Information visualizations are an important means through which we communicate knowledge. By considering visualizations as data-driven narratives, this thesis uses narrative thinking as an orienting concept to support the production and evaluation of information visualizations. It proposes a set of guides that are central to future developments in the visualization of information through the analysis of historical examples and a design-based research process resulting in a system called the Royal Society Network. This thesis investigates the themes of various types of objectivity, the layering of quantitative and qualitative methods, the parallel relationship between investigation and visualization, and the graphical nature of statistical thinking. It then identifies transparency, hybridity, and investigation as the central concepts to visualization, where transparency is the communication of underlying structures to end users and is expressed through the building of interface elements as equal components to visualization, the recording and visual incorporation of usage patterns, and the representation of uncertainty; where hybridity is-in terms of both method and form-expressed through the use of quantitative and qualitative methods to drive visualizations forward and the use of multiple graphical forms to aid in understanding and providing contextual information; and where the investigative quality of visualizations is based on the coordination of grain size and axis of representation with the author’s line of inquiry.