As a tour guide on London’s open-top buses, I used to look forward to passing the Florence Nightingale Museum. Just after Westminster Bridge, and before Waterloo station, it gave me the chance to waffle about the Crimean War, a historical interest in healthcare that culminated in the National Health Service and, of course, the fact that Florence Nightingale invented the pie chart.
Like most of what I used to say on those buses, this last point was untrue, although in this case I only realised years later. What the Lady with the Lamp invented was, in fact, ‘Nightingale’s rose’ or the polar area diagram. She used it to demonstrate how many more soldiers were dying from infection (shown in blue) than from wounds sustained in combat (in red) and all other causes (in black):
Now like most good data-visualisation geeks, I can’t stand pie charts. But as the great Edward Tufte says: “The only worse design than a pie chart is several of them, for then the viewer is asked to compare quantities in spatial disarray both within and between pieces.” (The Visual Display of Quantitative Information, p178). Not only did our Florence combine chunks of different pie charts, but she misrepresented her data, using the radius of each ‘petal’ to indicate quantity, rather than the area. This means that the outsides of each petal (the blue sections) are much larger than they should be and the figures, consequently, utterly distorted.
The effect of this has been demonstrated admirably well by Dynamic Diagrams, who have compared what Florence did (top image) with what she should have done (bottom image):
As Dynamic Diagrams go on to say themselves, the least distorting way to present this data would have been to ditch the cursed circular form altogether, and replace it with a stack chart. The central thesis remains as convincing as ever, and there isn’t the faintest whiff of pie in the air to distract us:
About the author
Dave Bevan is an Interim Analyst working mainly in the Education Team at FreshMinds Research. He is also studying for an MSc in Politics and Communication at the LSE. Before that he worked for the G77 (group of developing countries) at the Rome Chapter of the United Nations, and before that was a dessert chef, a translator and a tour guide on London’s open-top buses. Dave’s interests include this, this and this.
on Dec 3rd, 2009 at 10:10 pm
Dave Bevan says that Florence Nightingale misrepresented her data, using the radius of each ‘petal’ to indicate quantity, rather than the area. He is mistaken: she did not do so, as can be verified by examining the original data table in her “Sanitary History of the British Army”. Her diagram does not therefore distort (exaggerate) the data. As for his comment that pie charts are not as appropriate as a column chart for this, it is true as every management consultant knows that ‘time series data should be shown on a horizontal axis with time going from left to right.’ But if you understand the message that Nightingale was conveying which was that deaths from sickness were dramatically lower in the second year of the war than in the first, you will see that her format was better (though perhaps the second year should be on the right). Diagrams should not be used to show data: they should be used to support messages. I have explained the reasons for this diagram in my book ‘Florence Nightingale, Avenging Angel’ and in more detail in my article which can be found at
http://www.florence-nightingale-avenging-angel.co.uk/GraphicsPaper/Graphics.htm
on Dec 7th, 2009 at 1:48 pm
Dear Hugh,
Thank you so much for this comment. It’s great (and quite humbling) that an expert would comment here. When I get a moment, I will read through the points you make, and cross-reference them with those made at the Dynamic Diagrams site. I must admit I found theor analysis very convincing, and so did not bother to read much further (I did also check out a few other sources online but none were anywhere near as analytical). So I apologise in advance if it turns out that I was too hasty. I certainly should have taken a less authoritative tone.
As for your broader point, I agree that diagrams should be used to support messages, but I think the bar chart here does that just as well (if not better) than the roses. It is far easier to compare across months and causes of death, for reasons laid out in various places, and many are collected here:
http://simplecomplexity.net/pie-chart-arguments/
The general point is that it is much easier to compare the lengths of lines than the areas of ‘petals’. The rose I find even more difficult to interpret than a standard pie chart, as there is not a consistent radius. But do you feel that this is acceptable isf the message being supported is a just one?