For the last two decades, the topic of knowledge management (KM) has been spotlighted as knowledge emerged as a key source of competitive advantage in modern, knowledge-based societies (e.g., Grant, 1996; Von Krogh, Ichijo, and Nonaka, 2000; Hass and Hansen, 2007). Despite a proliferation of research on KM and relevant topics such as human resource management (HRM) and information systems (IS), the mechanisms through which knowledge networks are formed and the social connections’ resources&hibar;i.e., knowledge&hibar;are utilized have not been fully discovered yet. Particularly, research on the mechanisms through which high-level knowledge is shared and created in higher education institutions (HEIs) has been little explored. To fill this gap, this study explores two main research questions: (1) what are the determinants of the creation of knowledge networks in HEIs? and (2) when are social connections’ knowledge resources utilized to share and create knowledge collaboratively in the networks?

            Theoretically, this study adopts the institutional logics perspective (e.g., Thornton, Ocasio, and Lounsbury, 2012), which assumes interactive roles between structures and agencies. This means that we assume that actors’ decisions and behaviors are affected by both socially-generated, structural context&hibar;i.e., network structures&hibar;and actor-based attributes as well. Network structures are determined by actors’ choices in tie creation and dissolution. Along this line, our model takes into account social network factors, such as the tendency for networks to display reciprocity, transitive closure, and other common social network substructures, while simultaneously accounting for the effect of actors’ demographic attributes, such as gender, age, and ethnicity, and the perceptional costs and benefits of knowledge sharing and collaborative knowledge creation.

            To accomplish this, we use a stochastic approach for modeling social network data. Our analysis relies on a dataset collected from one or more colleges of a university. Two Exponential Random Graph Models (ERGMs) are utilized to analyze the mechanisms of the tie creation in (1) an acquaintanceship network and (2) academic mentorship and co-authorship networks, using StOCNet, a stochastic social network package developed by Sniders and his associates (Boer et al. 2006; Snijders 2001, 2002, 2005).

            The paper will begin with a brief literature review on the structure-agency debates, knowledge management, and social capital including homophily, followed by a theoretical framework and an outline of methods and procedure. The second section will present findings from analysis of data focusing on the structural and actor-based attribute factors discussed above. This paper will conclude with a discussion of implications for knowledge management policy in HEIs based on our modeling of factors that affect the creation of knowledge networks and the mobilization of knowledge that social connections possess.