Simplifying communication

I’m a fan of simplified communication; I don’t like to write more than what’s needed or use flowery language when straightforward and easy-to-understand sentences can do the job. (There are some rare cases where you can be more bombastic in words/visuals, but I’d leave that for another post.)

So, it was a nice feeling to see these data visualization examples by Stephen Few, where he gives before and after solutions to design problems in data viz.

For each example, he analyzes what’s wrong with the first visual, for example, using area circles to show differences in market capital size, when bar graphs are better — if you know what is the main message you want to communicate, why not show it?

In visual mediums like film and television, there is a golden saying: Show, don’t tell. (Disclaimer: I’m not using his images here because it’s his copyright; please visit his site for the linked examples.)

In many cases, bar graphs are great for communicating size/quantitative differences between companies or sectors. Don’t be shy to use them!

But there are also some pitfalls in using bar graphs that we need to be aware of; in his example, he warns about using a non-zero baseline (for the y-axis, which distorts differences in height) and unequal intervals (for the x-axis, which distorts how we perceive the timeline).

To show trends over time/dates, I agree with his intuitive solution of using line graphs.

In many of his examples, he uses bar graphs (or even just plain tables), or black and white images instead of color — to objectively present the data and also summarize the main message of the data analysis. In the case of data analysis and visualization, I believe these are the key points.