Understanding Shared Familiarity and Team Performance through Network Analytics

Abstract

In this article, we propose a network approach to understanding team knowledge with archival data, offering conceptual and methodological advantages. Often, the degree to which team members' possess shared knowledge has been conceptualized and measured as an aggregate property of a team as a whole. Rather than an aggregate property, however, we argue that shared team knowledge is more appropriately conceptualized as a network of knowledge overlaps or linkages between sets of team members. We created shared knowledge networks for a sample of 1,942 software teams based on members' prior experiences working with one another on different tasks and teams. We included metrics representing topological features of team shared knowledge networks within predictive models of team performance. Our results suggest that network patterning provides additional predictive power for explaining software development team performance over and above the effects of average level of knowledge similarity within a team.