Using Color in Data Visualization
Andrew and I spent three days last week at a workshop on Data Visualization. It was given by Stephen Few of Perceptual Edge. Stephen is a recognized expert in the design of information displays and he definitely earns that recognition. I would highly recommend to anyone who has job responsibilities of communicating any kind of data to others to take this course or, at a minimum, purchase his books. It is more than worth it to hear what he has to say about the topic and I can almost guarantee that it will change the way you do this.
There were many “Aha” moments for me over the three days, but one in particular really stood out. That was where he spoke about using color in data visualization.
Before going any further into this post, I do want to say that the graphics I’m using here and the information are all directly from Mr. Few. I’m not taking any credit for it here other than to convey that I was really influenced by the information.
Below is a list of the most useful visual attributes that we perceive preattentively. Meaning that we assign meaning to them before we are conscious that we are doing it.
These types of attributes can primarily be put into two categories: those that can be used for encoding categorical items, and those that can be used to encode quantitative values. Length, for example, is the primary attribute used to encode quantitative value. Think about a bar graph — does someone have to tell you that the longer the bar, the larger the value it is representing? Width, shape, intensity and position (as in a line chart) are other attributes that encode quantitative values.
But there are really only two of these attributes that can be readily used for categorical items. And only one of them is truly universal. That is color. If you have a list of items and some are red, some are blue, and some are green, you know instinctually that all the reds have something in common. The same goes for the items in green and blue. (By the way, the other attribute that can be used for categorical items is shape. Think of different shaped bullet points distinguishing the categories.)
That’s interesting stuff. We can use color to preattentively communicate something about our data. But what about the colors that we use? That was the really interesting piece for me.
Stephen says we should use color in data visualization the same way nature uses it. When do you see really bright, highly saturated colors in nature? It is nearly always either to draw attention or to warn away. It is rarely used and when it is, it is small amounts in strategic locations. (For attraction, think about birds. It is typically the male bird trying to find a mate that has the bright colors. And for warnings you only have to think about the many brightly colored poisonous frogs or the Black Widow with it’s bright red markings).
Below is a graphic of three nature pictures along with the color samplings from them.
Notice the low saturation of color. Very easy to look at. Also notice the lack of bright, bold colors. The low intensity colors are what our mind finds pleasant to look at. It’s wearisome to constantly look at bright, overly saturated colors.
The chart below does a good job of using bright color to draw attention to the main message. Instinctually, your eye is drawn exactly to the data you want to communicate with the chart.
A common mistake people often make when designing charts to communicate data is indiscriminate use of color — either using many different colors for things that should be grouped together, causing the brain to do unnecessary work trying to assign meaning to these differences, only to have to piece it back together into a whole; or using bright attractive colors everywhere, preventing the viewer from being immediately able to locate the things you want to stand out and instead having to scan through everything sequentially, while at the same time triggering a subtle stress response.
The next time you are communicating data and want to get your point across, remember to take it easy on the bright colors. It works for nature, it’ll work for you too.