The Practice of Visual Data Communication: What Works.

Abstract

The practice of data visualization is both a science and an art. There is science behind how humans’ eyes and brains process visual content, and statistical methods behind collecting, processing, analyzing, and preparing data to generate graphs, charts, and diagrams. But the art of data visualization is how we bring people into the visual, how we engage them, and how we make them care about the content we are communicating to them. The target article of this commentary, the thorough work by Franconeri et al. (2021), sets the stage to understand the academic underpinnings of data visualization— how people’s eyes and brains facilitate the understanding of visual content, how to design perceptually efficient and understandable visualizations, and how people use different platforms and technologies to interact with data and visual content. The practice of data visualization goes further than many of these concepts to consider how data are plotted, how to use colors and fonts, and how to facilitate engagement and understanding. The standard graphs that many of us have come to know and create, such as line charts, bar charts, and pie charts, are familiar to most readers and easy to read. But many other graph types can be used to communicate ideas and arguments. In the “How to Design an Understandable Visualization” section of their article, Franconeri et al. briefly discuss four alternative graph types (or what I call nonstandard graph types): connected scatterplot, parallelcoordinates plot, tree map, and node-link diagram. There, the authors focus on how people in specific fields use specific graphs—for example, engineers and economists use connected scatterplots—not the potential for these formats to engage audiences on a broader level. In some cases, such nonstandard graph types can be inherently better at communicating data and in other cases are simply more engaging, which can be a goal in and of itself. Whether you are a researcher, analyst, marketer, or journalist, you know that the amount of content people see every day makes grabbing and maintaining attention difficult; thus, engagement can be a crucially important part of the data communicator’s toolkit. Here, I present several alternatives to the standard ways of visualizing and communicating a relatively simple data set from the National Center for Education Statistics (NCES). From my perspective, these alternative graphs are not so far outside the experience of most readers that they cannot be used more frequently— in the language of Franconeri et al., the “schema” in these graphs are well known and consist of dots, lines, and icons. The goal of this commentary is not to argue that the presented graphs are somehow the “best” that can be created with these data. Instead, my goal is to demonstrate the array of visual options we have to communicate data and how those options enable us to highlight different patterns or values, and to draw out our own stories for readers and help them reach conclusions.