What does Elon Musk do every day?
A German obstacle course

Variance is a friend of dataviz

Seven years ago, I wrote a post about "invariance" in data visualization, which is something we should avoid (link). Yesterday, Business Insider published the following chart in an article about rising gas prices (link):


The map shows the average prices at the pump in seven regions of the United States. 

This chart is succeeded by the following map:


This second map shows the change in average gas prices in the same seven regions.

This design is invariant to the data! While the data change, the visualization looks identical. That's because the data are not encoded to any visual element - they are just printed as labels.



Feed You can follow this conversation by subscribing to the comment feed for this post.


I'm new here and was wondering what visual changes you would make the image more visually effective


Hi Ed: welcome! The easiest change is to code the data (average prices, average price changes) into the colors. So, a single color with different shades. But that is still more of a geography lesson then anything since color is a rather "coarse" representation of continuous values. A column chart is the next thing I'd try.

The comments to this entry are closed.