Part 1 of this article covered the four layers to understand before starting to design visualizations. Those layers are know the data, know the message, know the audience and know the options.
Once those layers are well understood, design principles behind visual perception and cognition can be applied.
These principles are not anything new to web designers and data visualization experts. The following information has been obtained through research and application. This article describes why some visual information is memorable while other such material is easily forgotten.
Cognitive Load
It is important to first understand how the brain stores and processes visual information. It is a complex process done naturally without any conscious thinking about how it all works.
Brains require vision to distill foreground from background, recognize objects presented in a wide range of orientations and accurately interpret spatial cues.
The eyes then break visual information into small chunks, and then both the eyes and the brain process the chunks in parallel. The brain tries to make sense of what it is seeing by searching its memory files for associations. This effort is called cognitive load.
One goal when designing visualization is to reduce the cognitive load by limiting the number of prominent chunks of information for the audience to notice and pack as much information as possible into each chunk. The more the cognitive load is reduced, the less likely it is that any information will be misinterpreted, misunderstood or overlooked entirely, and the more likely it will be stored deeper in long-term memory.
A way to reduce the extraneous cognitive load is to use the gestalt principles of form and perception. Understanding these principles is critical to well-crafted data visualization.
Gestalt Principles
Gestalt principles are theories proposed by German psychologists in the 1920s to explain how the human mind organizes individual elements into groups. These principles can be used to highlight patterns that are important and downplay other patterns. They are highly relevant to the design of charts, graphs and dashboards.
Law of Prägnanz (Also Known as the Law of Simplicity)
The law states that people will perceive and interpret ambiguous or complex images as the simplest form possible. This is the fundamental principle of gestalt. The charts in figure 1 considered in light of this question, “Which are the top five most expensive categories?,” provide a useful example. Most readers will find that answering the question based on the left chart requires more effort than the right chart, because brains prefer things that are simple, clear and ordered.
Law of Continuity
This law states that elements arranged on a line or curve are perceived to be more related than elements not on the line or curve. This approach to continuity entails supporting the natural viewing pattern according to one’s cultural background. European languages are read left to right, top down. Arabic and Hebrew are read right to left.
For example, in figure 1 the eyes follow a continuous path when processing the chart on the right; it makes the whole chart more readable because of the continuous downward direction.
Law of Similarity
This law states that things that are similar are perceived to be more related than things that are dissimilar. The human brain continually seeks patterns and naturally finds them by grouping elements that have similar properties. Practical applications include using a similar texture to associate countries in a map and the same colors for bar charts. Take, as an example, Figure 2: If the bars are the same color, viewers will know the bar values should be compared. The use of varying colors imposes additional cognitive load and might imply that each bar is representing unique categories that should not be compared.
Law of Focal Point
This law states that a point of interest, emphasis or difference will capture and hold the viewer’s attention. The law of focal point states that in a visual presentation, distinctive objects create a focal point. Distinctive characteristics (e.g., a different color or a different shape) can be used to highlight and create focal points (figure 3). Viewers’ attention is automatically directed to the second bar highlighted in dark blue. The different characteristics can also be included in a dashboard to create a flow by capturing the audience’s attention to the main chart and, from there, drawing attention to other charts.
Law of Proximity
This law states that things that are close to one another are perceived to be more related than things that are spaced farther apart. In figure 4, the sales of the South, West, East, Central and International regions form four distinct groups: Quarter 1, Quarter 2, Quarter 3 and Quarter 4.
It is worthwhile to try to anticipate the questions the audience needs to answer with the data and then create groupings through proximity to support that priority.
If the audience would like to compare the sales of a region by quarter to other regions, then it would be best to group them by quarter, as in figure 4. It becomes quite easy to compare the sales of each region within a quarter (which is the grouping according to the law of proximity) due to their location relative to one another.
However, if the audience would like to track the sales of each region across the four quarters, it would be much easier to do so when the law of proximity is applied to focus on region, rather than quarter. See figure 5.
Law of Isomorphic Correspondence
This law states that all images do not have the same meaning to us because we interpret their meanings based on our experiences. The cultural basis for color symbolism can be very powerful, and big mistakes can be made if the messages implied by colors are not understood. For example, the red color in North America generally represents excitement, danger, love, passion and stop. But in South Africa it is the color of mourning, and in China it represents good luck, celebration and summoning. For example, in figure 6, in North America, the bars in red would be intuitively interpreted as if there was something wrong when actually the opposite is true, as profit was above average in those months.
Law of Figure-ground
This law states that elements are perceived as either figure (the element in focus) or ground (the background on which the figure rests). The law of figure-ground captures the psychological fact that people need to determine what is important (figure) and what is secondary (ground). Therefore, the goal should be to ensure that there is enough contrast between the foreground and background so that charts and graphs are more legible.
Which of the charts in figure 7 is more readable? The chart on the left has additional cognitive load due to the lower contrast between the bars and background, meaning it takes the viewers’ brains more time to assess which elements are figures (carrying data and requiring immediate attention) and which are ground (not so important right now).
Law of Common Fate
This law states that elements moving in the same direction are perceived as being more related than elements that are stationary or that move in different directions. Elements that change at the same time group together. The law of common fate can be used to associate and connect lines and shapes. For example, in a line graph it is easy to interpret that the lines trending in the same direction belong to a common data set (figure 8). Conversely, lines and shapes placed in conflicting directions may indicate that the data are not related. When lines and shapes are placed in opposing directions, there may be a sense of uneasiness or confusion.
Colors
Colors are a tool, but they should never be the starting point. Instead, graphs and visualizations should first be conceived and planned in black and white whenever possible. Then color can be applied conservatively following the gestalt principles, mostly with the primary objective of highlighting and enhancing comprehension.
It is important to keep in mind that highly saturated colors (pure hues) are perceived as more dynamic. They attract attention. Too many saturated colors can compete and cause eye fatigue.
If the visualizations are likely to be printed, it is important for the colors selected to work in grayscale in case not all users have high-quality printers.
Also, it is important to understand color blindness and how it affects people. It is worthwhile to research more about the topic, but the following are some tips to keep in mind when designing:1
- Avoid the following color combinations: green-red, green-blue, green-gray, green-brown, green-yellow, blue-purple, orange-red, orange-yellow, orange-green, red-brown and gray-purple.
- Play with the contrast, hue and saturation.
- Ensure the design would be effective in grayscale or even try a monochromatic look by using shades of the same color.
- Use line thickness. (Note: The thicker the line, the more likely it will be perceived as the intended color.)
- Use varied line styles (e.g., dotted, dashed, double lines).
- Use patterns, textures and shapes.
- Use letters and numbers.
- Use labels.
- Underline links or add an icon next to them.
Font
Not all fonts that are legible are necessarily readable; that is, not all legible fonts are designed in such a way that supports and enhances reading so that the audience can easily absorb the text’s meaning without being distracted by the font.
Fonts for data visualization that do a good job of communicating the message without calling attention to themselves include Helvetica, Arial and Gill Sans. In general, it is best to use sans serif fonts for numbers in graphs of any sort because serif fonts tend to have more visual noise to be interpreted.
It is advisable to limit the use of fonts to emphasize small pieces of text, such as a title or heading, or to convey a specific style. In general, only one or two types of fonts should be used in the visualization because they create additional cognitive load.
Chartjunk
The term “chartjunk”2 was coined by Edward Tufte, who considers anything in a chart that does not represent data (or is a scale or label) as not just unnecessary, but harmful. Most of the time chartjunk happens when design principles are fully ignored.
Figure 9 contains an example of a chart that could be deemed chartjunk.
The chart’s presentation forces the audience to work harder to discover the meaning hidden in the data. Eyes must work back and forth between the bubbles and the legend. It is not intuitive. What is the order? When was the peak? Was the peak for IBM 50 percent or 40 percent less than Cisco? This chart seems to be making use of variety for the sake of variety. Compare this with figure 10. Which chart is easier to read?
A study conducted by the University of Saskatchewan (Canada) titled “Useful Junk? The Effects of Visual Embellishment on Comprehension and Memorability of Charts” concluded that the appeal of visual aids or nondata elements does not hinder chart interpretations when they are tightly connected to the subject and even the details of the chart.3
This study asked participants to look at 14 different information graphs created by Nigel Holmes (figure 11) and their equivalent, custom-made, plain versions.
Participants found the Holmes charts were more attractive, enjoyed them most, and found that they were easiest and fastest to remember.
An alternate definition of chartjunk is provided by Stephen Few, who defines it as “anything in your visualization not supporting the chart’s message in a meaningful way.”4 It may be useful to expand this definition to include any non-data elements that might hinder the audience’s ability to interpret the information. Figure 12 is an analogy to help understand meaningless nondata ink, that is, chartjunk.
Keeping figure 12 in mind, in order to reduce cognitive load and make it easy for the audience interpreting the visualizations, it is wise to aim for simplicity to start. An easy first step is to remove some of the items that could potentially be harmful, e.g., 3-D effects, grids, extravagant fonts, background images, clip art and animated icons.
Increased familiarity with the data and audience needs supports adding some elements, but always remember to ask, “Will this visualization suffer any loss of meaning or impact for the audience if this element is added?”
Conclusion
When data analytics are being covered in every business magazine and required by many audit and cybersecurity organizations, auditors and cybersecurity professionals must also understand that data analytics alone are not enough; data visualization is also key. There can be a big difference in acceptance and understanding when the results of analytics are delivered as visual dashboards rather than as spreadsheets. Effective data visualizations can assist in managing risk, promoting buy-in and adding value.
Endnotes
1 Martin, Angela; “Now You See It, Now They Don’t: Designing for the Color Blind,” December 2013
2 Tufte, Edward R.; The Visual Display of Quantitative Information, Graphics Pr, USA, 2001
3 Bateman, S.; R.L. Mandryk; C. Gutwin; A.M. Genest; D. McDine; C. Brooks; “Useful Junk? The Effects of Visual Embellishment on Comprehension and Memorability of Charts,” ACM Conference on Human Factors in Computing Systems (CHI 2010), Atlanta, Georgia, USA, 2573-2582
4 Few, Stephen; “The Chartjunk Debate. A Close Examination of Recent Findings,” Perceptual Edge, June 2011
Karina Korpela, CISA, CISM, CISSP, PMP, is the IT audit manager at AltaLink, a Berkshire Hathaway Energy Company and Alberta’s largest transmission provider. She has more than 13 years of experience with IT risk and controls, performing data analytics, and developing continuous controls monitoring applications for many different business processes. She began her career at Coopers & Lybrand as a system administrator and she was later invited to join its Computer Audit Assistance Group (CAAG) as an IT auditor. Korpela eventually became a manager in PricewaterhouseCooper’s Global Risk Management Solution practice. She can be reached at karina.korpela@altalink.ca.