Information diffusion through social networks: The case of an online petition

We modeled the core mechanisms of petition diffusion in online social networks.We added insights about awareness, interest, sharing, and forgetting mechanisms.We compared our model with the Bass Diffusion Model.Comparing the 'push' and 'pull' processes, spread is largely a pull process.Targeting the right population initially is the key to increase the sharing rate. People regularly use online social networks due to their convenience, efficiency, and significant broadcasting power for sharing information. However, the diffusion of information in online social networks is a complex and dynamic process. In this research, we used a case study to examine the diffusion process of an online petition. The spread of petitions in social networks raises various theoretical and practical questions: What is the diffusion rate? What actions can initiators take to speed up the diffusion rate? How does the behavior of sharing between friends influence the diffusion process? How does the number of signatures change over time? In order to address these questions, we used system dynamics modeling to specify and quantify the core mechanisms of petition diffusion online; based on empirical data, we then estimated the resulting dynamic model. The modeling approach provides potential practical insights for those interested in designing petitions and collecting signatures. Model testing and calibration approaches (including the use of empirical methods such as maximum-likelihood estimation, the Akaike information criterion, and likelihood ratio tests) provide additional potential practices for dynamic modelers. Our analysis provides information on the relative strength of push (i.e., sending announcements) and pull (i.e., sharing by signatories) processes and insights about awareness, interest, sharing, reminders, and forgetting mechanisms. Comparing push and pull processes, we found that diffusion is largely a pull process rather than a push process. Moreover, comparing different scenarios, we found that targeting the right population is a potential driver in spreading information (i.e., getting more signatures), such that small investments in targeting the appropriate people have 'disproportionate' effects in increasing the total number of signatures. The model is fully documented for further development and replications.