近年来，互联网技术的迅速发展和移动端设备的进一步普及，人们不仅可以 更便捷地通过各种社交媒体平台获取信息，而且还能借助点赞、转发、评论等多 种形式表达自身对某一社会议题的态度、情绪和具体意见，网络舆情也由此聚合 发酵。在网络舆论场中，用户态度极化显然不利于健康网络环境的塑造，用户个 体态度极化的影响因素因此成为亟需解决的问题。本文以此为切入点，沿着前人研究用户信息接触立场一致性(接触与自己立 场一致的信息、不一致的信息或二者皆有的平衡信息)与个体态度极化间关系的 路径，依据“偏见同化”理论和“回音室”理论，采用在线实验法重点探究个体 宽容度、选择性回避、接触到的留言情绪等对信息接触立场一致性与个体态度极 化关系间的中介效应与调节效应。研究发现，首先，信息接触立场一致性和个体态度极化之间的关系在不同的 信息环境和议题设置中均存在差异。在本次实验过程中，信息接触立场一致性与 个体态度极化呈负相关关系，即接触到立场越一致信息的被试们并没有进一步地 极化，反而出现了寻求更差异化信息的趋势。其次，宽容度在信息接触立场一致 性和态度极化间产生遮掩效应，这与可能影响这一传导路径的其他因素相关。再 次，选择性回避在信息接触立场一致性和个体态度极化间产生中介效应，即人们 越接触立场不一致的信息时，选择性回避会更高，态度极化程度进一步加深。最 后，积极或消极的留言情绪均能够调节信息接触立场一致性和宽容度之间的关系， 但是在信息接触立场一致性和选择性回避之间的关系中，仅有消极情绪能发挥调 节作用。本研究以一个相对完善的多变量模型，探究信息接触与个体态度极化转变的 复杂关系，希冀能够减少相关舆情的负面讨论发生，并指出，在公共领域的转型 过程中，人们接触到的信息不仅是由算法筛选，更有极大的人为因素，在选择性 回避发挥作用的同时，宽容度是否能够从另一路径弥补这一不足还需继续探讨。 当留言情绪会影响到个体态度极化时，个人可以通过使用积极的、正向的语言来 实现对自己的去极化，这也为后续的网络空间治理提供了新的解决方案。

In recent years, with the rapid development of Internet technology and the further popularization of mobile terminal devices, people can not only obtain information more conveniently through various social media platforms, but also express their attitudes, emotions and specific opinions on a certain social issue through various forms such as "like", "forward" and "comment", thus aggregating and fermenting online public opinion. In the network public opinion field, the polarization of user attitude is obviously not conducive to the shaping of a healthy network environment. Therefore, the factors affecting user attitude polarization become an urgent problem to be solved. Taking this as the starting point, this paper follows the path of previous studies on the relationship between users' information access stance consistency (access to information consistent with their own stance, inconsistent information or balanced information of both) and individual attitude polarization, and based on the theory of "bias assimilation" and "information cocoon". An online experiment was conducted to explore the mediating and moderating effects of individual tolerance, selective avoidance and exposure to message emotions on the relationship between stance consistency and individual attitude polarization of information exposure. The results show that, firstly, the relationship between information exposure stance consistency and individual attitude polarization varies in different information environments and issue settings. In this experiment, the information exposure stance consistency is negatively correlated with individual attitude polarization, that is, the subjects who are exposed to more consistent information do not further polarize, but instead have a tendency to seek more differentiated information. Secondly, tolerance has a masking effect between information exposure stance consistency and individual attitude polarization, which is related to other factors that may affect this transmission path. Thirdly, selective avoidance has a mediating effect between information exposure stance consistency and individual attitude polarization, that is, the more people are exposed to inconsistent information, the higher the selective avoidance will be, and the degree of attitude polarization will be further deepened. Finally, positive or negative message emotions can regulate the relationship between information exposure stance consistency and tolerance, but only negative emotions can play a moderating role in the relationship between information exposure stance consistency and selective avoidance. This study uses a relatively complete multivariate model to explore the complex relationship between information access and the change of individual attitude polarization, hoping to reduce the negative discussion of relevant public opinion. It also points out that in the transformation process of the public sphere, the information people are exposed to is not only filtered by algorithms, but also greatly influenced by human factors. Whether tolerance can make up for this deficiency in another way remains to be explored. When the mood of the message will affect the polarization of individual attitudes, individuals can use positive and positive language to achieve their own depolarization, which also provides a new solution for the subsequent governance of cyberspace.