Pie... or Death?
Hilary Spencer
Wednesday, 12 March 2008 17:02 UTC
I initially missed this short and sweet article that appeared in the December 19th (2007) issue of The Economist: Charts: Worth a thousand words
In addition to mentioning Minard’s now-famous map of Napoleon’s Russian advance and retreat, The Economist article discusses an interesting pie chart by Florence Nightingale. (Why do all the great visualizations of the 1800’s all seem to deal with death in some way?)
In order to support the argument that sanitary conditions in hospitals would help reduce the mortality rate of soldiers, Nightingale collected statistics on causes of death and presented them in an easy to read pie chart. The resulting pie chart shows 12 slices (one for each month), where the radius of each slice is proportional to the number of deaths that month. Color represents the cause of death (e.g. blue represents ‘Deaths from Preventable or Mitigable Zymotic Diseases’). One way to think about the pie chart is as a stacked bar chart, where the x axis has been condensed into a point.
The Economist article points out one flaw of Nightingale’s visualization, which is the same flaw found in many stacked bar charts. The area of each slice is intended to be proportional to the number of deaths. However, since all the wedges begin at the center, the red slices (‘Deaths from wounds’) obscure part of the black slices (‘Death from other causes’), which obscure the blue slices. Therefore the colored area is not directly proportional to the number of deaths, but the radius, when measured from the center, is. (It appears that Nightingale attempted to illustrate this by making the center of each chart blue.) This also creates a visual problem when illustrating causes of deaths in November 1854, where there were more deaths from wounds than from other causes. In the diagram, this means that the red slice completely obscures the black slice.
My other complaint about the graphic is that Nightingale places the chart for April 1855-March 1856 to the left of the chart for April 1854-March 1855, which makes it look as if the number of deaths increased exponentially. I think that most people expect to see time-based representations ordered from left (earlier) to right (later). Finally, I’d like to see the number of deaths written on each of the wedges for reference.
PS. I’m a big fan of Eddie Izzard, if that wasn’t obvious from the title of this post.
Updated 12 March 2008 17:09 UTC
-
Replies