Misleading Sector Graph

,

What-happen-in-the-internet-per-minute-300dpi

This graph entitled “What happens online in 60 seconds” was first published in November 2014 and has been widely redistributed on the internet. It is yet another example of an infographic designer misusing a statistical visualisation format so that it fails to communicate quantitative information accurately.

This format of graph first appeared in a report published by the nurse Florence Nightingale in the aftermath of the Crimean War and has various names including coxcomb chart, Nightingale rose chart or sector graph. The circle (or three-quarters portion thereof in the case of this infographic) is divided into sectors with equal angles. The relative size of the sectors are varied by changing their radii.

In Nightingale’s original graph, the areas of the sectors were proportional to the data they represented. This approach is far preferable to the alternative of making the radii of the sectors proportional to the values. Whichever convention is used to construct them, nowadays the graph is not considered a particularly good example of data presentation. Values can be much more easily compared by plotting the data on a common straight line axis.

So what are the particular issues with this infographic?

1. The sectors do not have a common scale. Hence it is down to the arbitrary choices made by the designer as to which sector is the largest and which is the smallest.

2. Each sector is divided into three portions, representing data for the years 2012, 2013 and 2014. The portions for 2012 are all the same size and those for 2013 are also identical in size. The 2013 portions are larger in area than the 2012 portions by a common factor, but their radial lengths are more or less the same. However, the relative differences between the 2013 and 2012 data values vary enormously, which the graph fails to communicate.

3. The designer of the infographic appears to have chosen radial length to plot the 2014 data in relation to the 2013 values where the data values they have collected make this possible. For example, the radial length of the 2014 portion for Google is about twice the radial length of the 2013 portion. However, our visual systems tend to compare the areas of these portions to each other, rather than their radial length, which causes us to mentally overstate the 2014 values.

4. The graphic contains a typo (the 2012 value for WhatsApp in labelled “20 billion messgaes”) and even more worryingly the data values for this sector simply make no sense at all. If 50 billion WhatsApp messages are sent every 60 seconds, that would mean everyone in the world would be sending about 7 messages a minute! This shows the importance of designers applying a common-sense check to all analyses that they create, as scaling mistakes and typos can clearly slip through the net.

Leave a Reply