Diffusion of Twitter Messages on Dallas Mass Shooting: Patterns and Factors
MetadataShow full metadata
Through analyzing Twitter data on the Dallas mass shooting, the objective of this thesis research is to add to our understanding about the diffusion process of social media messages regarding mass-scale events, which may reflect public response to an event of such. Descriptive statistics, geographic visualization, hot spot analysis, model fitting and logistic regression are used in this study to examine the spatial and temporal patterns of message diffusion on Twitter and the factors associated with these patterns. We found that tweets’ volume related to a mass-scale event grew very fast immediately following the occurrence of the event and decreased rapidly after a few hours, showing a negative exponential curve. Time lag to the event of interest and people’s daily routine were the two main factors highly related to the volume of tweets. Physical distance is less apparent in information diffusion on social media not only because of minimum friction for online communication but also due to information source other than local witnesses, like news report. Most of the hot spots distributed in places near Dallas. Twitter users close to the event played an important role in message diffusion. These people post related information at the earliest stage and continuously share significantly larger amount of information than people from other places. Findings of this study help us understand the process of messages diffusion on Twitter, which may be used to predict the public’s response on social media after emergencies or extreme events.