The proliferation of Internet-of-Things devices in urban environments empower citizens to appropriate data for civic purposes. Simultaneously, public visualization has shown to engage a wide audience with data by situating its graphical representation within the actual environment of its measurements. We thus propose a public visualization and polling system that enables residents to co-author a civically-motivated data-driven narrative and distribute it over multiple wireless displays located at different physical locations. Through an in-the-wild study, we studied how passers-by and residents engaged with the system by applying a user engagement evaluation model that maps the social and spatio-temporal context into interactions between the content, the environment and the infrastructure and two distinct user types, i.e. the residents who hosted the displays and the passers-by. Our findings show how the tacit social relationships between the user types, the social factors between passers-by, various temporal aspects, and several contextual factors affect user engagement with our spatially distributed public visualization and polling displays.