Scientific figures are hard!
Well now, part of the work I needed to get done this week involved making some new figures. Actually I had added work dumped on me yesterday that I got done just as quickly thankfully. However, there’s a bunch of other figures I need to make that are due… tomorrow! Ah! So of course instead of actually making them I’m here to talk about how big a pain figures are and why we (the scientific community) stress out about them so much.
It’s interesting that the further I get in my career the more I realize I should’ve taken art classes or maybe photography classes. Both would probably be more useful than I would’ve imagined when I started this journey. I’ve been digging through some of my old figures for things I’ve written and stuff I’m working on now to find some inspiration. Not just from the things I’ve done, but other published works that have interesting figures or ways of presenting things. Figures are hard, but the real problem is that good figures are harder to define than bad figures. A sort of, you’ll know it when you see it type deal. Using other figures for inspiration is helpful, but only to a certain point because most of the time it isn’t an exact translation to what you need.
There are two broad categories for figures, data and everything else. You wouldn’t think there would be so many different ways to visualize data, but there are! Bar, line, pie, donut, spider, scatter, heatmap, box and whisker, violin, topographic, Sankey, arc, network, cords, raincloud, and that’s just the ones I can name off the top of my head! I have literally spent hours adjusting color, font size, font, labels, and the little things that make an okay plot good (notice how I didn’t say great).
There’s a lot of things to consider when making a data plot. The most important is what are you trying to tell people? If I wanted to show that an electrical stimulation on one muscle causes twice the response as a stimulation on a different muscle a pie graph probably won’t be the best way to show you that relationship. If I wanted to show a budget breakdown however, a pie graph would help me illustrate that government military spending is out of control I would be able to make my point pretty clearly. So picking the correct way to show your data is important, but even that can be a challenge.
Sometimes there are multiple ways to show the point you are trying to make. For example we had a lot of data we were trying to show and my co-PI suggested a box and whisker, I suggested we add a scatter overlay (to show the individual data points, and our collaborator was suggesting a violin plot. To which I suggested a raincloud plot, but we had far too much data for that and when we plotted it, it looked cluttered. In the end after a somewhat short meeting yesterday we settled on a box and whisker plot (I believe, it was still kind of up in the air at that point). My co-PI is handling most of that though, I was tasked with a different type of figure.
One of the things in the “everything else” category is the experimental setup figure. Half the battle with figures is conveying what you did. It’s important when publishing that the reader have a clear idea of what you did to get the data you collected. Figures should be clear, not cluttered, and give the reader enough information about what you used that they would be able to make sense of the data. Ideally this would happen without the addition of a caption or text. Generally speaking I was taught that a figure and caption should be able to stand alone from the paper, so I try to keep that in mind as I develop my figures.
The problem comes in when you’re doing a very complicated experiment and want to show everything you did in one figure. You have to balance over crowding and complicated figures with presenting enough information for the reader that they can understand what happened. That takes a lot of work, but also takes creativity and a good understanding of art would probably be helpful. Same with photography and understanding composition. Photography would also be helpful in cases where you’re experimental setup figure includes actual photos (as the one I’m working on does!) so there are skills you may never have thought to learn that would be very helpful in the figure creating process.
I used to paint a lot growing up, so I do love art. I’m not very good at it so most of my “art” is just doodles of things or ideas I have. I’ve started doing little drawings of how I think my figures should look. Doing them does help me figure out the layouts for figures I want to make and also helps solidify otherwise abstract ideas like, “I want a uncluttered figure!” Well what do you need to show to make a figure and how do you plan to keep it from being too cluttered?
Doing a rough sketch helps, even in the case of data you want to plot. Sometimes it’s even more helpful to do that because if the ugly sketch conveys what you’re trying to show well, then chances are going to end up with something far more clear when you make the final thing. It’s just a small thing I’ve picked up over the years that I realized not only helped me improve my figures, but also it helped reduce the amount of time I was spending on making them.
I recently sent off a figure for a paper we’ve been working on in the lab and after some minor feedback I got the thumbs up from our collaborator (it was his study). Spending the few minutes to sketch out what I wanted saved me so much work in the long run. Plus I think it makes me look good when I can send out a figure that gets used right away like I’m some crazy good figure maker person. Yes, that’s the technical term we use in the industry.
That isn’t to say my figures are perfect, far from it. There’s always room for improvement and style that can be evolved over time. It’s like any art you do, the more you do it the better you get and the further you find what works for you. Like I said the other day, I’ve been digging around some of my older stuff on the blog and just reorganizing things on my computer and I’ve really had the chance to see how my style has evolved over the years. One thing holds true though, you’ll never get better at it if you don’t try. That goes for art in general, but also for figure making! We all start somewhere and while I know there are figures a million times nicer than the ones I make, I also know that I can get better if I keep trying.