心理学研究
Online ISSN : 1884-1082
Print ISSN : 0021-5236
ISSN-L : 0021-5236
ソシオメトリによる産業社会の集団凝集性の数量化について
北脇 雅男
著者情報
ジャーナル フリー

1956 年 26 巻 6 号 p. 386-392

詳細
抄録

Introduction : It is to be regretted that very little is known about the relation between the index of cohesiveness and the actual human relations existing in a group. The object of this paper is to examine the validity of the six kinds of indexes of cohesiveness obtained by various formulae. In order to determine whether an index can actually indicate the social relations in a group, it will be essential to ascertain first the hypothesis that the work-groups are the most efficient when the index is high.
Subjects : The subjects of this study were 211 girls (age : from 17 to 28) living in the workers' dormitory of the Kanegafuchi Spinning Company. The subjects were divided into thirty-one groups in size from 5 to 11 persons. The girls had lived together for at least six months and thus knew each other well enough so that they were supposed to be capable of making meaningful sociometric choices and rejections.
Method : 1. The writer administered a sociometric test asking the subjects to select as many friends as they would with whom they wanted most to live together. And then, the indexes of cohesiveness were computed respectively in each group, counting up the number of those who were selected by each girl in every group.
2. The subjects were measured for their intelligence and character for the purpose of clarifying the traits of these group members. Besides these traits, the wages, the length of service, and the number of absences of these subjects were also examined as indicators of the characteristics of group productivity.
Results : 1. The cohesiveness index of a certain group in the dormitory were not constant and were changeable according to the formula used in the procedure.
2. The inconstancy in these indexes seems to be due to the differences of the structure of formulae used. The partial correlations between the indexes were obtained. The results showed that “Icc”was quite different from any other index in terms of structure.
3. Trying to detect the correlation between sociometric index and personality variables, the writer correlated the indexes with the personality measures. Though a considerable variation was found between the correlations, the relations between “Icc”and intelligence score tended to be clearly positive, but those between “Icc” and V. Q. were negative as shown in Table 7. Similarly, statistical significancy was found in the correlations between “I” and intelligence score or in those between “I” and V. Q. But the correlations between the indexes obtained through any other formula and personality variables were not significant.
4. Comparing the indexes with the variables for the group productivity characteristics, the relations were found to be very low. In their recent paper, Schachter and Berkowitz have demonstrated that the group cohesiveness has not necessarily a close connection with the group productivity. Although “Icc” and “I” are not closely related with the characteristics of group productivity, the indexes computed here exhibited the objectivity of group cohesiveness, because the indexes computed showed a high degree of relation with personality variables of the group.
5. Why are those indexes effective and valid indicators of group cohesiveness? As one of the reasons, it may be pointed out that the distribution of “C” or “E” is approximately Poisson, whereas the distribution of personality variables is normal. Generally speaking, the coefficient of correlation is a measure showing the extent to which changes in one variable are associated with changes in the second variable.
If a series of variables are normally distributed, and if the second one shows Poisson distribution, it may be said that better prediction of the correlation can be made by using a correlation ratio.

著者関連情報
© 公益社団法人 日本心理学会
前の記事 次の記事
feedback
Top