British Journal of Social Psychology zyxwvutsrq (1997), 36, 241-262 zyxwvu 0 1997 The British Psychological Society zyxwvu SPO478 Printed in Great Britain 24 1 z Styles of group discussion in computer-mediated decision making Fraser J. M. Reid* Department of Psychology, University zyxwvut of Plymouth, Drake Circus, Plymouth PL4 zyxw 8AA, UK Linden J. Ball Division of Psychology, University zyxwv of Derby, UK Andrew M. Morley and Jonathan St B. T. Evans Department of Psychology, University of Plymouth, UK This study investigated the claim that keyboard-based computer-mediated group dis- cussion suppresses normative influence relative to informational influence. In a simu- lated panel of inquiry presented with computer databases containing inconsistent and incompletely shared information, four-person groups attempted to reach decisions either in face-to-face (FF) discussions or in real time computer-mediated (CM) discussions via a network linking computers at separate locations. CM groups reported greater difficulty communicating ideas than FF groups and took longer to reach a decision. Contrary to previous research, CM groups exhibited a preference for a normative style of discussion, exchanging proportionally more positional and value statements and proportionally fewer factual and inferential statements than FF groups, but only in the final stage of the experiment. This discussion style led in turn to lower levels of private post-decision agreement and decision satisfaction among CM groups. Results are discussed in terms of the restrictions imposed by computer mediation on group discussion, and how these combine with other task conditions to determine group goals and discussion style. With the recent growth in interactive computer networks, electronic messaging and com- puter conferencing are becoming accepted modes of group communication, almost as commonplace as a phone call. However, the effects of computer-mediated communication (CMC) on group performance are only now beginning to be understood, and are occasionally contrary to our expectations. We now know, for example, that conventional, keyboard-based CMC can have unfavourable as well as favourable effects, and that the out- come largely depends on the nature of the group task: computer mediation typically helps idea generation and sharing, but hinders groups performing tasks requiring agreement on judgmental matters (Straus & McGrath, 1994; Valacich, Dennis & Connolly, 1994). The single most widely held explanation for these effects is based on the idea that CMC filters out the social and contextual cues normally present in face-to-face conversation. The * Requests for reprints.