The Why of Data Visualization

I want to disavow the pie chart you see accompanying this post.

It's hideous. (My source for the graph is the Telegraph article: "Shakespeare's 74 death scenes in a single play more gory than Game of Thrones")

And if you were concerned with some of the unlabeled slices (one of the reasons this chart is hideous), here is a list, starting at the top and going counterclockwise:

• black, on top: pursued by a bear (1)
• red: indigestion (1)
• orange: shame (1)
• blue: drowned (1)
• teal: broken heart (1)
• mustard: poisoned (4)

I find this graphic of Shakespearean deaths (omitting the History plays) to be better:

That "pursued by a bear" is always a favorite.  I don't think that the death was "caused" by "pursued by a bear". I think "mauled by a bear" is more accurate, though not portrayed on stage.

My own stab at displaying Shakespearean bloodiness:

It's not fancy, but it sure lets you know which plays to attend for the gore.  

Which leads me to my topic: when trying to display quantitative data, you need to think of the message you're trying to convey. 

The pie chart is a failure. You can tell that stabbing is the most common cause of death in a Shakespearean play, but the labels are all over the place, there's no good ordering of the causes, and some slices don't even have labels!  Even when a pie chart is executed well, it still has many faults.

The stick-figure-cartoon is more successful, giving detail by play, letting you know which characters die, and the manner in which they meet their ends. But it sure is a lot to look at. 

My bar chart is much simpler, and conveys a coherent story: for a bloody time, go see Titus Andronicus (if you can stomach it).

For more on what one should think about in choosing the manner in which to display data, check out my article "The Why of Data Visualization" in this month's CompAct.

And for the Bard's sake, don't use pie charts.