The spread of narratives online has been extensively studied. However, prior research typically relies on metadata and often overlooks the dynamics of post stances. In this paper, we introduce a novel stance-based epidemiological model, which explicitly incorporates the stance of posts—a critical element often ignored by models that focus solely on isolated narratives. Our model, \(SEI_A I_D Z\) ( \(S\) = susceptible, \(E\) = exposed, \(I_A\) = agree with the narrative, \(I_D\) = posting disagree or alternative narrative, \(Z\) = skeptic), captures the dynamic interactions among competing narratives, including the introduction of countermeasures or opposing viewpoints, thereby offering a more comprehensive framework for analyzing online narrative dissemination. Comparative evaluations demonstrate that our model outperforms the baseline \(SEIZ\) model, achieving lower predictive error rates. Furthermore, we identify two key parameters: \(\beta\) (transmission rate) and \(\psi\) (the rate at which exposed individuals transition to the Agreed and Skeptic compartment). Both parameters significantly influence the basic reproduction number ( \({\mathcal {R}}_0\) ), a measure of transmission potential. Our findings indicate that β plays a significant role in driving propagation across all platforms, underscoring the need to control it in order to mitigate the spread of misinformation. To ensure the generalization of our model, we validate our approach using three distinct platforms and scenarios: Health-related conspiracy theories about COVID-19 and 5G on the text-based platform X (formerly Twitter), Geopolitical narratives related to the Russia-Ukraine conflict on the messaging platform Telegram, and Election misinformation campaigns during Taiwan’s 2024 elections on the multimedia-based platform TikTok. Our research provides critical insights into the role of stance and content dynamics in online narrative dissemination, paving the way for more effective strategies to combat harmful narratives and inform policymaking.
