The Science of Visual Data Communication: What Works

The Science of Visual Data Communication: What Works. Steven L. Franconeri et al. Psychological Science in the Public Interest, December 15, 2021. https://doi.org/10.1177/15291006211051956

Abstract: Effectively designed data visualizations allow viewers to use their powerful visual systems to understand patterns in data across science, education, health, and public policy. But ineffectively designed visualizations can cause confusion, misunderstanding, or even distrust—especially among viewers with low graphical literacy. We review research-backed guidelines for creating effective and intuitive visualizations oriented toward communicating data to students, coworkers, and the general public. We describe how the visual system can quickly extract broad statistics from a display, whereas poorly designed displays can lead to misperceptions and illusions. Extracting global statistics is fast, but comparing between subsets of values is slow. Effective graphics avoid taxing working memory, guide attention, and respect familiar conventions. Data visualizations can play a critical role in teaching and communication, provided that designers tailor those visualizations to their audience.

Keywords: visual communication, graph comprehension, reasoning, statistical cognition, uncertainty communication, data visualization

Summary of Key Guidelines

• A viewer’s visual system can extract broad statistics about the data within a display, such as the mean and extrema, within a fraction of a second. Visualize your data with histograms and scatterplots before trusting statistical summaries.

• Beware common visual illusions and confusions. Failing to start axes at zero can cause viewers to overestimate differences. When plotting data with circles or squares, map the data to their areas, not their diameters. The differences between lines in a line graph are increasingly perceptually distorted as the lines increase in slope. Do not plot intensities on intensities, which causes contrast illusions. Mapping a continuous set of numbers to a spectrum of different hues exaggerates differences that happen to straddle the hue boundaries. For accessibility of color-blind viewers, pair red with blue instead of green.

• Although extracting global statistics is fast, comparisons between subsets of values are slow—limited to only a handful per second. So use visual grouping cues to control which set of comparisons a viewer should make, and use annotation and highlighting to narrow that set to the single most important comparison that supports your message. In a live presentation, rely on language and gesture to illustrate what you see. Do this even when you feel it is not needed: Presenters suffer from a “curse of knowledge” that causes them to overestimate how well others see what they see.

• Avoid taxing working memory by converting legends into direct labels. When possible, integrate relevant text into visualizations as direct annotations. Avoid animations, which typically lead to confusion. Graphical embellishments, sometimes derided as “chart junk,” can distract if unrelated to the data, but if they are related, they can improve viewers’ memory and engagement.

• New visualization formats must be learned, so try to rely on formats that are familiar to your audience. Respect common associations, such as “up” mapping to “more” for vertical position and “more opaque” mapping to “more” for intensity.

• Graph comprehension depends on both bottom-up and top-down factors. Use bottom-up visual salience and top-down direct labels to drive attention to relevant features. Use a graph format that guides viewers to the conceptual message you are trying to convey, respecting their previous experience with graphs.

• When communicating uncertainty to a lay audience, avoid error bars, which can be misinterpreted as data ranges. Instead, show examples of discrete outcomes, either simultaneously or over time.

• When communicating risk to low-numeracy audiences, rely on absolute instead of relative rates, convey probabilities with frequencies (e.g., 3 out of 10) instead of percentages (e.g., 30%), and use well-constructed icon arrays with the same denominator.

• Supporting comprehension and understanding is especially important when the intended audience may have low domain knowledge, knowledge about graphing conventions, numeracy, or working memory capacity.

•