Team Boundary Spanning through Enterprise Social Media: Exploring the Effects of Group-Level Diversity Using a Data Science Approach

Effective work groups engage in team boundary spanning, that is, the use of communication ties as conduits to critical external resources. With the proliferation of social media technologies in enterprise settings and the associated increase in visibility of communication ties, understanding their impact on boundary spanning becomes imperative to improving cross-boundary knowledge creation and management inside organizations. In this paper, drawing on log data from 415 unique work groups in an enterprise social media (ESM) system, we use a machine learning approach to automatically detect three distinct team boundary-spanning activities. Using zero-inflated poisson regressions, we further show the effect of group visibility as well as three distinct sources of group structural diversity -- geographic, functional, and hierarchical -- on the extent to which teams engage in boundary spanning through ESM. Implications for theory and practice around the use of data science approaches as well as visibility and diversity constructs for understanding team boundary spanning are discussed.