Communication style drives emergent leadership attribution in virtual teams

Abstract

Leader selection plays a key role in how human social groups are formed and maintained. Leadership is either assigned through formal processes within an organization, or emerges informally through interactions with other group members-particularly in novel contexts. COVID-19 has accelerated the adoption of virtual meetings and more flexible team structures. However our understanding of how assigned leadership influences subsequent leadership emergence in virtual settings is limited. Here we examine the relationship between assigned leadership within an existing organization and subsequent emergent leadership attributions as members engage in virtual interactions. To do so, we created and implemented a novel virtual group decision-making task designed to support quantification of a more comprehensive set of communication style elements, such as speech dynamics and facial expressions, as well as task behaviors. Sixteen members of a real world organization engaged four repeated rounds of a group decision making task with new team members each time. We found participants made novel attributions of emergent leadership rather than relying solely on existing assigned leadership. While assigned leadership did influence leadership attributions, communication style, including amount of speech but also variability in facial expressions, played a larger role. The behavior of these novel emergent leaders was also more consistent with expectations of leadership behavior: they spoke earlier, more often, and focused more on the correct decision than did assigned leaders. These findings suggest that, even within existing social networks, virtual contexts promote flexible group structures that depend more on communication style and task performance than assigned leadership.

Keywords: facial affect; group decision making; leadership selection; speech analysis; synchrony; virtual teams.

Figure 1

Experimental approach and analysis pipeline: (A) Quantified social network, (B) Leadership scores of participants in social network. (C) Hidden profile task schema. (D) Ratings of participants on warmth and competence from other participants. (E–G) Example data from session 10, (E) Raw speech waveforms (F) cumulative speech over session durations (G) Percentage of total speech by each participant. (H) Percentage of speech by each participant in each session. (I) Facial affect coding shema. (J) Example facial affect time course from session 10, Engagement, Joy, Sadness and Anger (K) pairwise synchrony between interacting participants, black squares indicate participants that did not interact. Faces created with AI software (https://stablediffusionweb.com/) License: CC0 1.0 Universal Public Domain Dedication.

Figure 2

Communication style predicts emergent leadership attribution. (A) Proportion of 16 sessions that reached the correct consensus. (B) Standard deviation of speech percentage between correct consensus and incorrect groups. (C) Log likelihood ratio of reduced models compared to best models. (D) Cumulative speech for each participant session color temperature indicates emergent leadership score. (E,F) Relationship between percentage of speech and (E) emergent or (F) assigned leadership by each participant in a session split into five (i–v) quintiles. (F) Percentage of speech and assigned leadership proportion by each participant in a session split into five quintiles. (G,H) Relationship between each proportion of each group’s discussion of the correct outcome by each participant and emergent and assigned leadership.