Developing a Stance-induced Epidemiological Model to Examine Polarized Information Contagion

The spread of narratives in online environments, particularly those involving misinformation, poses significant challenges in under- standing and mitigating their impact. In this paper, we propose a novel stance-induced epidemiological model (SEIAIDZ) that incor- porates the stance of posts-an aspect often overlooked in previous models that primarily focused on metadata (such as hashtags) and single narratives in isolation. To demonstrate the utility of our model, we apply it to real-world data from X (formerly, Twitter) related to COVID-19 and 5G conspiracy theories, where some users spread misinformation (i.e., agreed with the narrative) while others actively debunked it (i.e., disagreed with the narrative), capturing the dynamic interactions between competing narratives. Our re- sults show that the proposed model outperforms the baseline SEIZ model by achieving lower error rates in predictive accuracy. Ad- ditionally, we identify two key parameters-β (transmission rate) and ψ (rate at which exposed individuals transition to the ''agree'', ''disagree'', and ''skeptic'' compartments)-as critical factors significantly influencing the basic reproduction number (ℜ0), a measure of transmission potential. These findings provide valuable insights into the role of stance and content dynamics in online misinforma- tion dissemination, paving the way for more effective strategies to model competing narratives and combat harmful ones, thereby informing better policymaking