| This thesis investigates the effect of computer-mediated communication, as compared to face-to-face discussion, on inductive learning by groups. An experiment was conducted in which four-person groups attempted to induce a rule which partitioned a deck of playing cards into exemplars and nonexemplars. All data for inducing the rule were presented via computer in an iterative process consisting of several rounds of hypothesis generation and testing. In half of the groups, members communicated with each other in face-to-face discussions. In the other half, groups members were physically isolated from each other and communicated through a network of personal computers.;Social combination processes, which describe the mappings to a group decision from individual members' views prior to discussion, and problem solving performance were compared between communication conditions. In the computer-mediated and face-to-face condition, social combination model testing indicated that group hypotheses concerning the rules determining exemplars and nonexemplars were similarly predictable from the individual members' hypotheses recorded prior to group discussion. The percentage of correct rule inductions was also the same in both conditions. However, groups in the computer-mediated condition entered significantly more hypotheses which were inconsistent with the available evidence than did groups in the face-to-face condition.;Because of the greater time and effort required to type and send comments, relative to speaking, groups in the computer-mediated condition may not have reviewed proposed hypotheses as thoroughly as did groups in the face-to-face condition. In fact, subjects in the computer-mediated condition required more time, made fewer comments, and were less satisfied with their groups than were subjects in the face-to-face condition.;That groups in the two communication conditions were equally likely to induce the correct rule and that the same social combination processes described the group decision in the two conditions provide support for the substitution of computer-mediated communication for face-to-face discussion. Compared to face-to-face discussion, though, there were more errors and lower levels of satisfaction with computer-mediated communication. Modifications to the group decision support system which might reduce the potential for errors are suggested. |
