#28: Limiting the categories for clarity
People LOVE putting things into categories! If there is a lot of information, we will naturally create categories to make it easier to understand. This is one of the 100 things that every designer needs to know about people. Susan Weinschenk explains all of them in her book under the same title. What does it mean for us, the designers? In a nutshell, we should always organize information for our audience as much as possible. And while doing that, we should keep in mind the four-item rule.
I’ll show you how to implement this using Radio Free Europe/Radio Liberty data visualization based on Ipsos Public Affairs publication. The chart focuses on what concerns people around the world — showing both the most concerning problem and the share of people believing that their country is on the wrong track. Conceptually, this is decent work, but we can improve a few areas.
Elements that work in this chart
Providing additional information
This chart nicely combines two data types and pieces of information. First, the share of respondents who say their country is on the wrong track is numerical data encoded with bar length. Second, the biggest problem is categorical data encoded with color. We can quickly switch between analyzed aspects.
Meaningful sorting
Presented data could be sorted in several ways, each stressing slightly different information. We could sort globally by the share (the chosen option); alternatively, we could group the countries based on the most concerning problem or the geography and then sort. The second option would focus on the problem, and the third on geographic diversification. The chosen order focuses on displaying the population’s mood, which is the most important information here.
Elements that don’t work in this chart
Using too many categories
The rule of thumb says we should limit the number of categories to four. Some exceptions exist for well-established classifications like geographical regions, races, etc. But if we create our classification, it definitely should have limited categories. Otherwise, it will take additional work to analyze. And there is a practical aspect to it as well. By limiting categories, we reduce the number of colors to encode them. This means a cleaner chart that is easier to scan and analyze.
Adding flags
There is a tendency to “brighten up” charts using flags, which has at least a few disadvantages. The first and most important one is people don’t know the countries’ flags by heart. They probably know a few — their own, some of their neighboring countries, and the important countries for the region. But the rest is a decoration that remains unfamiliar without description. The second disadvantage is readability — even if people are familiar with the flag, their tiny size makes them illegible. Good luck distinguishing Luxembourg from the Netherlands, Chad from Romania, New Zealand from Australia, or Venezuela from Ecuador. And lastly, flags are pretty colorful, and using them brings a lot of noise and chart junk.
Step-by-step improvements
Remove the chart junk
We should start with removing all unnecessary and distracting elements — flags, coloring, and categorization. We will work on the latter in the following steps. For now, let’s have a clear canvas for a further redesign.
Clean up the layout
The original orientation of the chart misleadingly suggests negative values. We can quickly fix the problem by flipping the chart horizontally. The next step is removing the top axis marks, which simplifies the layout and allows better align the headers.
Provide additional information
In this step, we should rethink the color coding. Generally, encoding categories with color is a good approach; the only problem is the amount. Instead of assigning eight colors, we can create another hierarchy level and group the concerns into fewer categories. That way, we can maintain the original granularity (actual name of the biggest problem) while simplifying the analysis (only four categories). The next change is adding labels with the share of respondents. After removing the flags, we regained some space that we now reuse to enrich the chart.
Work on formatting
The last step is formatting. We can create a visual hierarchy by de-emphasizing the axis, which is secondary information. Adding the category color next to the subcategory label eases the understanding of the hierarchy and works as an extension of the bar. Optionally, we can put the grid in front of the bars — this incorporation into the chart makes it more subtle while still functional.