it may restrict the observation. For the purpose to observe the n-dimensional phenomenon intuitively, the two-dimensional component space was approximated to the n dimensional variable space. In order to extract the two components, principal component analysis was applied and, the first and the second principal component were used. Then, (C-C)V^<-1> (C-C)' = X^2 (β)expresses the β% probability ellipse in the component space, where C stands for component score vector, C the mean vector of component score and V the variance covariance matrix of component score in each grade group, so the ellipse of each grade group was computed and drawn in the two dimensional component space. The change of ellipse area, position and the overlapping area between the ellipses of different grade groups were investigated. This approach can not be applied to all the situations. If considerable portion of total variance can be explained by these two components, this technique can be successfully applied. Then, this technique was applied to investigate the physique growth of elementary school sample in some rural area in Kyoto. Then, several interesting findings were extracted, such as described in this paper.
