We are moving on to topics beyond GIS this week. Our focus is now turned to design, and Yani brings us the following summary and questions from the work of Edward Tufte:
- Edward R. Tufte, The Visual Display of Quantitative Information
Printed tables of data poorly impress themselves on a reader, “like a figure imprinted on sand” (32), time easily erases their impact, whereas graphics powerfully represent data in an easily-consumable way. Edward Tufte discusses the surprisingly brief history of visual displays of data, the best practices for creating graphics ethically and effectively, and theories for improving data graphics in The Visual Display of Quantitative Information. Tufte underscores his primary claims with purposefully curated examples of maps, graphs, and figures.
Part I begins with an outline of the characteristics of “excellent graphics.” Excellent graphics make data easily accessible without compromising meaning, but many graphics end up distorting information. Tufte warns that the authority of visuals encourages the consumer to overlook fallacies. The author then examines the history of visualization of data. The first data maps originated in the late 17th century, showing information like monsoon winds, patterns of disease, or trade. Time-series charts, graphics that plot time along one axis, appeared in scientific writings by the late 18th century. Complexity of time-series charts increased over time in several ways, as through the addition of a spatial dimension like in Minard’s graphic (40). By the 19th century, graphical designs no longer depended on an analogy to the physical world (time or space); any variable could be measured against any other variable.
The following section considers graphical integrity. Tufte posits that people associate visualizations with lies more often than they might text, despite that individuals lacking integrity can twist text into shapes just as deceitfully. Creators of graphics should employ as much principle in their practice as ethical creators of text. The author claims that a creators of reliable graphic should (1) make physical representations of numbers on the graphic directly proportional to the numbers themselves, (2) use clear, detailed labels to discourage distortion and ambiguity , (3) understand that inconsistency of scale or variations in design often create distortion, which implies data variation that might not exist, (4) display data related to money over time in real prices, or risk demonstrating inaccurate patterns, (5) avoid using graphics of area in two or three dimensions to represent one-dimensional data, which exaggerates changes, and (5) create a fuller truth by showing context because graphics often lie by omission. People can use the following formula to calculate the truthfulness of a visual: lie factor= size of effect shown in graphic/size of effect in data.
Several perspectives and practices have eroded graphical integrity, but diligence can rebuild it. Many artists produce distorted graphics because they lack understanding or respect for quantitative information; their bosses hired them to beautify data, and statistics bore consumers. An assumption that graphics exist for people who find text too difficult led to over-simplified, over-decorated visuals that rarely concern more than one variable. The theory of data graphics states: “Above all else show the data” (92). Maximizing data-ink creates clearer graphics. Individuals can use this formula to determine the proportion of data-associated ink use: Data-Ink Ratio= data-ink/total ink used to print the graphic (93). Identifying the data-ink, effective artists prune non-data-ink and redundant data-ink and face editing without apprehension. Chartjunk refers to any decoration that distracts the viewer from information in a graphic. Computer programs like excel make adding chartjunk to a figure easier, increasing its occurrence. Muted or removed grids reduce clutter in graphs. Artists should avoid “duck” graphics, in which graphical style supplants quantitative reasoning in importance.
Drafters of figures can harness the principles Tufte outlines as best practices to reinterpret traditional forms of visualization and innovate new ones. For example, he prunes redundant data-ink in the box-and-whiskers plot until it consists of a dot with two lines, and yet it still communicates the same information. Tufte explains several other examples of emphasizing data while removing unnecessary ink, such as using a white grid or transforming axes into a range-frame. Multifunctioning graphical elements fulfill more than one graphical purpose. For example, data measures like bars of bar charts, points of scatter plots, or blots on blot maps can represent more than one measure using position and color or shape. Data-based grids create a design element while also representing data at distinct measurement points. Double-functioning labels include elements like range-frames. Shades of gray convey clearer meaning to hierarchical data than colors and allow a viewer to understand data in both a broad and a specific sense. The human eye can distinguish between data as small as .1 mm and statistical visualizations can take advantage of this trait. High-density data displays encourage the viewer to engage with the data presented. Maps consistently achieve excellent data density. Graphics that strive for maximum data density present a more complete picture. Additionally, the intent of graphics may persist even if they are shrunken down, as in sparklines or small multiples, which often show the same graph shown repeatedly over time.
The concluding chapter speaks to aesthetics. Tufte correlates graphical elegance with “simplicity of design and complexity of data” (177). He suggests that an aesthetically pleasing design for quantitative information must have an intentional format, often integrates words, numbers, and drawings together, reflect relevant scale and proportion, displays complexity accessibly, often tells a story, and avoids chartjunk. In closing, Tufte reminds the reader that displays of data should draw clarity from of complexity, but also that the theories he outlines are guidelines, for design is always choice.
- Edward Tufte presents many types of visualizations in this book that are not maps. Do you think any of these methods might be useful for presenting information about your study on your fact sheet?
- How useful do you find Tufte’s mathematical equations for approximating graphical “excellency?” Consider the “lie factor,” “data-ink,” and “data density of a display” equations.
- Please link to an image of a “duck” visualization (as described by Tufte, please). How could the artist have presented their data in a more intentional or appropriate format to add clarity?
- Guchev, Vladimir, Massimo Mecella, and Giuseppe Santucci. “Design Guidelines for Correlated Quantitative Data Visualizations.” Proceedings of the International Working Conference on Advanced Visual Interfaces, 2012, 761-64.
- Enikeev, Ruslan. “The Internet Map.” The Internet Map. Accessed October 19, 2017. http://internet-map.net/. Visualization of the “Semantic Web”
- Halloran, Neil. “The Fallen of World War II- Data-Driven Documentary About War & Peace.” May 25, 2015. Film. http://www.fallen.io/ww2/.
- Keating, Joshua, and Chris Kirk. “Confused About Syria? A Guide to the War’s Friends, Enemies, and Frenemies.” Slate Magazine. October 06, 2015. http://www.slate.com/blogs/the_slatest/2015/10/06/syrian_conflict_relationships_explained.html. An example of using cells of grid as meaningful points, like the fear/rage dog visualization
- Lysy, Christopher, Azzam, Tarek, and Evergreen, Stephanie. “Developments in Quantitative Data Display and Their Implications for Evaluation.” New Directions for Evaluation 2013, no. 139 (2013): 33-51.
- Popova, Maria, and Gareth Cook. The Best American Infographics 2015. Boston: Mariner Books, 2015.
- Trouvé, Antoine, and Kazuaki Murakami. “Interactive Visualization of Quantitative Data with G2D3.” Proceedings of the 8th International Symposium on Visual Information Communication and Interaction, 2015, 154-55.