Wednesday, May 29, 2019

The “me” layer in visualization

Since The Truthful Art was published I've been asking people attending my talks and workshops to add more “me” features to visualizations. Following Hadley Wickham's “A Layered Grammar of Graphics”, we could envision visualizations as consisting of stacking several layers: a scaffolding layer (axes, legends, and so forth,) an encoding layer (the actual representation of the data,) an annotation layer (explainers, footnotes, and the like,) etc. The “me” layer would be features that let readers see themselves in the data, or that let them manipulate the data to create scenarios that speak to them.

For instance, if you're visualizing the income of a country, you can simply show a histogram or density plot. But if before that you ask readers to input their own income, you can show them where they are in the distribution and in comparison to everyone else. That adds interest and likely increases engagement.

This is what WSJ's Yaryna Serkez and Theo Francis did in this beeswarm plot published more than a month ago. Write down your income and see how you compare to the typical employee of 1,000 companies in different sectors. Kaiser Fung has a nice post about this project.