Information design: Five ways to improve your bar graph
This is the second post in our series on information design. You can find the first post here.
We at Prometheus like to think about the most effective way to display data. We don’t often turn to bar charts, but since they are one of the most commonly seen data displays, they deserve some attention. Here are some ways we’d apply our thinking about information design to improve the typical bar chart.
Let’s start with this chart showing the area of twelve US states (these happen to be some of the states where Prometheus has had a client):
How can we improve the design? Here are a few things we can do to make the data display more honest, efficient, and useful.
#1. Remove chartjunk (3d effects, shadows, gradient color fills, etc.)
The 3D effect is sort of cool, eh? It’s a nice way for software programs to say, “hey, look what I can do!”. But it’s poor design–it doesn’t make it any easier to understand the content, in fact, it may misrepresent the data in some ways (notice how Texas appears to reach 300,000 square miles while the actual area is 268,580).
The above is a more honest approach. I removed color as well–with charts, I find it best to start simple and add color if it’s necessary.
#2. Order data points in a non-arbitrary fashion
We tend to default to alphabetical order, but that’s usually not the best way to sort things. Sorting the data by a value (in this case land area) makes it easier to infer additional information (namely, rank order).
#3. Indicate authorship and cite your sources
Why would you trust a Promethean to tell you how big Pennsylvania is? Your audience deserves to know the source of the data.
Indicating authorship also adds credibility to a data display–by including your name, you indicate that you stand behind the data. Use the opportunity to explain the purpose of the display: who created this, when, and why?
#4. Make Axes more informative
Rather than using an arbitrary value as the maximum for the y-axis, consider using the value of the largest data point. There’s usually no need to go any higher. This also provides additional information by indicating the exact value of the largest data point.
I also suggest displaying axis labels horizontally–they’re much easier to read than vertical text.
#5. Consider plotting additional information
The typical bar chart isn’t very data dense. If the story you’re telling with the chart can be enriched by including additional information, don’t hesitate to do so. For example, we could add population data to our chart. We could also represent the amount of area accounted for by land versus water. Here’s an opportunity to make effective use of color.
Don’t forget to ask: is a bar chart the best way to display this information?
In many cases, it’s not. A data table can be more useful, especially because it displays the actual values rather than providing a visual representation of the values.
This data set, which contains only 60 values, could be displayed pretty easily using a table.