The Science of Graphs
What does the following image tell you?
The first thing that popped into my head was: perhaps this is someone’s attempt(s) at a 3-dimensional mountain-scape. Knowing that was absurd, I read through the dense paragraph describing these graphs and was somewhat overwhelmed by an overload of information. As it turns out, this graphical representation of data has been nominated as the worst graph ever published according to this list: https://www.biostat.wisc.edu/~kbroman/topten_worstgraphs/
So what makes this image so terrible? Let us begin our analysis with a quote from an article by Gelman & Unwin (2013):
Ultimately the interpretation of a graph is a joint product of the data, the designer, and the viewer. With the increasing prominence of innovative infographics in the news media and the Web, viewers are changing their expectations of data presentation, and, more than ever before, statisticians should consider the diversity of means to achieving the very different goals of attracting attention, displaying patterns, displaying data in a way that allows for discovery, and getting viewers intellectually involved with data. As is illustrated in the historical reviews such as Wainer (1997) and Friendly (2006), there is a centuries long tradition of data graphics that are both informative and beautiful. We should seek to continue this collective endeavor, and we hope the present article sparks a discussion among statisticians, computer scientists, graphic designers, psychologists, and others who are interested in the graphical presentation of data and inferences.
I apologize for the long block of text, but the authors articulate the idea perfectly. The purpose of a graph is to help the reader understand a point you are trying to make. Or, perhaps better said, it helps them understand the significance of specific aspects of your data. And, as the authors point out, this can be done in a way that is pleasing to the eye. In fact, it works better if the graphs are beautiful. So let’s look at some of the ground rules for graphical data and how the above image violates those rules.
First (because it bugs me): 3D isn’t cool. 3D graphics came at a time when computers were young and 3D was all the rage, but, as it turns out, when you want to be taken seriously you don’t use 3D. Doing so only clutters up your graph and makes the data difficult to read. The above graph is an excellent example of this.
Second, have a single, clear purpose to your graph. This may seem trivial, but without a clearly defined purpose the graph will just turn into a mess as you try to turn it into everything all at once. A graph should be a communicative tool that conveys an idea to your audience, not a representation of every idea you’ve ever had about the dataset. The example above doesn’t violate this one quite as severely, but it could certainly due with tidying up.
Third, use colors carefully. Colors should be about making the data more clear, not about decoration. The above graph is a great example of how decoration can make a graph difficult to read.
Fourth, make sure your units are clear. The above image has, quite literally, no labeled units. In order to understand what the numbers mean, you have to dive into the paragraph that is filled with numbers.
Fifth, and we’ll call this Mikey’s golden rule, make sure your readers can tell what your graph is trying to say without too much work. A reader should be able to use your graph to familiarize themselves with your point. It should not be something they have to read your article through three times before they know how to interpret it.
Graphical representations are supposed to enhance your article and enlighten your readers. Their eyes will naturally be drawn to the graphic, and it should be given effort requisite of its prominence. Make sure your graphics have a clear purpose, are concise and uncluttered, are designed to be pleasing to the eye, and are easy to comprehend.
Don’t do what these authors did and make mountain-scapes.
References:
Gelmen, A., & Unwin, A. (2013). Infovis and statistical graphics: Different goals, different looks. Journal of Computational and Graphical Statistics, 22(1), 2–28. DOI: 10.1080/10618600.2012.761137
