Inferential Statistical Analyses on Information Network Generation

Information technology has made massive networked data increasingly available. Information networks, which include various digital artifacts and social structures, provide insights into relationship structures. The formation of these information networks results from users’ activities in information systems or on electronic platforms. This dissertation studies the formation processes of two types of information networks using inferential statistical analyses: a user-item preference network embedded in recommendation systems, and a brand associative network of user co-engagement with brands on Facebook. The studies in this dissertation contribute to understanding information networks in three ways. First, statistical inferences are deduced to explain the generation mechanisms of information networks. Second, the dynamic aspects of information networks are studied using inferential statistical analyses. Third, insights are inferred in the studies to improve business practices. The findings in the analyses on user-item preference networks provide insights to improve current movie recommendation systems, and to inform the design of recommenders for different types of products across different E-business sites. The findings in the analyses on brand associative networks help establish the potential of such brand networks for future research on varied issues of importance to marketing and brand managers.