teams' dynamics and success
Since many human enterprises take place in formal/informal teams, the study of what makes them effective, efficient, and performant percolated across disciplines producing a very large body of literature. Regular communication, distinctive roles, interdependent tasks, shared norms, coordination, personality traits, diversity, particular structural properties of the underlying social networks have been listed as key factors affecting teams success. Yet, our understanding of how teams form, evolve, and how the attributes of individuals affect processes, emergent states, and their performance is still very limited. The first fundamental issue is how to model and represent a team. In particular, how do we fuse the features of single individuals such as skills, socio-demographic indicators, past experiences at the team level? The simplest answers are offered by the so-called compositional models. They rely on the assumption that the contribution of each team member is equal. As a result, attributes of single individuals are considered additive and averaged in a summary index. Diversity indices, that consider the variance of members’ individual attributes, are also compositional in nature. Despite their popularity, compositional approaches are an extreme simplification of the dynamics at play. In fact, from sports to businesses, groups able to embrace team-work, multiply the skills using effective networks of interactions and communication, are likely to outperform others that might be even stronger at the individual level. This intuition implies that teamwork is an emergent phenomenon that can be captured only using holistic rather than reductionistic perspectives. To this end, and in clear contrast with compositional approaches, we find compilational models where team-level attributes are considered as a complex combination of individual-level properties. These models often consider explicitly the social network that encodes the interactions, roles, and positions of members. However, our understanding of teamwork as an emergent and complex phenomenon is still very limited. In fact, many questions remain unexplored: how do we learn optimal high-order compilational models directly from data? How do they differ across different tasks/contexts? How do we account for time? My research in the area aims to provide some answers to these challenging questions
