In the context of bridging the so-called externalist and cognitive perspectives on the growth of research communities, a cancer problem domain is examined (1) to distinguish a growth in knowledge from a proliferating research literature, and (2) show how measurement of formal communication, uninformed by the historical record, clarifies or distorts sociological interpretations of innovation and growth in biomedicine. Specifically, coauthorship and citation networks are analyzed for reverse transcriptase researchers, 1970–74. This analysis reveals the visibility of large National Cancer Institute laboratories in the research literature, but demonstrates the need to augment disaggregated network data with intellectual and social (policy) history to explain the growth and structure of the domain.
Social science network analysis originated in the small group sociometric tradition, thus many of the common assumptions of network models are inappropriate theoretically and formally for the analysis of open systems of social relationships. Five common assumptions of network analysis are identified, discussed and criticized: (a) generators are homogeneous, (b) relationships are dichotomous, (c) groups have fixed boundaries, (d) relationships are symmetric, and (e) networks are static. It is suggested that an open input-output model overcomes many of the difficulties inherent in the more common network analytical techniques. After a formal treatment of input-output analysis, and its relationship to network analysis, some interpretations from exchange theory are suggested. This model helps the analyst overcome many of the theoretical difficulties encountered in other models and allows the researcher to specify how subsets of individuals are embedded within larger social contexts. Specifically, because society is comprised of numerous interacting subsystems, this model is particularly beneficial in describing how groups of scientists interface with each other and with the larger social domains.