专家研讨中从众与迷信权威现象的动态网络模型-实证社会科学研究所.PDF

专家研讨中从众与迷信权威现象的动态网络模型 赵 峰 清华大学自动化系 北京 100084 摘 要：针对群决策中容易出现从众和迷信权威现象，将人员集合和观点集合抽象为观点—人员 2-模网， 并分别针对这两种现象构建初步的动态网络分析算法，进行由聚类分析、对应分析等方法组成的定性定量 综合分析，探索群决策过程中人员间观点支撑关系的动态变化情况，并根据两种算法的混合实验，对群决 策过程进行改进。仿真实验初步表明：从众现象中观点的入度分布符合幂律分布，迷信权威现象无明显统 计规律；使用以上方法能辅助实时掌握群决策情况、选择会议焦点、安排发言次序与分组等方面的工作， 有助于减少从众与迷信权威现象的影响，提高群决策效率。 关键词：群体决策；多元统计分析；社会网络分析；从众；迷信权威 Dynamic Network Analysis of the Conformity and Fetish in Group Decision ZHAO Feng Department of Automation in Tsinghua University, Beijing 100084, China Abstract: To study the conformity and fetish in the group decision, this paper proposes a 2-mode network model for the opinions and people, and gives two primary algorithms with dynamic network analysis method for the conformity and fetish. It gives a qualitative and quantitative procedure which includes a series of analytical methods such as correspondence analysis and cluster analysis, and studies the network in static and dynamic way. The paper improves on the group decision procedure based on the experiments which use the above two algorithms together. Case study proves that the statistic rule of conformity conforms to the power law, but fetish phenomenon doesn’t have obvious law. The application proves that those methods can help us know the evolution of group decision, catch the innovative ideas, decrease the influence of conformity and fetish, and help us improving the efficiency of the procedure. Keywords: group decision; multivariate statistical analysis; social network analysis; conformity; fetish 群体决策是进行社会问题讨论、某领域规划计划、发展战略研究等很多工作中经常采用 的办法。群体成员作为一个圈子中的个体，其行为受社会习惯、行内规则等的潜在影响。根 据实践经验，往往有以下两个现象。 现象 1：得到响应比较多的观点建议往往新颖性不强，得到响应比较少的观点建议可能 有创造性。 对于很多群体决策问题，例如在论证未来不确定性较强的任务