Clusters, recipients and reciprocity: Extracting more value from email communication networks

A common problem in the empirical research of communication networks is how to construct network models from raw data. Email communication datasets are typically transformed into networks in which nodes designate email users and ties connect nodes if an email has been exchanged between the corresponding users. Unfortunately, this method of extracting sender-recipient dyads from multiple recipient emails conceals information regarding the nature of affiliations and the process by which they come about. Specifically, the findings of this paper indicate that recipient lists of a user's outgoing emails are not arbitrary subsets of the user's contacts. Instead, recipient lists delineate meaningful organizational units so that being co-recipients of the same email(s) is a stronger indication of affiliation than being neighbors of the same node(s). Moreover, the number of recipients in an email reveals not only the nature of affiliations within the network but also the way in which these relationships are likely to unfold. Methodologically, this paper engages in a search for new ways to construct network models that extract more value from existing data. Theoretically, the findings could inform the debate between interaction approaches and network sciences.