Online discussions are performed in the gaze of fellow users. To increase engagement, platforms typically let these users evaluate the comments made by others through rating systems (e.g., via Likes or Down/Up votes). Understanding how such ratings shape, and are shaped by, features of the underlying discussion is important for our understanding of online behavior. In this study, we focus on an increasingly concerning aspect of online discussions: incivility. We draw on the theory of normative social behavior to analyze a data set of over 6,000 online newspaper comments. We find that repeated incivility by the same person is more likely when their initial incivility was affirmed by both descriptive norms (incivility in nearby comments) and injunctive norms (Up votes). Repeated incivility receives more Up votes if nearby comments also include incivility but fewer Up votes if they do not, suggesting that injunctive norms are contextual and shaped by descriptive norms. We conclude that online incivility is a dynamic, normative process that is responsive to both positive feedback and proximate incivility.

在线讨论是在其他用户的注视下进行的。为了增加参与度，平台通常允许这些用户通过评分系统（例如，通过点赞或上下投票）评估其他人的评论。理解此类评级如何影响基础讨论的特征以及如何影响这些特征对于我们对在线行为的理解非常重要。在这项研究中，我们专注于在线讨论中一个越来越令人关注的方面：不活跃。我们利用规范性社会行为理论分析了6,000多个在线报纸评论的数据集。我们发现，当描述性规范（附近评论中的不活跃性）和禁令性规范（上投票）都确认了他们最初的不活跃性时，同一人反复出现不活跃性的可能性更高。如果附近的评论也包含不活跃性，则重复的不文明行为会获得更多的赞成票，如果附近的评论中也包含不受欢迎的话，则重复投票的人会获得更多的赞成票，这表明禁令规范是与上下文相关的，并且由描述性规范决定。我们得出的结论是，在线不活跃是一个动态的，规范的过程，对积极的反馈和最接近的不活跃做出响应。