“Information visualization” vs. “Statistical graphics”

By now you all must be tired of my one-sided presentations of the differences between infovis and statgraphics (for example, this article with Antony Unwin).

Today is something different. Courtesy of Martin Theus, editor of the Statistical Computing and Graphics Newsletter, we have two short articles offering competing perspectives:

Robert Kosara writes from an Infovis view:

Information visualization is a field that has had trouble defining its boundaries, and that consequently is often misunderstood. It doesn’t help that InfoVis, as it is also known, produces pretty pictures that people like to look at and link to or send around. But InfoVis is more than pretty pictures, and it is more than statistical graphics.

The key to understanding InfoVis is to ignore the images for a moment and focus on the part that is often lost: interaction. When we use visualization tools, we don’t just create one image or one kind of visualization. In fact, most people would argue that there is not just one perfect visualization configuration that will answer a question. The process of examining data requires trying out different visualization techniques . . .

http://statmodeling.stat.columbia.edu/wp-content/uploads/2011/07/Screen-shot-2011-07-22-at-5.43.06-PM.png

Antony Unwin and I give a statisticians’ view:

Quantitative graphics, like statistics itself, is a young and immature field. Methods as fundamental as histograms and scatterplots are common now, but that was not always the case. More recent developments like parallel coordinate plots are still establishing themselves. Within academic statistics (and statistically-inclined applied fields such as economics, sociology, and epidemiology), graphical methods tend to be seen as diversions from more “serious” analytical techniques. Statistics journals rarely cover graphical methods, and Howard Wainer has reported that, even in the Journal of Computational and Graphical Statistics, 80% of the articles are about computation, only 20% about graphics.

Outside of statistics, though, infographics and data visualization are more important. Graphics give a sense of the size of big numbers . . .

http://statmodeling.stat.columbia.edu/wp-content/uploads/2011/07/Screen-shot-2011-07-22-at-5.45.26-PM.png

(True to form, Kosara’s graph looks more interesting than ours.)

Both articles are in the current issue of the newsletter (pages 5-8 for Kosara and 9-12 for Antony and me).

I’ll give my reactions to Kosara’s article in a future post. For now, I wanted to link to this discussion, which I think should interest many of you. You can comment at Martin’s blog.