Community outreach: pie-making
Aug 22, 2009
Frederic S. pointed us to a set of pie charts he made for the Wikipedia page on pie charts showing why they are misleading.
The Wiki write-up points to something called Stevens' power laws which purport to quantify our perception of areas, etc. Quite an interesting read.
As always, thanks for the contributions.
To give back to Caesar the things that are Caesar's... the original idea for the plot comes from Stephen Few's document Save the pies for desert (page 11) -- although I chose to use vertical barcharts instead of horizontal ones, since it makes comparing them easier.
Posted by: Frédéric | Aug 23, 2009 at 04:32 AM
This is, of course, the same S. S. Stevens who gave us the language of nominal, ordinal, interval, and ratio to distinguish between the different types of data we may graph and to suggest how we ought to treat them.
Posted by: derek | Aug 23, 2009 at 04:50 AM
The graphs are copies from William Cleveland's book the Elements of Graphical. I had made a version earlier and wikieditor Schultz had done a nice clean up of them. See the discussion page for Pie Charts for more details.
Cleveland's references regarding the readability of various charts probably go back to primary work from the 1970s and early 1980's. Given the explosion of spreadsheet graphics which may or may not have created a smarter user base, it may be that pie charts are not nearly as bad as they once were.
Posted by: Chris P | Aug 23, 2009 at 10:26 AM
Hi Chris -- indeed, I just had a look at Cleveland's book and noticed the graph, which I did not remember at all (in my edition it's on pages 262-263). Few's document, however, is the first place where I saw such a plot with several pie charts and barplots, showing how hard it is to compare the data across pie charts.
Posted by: Frédéric | Aug 23, 2009 at 12:55 PM