Does Social Pressure Drive Persuasion in Online Fora?

Proceedings of the 2021 Conference on Empirical Methods in Natural Language Processing, pages 9201–9208 November 7–11, 2021. c⃝2021 Association for Computational Linguistics 9201 Does Social Pressure Drive Persuasion in Online Fora? Ayush Jain Indian Institute of Technology, Kanpur ayushj@iitk.ac.in Shashank Srivastava UNC Chapel Hill ssrivastava@cs.unc.edu Abstract Online forums such as ChangeMyView have been explored to research aspects of persua- sion and argumentative quality in language. While previous research has focused on argu- ments between a view-holder and a persuader, we explore the premise that apart from the mer- its of arguments, persuasion is influenced by the ambient social community. We hypothe- size that comments from the rest of the com- munity can either affirm the original view or implicitly exert pressure to change it. We de- velop a structured model to capture the am- bient community’s sentiment towards the dis- cussion and its effect on persuasion. Our ex- periments show that social features themselves are significantly predictive of persuasion (even without looking at the actual content of discus- sion), with performance comparable to some earlier approaches that use content features. Combining community and content features leads to an overall performance of 78.5% on the persuasion prediction task. Our analy- ses suggest that the effect of social pressure is comparable to the difference between per- suasive and non-persuasive language strategies in driving persuasion and that social pressure might be a causal factor for persuasion. 1 Introduction Recent interest in analyzing online discussion fo- rums has led to growing body of work on explor- ing aspects such as argument quality and persua- sion in comments, posts, and tweets (Habernal and Gurevych, 2016; Toledo et al., 2019). In partic- ular, Tan et al. (2016)’s seminal work used lin- guistic features of comments and their interplay with a view-holder’s stance to predict persuasion. Priniski and Horne (2018) investigate differences in persuasion between sociomoral vs non-sociomoral topics. Other work has explored the role of gram- matical (Khazaei et al., 2017) and argument struc- ture (Li et al., 2020; Wei et al., 2016; Ji et al., 2018) in persuasive language. Finally, Lukin et al. (2017) CMV : In certain forms of motorsport t = 0 These motorsports aren't very techni… I think you'll find the team… t = 1 I agree with you … t = 2 Δ , I understand what you… t = 3 OP RR Community Figure 1: A sample discussion in CMV. The red path in- dicates the OP-RR interaction. We explore the premise that comments from the ambient social community in- fluence whether the OP is persuaded by a RR. and Wang et al. (2019) study the role of prior be- liefs and personality in influencing persuasion. In this work, we investigate the ties between so- cial pressure and persuasion. Social influence and peer pressure are highly pervasive in online discus- sions (Hui and Buchegger, 2009; Huffaker, 2010). Thus, apart from the merits of a discussion, peer pressure can be a strong force that can make one susceptible to change (Cascio …

Content truncated to 500 words