Assessing the Impact of Design Decisions on the Usability of Uncertainty Visualization: Noise Annotation Lines for the Visual Represen- tation of Attribute Uncertainty

We report findings from a web-based experiment on noise annotation lines, a method to represent attribute uncertainty. We tested and compared three design aspects of noise annotation lines and evaluated how different design variations influence user performance. We systematically varied the number of uncertainty categories, noise width, and noise grain. Our results show that the number of uncertainty categories significantly influences user performance but that certain design characteristics can counterbalance the negative effect of an increased number of categories. Additionally, we were able to show that performance decreases if uncertainty changes continuously.