Most of my rants about the poor state of data visualization are aimed at business intelligence software vendors that dabble in the field without understanding it, resulting in poorly designed and ineffective software. Every once in awhile, however, circumstances prompt me to redirect my aim toward the academic visualization community. Even though academic researchers are responsible for most of the breakthroughs in the field of visualization, they are sometimes responsible for downright clueless visual communication and for colossal wastes of time and effort.
In the last two days I’ve noticed several blog entries about something called “A Periodic Table of Visualization Methods.” This is the work of two fellows at the Institute of Corporate Communication, University of Lugano, Switzerland. It was recently published in the paper “Towards a Periodic Table of Visualization Methods for Management.” Coming from an institute that focuses on communication, it is especially alarming that this presentation of visualization methods communicates so poorly.
Before commenting further, let me show you the table.
The periodic table of elements, unlike this table that was modeled after it, is quite effective in design. It works because its organization reflects the meaningful attribute of each element’s atomic number (the number of protons in the elements atomic nucleus). The two-character abbreviations that it uses to label each element work because they are the standard abbreviations that are used by scientists. This new periodic table of visualization methods, however, exhibits no such organizing principle. What is the point of doing an entire research project to force a list of visualization methods into a paradigm that doesn’t fit it? Apart from breaking the visualization methods into general categories (information visualization, concept visualization, etc.), this table exhibits no useful organization.
It also fails in its printed form to use the most obvious means of presenting methods of visualization: images that illustrate the methods. The web version provides access to images by hovering the mouse over an entry in the table. This is an effective way to give people access to details on demand, which are needed infrequently, but the images work even better than the names to identify the visualization methods, and therefore ought to be visible at all times. A simple table with one row per visualization method, organized into categories (information visualization, concept visualization, etc.), would present this information more clearly, especially if it included the following:
- Column 1: A full name for each method, without the two-character abbreviations that nobody knows and nobody will ever be inclined to use
- Column 2: A simple image to illustrate the method
- Column 3: A description of the method, including what it is used for and what differentiates it (which is missing entirely in the display above)
If there are particular attributes that ought to be identified for each method, such as whether it supports overview, detail and overview, or detail views of information, then a column could be included for each, perhaps with a simple check to indicate if it applies. If included, these designations ought to be clearly defined and separated into unambiguous values. I only got as far as looking at the second entry–Table–which was designated as a means to present overview information only, before giving up on this particular distinction as useless. Tables can be used to display overview (summary) information, detail information, or a combination of overview and detail information. I also found the distinction between “divergent thinking” (“adding complexity”) and “convergent thinking” (“reducing complexity”) confusing. I believe that good visualizations neither add to nor reduce complexity, but accurately and as simply as possible represent the level of complexity that exists in the data.
I am sorry to single out the authors of this particular research paper, because the problems found in their paper are not uncommon. I welcome their efforts to create a better or more comprehensive taxonomy of visualizations. What I’m saying is that researchers in the field of information visualization ought to present their work in a manner that exhibits practices of effective visual communication, which are well established in the field. You don’t get a pass just because you assume your audience consists of other academics in the field. Even academics understand information better when it is presented clearly.
P.S. Anyone looking for an encyclopedic reference for data visualization should consider Information Graphics: A Comprehensive Illustrated Reference, by Robert L. Harris.