04 November 2009

Does embellishment improve graphs?

A new article on the Science Daily website (seemingly based on this press release) provocatively claims:

Those oft-maligned, and highly embellished, graphs and charts in newspapers and other media outlets may actually help people understand data more effectively than traditional graphs, according to new research from North Carolina State University.

Unfortunately, some people will take this article to mean that those who have argued against such embellishments, notably Edward Tufte, are wrong, and that it’s okay to make cutesy infographics that so many people think is the only way to make people pay attention to numbers. That’s premature.

The research discussed in the article was presented at a conference. Anyone who has been to academic conferences knows that a lot of research discussed there is often incomplete. Sometimes, the work never finds a home in a journal. But I thought it would be worth looking to see if there was an abstract. (Later found it at the bottom of the press release.)

A little searching turned up the online program (PDF). More complete proceedings of the meeting are available, but I don’t think I’m curious enough about this one bit of research to shell out the cash to buy it.

Searching it for the author, Doug Gillan turned up five papers he gave at this conference. It seems the one being discussed here was titled, “Minimalism and the Syntax of Graphs: II. Effects of Graph Backgrounds on Visual Search.”

In research, the devil is in the details. The abstract is so spare that it’s almost impossible to say anything about this research except that it sounds interesting. For instance, it’s not clear how reading was measured. It’s not clear how many different backgrounds were tested. It sounds like eighteen people looked at three graphs. If it was only three graphs, I would really want to see those actual graphs and how they differed. It sounds a bit like the researchers have shown an effect of contrasting shapes that is analagous to colour contrast. It may well be that a rectangle stands out against circles just that as the red petals of a rose “pop” against green leaves.

It looks like the opening (quoted above) overreaches what the study actually does. The research only looks at backgrounds, but “chart junk” comes in many other forms: pointless 3-D effects, crazy colour schemes, excessive gridlines, cutesy cartoons, and more. The summary of this research in no way provides a scientific basis to argue, “I like the 3-D effect, and science supports it’s easier to read!”

This research might provide some interesting suggestions for improving graphs. I look forward to seeing the final published paper.

Additional, August 2012: Another conference paper makes similar claims. I am not sure if this is a peer-reviewed paper, though.

No comments: