Abstract: Based on the studies of Stasser and Larson, this study investigated the effects of task difficulty, decision training and the role of group leaders on the quality of decision-making by groups with laboratory experiments. In this study, the 3-person group was asked to decide which one among 3 candidates was best suited for the position of student body president or which one among 6 candidates was best suited for the position of university English teacher. The results showed: (1) The information sampling model was partly verified, and the personal preference was the mediated factor affecting the decision making of groups, but only when the task was less complex was the model correct; (2) Group leader managed and colligated already-pooled information by repeating information and questioning, but such a group structure had no major effect on the information sampling and its effect was shown only by combining with decision training and task complexity; (3) Training of the strategy planning and information vigilance enhanced the proportion of unshared information discussed. But only when task complexity was low or moderate did the effect appear, and it happened only at the beginning of discussion, and had impact on personal preference of members’ before discussion; (4) There was an interaction between the effects of a group leader and decision training, i.e., the leader could prompt information sampling of a group in the decision-making training condition, and the effect of decision-making training was more prominent in the condition of having a group leader. These two factors were both affected by task complexity

Key words: task complexity, decision training group structure, information sampling