Graph gripes

Graphs are a great way to summarize your data. In fact, I often start reading a paper by scanning through the graphs, as do several of my colleagues.

But not all graphs are created equal. I find myself making the same comments on graphs, figures, depictions of data over the years (and, okay, I made these mistakes myself for a long time) so I thought I would condense them here. If you aren’t my student, you can write this off as the rantings of a (literally) myopic academic, but I suspect you’re going to hear it from someone else. Unless you’re a grapherator for the New York Times, in which case you may be shielded from myopic academics somewhat.

(1) Tiny font. There, see how annoying that is? The default font sizes for most programs (Excel, R, others) are often too small, especially when you reduce the figure to fit into your paper. Same goes for blurry or pixelated font.

(2) Mysterious labels. Numeric labels (“1”, “2”) or shorthand labels that made sense to you and maybe your professor (“COND1”, “REV”) do not make for easy interpretation of figures. Choose clear names that communicate your meaning without anyone having to read through the text to figure it out.

(3) Different scales on otherwise-identical figures. Yes, I know it looks prettier if you rescale the y axes across figures so that the results appear more similar, or so that the data take up the maximum amount of space in the plot. The problem is that it sets up false similarities, or perhaps false differences, across figures. Your duty is NOT to make things pretty. The duty of every Starfleet officer scientist is to the truth [1].

Good luck making some sweet graphs!

Leave a Reply

Your email address will not be published.