In this study, we examined the process of problem solving in mailing list discussion groups. We analyzed subjects' participation in discussions and e-mail messages when solving two types of tasks (a mechanical spring problem and a catch phrase making task) by groups of three sizes (small, medium, large). Group performance was compared with the performance of non-collaborating artificial work groups. Results indicate that group size and task type influence the degree of participation and group performance. We found a significant contrast in large-groups. In the `spring problem', which is a difficult scientific task, the group process was inactive and group performance got worse as group size increased. In the `catch phrase' task, communication was active and various opinions were proposed, and group participation and performance did not drop even in large groups.

Key words: CMC, group process, problem solving, communication