Three models of change and continuity in group structure are tested using existing longitudinal data on 20 small groups. Groups met face to face or via a computer-mediated communication system for 13 weeks. Computer-mediated groups fit the robust equilibrium pattern best, with initial fluctuations in the influence hierarchy followed by a more stable structure that persisted despite changes in operating conditions. Face-to-face groups fit a bistable punctuated equilibrium pattern best, retaining their initial influence structure until an environmental cue triggered a shift. Contrary to die predictions of this model for radical change, adjustments were modest. Poor performance on tasks failed to trigger changes predicted by the adaptive response model, probably because outcomes were not very important to group members.
