The Curious Case of Combining Text and Visualization

Abstract

Visualization research has made significant progress in demonstrating the value of graphical data representation. Even still, the value added by static visualization is disputed in some areas. When presenting Bayesian reasoning information, for example, some studies suggest that combining text and visualizations could have an interactive effect. In this paper, we use eye tracking to compare how people extract information from text and visualization. Using a Bayesian reasoning problem as a test bed, we provide evidence that visualization makes it easier to identify critical information, but that once identified as critical, information is more easily extracted from the text. These tendencies persist even when text and visualization are presented together, indicating that users do not integrate information well across the two representation types. We discuss these findings and argue that effective representations should consider the ease of both information identification and extraction.