Let the data speak for itself.
This morning, an article about children’s books reminded me once again of how sometimes in our desire to pique interest we overreach and unintentionally do an injustice to our cause. The article is about a study done with toddlers to evaluate the effectiveness of learning through reading lift-the-flap children’s books vs. traditional children’s books. In this study, children were introduced to a new fruit, the starfruit, via both kinds of books. Around 68% of the children who looked at the book without flaps correctly remembered the starfruit at the end of the session compared to just 30% of those who were given books with flaps.
Pop-up pages were created to make the learning experience more fun and interactive, but the 3D features of these books actually inhibit learning. It turns out that kids see the lift-the-flap books as toys rather than a tool for learning. A similar manifestation of this kind of distracting over-embellishment can be seen in the realm of data visualization. In 1983, Edward Tufte coined the term “chartjunk” to refer to all the extraneous visual elements in charts and graphs that may have been designed to add visual appeal, but instead detract from comprehension of the data. Rather than being a question of individual preference and taste, the principles that govern designing for visual perception are rooted in human biology. If we understand how our eyes and brains process information, we can leverage that knowledge to design dashboards that can be processed more quickly and efficiently.
Pre-attentive processing is our amazing ability to process certain environmental stimuli in parallel and without conscious thought. These are detected through all five senses, for our purposes we will focus on visual stimuli only. One example of this is our inability to correctly estimate differences in 2D areas, as in pie charts. The proliferation of round gauges and charts on dashboards today is a pet peeve of mine because frequently the data that is presented in a round form, which seems to be considered more modern and sleek, would be processed much more quickly by our brains if it was encoded in a bar chart. Stephen Few describes this phenomenon in Visual Pattern Recognition as below:
“People generally use pie charts to display part-to-whole relationships, such as each product’s percentage contribution to total sales, but it is much harder to compare the 2D areas of pie slices (a comparison that visual perception can only approximate) than it is to compare the lengths or heights of bars (illustrated in Figure 6). Bars may always be used in place of a line when you wish to emphasize the individuality of the values and to compare one to another, instead of examining the overall shape of the values.”
The Gestalt Principles provide another framework that can help us weed out chartjunk. In the example below, the principle of closure is at work. Closure says that we perceive open structures as closed ones, and when shapes are not finished our mind subconsciously fills in the blanks. Applying that to the images below, we perceive the contents of image at right to be grouped together with just the X and Y axes displayed, and the full enclosure (and background color) in the image at left is unnecessary and distracting clutter.
These may seem like small principles, but a primary challenge when designing charts and dashboards is making the right data points stand out in meaningful ways and displaying the supporting data in a way that is visible but not too loud; these and many other fundamentals are of great practical use for this task. As the fields of big data and analytics continue to grow, data visualization techniques evolve as well. As we delve into the complexities of separating signal from noise so that we can find the insights that will lead to informed decision making for our customers, let’s not forget the basics of human perception.
.
