Assessing Effects of Task and Data Distribution on the Effectiveness of Visual Encodings

• DOI: 10.1111/cgf.13409
• Publication Date: 2018-07-10T12:38:40Z
• Authors: Younghoon Kim; Jeffrey Heer
• Keywords: Categorical variable; Statistical graphics

Abstract In addition to the choice of visual encodings, effectiveness a data visualization may vary with analytical task being performed and distribution values. To better assess these effects create refined rankings we conduct an experiment measuring subject performance across types (e.g., comparing individual versus aggregate values) distributions varied cardinalities entropies). We compare 12 encoding specifications trivariate involving 1 categorical 2 quantitative fields, including use x, y, color, size, spatial subdivision (i.e., faceting). Our results extend existing models suggest improved approaches for automated design. For example, find that colored scatterplots (with positionally‐coded quantities color‐coded categories) perform well points, but poorly summary tasks as number categories increases.