Abstract: This article examined the factors that affected the decision quality in unstructured discussion of groups and tried to further confirm the information sampling model of group decision making proposed by Garold Stasser et al., which was done by simulated experiments in the laboratory. Two kinds of materials for experiments were used in this study. One was the profiles of three hypothetical candidates for student body president (less difficult decision level) and the other was the profiles of six hypothetical faculty candidates (high difficult decision level). Subjects met in three person groups (small size) or six person groups (large size) to decide which of the candidates was best suited for the position of student body president or would be the best person to teach English course. Each profile for candidate contained five positive, three neutral and one negative item. The actual items member in a group received were not identical, 1 33% of the total items was shared (each member received the same items) and the remained items were unshared (only one member received). 144 Zhejiang University undergraduates were involved in this study. Assignment to experimental conditions and to decision making groups within each condition was random. Subjects individually studied written descriptions of candidate profiles and then met to make the decision. Several questionnaires(such as preferences on a private questionnaire and questionnaire of free recall task) were fulfilled by subjects during pro and post discussion. The results showed:(1)Partially certified Stasser's information sampling model. When information that members held was not ample and if group member's preference before discussion was consistent, the group discussion tended to focus on information that members already shared before discussion and information that supported the predominant sentiment within the group. But if the preference before discussion was not consistent or if the task was relatively easy, this conclusion would not be confirmed. There might be two kinds of different ways through which the shared and unshared information affected group decision making. (2)In the unstructured discussion,only the discussion of shared information increased as group size increased and only when the task was more difficult could we find the preponderance of shared information. Variables of personality of members and culture might affect sampling information processes of group decision making also, which should be explored further.

Key words: unstructured discussion, shared information, decision quality of groups