Recent studies on leadership do characteristically employ the situational approach which, unlike the trait approach, investigates the leadership not as a personal trait of the leader but as a function of the group situation. Although many studies(e.g., Jennings, H.H., Gibb, C.A.)showed the relationship between the person's status as represented by the positive choice in sociometry and the leadership structure, more thorough analysis of the total structure of the choice process is needed. Hence, Kano conducted an experimental study based on the postulate of multi-stage structure of the group. This multi-stage structure is the structure which takes both direct and indirect choice into account. Based on the data of 200 junior high school pupils, it was found that the influence of a pupil as judged by the teacher did highly correlate with his second-stage status. In conventional studies of the types and functions of leadership, a bipoler dimension with P(erformance) function and M(aintenance) function pole was assumed. We proposed to take P and M functions as different dimensions. Assuming three types of leadership, P, M, and PM, we experimentally analyzed their relationship with the productivity of the group. P-type corresponds to conventional "autocratic" or "work-centered" type. M-type corresponds to "democratic" or "human-relation-centered" type, and PM-type is both work and human relation centered. The results obtained at a governmental training institution showed that PM type leadership was most effective in terms of both productivity and group moral. Another study conducted witheight groups of coal mine workers confirmed this finding.