Radial plots
In my recently published article in the American Journal of Agricultural Economics, in which we measure the seasonal impact of year-on-year price inflation on violence against civilians in the croplands of Africa, I had to present twelve parameter estimates—one for each month—and I was, sort of, dreading it.
When it comes to presenting the results, I favor figures over tables, but in this instance, presenting the estimates (and standard errors) using errorbars, for example, didn’t feel right. Partly because of the possibility of the temporal displacement of the effect within a crop year. So, where do you start the horizontal axis: harvest month? just before the harvest month? sometime midway into the crop season?
There was no clear answer, particularly as we were agnostic about the seasonal pattern. That led me to further investigation of possibilities, and lo and behold, I realized, a radial plot was the answer.
What’s really neat about it is that you can make a plot ‘turn radial’ with just a single line (really, a handful of characters) of code… that you add to your ggplot function. Below I replicate the main graph from the paper (using a different color scheme and a slightly altered axis range).
You can see the benefits of presenting the results in this way. But you can also sense a potential shortcoming of a radial plot. It is a bit difficult to immediately get a sense of the axis (presented on the top-left of the graph).
In my opinion, radial plots work best when there is no immense need to observe the exact values of the presented figures. They work better when the goal is to show relative differences. Here the exact values matter, of course. But the main takeaway message from this graph is that violence peaks around the harvest time but is largely absent during the rest of the crop year. For that, the radial plot just does the trick.
Replication material is available here.
References:
Ubilava, D., Hastings, J.V., and K. Atalay (2023). Agricultural Windfalls and the Seasonality of Political Violence in Africa. American Journal of Agricultural Economics (in press).