The creation of a data visualization depends on both analytic design and graphic design. Analytic designs focus on the structure of a chart and how data are encoded onto structural pieces such as length, area, and angle. Graphic design focuses on the use of supporting visual elements such as colors, patterns, and supporting context. Our work tests how viewers interact with three elements of modern data visualization: structure, aesthetics, and interpretation. Unlike studies from the past, which used small convenience samples, we utilize NORC’s AmeriSpeak panel, a nationally-representative probability sample, to test how well viewers make decisions from data visualizations with varying structure and aesthetics. In this talk, we will discuss previous work testing graphical perception of structural elements, describe our approach to testing, and share some initial results.
This is joint work with Nola du Toit, Ed Mulrow, and Heike Hofmann.