Fig. 2. Solid square is suggested as a general representative value such as mean of overall data (A). It looks reasonable to divide the data into three groups and explain the data with three different means of groups (B). To evaluate the efficiency or validity of dividing three groups, the distances from group means to overall mean and the distances from group means to each data are compared. Distance between group means and overall mean (solid arrows) stands for the inter-group variance and distance between group means and each group data (dotted arrows) stands for the intra-group variances (C).