In defense of the bar chart

Posted on 02/22/2011 by


From the New York Times comes a visualization that is not all that surprising, but still unnerving.  This is not a bar chart.

(Via NYT)

Is this effective?  Yes and no.  For one, the coloration is good; red and darker colors stand out, drawing our attention to our dismal statistics.  But let’s compare this to a CDC graphic of infant mortality:

(Via CDC)

That, however, is a bar chart. What I find so effective is the relative size of the bars (and this is why log transformations are so useful…), which makes comparison intuitive and easy.  Yet it is still powerful:

(Via NYT, again)

A while back, Seth Godin, a well-known marketing guru, posted an invective against bar charts, arguing that:

If you want to prove some deep insight or give people textured data to draw their own conclusions, DON’T put [a bar chart] in a presentation.

As Zach Gemingnani responded on the Juice Analytics blog, bar charts are actually quite useful, showing both trends and magnitudes at the same time.  This is not a trivial point.  The ability for a visualization to encode more information in a simpler way is a powerful asset that shouldn’t be overlooked.  As I have argued before, the brain is a very good tool for decoding pattern information, a point that Godin seems not to recognize.

So let’s not forget about the bar chart—it is the unsung hero of data visualization, and deserves some respect.