Expertise, trust, and virtual team performance : leveraging team composition for intelligent agent design

Abstract

With technology-mediated communication becoming more prevalent in modern organizations, research efforts have turned toward mechanisms for improving virtual team performance. One approach considers incorporating intelligent machine agents into teams of humans as a means of influencing behaviors, ultimately leading to better performance outcomes. Such agents would require a robust representation of the factors influencing team performance in order to determine which behaviors are effective for a given context. At present, these factors are not well understood, and additional research is required in order to reliably predict their effects on team performance. In this thesis, we contribute to the virtual team literature by investigating the effects of team composition on performance, including (1) moderating effects on the relationship between team communication and performance, and (2) interaction effects between features of team composition and team performance. To this end, we conducted an experiment in which teams of participants competed in a data science competition. Two features of team composition, domain expertise and dispositional trust, were used as treatment factors for team assignments, and we measured the processes, emergent states, and performance outcomes of each team. Our results indicate that domain expertise moderates the relationship between communication and team performance, and there was evidence to suggest further interaction effects may exist which merit additional study.