Dimensions of Social Networks: A Taxonomy and Operationalization
Michael Genkin, Nicholas Harrigan, Rajee Kanagavel, Janice YapWhat are the basic types of social network ties and what are the optimum name generators to capture them? While there have been several network classifications proposed and a large proliferation of name generators capturing various tie content, there is no definitive taxonomy of network tie types, and no clear mapping of name generators to tie types. Building on previous research, this paper proposes a framework for doing so in a principled way. We collect and analyze a dataset of 1,412 respondents and 24 name generators. We conduct multi-dimensional scaling and hierarchical cluster analysis on the conditional probability of an alter receiving each type of tie. From this we identify a taxonomy of three positive tie types (admiration, closeness, socialize); and three negative tie types (contempt, passive conflict, active conflict). Next, we use Youden's J statistic as a metric to identify the name generator that best maximizes sensitivity and selectivity when classifying ties into one of the six tie types. Finally, we analyze the roles captured by the resulting name generators to better understand the composition of the networks that they identify. These results provide a theoretical classification of tie types and of their name generators and speak to several debates in the social networks literature.