Abstract: Efforts to enhance the public sector's performance through cost-saving measures and efficiency initiatives have not resulted in improved citizen relations. Scholars argue that these reforms have failed to alter perceptions due to the absence of strategic communication that could counteract cognitive biases and address misaligned expectations. Communication is deemed crucial in building reputation, legitimacy, satisfaction, trust, and citizen participation. Overall, such communication efforts are expected to contribute to a better democracy. The birth of Web 2.0 brought new ways for public organizations to engage with citizens. Many government organizations, in addition to general e-government initiatives, have invested in an active social media presence to interact with large audiences and provide real-time updates on services. Despite the vast potential, very few studies have attempted to measure the impact of such a social media presence. This dissertation set out to contribute to this growing literature by applying innovative methods to study whether a Twitter presence can enhance citizens’ perceptions of a public organization. The focus was on a public organization that makes optimal use of social media to interact with citizens, the Belgian Railway Company (NMBS). Supervised machine learning was employed to study both the immediate and long-term effects of the NMBS social media presence on the sentiment of Tweets. Furthermore, the study measured the effect of Twitter interactions on passengers' satisfaction with railway services. Based on empirical findings and normative reflections, this dissertation argues that public organizations should actively engage with citizens on social media to bridge the gap between the state and its citizens.