Paper: ISOTYPE Visualization – Working Memory, Performance, and Engagement with Pictographs

Unit charts are not common in visualization, and they are often considered a bad idea. The same is true for using shapes other than rectangles. Neither is based on much actual research, however. In a new paper, we look at the specific example of ISOTYPE-style charts – and find them to be quite effective.

I have written about ISOTYPE before: Otto and Marie Neurath developed the idea in the 1920s, Gerd Arntz created the iconic shapes. Neurath’s idea was to communicate facts about the world in terms of numbers that would be easy to understand. The charts they produced showed data like the kinds of technology used by people (radio receivers, cars, telephones), changes in the way people worked through the course of the industrial revolution, etc.

For a paper presented this week at CHI 2015, Steve Haroz, Steven Franconeri (both at Northwestern University), and I conducted a number of studies to gauge how well people could read these charts, how well they would remember what they had seen, and how engaging they found them. The different experiments had varying numbers of participants, mostly in the range of 20-50.

Here is the kind of image we used in the study to represent the ISOTYPE style. We focus on just the idea of repeating small icons. Icon shapes were drawn from a large number of different types of things, animals, etc.

We compared this to four others: basic bar charts, stacked circles (to see if stacking alone would be better), scaled objects, and a superfluous image in the background.

In broad strokes, it turns out that repeated objects are easier to read and compare, as long as the number is low. But since ISOTYPE icons always represent some multiple anyway, that is not a significant limitation.

Memory is also improved when using icons instead of generic shapes. This is not surprising, though it is worth pointing out that this did not come with a decrease in reading speed of the ISOTYPE charts when compared to bar charts. Also, using icons as labels (instead of words) for bars did not work nearly as well.

Lest you think that this study can be used to justify chart junk, we also found that the superfluous object in the background (bottom right in the image above) was highly distracting and interfered with memory and reading performance.

These are just some of the results, there are many more in the paper. Steve Haroz has put together a nice little landing page for the project with key take-aways and design tips, as well as the interactive playground to create ISOTYPE images. You can even try the studies yourself!

Steve Haroz, Robert Kosara, Steven L. Franconeri, ISOTYPE Visualization – Working Memory, Performance, and Engagement with Pictographs; Proceedings CHI, pp. 1191–1200, 2015