Collaborative Research: Moderator Interventions and the Automated Coding of Status Hierarchies in Task Groups

The formation of hierarchies in task groups is a ubiquitous process, but these hierarchies are not always stable. Group hierarchies can evolve with the contributions of members. If a comparatively low status group member makes a positive contribution to the group, for example, they are likely to increase their standing in the overall group. Conversely, if a relatively high status group member makes a poor contribution, they are likely to decrease their standing. This research systematically varies moderator interventions to study their effects on the contributions of various group members to evaluate whether group moderators can enhance or flatten the emergence of hierarchies in task groups and what the effects of this are for task success. Drawing on existing research on status dynamics, moderators intervene in task discussions to enhance the standing of comparatively low status group members. To study these processes, the research uses experimental and computational methods. First, moderator interventions are experimentally varied. In a control condition, moderators do not intervene in the emergence of the group’s status hierarchy. In two experimental conditions, moderators use cues and interventions from the extant literature on status dynamics to enhance the perceived standing of comparatively low status group members. Influence over the group is measured and used to evaluate whether moderator interventions are successful in mitigating task irrelevant characteristics from shaping the power and prestige order of the group. Also examined is what this means for task success. The research further develops a computational pipeline that uses Natural Language Processing to track the group’s status hierarchy. The group’s conversation for status-relevant information is coded and machine learning algorithms “learn” the codes that humans assign to the text. This enables automatically coding of subsequent conversations for status-relevant information. Collectively, the work evaluates new moderator interventions to make task groups more egalitarian, and develops computational tools to automate the coding of hierarchy formation in task groups. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.