The Susceptibility Paradox in Online Social Influence

TL;DR

This paper investigates susceptibility to online influence by separating influence-driven adoption ($IAR$) from spontaneous adoption ($SAR$) and demonstrating that susceptibility is highly structured by social networks. Using two large Twitter datasets, the authors show strong homophily in susceptibility and establish a Susceptibility Paradox: friends are generally more susceptible to influence than the focal user ($IAR$), but not necessarily more likely to engage in spontaneous sharing ($SAR$). They operationalize exposure and adoption to derive $IAR$ and $SAR$, and show that $IAR$ can be effectively predicted from friends' $IAR$ alone, whereas $SAR$ requires broader user and metadata features. The study uses linear regression, random forests, and SHAP explanations to reveal feature importances and demonstrates robustness across sampling thresholds, with implications for moderation and protective interventions against misinformation and manipulation.

Abstract

Understanding susceptibility to online influence is crucial for mitigating the spread of misinformation and protecting vulnerable audiences. This paper investigates susceptibility to influence within social networks, focusing on the differential effects of influence-driven versus spontaneous behaviors on user content adoption. Our analysis reveals that influence-driven adoption exhibits high homophily, indicating that individuals prone to influence often connect with similarly susceptible peers, thereby reinforcing peer influence dynamics, whereas spontaneous adoption shows significant but lower homophily. Additionally, we extend the Generalized Friendship Paradox to influence-driven behaviors, demonstrating that users' friends are generally more susceptible to influence than the users themselves, de facto establishing the notion of Susceptibility Paradox in online social influence. This pattern does not hold for spontaneous behaviors, where friends exhibit fewer spontaneous adoptions. We find that susceptibility to influence can be predicted using friends' susceptibility alone, while predicting spontaneous adoption requires additional features, such as user metadata. These findings highlight the complex interplay between user engagement and characteristics in spontaneous content adoption. Our results provide new insights into social influence mechanisms and offer implications for designing more effective moderation strategies to protect vulnerable audiences.

Figures (13)

• Figure 1: An illustration of our Susceptibility Framework.
• Figure 2: Differentiation between exposure and friendship. For User X, the unidirectional interaction with User A is modeled as exposure, while the bidirectional interaction with User B is modeled as friendship.
• Figure 3: Homophily of influence‐driven adoption rates (IAR) in the Retweet network of the Election dataset. Nodes are colored by IAR score (blue $=$ low susceptibility, red $=$ high susceptibility), scaled in size by node degree, and connected by edges representing mutual retweet interactions. Panels (a–c) display communities in which connected users share similar IAR scores, illustrating homophily.
• Figure 4: The susceptibility metrics (IAR and SAR) exhibit homophily within different friendship networks, as evidenced by the weighted average correlation (Wgt. Avg. Corr.) between users and their friends. All correlation coefficients are significant ($p < 0.001$).
• Figure 5: Paradox holding probability $p(k, s)$ at a varying degree $k$ within the Interaction network of the COVID dataset.