I came across this plot the other day while reading an article about LinkedIn on fastcompany.com. What struck me (other than the poor labeling on a vertically oriented phone) was the lack of significance of the vertical sort on the plot. Theres a huge missed opportunity here to provide meaning to the vertical axis by sorting this data on 2015 graduates hired into different industry categories:
Most readers tendency would be to sort this data from largest to smallest. Instead we have a sort with no apparent order. Additionally the x-axis label is at the bottom of the plot, forcing you to scroll to the bottom to understand what the values are for the strong vertical gridlines. Given the data labels are already present, the grid lines and values are really not necessary. We also have a rainbow of colors with no apparent significance. The significant digits are, in my opinion, a bit extreme, with percentages calculated out to the hundredths. Overall, this is, well, a pretty bad plot.
If it were me, I’d take a more minimalist approach and focus on the data, and not colors, random sorts and unnecessary chart clutter. Here’s a relatively straight forward representation of the same data, this time with significance to the vertical sort, no crazy double rainbow of colors, and sans chart clutter:
We need to be more mindful of our plot designs and what is truly necessary to best represent the data we are presenting.