Using Social-Media-Network Ties for Predicting Intended Protest Participation in Russia

Previous research has highlighted the importance of network structures in information diffusion on social media. In this study, we explore the role of an individual’s social network structure in predicting publicly announced intention of protest participation. Using the case of ecological protests in Russia and applying machine learning to publicly-available VKontakte data, we classify users into protesters and non-protesters. We have found that personal social networks have a high predictive power allowing user classification with an accuracy of 81%. Meanwhile, using all public VKontakte data, including memberships in activist groups and friendship ties to protesters, we were able to classify users into protesters and non-protesters with a higher accuracy of 96%. Our study contributes to the political-participation literature by demonstrating the importance of personal social networks in predicting protest participation. Our results suggest that in some cases, the likelihood of participating in protests can be significantly influenced by elements of a personal-network structure, inter alia, network density and size. Further explanatory research should be done to explore the mechanisms underlying these relationships