Abstract
Data visualizations help people understand numbers that describe our world, across research, public policy, education, and business. Computer screens allow for the animation of visualizations, which can effectively show how data change over time, and they can be engaging to watch. For example, countries could be shown as points in a scatterplot, with GDP and life expectancy on the X and Y axes, and the points could move to show changes in those values over time. But animated data visualizations can easily overwhelm visual processing capacity, and designers rarely have clear rules that help them see when and why that might occur. To produce these rules, we reviewed examples of real-world animations, and created a taxonomy of perceptual tasks that those animations tend to demand. The identified tasks include holistic judgments, such as extracting optic flow or ensemble statistics, or evaluating changes to the shape envelope surrounding a set of points, and tracking tasks across individual objects, such as tracking objects with a common feature, noticing the removal/addition of objects, and tracking objects without (easier) or with (tougher) remembering their features. We then described the known processing limits of each of these perceptual tasks, so that designers can create animations that respect those limits. We catalog design techniques for preventing information overload, such as making a subset of objects more salient, or animating only a subset of objects at once, and explain why they work. Together, the resulting guidelines will help designers engage and explain with animations while avoiding or mitigating viewer overload.