Data-ink ratio
The ink in a chart that shows data, divided by all the ink used. Tufte called decoration that doesn’t inform chartjunk.
What problem it addresses
It explains why gridlines, shadows and drawings can hide the data. It gives design a question for every mark: does it inform or only decorate?
An example
An illustrative example: the same bar chart, before and after cleaning it up.
Removing ink isn’t removing information: it lets you see what was already there.
Why it happens
Every mark asks for attention, even the one that says nothing;
decoration competes for the eye with the bars, the lines and the numbers;
when you erase what’s extra, the data come to the front.
Applications in design
Use it like this
Go over every element of the chart and remove whatever neither shows data nor helps read it.
Avoid
3D effects, shadows, backgrounds and dense gridlines that only decorate.
Don’t fall for it
Don’t confuse plain with better: an image tied to the topic can help people remember, as long as it doesn’t distort the data.
2–3 weeks
What the research says
In a study with 20 people, those who returned 2 or 3 weeks later remembered illustrated charts better than their plain versions, and had understood them just as well (Bateman et al., 2010).
Debated evidence. It is a small study, with charts by a single illustrator and general description tasks: it doesn’t endorse decorating for its own sake. The authors themselves warn that images can bias the message.
Connections
Authors and works
- 1983
Edward R. Tufte
Defined the data-ink ratio, coined chartjunk and proposed erasing ink that doesn’t show data.
- 2010
Scott Bateman et al.
Found that charts with illustrations were remembered better weeks later, without being understood worse.
Sources
Original sources
- Tufte, E. R. (1983). The visual display of quantitative information. Graphics Press.
- Bateman, S., Mandryk, R. L., Gutwin, C., Genest, A., McDine, D., & Brooks, C. (2010). Useful junk? The effects of visual embellishment on comprehension and memorability of charts. In Proceedings of the SIGCHI Conference on Human Factors in Computing Systems (pp. 2573–2582). ACM.
Last reviewed: