ABSTRACT
Sociometry, the study of interpersonal relations within groups, is an area that has been of much interest to psychologists and sociologists since its inception. Yet despite its long history and widespread popularity, sociometry has benefited relatively little from formal measurement or statistical innovations. Perhaps the area where this is most acutely felt is in the representation of the results of a sociometric study. Usually the social interaction network is summarized in a sociogram, which is a spatial representation of the group of interest with arrows indicating choices and/or rejections between members. A problem with the sociogram is that it is not entirely empirically derived. The many techniques in use for drawing sociograms all rely to some extent on judgments made by the researcher. This lack of objectivity and formality is often seen as a serious shortcoming. As Lindzey and Byrne (1968, p. 462) put it,
Non-graphical representations of the social interact network are available. Usually these take the fonn of an index or a series of indices reflecting such
things as groups coherence (Criswell, 1946, 1947). Although such statistical approaches can be useful and even essential, most sociometric studies report a sociogram even if a nongraphical representiltion is used as well. This is done simply because no array of indices can present as compelling a picture of a group interaction network as the sociogram.
