ABSTRACT: Twenty-four groups of five professionals and managers used computer conferences to reach agreement on the best solution to a complex ranking problem. Two software tools for structuring the conferences were employed in a two-by-two factorial design. Groups with "designated leadership" (DL) used software support to elect a discussion leader. Groups with"statistical feedback" (SF) were presented with tables periodically that displayed the mean rank and degree of consensus for each item. DL improved levels of consensus; in the absence of a leader, SF improved level of agreement slightly. Statistical feedback as operationalized in this experiment was detrimental to the ability of a group to achieve "collective intelligence," defined as a group decision better than the prediscussion decision of any of its individual members. Characteristics of the individuals and groups were also associated with variations in outcomes.
Key words and phrases: computer-mediated communication systems, group decision support systems