Collaboration + the best graph ever drawn (maybe)
This week’s reading by Unwin and Gelman (2013) was concerned with the divide between information visualization (Infovis) and statistical graphics. Their immediate concern is that many seem to be parting ways with statistical graphics in favor of eye-catching visuals and flair. The authors understand that people will obviously gravitate towards better-looking graphics, but their worry is that these Infovis graphics are focusing more on visual appeal than conveying information. However, they are also sure to note how dull, uninspiring, and trite conventional statistical graphics can be. Thus, the problem arises: we have different goals and methods that need to find the middle ground on the spectrum. There should be no graphics without analysis, and there should be no analysis without graphics. “Both approaches have their value, and it would probably be best if both could be combined” (Unwin & Gelman, 2013).
I think the authors do a great job at acknowledging the issue overall. They acknowledge that there is actually a problem and disagreement, but they also acknowledge that there needs to be a forthright discussion and understanding in order to move forward. I think this is a great way to approach it and is naturally more sympathetic than our other material this week (Broman, 2013), which calls the work of some researchers an “embarrassment”. Regardless, the point remains that there is a discussion to be had between the two sides: statisticians and designers.
I think the authors do a great job framing the inherent divide between these two groups. I also think there are moments when we tend to overly-pigeonhole each side into their respective knowledge areas: statisticians just worship objectivity without an eye for visuals and designers just obsess over aesthetics without caring about what it means. I don’t think this is the case at all, and as far as I’m concerned (and I really do mean this as endearingly as possible), statisticians are nerds, designers are geeks, and they’re all in the same boat together.
In regards to terms that may be more relevant to us as proponents of human factors, I think this same issue arises among UX researchers and designers. Researchers want to convey the context, connectedness, and meaning behind the data; designers want to attract attention and display the complexity of the whole. There is a battle between trying to represent data and trying to tell a story. But these are fortunately not mutually exclusive (Unwin & Gelman, 2013). Where the UX practitioners have an advantage and can succeed is with collaboration for the sake of the user because this eliminates the issues of competing/different goals. What does our user need? What does our audience need? That is where the common ground can be met.
Before I shift focus, there is a small section that I wish the authors would have elaborated on more: “…the first consumer of any graph is the person who makes it” (Unwin & Gelman, 2013). I think this is a tremendously underrepresented point and gets at some of the problem in making graphs. When I make a graph for my data, I have an immediate understanding and context that my audience does not and thus my graph is likely to fit that understanding and not my intended audience’s. We need to understand our user.
The Best Graph Ever Drawn…maybe
Several times throughout the reading the authors mention The Visual Display of Quantitative Information by Edward R. Tufte (1983). There is one particular graphic that I wanted to bring attention to because Tufte notes (the hyperbole is not mine), “it may well be the best statistical graphic ever drawn” (Tufte, 1983):
It looks archaic because…it is. It’s from the mid-19th century and was created by Charles Joseph Minard. It shows the fate of Napoleon’s army in Russia from 1812–1813. While it may just look like a series of jagged lines and antiquated writing, it actually plots six variables: “the size of the army, its location on a two-dimensional surface, direction of the army’s movement, and temperature on various dates during the retreat from Moscow” (Tufte, 1983).
The width of the red band indicates the size of the army at each place on the map as they move into Russia and the black band represents their subsequent retreat and dwindling numbers. On the lower half of the graph temperature is plotted. This is a (very) simple explanation of the graphic but I think it does a superb job at describing a narrative and communicating information. In fact, it could probably be argued that it satisfies all three discovery goals and all three communication goals from Unwin and Gelman (2013): it gives a qualitative overview, conveys complexity and scale, allows for unexpected discoveries in the data, communicates to others, tells a story, and is visually stimulating.
This graphic is 200 years old, and Tufte’s analysis is almost 35, so it seems with the advancement in technology and increased emphasis on design over the years that there is probably a better graphic somewhere out there, but given the constraints and sheer amount of information it conveys, I think Minard’s graphic has certainly earned its place in the annals of great visuals.
References
Broman, K. (2013). The top ten worst graphs. Retried from https://www.biostat.wisc.edu/~kbroman/topten_worstgraphs/.
Gelman, A., & Unwin, A. (2013). Infovis and Statistical Graphics: Different Goals, Different Looks. Journal of Computational and Graphical Statistics, 22, 2–28.
Tufte, E. R. (1983). Narrative Graphics of Space and Time. The Visual Display of Quantitative Information. Chesires, CN: Graphic Press.
