In this series, we are analyzing and remaking poorly designed data visualizations found on the internet. Today’s chart has been ridiculed a lot on Twitter and surprisingly there seems to be more than version of this circling around.^1,2 We will go with the graph constituting the original sin:

Furthermore, we add the growth of height over the past century beneath the original chart. This puts the variation of height over countries into the context of the variation of height over time. This time we seemingly break Tufte’s principle and not include zero on the y-axis. Notice however, that we are using the same scaling as in the graph above, thus aligning the growth in height with the impression of height obtained from the first graph in context.

One error, however, cannot be fixed easily: Here we try to display the one-dimensional data of heights, while in fact plotting two dimensional areas. The areas somewhat misrepresent variation in the data since the proportions of a human body do not generally stay the same over varying heights. Ideally, one would want to vary height while altering body proportions. Not an easy task and a hint that those kinds of representations should generally be avoided.