For a while now, I’ve been interested in trying to figure out how different research fields think about and use data visualization in their work. As an economist, I visualize data all the time–some of it makes it into the final product, some doesn’t. But what about the economics profession as a whole? How do other economists present their work visually?
The idea has stuck with me for a long time, so I finally decided to try my hand at measuring it. In this new working paper, Categorizing and Ranking Graphs in the American Economic Review Over the Last Century, which I’m posting here, I collect, categorize, classify, and rate more than 2,600 graphs that appear in the first issue of the American Economic Review since 1911.
I’m publishing this on my site now because this is not my typical research area and am looking for feedback. It’s the first time I’ve used Mechanical Turk (thanks to Lane Harrison for all the help!) and want to make sure I’ve covered my bases before sending this off for peer review (likely at an economics journal). If you have any specific comments about the research, please send me an email. If you just want to share your general thoughts about how you or people in your field use data visualization, please reach out.
Here’s the full abstract:
The American Economic Review (AER) is one of the most prestigious journals in the field of economics. First published in 1911, the journal has published articles covering every aspect and topic in the field. AER articles are not just in-depth prose; they might also include tables, diagrams, and graphs. In this paper I ask two primary questions: First, do most graphs in the AER use data or are they some kind of diagram or illustration of a theory or concept? Second, what kinds of graphs—lines, bars, pies, etc.—do economists use to help visualize their arguments in the AER? And third, are those graphs of generally high quality? To help shed some light on those questions, I collect, catalog, and—using Amazon’s Mechanical Turk platform—rate every graph in the first volume of the AER from 1911 to 2017. I find that the share of graphs that use data fell over the first half of the century and then increased from about the early 1980s to today. I also find that economists use a lot of line charts—of the more than 2,600 graphs in total, more than 80% are line charts. Finally, I find a U-shaped curve in perceived graph quality, falling to a low in the early-1960s and rising over the past several decades, on average reaching a level only slightly higher than in the first issues. This research is the first step in understanding how economists use data visualization to communicate their work and can help provide a basis for effective strategies that will enable better communication of that work.
Update: I have added this working paper to the OSF Preprint archive. You can find it here. If you’re interested in citing the preprint, here is the DOI number: 10.31219/osf.io/rakpy.
Oh, and if you want to see every graph in the first issue of the AER from 1911-2017, check out this video:
Data for this project: