¡§Individual¡¨is the basic analytic unit in a pluralistic society. Especially, phenomenon of public affairs is essence of the problem and is based on individual cognition, hidden in group behavior. The individual cognition forms group judgment and interpersonal influence in the group. This interpersonal influence process may simplify as the power relations between group members, the communication network and the interaction form in the group, and the opinion relations within the group (French, 1956). The conflict is the essential situation of interpersonal influence and also is one kind of relational form and phenomenon for group judgment. Therefore, it is necessary to understand the interpersonal influence and conduct interpersonal mutual cooperation in the public affairs area. The group decision-making (or judgment) can be explored based on observable individual preference and group decision-making. That can infer unobservable interpersonal influence. This interpersonal influence process transforms the individual judgment into the group decision-making. There are two stages in interpersonal influence process, including opinion formation for individual members, and compromise among group members. Namely, the individual might revise his or her preference voluntarily. The group integrates the revised members¡¦ judgments into group decision-making. That is, the group process resembles the process which individual integrates multi-cues like information integration theory (IIT) (Friedkin, 2005; Sniezek et al., 1989). The average cognitive algebra pattern in IIT may measure the interpersonal influence effectively. The theories of group decision-making are still insufficient. For example, social power theory and social comparison theory explain the concept of choice shift, but not positivism. Social decision scheme (SDS) employs decision scheme to predict the group decision-making, but it is insufficient for explaining the group decision-making process. Although social dilemma explores both individual level and group level, it cannot provide the weighting method. Cognitive conflict paradigm (CCP) discusses judgment policy shift, but preference shift is still not mentioned. Although CCP focuses on interpersonal learning, it does not propose how to weight interpersonal influence. The functional measurement theory in IIT may supplement insufficiencies in these theories.. The research utilizes the concept and the method of IIT, which prodivides experimental validity for explaining the complex interpersonal influence process by using social weight. This research uses budget allocation as discussion cases. Interpersonal conflicts are divided by the cognitive conflict and the interest conflict. By using social judgment theory (SJT), this research can analyze cognitive difference in the case of cognitive conflict. Besides, using quasi-experimental procedure in IIT, the findings of this research include: 1. In the group process, group members¡¦ judgments are integrated to group decision-making based on unequal-weight rule mostly. Members¡¦ social weights are different and depend on the level of members¡¦ preferences. 2. The members in different groups have the same preferences initially. Although social weights of these members are not significant difference in statistic, these members still appear the differences between individuals. 3. In cognitive conflict case, the group consensus is not consistently accompanied by cognitive consensus. 4. The group influence results from normative social influence, rather than from informational social influence. 5. It shows that there is negative correlation between social weight and normative effect. Besides, social weight and comprise degree also show negative.correlation 6. The cognitive feedback and the outcome feedback don¡¦t affect decision-making result. 7. The relationship between social weight and the degree of satisfaction is not supported. The social weight and the fairness of decision-making process show significant correlation 8. The subjects¡¦ decision-making performances in the study don¡¦t show significant difference