ABSTRACT
In this chapter, the authors illustrate how most extant sociograms distort the networks of relationships they are presumed to represent. Sociograms distortion cannot be removed or rendered less obscuring by throwing away data or by applying more complex structural models. The authors explore the question whether sociometrics could serve as a data source for sophisticated analyses of social structure. They discuss the meaning of measurement error in sociometric data and shows how it is generated in a variety of common sociometric procedures. The authors examine the likelihood that fixed-choice sociometric tests, tests in which a set number of responses is requested from every group member, yield sociograms of low compatibility. The importance of sociometric data for the study of small-scale social systems leads us to suggest two approaches that may help to make these data more useful. A more subtle generator and major source of measurement error resides in the formalities of common sociometric procedures.
