Spatiotemporal analytics for uncovering human responses to social and disaster events

People react to social and disaster events in different ways depending on various factors such as their socioeconomic status, social norms, and political polarization. These factors influence not only psychological responses, such as perception, emotions, and sentiments, but also behavioral responses, such as movement and communication, which vary across different geographies and evolve over time. This dissertation contributes to advancing our understanding of dynamic human responses across space, time, and the contexts of social and disaster events through three studies. Using Twitter data from the 2016 presidential debates, the first study reveals how people’s sentiments towards specific topics vary across geographic space and how those variations change over time. Close places do not always have similar sentiment trends, and the groups of places sharing similar sentiment trends also vary across different topics and time. Then, by analyzing human mobility behaviors during the COVID-19 pandemic, the second study examines how disaster severity and mobility behaviors influence each other simultaneously and how this association varies across space and time. The findings suggest that overall, the pandemic severity had a stronger impact on mobility behaviors than vice versa, but the strength and direction of their relationships varied across space and throughout the different waves. The third study identifies the unequal impact of the COVID-19 pandemic on socially vulnerable neighborhoods by evaluating the changes in diverse mobility behaviors. The results show that in more socially vulnerable neighborhoods, the strength of connections among neighborhoods became weaker, and the visits from more socially vulnerable neighborhoods to other neighborhoods became less diverse compared to the pre-pandemic period. Combining three studies together, this dissertation could help design more tailored and equitable policies that address the needs of populations disproportionately affected by events.