Legends explain graphics; at least, this is what we hope for. Humans are spatial animals very much dependent upon perceptual information. Keeping this in mind, graphics should use both perceptual and spatial relationships to make efficient use of our traits. Unfortunately, graphs that use arbitrary symbols require substantial cognitive overhead that takes away mental power we would otherwise use on understanding the main facts in the graph. This diversion of “brain power” may lead to errors and misinterpretations. To keep the naturalness principle* intact, ask yourself do I need a legend or can I design my graph without a legend and its arbitrary symbols, colors or shades? If you decide a legend is necessary, make the representation as meaningful as possible.
*More on naturalness can be found in The Design of Everyday Things by Donald A. Norman. Cambridge, MA: Perseus Books, 1990
Tufte points out that the “frame of a graphic can become an effective data-communicating element simply by erasing part of it.*” There is no need to start your vertical axis at zero if your smallest value is 13, for instance. On the other hand, starting the axis with a tick mark labeled 13 gives the reader an additional piece of information that is only possible by extrapolation and visual estimation if the graph is designed in Excel. The same applies for the maximum value and the upper limit of the axis. See Figure 1 below for an example.
Figure 1: Range frame example (click on image to enlarge)
When the data are magnitudes, it is helpful to have zero included in the scale so we can see its value relative to the value of data. But the need for zero is not so compelling that we should allow its inclusion to ruin the resolution of the data on the graph. There has been much polemical writing about including zero when graphs are used to communicate quantitative information to others. Too frequently zero has been endowed with an importance it does not have. Darrel Huff in his book How to Lie with Statistics goes so far as to say that a graph magnitudes without a zero line is dishonest**. Read more about range frames and how to create them in the booklet
Data Visualization!
*Read on range frames in
The Visual Display of Quantitative Information, 2nd edition by E. Tufte, Cheshire, Connecticut: Graphics Press. 2001. ISBN-13: 978-0961392147
** Cleveland,W.S.: The Elements of Graphing Data. Wadsworth & Brooks/Cole Advanced Books & Software. Pacific Grove, CA, 1985, p. 68-89
Recently, I developed with my colleagues, Dr. Schleyer and Mei Song, a poster about a research study in relation to social networking. The task was to show researchers and how they use an electronic infrastructure in order to find collaborators. An early idea was to let them “walk” from unconnected via the system to connected groups. The lead researcher on this project, Dr. Schleyer, created a low-fidelity drawing on a white board which he then photographed to share with all collaborators. We use this method often which blends well a traditional low-tech approach (drawing with markers) with the high-tech needs of email (digital file) necessary to communicate with remote collaborators (see Figure 1).
Figure 1: Dr. Schleyer's Draft (click on image to enlarge)
A first digital mockup version was created which included mainly dummy text and the partially executed visual idea for evaluation purposes. We identified that the people were dominating too much in this rendition. During the creative session, it was decided that the people need to be more or less “background.” We also wanted a layout that is not completely angular (see Figure 2).
Figure 2: First digital mockup (click on image to enlarge)
The final version was developed and finetuned for consistency and proper alignment of all image elements (see Figure 3).
Figure 3: Final version (click on image to enlarge)
Bibliographical information about the poster:
Schleyer T, Spallek H, Butler BS, Subramanian S, Weiss D, Poythress ML, Rattanathikun P, Mueller G. Requirements for expertise location systems and the Semantic Web. NCI/NCRI Joint Conference “Biomedical informatics without borders: Enabling collaboration to strengthen research and care”, September 2-3, 2008, Bethesda, MD
This statement by Edward Tufte is the leading principle for the new booklet about data visualization using Photoshop CS3. Excellence in statistical graphics and visuals, Tufte believes, “consists of complex ideas communicated with clarity, precision and efficiency.” You will learn how to display quantitative data using Tufte’s design principles, which guide all descriptions and tutorials in this booklet. The tutorial will take a previously published data graph and provide you with the steps to meet the design goals and create the final visual. The example is contrasted with the conventional graphic approach if you were to use Excel. Get the booklet Data Visualization! Discover how to use Photoshop to display quantitative data and produce graphs that “sing” your message to colleagues and reviewers alike. After reading this booklet and completing the tutorial you will smirk at Excel template-based 3D bar charts.
